High performance electrodes are crucial for efficient electrochemical systems. Besides optimizing composition and surface conditions of electrode materials, the architecture effect is increasingly relevant, since porous materials have shown promising applications in fuel cells, supercapacitors, electrochemical sensors, pollutant degradation and so forth [1–4]. Various macro-, meso-, and micro-porous materials have been synthesized [5–8], and the large surface area derived from their high porosity provides a large interface between electrolyte and active electrode surfaces . However, the promotion of electrochemistry from porous architectures remains ambiguous and difficult to measure exclusively, because the accessibility of active electrode surfaces and mass transport in the architecture also play an important role in electrochemical reactions [10, 11]. Unfortunately, detailed connection between electrochemical and mass transport behaviors are still far beyond expression due to the complexity and irregularity of these synthesized porous architectures. Recent efforts on diffusional mass transport analysis on architectured electrodes are mainly limited to theoretical simulation such as the rough and scratched model [12, 13], without relevant experimental data to test the calculated results. Also, models characterized by arrays of vertically aligned cylinders or cylindrical pores built to investigate porous carbon nanotube (CNT) film modification were only approximate models, since the real CNT layers were randomly oriented [14, 15]. It is desirable to develop regularly arranged porous architectures with precise geometry for both experiment and modeling to probe into the knack of porous architecture assisting effect and hence provide a guideline for future architecture design of efficient electrochemical systems.
Modern techniques to build well defined and dimensionally controlled architectures, such as lithography and nanoprinting, are often costly, time-consuming, low-throughput, and can only achieve relatively simple surface patterns for electrodes [16–18]. Here, we turned to the great configuration gallery of nature which possesses hierarchical architectures scaled from macro-, micro- to nano-sizes for inspiration [19–21]. Searching for highly ordered porous bio-templates and applying them into electrodes would be of great strategic significance. Butterfly, nearly 20 000 species of which can be identified solely by the color patterning of their wings, is one of the most fascinating creatures in the natural world [22, 23]. Porous arrangement occupies a great proportion of butterfly-wing scale architectures [24–29]. We believe that these elaborate porous architectures may exhibit beneficial influence on mass transport and at the same time may offer a vast structural pool for electrode design .
In this paper, we constructed carbon electrodes with ridge/pore array hierarchical architecture (ridge/pore-C) using the black forewing of Atrophaneura horishana butterfly, to explore the architecture assisting electrochemistry from mass transport perspective. A simple carbonizing-graphite coating method was used to transfer butterfly wings into carbon samples (Scheme 1), for the original butterfly wing composition is chitin which can be easily converted into amorphous carbon by the vacuum pyrolysis method [31, 32]. Also, carbon as a popular electrode material possesses properties of good conductivity, low cost and good chemical stability in solutions and is used extensively in both analytical and industrial electrochemistry [33–36]. The purpose of graphite coating was to enhance the surface activity of carbon samples [37–40]; the electron transfer rate constant k0 ranges from <10–9 m/s to ∼10–3 m/s for ferri/ferrocyanide at graphitic surfaces [33, 37].
A most straightforward yet accurate way was chosen to investigate the electrochemical behaviors on electrodes, by conducting a basic one-electron transfer process using the redox couple ferri/ferrocyanide as a benchmark under cyclic voltammetric conditions [41–44]. The criterion for efficient electrochemistry is based on smaller peak potential separation and higher peak current density of the exemplary redox couple. Despite its surface sensitivity and other redox systems being used such as Ru(NH3)62+/3+, IrCl62–/3– and ascorbic acid, ferri/ferrocyanide is still the most prevalent electrochemical indicator due to its low cost and easy accessibility [38, 43–45]. Another reason we chose this typical object is its fundamental reaction behavior and the abundant reference parameters for electrochemical analysis especially for theoretical calculations which can hardly be reached by other electrochemical systems involving complex processes that cannot be depicted accurately. Finite element modeling and simulating were carried out using Comsol Multiphysics software to further analyze the mass transport behaviors in the ridge/pore array hierarchical architecture and to demonstrate the porous architecture effect on diffusional electrochemical performance.
Butterfly wing samples
Butterfly specimens were provided by Shanghai Entomological Museum, China. As has been universally verified, butterfly wing configuration can be viewed as a flat membrane covered by scales on both the dorsal and ventral side (Fig. S2). The black forewings of Atrophaneura horishana butterfly with ventral scales removed as the ridge/pore array architectured sample (ridge/pore-S) and the wing membrane with both side scales removed as the non-architectured sample (flat-S) were prepared for further carbonization (Scheme 1a, Fig. S3). Additionally, a ridge array architectured sample (ridge-S) without pore structure component was also prepared as a contrast sample using the yellow hind-wing of Troides aeacus with ventral scales removed (Fig. S4).
A carbonizing-graphite coating process was carried out to obtain carbon samples (ridge/pore-C, flat-C, and ridge-C) from corresponding butterfly wing samples. All butterfly wing samples were clipped by two alumina slides and put into a vacuum furnace for carbonization. Temperature was set to rise gradually from room temperature to 350 °C with a heating rate of 2 °C min–1, and then held at 350 °C for 1 h in order to remove the structural water molecules. After that, the temperature rose slowly to 800 °C with a heating rate of 1.5 °C min–1 to make a moderate carbonization process and avoid collapse of the elaborate architectures. Then the temperature was held at 800 °C for 3 h to ensure a fully carbonized material. Tin-doped indium oxide (ITO) glasses were cut into slides of 24 mm long and 3 mm wide, cleaned ultrasonically in absolute ethanol and deionized water, and then dried in air. Squares with lengths and widths of 3 mm were cut from the carbonized wing samples and carefully pasted on the end of the conductive side of the ITO slide using inactive carbon conductive adhesive with dorsal scales facing outside. A clean atmosphere was required for all procedures to ensure uncontaminated samples.
To introduce enough active sites on the surface, graphite coating was carried out using a Carbon Coating Unit (CCU, E-1045). The graphite rod was fixed with its two terminals contacted intensely with the electrifying tips of the Carbon Coating Unit, and after vacuumization the current was gradually increased to 20 mA and held for 30 s to ensure enough graphite evaporation. All samples were processed concurrently in one batch to ensure the utmost homogeny of graphite coating factors and surface conditions. The final graphite coating thickness by vapor deposition was about 15 nm. Other conductive areas of ITO slides were wrapped by elastic insulating film to insulate from electrolyte (Scheme 1b). Electrochemical experiment were conducted on these freshly graphite coated samples promptly to avoid changes for the surface conditions.
Morphology and composition characterization
Digital photographs of butterfly samples were taken by a Canon EOS 350D digital camera. Optical microscopic photos were taken by a digital microscope (VHX-100, KEYENCE). FESEM images were obtained by observing samples under low vacuum field emission scanning electron microscope (LV FESEM, NOVA NanoSEM 230) operated at 5 kV.
X-ray diffraction (XRD) patterns were obtained using the X-ray diffractometer (D8 ADVANCE) with Cu Kα radiation. Raman measurements were conducted using a Dispersive Raman Microscope (Senterra R200-L) under excitation wavelength of 523 nm. X-ray photoelectron spectra (XPS) were obtained on an X-ray Photoelectron Spectroscopy (AXIS UltraDLD), using Al Kα radiation as the X-ray source.
Electrochemical experiment and simulation
All electrochemical experiments were performed on a Parstat2273 Potentiostats-Electrochemistry Workstation at room temperature. A three electrode cell was used with the fabricated carbon electrodes as the working electrode, a large platinum plate (20 × 20 mm) as the auxiliary electrode, and saturated calomel electrode (SCE) as the reference electrode. The aqueous solution of 1 mM potassium ferricyanide with 0.1 M KCl as supporting electrolyte was prepared using deionized water. The solution was purged with nitrogen for 30 min to remove dissolved oxygen in the electrolyte before use. To eliminate possible inaccuracy derived from surface sensitivity of ferricyanide, all experiments were conducted on the freshly graphite coated samples and cyclic voltammograms from the second cycle were recorded for further analysis. Finite element modeling and simulation were conducted on Comsol Multiphysics 4.2 software from CnTech.
Results and discussion
After carbonizing and graphite coating, architectures retained perfectly with certain shrinkages compared to original butterfly-wing scale dimensions (Fig. 1, Fig. S3). As shown in Fig. 1a,b, the architecture of ridge/pore-C was characterized by evenly spaced inverse-V type ridges separated by double-row stagger-patterned pore arrays. The periodic distance between two ridges was 1.34 μm with a ridge height of about 600 nm; dimensions of the micro/nano-scale pores were approximately 460 nm long and 330 nm wide, with a depth of 670 nm. Detailed dimensions are exhibited in Fig. S5 & Table S1. In contrast, flat-C showed an excellent flat morphology as an non-architectured sample (Fig. 1c,d).
XRD patterns in Fig. 2 verified the amorphous carbon composition for both ridge/pore-C and flat-C. Raman spectra were conducted and analyzed by curve fitting with band combination (Fig. 3). Both spectra were divided into 5 bands (G, D1, D2, D3, D4)  with approximately the same band position, band intensity ratio and full width at half maximum, which indicated the same composition of both carbonized samples (Table S3,S4).
To examine the variation of surface conditions before and after graphite coating, XPS were carried out on both sample surfaces. Figure 4 shows the C1s high resolution spectra before and after graphite coating. The fitting B.E. values were fixed at 284.6 eV (sp2), 285.4 eV (sp3), 286.3 eV (hydroxyl), 287.7 eV (carbonyl), 289.4 eV (carboxyl) and 285.0 eV (aliphatic carbon) . Detailed fitting information (B.E. values, full widths at half maximum, intensity ratio and peak shape) were approximately the same for ridge/pore-C and flat-C, indicating the same surface conditions for both samples (Table S6, S7, S9, S10). Compared to spectra before graphite coating, the C1s spectra showed great variation with obviously higher 284.6 eV (sp2) intensity ratio and lower 285 eV (aliphatic carbon) intensity ratio indicating the successful vapor deposition of graphite. Also, according to the full XPS spectra (Fig. S10, S11), the O1s intensity was greatly increased after graphite coating because of the oxidation of graphite steam in vapor deposition process which was beneficial for enhancing surface activity [48, 49].
Experimental cyclic voltammetry
Figure 5 depicts the cyclic voltammograms on carbon electrodes at 25 mV s-1. The peak potential separations (ΔEpp) were 81 mV and 198 mV for ridge/pore-C and flat-C, respectively. The oxidative peak current densities (Iop) were 64.6 μA cm–2 and 54.7 μA cm–2 for these two electrodes, respectively. A false impression of “electrocatalysis” was caused by the ΔEpp reduction and Iop augment of the cyclic voltammogram for ridge/pore array architectured samples compared to the non-architectured one. However, there were no real catalytic factors here, and this phenomenon should be solely attributed to the ridge/pore array hierarchical architecture with enhanced surface area and mass transport behaviors rather than a direct alteration of fundamental electrode kinetics, considering the same composition and surface conditions of both electrodes .
To further investigate electrochemical kinetics on the two electrodes, cyclic voltammograms at different potential scan rates were performed. As is shown in Fig. 6a, ΔEpp increased with the increase of potential scan rate indicating a quasi-reversible electrochemical reaction process on the carbon electrodes. According to Fig. 6b, Iop for ridge/pore-C electrodes was obviously higher than that of flat-C electrode. Despite the semi-infinite diffusion of reactants to the macro electrode surface, diffusion and depletion of reactants in the porous geometry of ridge/pore-C also contributed to the overall current .
The results above have clearly demonstrated the architecture effect on promoting electrochemical performance for ridge/pore-C. To obtain a deeper understanding of this enhancing effect, we took advantage of the highly ordered hierarchical architecture, constructed models and analyzed the mass transport behavior with finite element simulation method. Here, a diffusion domain approach  was used in the simulating process with one ridge/pore array period as a domain unit for ridge/pore-C electrode, while flat-C was idealized as a non-architectured plane. Equation (1) shows the electrochemical reaction investigated in this work, with detailed modeling and simulating the process provided in S2.
Simulated concentration profiles demonstrate that the consumption of reactant was more efficient on ridge/pore-C electrode than on its non-architectured counterpart (Fig. S19). In order to find out the detailed diffusion behavior in the ridge/pore array hierarchical architecture, enlarged views of cross sectional concentration profiles were drawn. Compared to the simple one direction planar diffusion for flat-C, mass transport for the ridge/pore-C electrode displayed a more efficient and complex signature (Fig. 7). The high efficiency was demonstrated by the additional lateral direction diffusion to ridge walls and the even more effective multi-direction diffusion within the pores of the ridge/pore array hierarchical architecture. This multi-direction diffusion signature was in fact parallel to the thin layer effect constantly discussed in porous modification layers for electrode . The trapping and depleting of the solution within the porous pockets greatly enhanced mass transport efficiency and caused more efficient current rising and dropping on the voltammetric timescale leading to a decrease for the peak potential separation in cyclic voltammograms. According to theoretical calculation , the pore diameter for thin layer limiting behavior should be less than 20 μm in the ferri/ferrocyanide system and the submicrometer size of pore arrays in ridge/pore-C totally coincides. So the mass transport for ridge/pore-C electrode should be interpreted in terms of a combination of semi-infinite planar diffusion to the macroelectrode surface and additional lateral diffusion and thin layer diffusion within the ridge/pore array hierarchical microarchitectures.
Simulated cyclic voltammograms are also displayed in Fig. 8, with a ΔEpp of 83 mV and 200 mV for ridge/pore-C and flat-C model, respectively. The simulating results exhibited good coincidence and identical trend with experimental curves. However, there were still some disagreements between experimental and simulating results especially at relatively high potential values. This may be attributed to the additional radial/convergent diffusion effects derived from unideal uniformity of graphite coating for actual electrode surfaces, while a complete surface uniformity was assumed in the simulation process. What’s more, the highly ordered electrode architectures were not as ideally infinite as the simulation assumptions in the actual electrochemical experiment, and additional thin layer cell effects may also take place due to the overlapping of wing scale edges and other unideally blocked areas.
Pore array architecture component analysis
In order to obtain more comprehensive understanding of the effect of pore array architecture component, we also constructed ridge-C electrode without pore structure as a contrast sample using the yellow hind-wing of Troides aeacus butterfly (Fig. S4,S6). Raman and XPS spectra indicated the same carbon composition and surface conditions of ridge-C with ridge/pore-C and flat-C (Fig. S8,S9). Table 1 shows the summary of cyclic voltammetric experiment and simulation results for all samples. It is obvious that the promotion effect of mere ridge architecture is less significant than the ridge/pore array architecture for electrochemical performance, which is verified by the larger ΔEpp (98 mV) and lower Iop (60.7 μA cm–2) of ridge-C compared to ridge/pore-C. The coupling of submicro-scale pore arrays into ridges leads to increased surface areas, together with partial thin layer diffusion within the pore array domains which is a more efficient mass transport behavior than simple lateral diffusion in mere ridge domains (Fig. 7, Fig. S20). Besides ridge/pore array architecture, this methodology may also be extended to other architectures in butterfly wings and other porous biological configurations from nature to further explore the porous architecture effects on optimizing electrochemical performance.
In summary, we demonstrated that porous architecture assisted electrochemistry, by converting butterfly wing samples into carbon electrodes and investigating the cyclic voltammogram of a basic one-electron transfer process using ferri/ferrocyanide as a benchmark on the ridge/pore-C electrode. Compared to the non-architectured counterpart (ΔEpp = 198 mV), we found that the ridge/pore array hierarchical architecture exhibited a remarkable promotion for electrochemical response on the ridge/pore-C electrode with a declined ΔEpp of 81 mV and increased Iop of 64.6 μA cm-2. The essence of the architecture assisting effect lies in the increased surface area together with enhanced rate of mass transport, achieved by additional lateral direction diffusion within ridge domains and partial thin layer diffusion within submicro-scale pore array domains. This work may provide a typical prototype for developing optimized electrochemical configurations and could be extended to other catalytic components besides carbon such as noble metals, semiconductor oxides and so forth to meet the demands of efficient electrochemical performance in various applications such as fuel cells, biosensors, pollutant degradation and so forth.
The authors acknowledge the consistent financial supports of National Natural Science Foundation of China (No. 51425203 and No. 51172141) and National Basic Research Program of China (No.2011CB610301).
J. Ding, K. Y. Chan, J. Ren, F. S. Xiao. Electrochim. Acta50, 3131 (2005).Google Scholar
L. Hu, M. Pasta, F. L. Mantia, L. F. Cui, S. Jeong, H. D. Deshazer, J. W. Choi, S. M. Han, Y. Cui. Nano lett.10, 708 (2010).Google Scholar
Y. Li, Y. Y. Song, C. Yang, X. H. Xia. Electrochem. Commun.9, 981 (2007).Google Scholar
J. Fan, G. Zhao, H. Zhao, S. Chai, T. Cao. Electrochim. Acta94, 21 (2013).Google Scholar
P. Losch, M. Boltz, K. Soukup, L. H. Song, H. S. Yun, B. Louis. Microporous Mesoporous Mater.188, 99 (2014).Google Scholar
L. J. Fu, T. Zhang, Q. Cao, H. P. Zhang, Y. P. Wu. Electrochem. Commun.9, 2140 (2007).Google Scholar
H. P. Liang, H. M. Zhang, J. S. Hu, L. J. Wan, C. L. Bai. Angew. Chem. Int. Ed.43, 1540 (2004).Google Scholar
R. W. Murray. Chem. Rev.108, 2688 (2008).Google Scholar
S. W. Kim, T. H. Han, J. Kim, H. Gwon, H. S. Moon, S. W. Kang, S. O. Kim, K. Kang. ACS Nano3, 1085 (2009).Google Scholar
Y. E. Seidel, A. Schneider, Z. Jusys, B. Wickman, B. Kasemo, R. J. Behm. Faraday Discuss.140, 167 (2008).Google Scholar
A. Schneider, L. Colmenares, Y. E. Seidel, Z. Jusys, B. Wickman, B. Kasemo, R. J. Bechm. Phys. Chem. Chem. Phys.10, 1931 (2008).Google Scholar
D. Menshykan, R. G. Compton. J. Phys. Chem. C.113, 15602 (2009).Google Scholar
D. Menshykau, I. Streeter, R. G. Compton. J. Phys. Chem. C.112, 14428 (2008).Google Scholar
M. J. Sims, N. V. Rees, E. J. F. Dickinson, R. G. Compton. Sens. Actuators, B144, 153 (2010).Google Scholar
K. R. Ward, L. Xiong, N. S. Lawrence, R. S. Hartshorne, R. G. Compton. J. Electroanal. Chem.702, 15 (2013).Google Scholar
G. S. Attard, P. N. Bartlett, A. R. B. Coleman, J. M. Elliott, J. R. Owen, J. H. Wang. Science278, 838 (1997).Google Scholar
M. Sawangphruk, J. S. Foord. J. Vac. Sci. Technol. B.27, 1484 (2009).Google Scholar
B. A. Grzybowski, K. J. M. Bishop, C. J. Campbell, M. Fialkowski, S. K. Smoukov. Soft Matter1, 114 (2005).Google Scholar
A. Sweeney, C. Jiggins, S. Johnsen. Nature423, 31 (2003).Google Scholar
M. Srinivasarao. Chem. Rev.99, 1935 (1999).Google Scholar
P. Vukusic, I. Hooper. Science310, 2151 (2005).Google Scholar
R. A. Potyrailo, H. Ghiradella, A. Vertictchikh, K. Dovidenko, J. R. Cournoyer, E. Olson. Nat. photonics1, 123 (2007).Google Scholar
V. Saranathan, C. O. Osuji, S. G. J. Mochrie, H. Noh, S. Narayanan, A. Sandy, E. R. Dufresne, R. O. Prum. PNAS107, 11676 (2010).Google Scholar
A. L. Ingram, A. R. Parker. Phil. Trans. R. Soc. B363, 2465 (2008).Google Scholar
P. Vukusic, R. Sambles. Proc. SPIE.4438, 85 (2001).Google Scholar
X. Guo, H. Zhou, D. Zhang, T. Fan. RSC Adv.4, 3748 (2014).Google Scholar
Q. Zhao, T. Fan, J. Ding, D. Zhang, Q. Guo, M. Kamada. Carbon49, 877 (2011).Google Scholar
R. L. McCreery. Chem. Rev.108, 2646 (2008).Google Scholar
E. Frackwiak, F. Beguin. Carbon39, 937 (2001).Google Scholar
A. N. Patel, M. G. Collignon, M. A. OConnell, W. O. Y. Hung, K. McKelvey, J. V. Macpherson, P. R. Unwin. J. Am. Chem. Soc.134, 20117 (2012).Google Scholar
R. S. Nicholson. Anal. Chem.37, 1351 (1965).Google Scholar
R. J. Rice, N. M. Pontikos, R. L. McCreery. J. Am. Chem. Soc.112, 4617 (1990).Google Scholar
J. Liu, A. Chou, W. Rahmat, M. N. Paddon-Row, J. J. Gooding. Electroanalysis17, 38 (2005).Google Scholar
A. Sadezky, H. Muckenhuber, H. Grothe, R. Niessner, U. Poschl. Carbon43, 1731 (2005).Google Scholar
S. Kaciulis. Surf. Interface Anal.44, 1155 (2012).Google Scholar
P. Chen, M. A. Fryling, R. L. McCreery. Anal. Chem.67, 3115 (1995).Google Scholar
W. Gao, L. B. Alemany, L. Ci, P. M. Ajayan. Nat. Chem.1, 403 (2009).Google Scholar
I. Streeter, G. G. Wildgoose, L. Shao, R. G. Compton. Sens. Actuators, B133, 462 (2008).Google Scholar
The online version of this article (DOI: 10.1515/pac-2014-1201) offers supplementary material, available to authorized users.