Report

Liquid-Mediated Dense Integration of Graphene Materials for Compact Capacitive Energy Storage

See allHide authors and affiliations

Science  02 Aug 2013:
Vol. 341, Issue 6145, pp. 534-537
DOI: 10.1126/science.1239089

Keeping Electrolytes in Porous Electrodes

Electrochemical capacitors (ECs) can rapidly charge and discharge, but generally store less energy per unit volume than batteries. One approach for improving on the EC electrodes made from porous carbon materials is to use materials such as chemically converted graphene (CCG, or reduced graphene oxide), in which intrinsic corrugation of the sheets should maintain high surface areas. In many cases, however, these materials do not pack into compact electrodes, and any ECs containing them have low energy densities. Yang et al. (p. 534) now show that capillary compression of gels of CCG containing both a volatile and nonvolatile electrolyte produced electrodes with a high packing density. The intersheet spacing creates a continuous ion network and leads to high energy densities in prototype ECs.

Abstract

Porous yet densely packed carbon electrodes with high ion-accessible surface area and low ion transport resistance are crucial to the realization of high-density electrochemical capacitive energy storage but have proved to be very challenging to produce. Taking advantage of chemically converted graphene’s intrinsic microcorrugated two-dimensional configuration and self-assembly behavior, we show that such materials can be readily formed by capillary compression of adaptive graphene gel films in the presence of a nonvolatile liquid electrolyte. This simple soft approach enables subnanometer scale integration of graphene sheets with electrolytes to form highly compact carbon electrodes with a continuous ion transport network. Electrochemical capacitors based on the resulting films can obtain volumetric energy densities approaching 60 watt-hours per liter.

Electrochemical capacitors (ECs) store energy by charging electrical double layers through highly reversible ion adsorption on the surface of high-surface-area electrodes, generally made from porous carbon (13). They are attractive for energy storage because of their fast charging capability and long life span (4, 5). The energy density (or the amount of energy stored per unit volume) of most commercially available ECs to date is close to 5 to 8 Wh/L (6), still much lower than that of lead-acid batteries (50 to 90 Wh/L) (7).

The efficiency of a material for EC energy storage can be described by its specific volumetric capacitance in a single electrode (Cvol) and energy density against the volume of two EC electrodes (Evol-electrode); the volumetric energy density against the whole EC stack (Evol-stack)—including two electrodes, electrolyte, a separator between two electrodes, and current collectors—was recently recommended to be a more reliable parameter than the gravimetric one to evaluate the real potential of a porous carbon for ECs (810). Evol-stack relates to gravimetric capacitance of the active carbon component in a single electrode (Cwt-C), packing density of the carbon (ρ), volume fraction of the electrodes in the device stack (felectrode), as well as the nominal voltage (U) of the EC as follows (6).Embedded Image(1)Embedded Image(2)Embedded Image(3)Improvement in Evol-stack requires that Cwt-C, ρ, and felectrode of electrodes all be maximized, but increasing ρ and felectrode (or the thickness of electrodes) is difficult without substantially compromising the Cwt-C. As an example, graphite is possibly the most compact conductive carbon materials (~2.2 g/cm3 at the ambient condition) but delivers little capacitance because ions cannot access the interplanar space. Even if the packing density of the graphene stack was reduced by half to 1.1 g/cm3, the interplanar spacing would be only ~0.67 nm (fig. S1). The ion diffusion and adsorption in such small-size channels is very sensitive to pore size, surface wettability, and particularly the pore interconnectivity (4, 11, 12). Impaired ion transport in subnanometer pores becomes even more pronounced when the electrodes are made thicker, further limiting the achievable value of Evol-stack (13, 14). Although progress has been made to achieve high Cwt-C values through a variety of methods, such as laser scribing or chemical activation of graphene (1519), the packing density of the electrodes obtained was rather low, ranging from 0.05 to 0.75 g/cm3 (810).

Recent advances in graphene chemistry (20, 21), particularly the colloidal chemistry of chemically reduced graphene oxide (22), also called chemically converted graphene (CCG), open up new ways to address this challenge. We have shown that CCG can be well dispersed in water without the need for any surfactants by controlling its colloidal chemistry (23). The resulting CCG sheets, being microscopically corrugated (22), could self-assemble to form an oriented hydrogel film through a simple directional-flow–induced bottom-up assembly process (24). As a result of microcorrugation and repulsive intersheet solvation/electrostatic forces, CCG sheets in the hydrogel film remained largely separated, which gave rise to a high Cwt-C of over 200 F/g (25). However, the as-formed gel film exhibited a low packing density (~0.069 g/cm3), resulting in a mediocre Cvol (~18 F/cm3).

In this work, we show that in contrast to porous carbon films prepared by the traditional techniques that generate fixed pore size and “hard” texture (26), the CCG hydrogel films, with a metastable and adaptive pore structure, can be compressed irreversibly by capillary pressure to increase the packing density through controlled removal of volatile solvent trapped in the gel. The graphene sheets in the films stacked in a nearly face-to-face fashion, so the packing density can be increased up to ~1.33 g/cm3, nearly double that of the traditional activated porous carbon (0.5 to 0.7 g/cm3) (6). More important, the liquid electrolyte-mediated CCG (EM-CCG) films created a continuous ion transport network that led to exceptionally high Cvol and Evol-stack.

The CCG hydrogel films obtained by filtration of CCG dispersion were exchanged with a miscible mixture of volatile and nonvolatile liquids and were then subjected to removal of the volatile liquid by vacuum evaporation (fig. S2) (27). The gel film reduced in the thickness direction as a result of selective removal of the volatile liquid, while the CCG sheets remained solvated by the nonvolatile liquid during the whole process. The packing density of these flexible films (Fig. 1A) was controlled from 0.13 to 1.33 g/cm3 by changing the ratio of volatile and nonvolatile liquids (Fig. 1D and fig. S9G). The nonvolatile liquid electrolytes, sulfuric acid, and 1-ethyl-3-methylimidazolium tetrafluoroborate (EMIMBF4) were used in these studies.

Fig. 1 Characterization of liquid electrolyte-mediated CCG (EM-CCG) films.

(A) A photograph showing the flexibility of the film. (B and C) SEM images of cross sections of the obtained EM-CCG films containing (B) 78.9 volume percent (vol. %) and (C) 27.2 vol. % of H2SO4, respectively, corresponding to ρ = 0.42 g/cm3 and ρ = 1.33 g/cm3. (D) The relation between the volumetric ratio of incorporated electrolyte and the packing density as well as the estimated intersheet spacing.

Scanning electron microscopy (SEM) analysis revealed that the as-prepared EM-CCG films had a rather uniform cross section (Fig. 1, B and C, and figs. S3 and S9), which was also confirmed by energy dispersive x-ray spectroscopy (EDX) mapping of sulfur (fig. S3). The thicknesses were nearly proportional to the volumetric fraction of incorporated nonvolatile liquids trapped in the gels. X-ray diffraction (XRD) analysis showed that the as-compressed EM-CCG films displayed a nearly amorphous structure (fig. S4). Only a small and broad peak at around 23°, corresponding to a d002 distance of 0.39 nm, was detected when ρ > 0.76 g/cm3 but the peak was much weaker than that of the dried CCG film. This indicated that the majority of CCG sheets did not restack back to graphite, despite being substantially compressed.

To investigate the effect of ρ on capacitive energy storage, we fabricated a series of prototype ECs with the same areal mass loading of CCG sheets (27). Fully dried CCG films with a ρ of 1.49 g/cm3 were also tested for comparison. Figure 2 presents typical EC characterization of EM-CCG film–based ECs in 1.0 M H2SO4 electrolyte. The cyclic voltammetry (CV) curves (Fig. 2, A and B) showed nearly symmetrical rectangular shapes, indicative of an ideal capacitive behavior. As shown in Fig. 2C, a higher ρ led to a lower Cwt-C at a given charging rate. However, the extent of decrease in capacitance was strongly dependent on whether the CCG film was preincorporated with an electrolyte during the preparation process of the film. At a low operation rate of 0.1 A/g, when ρ of the EM-CCG films was increased from 0.13 to 1.33 g/cm3, Cwt-C only dropped from 203.2 to 191.7 F/g (fig. S6A). In contrast, the completely dried CCG film (1.49 g/cm3) only gave a Cwt-C of 155.2 F/g. The difference became even more pronounced when the films were more rapidly charged (fig. S6, B to D). All of the EM-CCG films delivered a Cwt-C > 100 F/g at an operation rate of 100 A/g, whereas the dried CCG film yielded only 10.2 F/g. The data obtained with an organic electrolyte [EMIMBF4/acetonitrile (AN)] at an operation voltage of 3.5 V showed a similar result (figs. S10 and S11).

Fig. 2 Electrochemical characterization of EM-CCG films.

Electrochemical characterization of EM-CCG films in 1.0 M H2SO4 (ρ = 0.13, 0.42, 0.76, and 1.33 g/cm3, respectively). CV curves at (A) 50 mV/s and (B) 500 mV/s. (C) Gravimetric and (D) volumetric capacitances with varied charging/discharging current densities. (E) Nyquist plots and (F) Bode plots of phase angle versus frequency. The results for the dried CCG film (ρ = 1.49 g/cm3) are also shown for comparison.

The influence of ρ on Cvol was very different. The Cvol of the EM-CCG films was nearly proportional to ρ (Fig. 2D and figs. S11E and S13). The highly compact EM-CCG films (1.25 to 1.33 g/cm3) yielded a Cvol of 255.5 F/cm3 in aqueous electrolyte and 261.3 F/cm3 in organic electrolyte at 0.1 A/g. These values were much higher than those of the existing porous carbon materials (table S2). The EM-CCG films showed a very high Cvol at a broad range of charging rates from 0.1 (0.133 A/cm3) to 200.0 A/g (266.0 A/cm3). Even though the dried CCG film was able to deliver a Cvol of around 231.5 F/cm3 at 0.1 A/g, the value decreased rapidly with the increase of operation rate (Fig. 2D) as a result of a rapidly declining Cwt-C.

The above results showed that preincorporation of electrolytes in the CCG films ensured excellent rate performance of compact EM-CCG films. We performed EC impedance spectroscopy analysis to further ascertain the role of the preincorporated electrolyte. A frequency response analysis at open circuit potential over the frequency range from 100 kHz to 10 mHz yielded the Nyquist plots shown in Fig. 2E. The plot featured a vertical curve, indicating a nearly ideal capacitive behavior of the cell. At high-frequency regions, a transition from a vertical curve feature to a –45° line followed by a semicircle was observed, and this transition was pushed to higher-frequency regions as a decrease in the packing density of EM-CCG film. This variation was small between the different EM-CCG films, but it showed a pronounced difference when they were compared with the dried one. The corresponding time constant τ0 (the inverse of the characteristic frequency at which –45° is reached in the Bode phase plots) also responded in a similar fashion (Fig. 2F). The τ0 increased from 0.51 to 0.73 s as the ρ of the EM-CCG films increased from 0.76 to 1.33 g/cm3. A doubling of ρ only led to a change in τ0 by a factor less than 1.45. By contrast, less than 20% change in ρ from the EM-CCG film (1.33 g/cm3) to the dried film (1.49 g/cm3) gave a shift of τ0 by a factor more than 5.27. The frequency response of the CCG films in organic electrolyte followed a similar trend (fig. S10C).

The above results indicated that highly efficient ion transport channels were retained in the EM-CCG films, even when they were packed quite densely. Due to the fluid nature of liquid electrolytes, the continuous liquid network was likely to remain within the whole film during the capillary compression process. Furthermore, as the electrolyte became integrated within the film from the start of the assembly process, there was no subsequent wettability issue for these EM-CCG films, which remained a serious problem for the dried CCG film (28).

The indispensable role of the preincorporated electrolyte in ion transport was further evidenced by the effect of areal mass loadings (or electrode thickness) on the electrochemical performance (Fig. 3, A and B, and fig. S7). Both Cwt-C and Cvol generally decreased with increasing thickness of the electrodes (4, 13, 29); this effect was particularly prominent when the pore size was in the subnanometer range (13). Figure 3 showed how the EC performance of the EM-CCG film (ρ = 1.25 g/cm3) and the dried CCG film (ρ = 1.49 g/cm3) responded to mass loading in the organic electrolyte. As with other porous carbon (13), both the Cvol and Evol-electrode of the two types of materials decreased with thickness. Nevertheless, the EM-CCG film displayed a much slower rate of decrease than the dried film, reflecting superior ion transport. When the effect of thickness on Evol-stack was considered, the volume fraction of the active electrodes (felectrode) needed to be taken into account (see Eq. 3). Because the felectrode increased with the electrode thickness and the other components of ECs remained unchanged, the reduction in Evol-electrode with thickness was not substantial in the case of the EM-CCG film; however, the EM-CCG film displayed an overall increase in Evol-stack with thickness (see the red dashed line in Fig. 3B). The EC device based on two CCG/electrolyte films of 80 μm (containing 10 mg/cm2 of CCG, comparable to the amount of carbon contained in many commercial EC devices) yielded an Evol-stack of 59.9 Wh/L. By contrast, the Evol-stack of the dried CCG film initially increased with thickness, then decreased by a mass loading of 5 mg/cm2 because of substantial drop in Evol-electrode (see the black dashed line in Fig. 3B). The maximum Evol-stack using 10 mg/cm2 of the dried CCG film as electrodes was only ~18.4 Wh/L, further highlighting the crucial role of low ion transport resistance for achieving a high Evol-stack.

Fig. 3 Volumetric performance of ECs based on EM-CCG film electrodes.

(A) Volumetric capacitance and (B) energy density as a function of the areal mass loading of EM-CCG film (ρ = 1.25 g/cm3) and the dried CCG film (ρ = 1.49 g/cm3) at the current density of 0.1 A/g. (C) Ragone plots of representative EM-CCG films showing how Evol-stack and Pvol-stack depend on both the packing density and areal mass loading of CCG: (a) 1.25 g/cm3, 10.0 mg/cm2; (b) 1.25 g/cm3, 5.0 mg/cm2; (c) 1.25 g/cm3, 1.0 mg/cm2; (d) 0.39 g/cm3, 1.0 mg/cm2; the dried CCG films (1.49 g/cm3) were also presented for comparison with the areal mass loading of (e) 1.0 mg/cm2 and (f) 5.0 mg/cm2. The data were obtained from the prototype ECs with EMIMBF4/AN as electrolyte and an operation voltage of 3.5 V.

The ECs based on our compact EM-CCG films also delivered a high volumetric power density. As shown in the Ragone plots (Fig. 3C), both the Evol-stack and the Pvol-stack (maximum power density) were strongly dependent on the packing density and areal mass loading of CCG. The Pvol-stack corresponding to the highest Evol-stack (59.9 Wh/L) was ~8.6 kW/L. With the same areal mass loading of 1 mg/cm2, EM-CCG film (1.25 g/cm3) could deliver the maximum power density of ~75 kW/L, higher than that of the dried CCG film (1.49 g/cm3) as well as the low packing–density EM-CCG film (0.39 g/cm3). These results also suggested that the key technical specifications of CCG-based ECs (e.g., Evol-stack, Pvol-stack, and τ0) could be readily customized to suit different applications by simply tuning the packing density of the CCG electrodes.

These EM-CCG films were very stable under repeated charging/discharging or under the application of a constant voltage (figs. S8 and S12). Over 95% of the initial capacitance was retained after a 300-hour constant voltage holding at 3.5 V in a neat EMIMBF4 electrolyte (fig. S12). In addition, the fabrication of CCG gel films and subsequent compression are essentially compatible with the traditional cost-effective paper-making process and can be readily scaled up. All these attractive features make this class of graphene materials promising for large-scale real-world applications.

Supplementary Materials

www.sciencemag.org/cgi/content/full/341/6145/534/DC1

Materials and Methods

Figs. S1 to S13

Tables S1 and S2

References (3039)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: The authors acknowledge the financial support from the Australian Research Council and G. P. Simon for helpful discussion. This work made use of the facilities at the Monash Centre for Electron Microscopy. D.L. and X.Y. filed an Australian provisional patent application (application no. 2013900820) on the basis of the work reported in this paper.
View Abstract

Navigate This Article