Development of easy made low cost bindless monolithic electrodes from biomass with controlled properties to be used as electrochemical capacitors

Bioresource Technology 102 (2011) 2781–2787

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Development of easy made low cost bindless monolithic electrodes from biomass with controlled properties to be used as electrochemical capacitors J.M. Valente Nabais ⇑, Jorge Ginja Teixeira, I. Almeida Centro de Química de Évora & Departamento de Química, Universidade de Évora, Escola de Ciências e Tecnologia, Rua Romão Ramalho n° 59, 7000-671 Évora, Portugal

a r t i c l e

i n f o

Article history: Received 27 September 2010 Received in revised form 18 November 2010 Accepted 19 November 2010 Available online 25 November 2010 Keywords: Activated carbon Supercapacitor Electrochemistry Biomass Chemical structure

a b s t r a c t The aim of the work now reported is the development of low cost electrodes in the monolithic shape without the need for a pos-production step with potential to be used in supercapacitors. The tested materials were activated carbon fibres prepared and activated carbons made from coffee endocarp. The main functional groups identified were quinone, lactone, Si–H, phenol, hydroxyl, carbonyl and ether for activated carbon samples and amine, amide, pyrone, lactone, carbonyl and hydroxyl for activated carbon fibres samples. The nanostructure of the materials is predominantly microporous but with a significant variety of porosity development with BET surface area and pore volume given by as method range from 89 to 1050 m2 g1 and 0.04 to 0.50 cm3 g1, respectively. The electrochemical properties of the materials were investigated using classic cyclic voltammetry, chronopotentiometry and electrochemical impedance spectroscopy. The higher specific capacitance achieved was 176 F g1. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction The use of carbon materials as supercapacitor electrodes arises from a unique combination of chemical and physical properties, namely: (1) high conductivity; (2) high surface area and porosity; (3) good corrosion resistance; (4) high temperature stability; (5) controlled pore structure; (6) processability and compatibility in composite materials (Pandolfo and Hollenkamp, 2006). The development of carbon materials tailored for electrochemical application is actually the subject of a considerable amount of papers (Frackowiak and Béguin, 2001; Pandolfo and Hollenkamp, 2006; Obreja, 2008). However, more research is needed in order to produce better materials and to increase the knowledge about the critical parameters and the mechanisms of the processes. The charge accumulation, or capacitance, can be acquired via the electrical double layer formed at the electrode/electrolyte and enhanced through pseudocapacitance effects by additional faradaic reactions determined by the presence of heteroatoms. The performance of carbon materials depends on a complex combination of several factors which includes carbon material properties (porosity, surface chemistry, electrical conductivity, wettability), electrode preparation method and electrolyte characteristics (Frackowiak and Béguin, 2001). The production of the electrode typically involves an activated carbon and a suitable binder done in a pos-production step subse⇑ Corresponding author. Tel.: +351 266745318; fax: +351 266745378. E-mail address: [email protected] (J.M. Valente Nabais). 0960-8524/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2010.11.083

quent to the activated carbon preparation. The most frequent binders are a mixture of carboxymethylcellulose and poly(tetrafluoroethylene) (Gamby et al., 2001) and the use of polyvinylidene fluoride (Takeuchi et al., 2006; Ismanto et al., 2010). The electrode production step makes the process more expensive and environmentally less friendly. In the present work all materials were prepared in monolithic shape without the use of binders, which can be considered a major advantage over other reported methodologies. Additionally, it should be point out that the used materials were never tested before for such applications. The activated carbon fibres (ACF) have several advantages over activated carbons (AC) such as high rates of adsorption/desorption and uniform and controllable micropore structure (Carrott et al., 2002). In addition, it is possible to construct continuous forms, such as fabrics and composite membranes and filters that can significantly simplify process design and reduce costs. The main difference between activated carbons and activated carbon fibres is the microscopic shape of the particles, which can be observed with scanning electron microscopy (Carrott et al., 2002). The ACFs are in the form of filaments where the microporosity opens directly to the exterior. In ACs, the microstructure is normally spherical and the porosity decreases from macro (dp > 50 nm) to meso (2 nm < dp < 50 nm) to microporosity from the external surface towards the interior of the material (Valente Nabais and Carrott, 2006). The use of biomass; for example wood, fruit shells and stones; as precursors for producing activated carbons (AC) has been studied with excellent results, which proves the aptitude of lignocellulosic materials to produce carbon materials (Crini, 2006; Amaya

2782

J.M. Valente Nabais et al. / Bioresource Technology 102 (2011) 2781–2787

et al., 2007; Suhas Carrott and Ribeiro Carrott, 2006). The precursor used in this work to produce the activated carbon samples is an industrial residue from the Coffee Industry. Therefore, the study of possible applications for this material is also relevant to protect the environment by the reuse of industrial residues. The objective of this work is the development of low cost electrodes with potential to be used in supercapacitors based on activated carbons and activated carbon fibers produced from biomass and industrial residues. The aim is to produce the electrodes in the monolithic shape using an environmentally friendly process without the need for a pos-production step.

2. Experimental 2.1. Materials The activated carbon samples were produced by physical and chemical activation from a lignocellulosic precursor, coffee endocarp, which is an industrial residue of the coffee industry given to us by NovaDelta-Comercio e Industria de Cafes (Campo Maior, Portugal) in the form of cylindrical pellets. The physical activation with carbon dioxide as activating agent was done in a horizontal tubular furnace with heating rate of 10 °C min1. Carbonisation was carried out for 1 h by heating to 600 °C under a constant N2 flow. Activation was carried out at 800 °C under a CO2 flow for 1, 2 and 3 h in order to obtain burn-offs 23, 40 and 63 wt%, indicated in the samples designation after C8. The carbonised sample was designated by CC. For the AC production by chemical activation about 10 g of precursor and a horizontal tubular furnace were used. The precursor was mixed with powdered KOH to obtain impregnation ratios of 1:0.5 and 1:2 (precursor:KOH), samples AQ605 and AQ62, respectively. The pyrolysis of the mixtures was done for 2 h under a constant N2 flow at 650 °C using a heating rate of 5 °C min1. More experimental details can be found on Valente Nabais et al. (2008a,b). The activated carbon fibre samples in cylindrical monoliths were produced by pressing the precursor uniaxially at 10,000 kg for 20 min in a 13 mm Specac die, in vacuum. The resulting green monolith was heated in a horizontal tubular furnace to 300 °C at a rate of 1 °Cmin1 under a constant N2 flow and maintaining for 2 h. After the stabilisation step the temperature was raised at a rate of 5 °C min1 to 800 °C and maintained at the carbonisation temperature for 1 h. Activation was carried out at 900 °C for 1, 2 and 3 h under a carbon dioxide flow, samples F920, F932 and F993, respectively. As precursor a commercial acrylic textile fibre (provided by FISIPE, Barreiro, Portugal) was used. According the manufacturer the fibre with bright three denier filaments was produced from acrylonitrile (90%) and vinyl acetate (10%) monomers. More experimental details can be found on references (Carrott et al., 2001) and (Valente Nabais et al., 2006). The sample F932 was Table 1 Samples designation and production method. Sample

Precursor

Production method

CC C823 C840 C863 AQ605

Coffee endocarp

Carbonisation at 600 °C for 1 h Activation with CO2 at 800 °C for 1 h Activation with CO2 at 800 °C for 2 h Activation with CO2 at 800 °C for 3 h KOH activation 1:0.5 (precursor:KOH) at 650 °C KOH activation 1:2 (precursor:KOH) at 650 °C

AQ62 F920 F932 F932oxi F993

Commercial acrylic textile fibre

Activation with CO2 at 800 °C for 1 h Activation with CO2 at 800 °C for 2 h F932 oxidised with nitric acid Activation with CO2 at 800 °C for 3 h

oxidised in liquid phase with concentrated nitric acid during 1 h in a hot plate with stirring at 80–90 °C. The monoliths were removed and washed with distilled water until the wash water attained the same pH value as the distilled water employed in the wash. The oxidised fibre was designated as F932oxi.All the conditions used are indicated in Table 1 as well as the sample designations. 2.2. Characterisation 2.2.1. Textural, microstrutural and chemical Nitrogen adsorption isotherms at 77 K were determined using a CE Instruments Sorptomatic 1990 after outgassing the samples at 400 °C to a residual vacuum of 5  106 mbar. FTIR spectra were recorded with a Perkin Elmer model Paragon 1000PC spectrophotometer, using the KBr disc method, resolution of 4 cm1 and 100 scans. Elemental analysis of carbon, hydrogen, sulfur, nitrogen and oxygen was carried out using a Eurovector EuroEA elemental analyser. X-ray powder diffraction patterns were determined using a Bruker AXS D8 Advance diffractometer using Cu Ka radiation (40 kV, 40 mA) at a step size of 0.020° between 5.000° and 60.020°. The point of zero charge was determined by mass titrations, details given elsewhere (Valente Nabais et al., 2004). 2.2.2. Electrochemical measurements The electrochemical properties of the materials in solution, such as the electric double-layer capacitance, were investigated using classic cyclic voltammetry (CV), chronopotentiometry (CP), and electrochemical impedance spectroscopy (EIS). The electrochemical measurements were carried out with an Autolab PGSTAT 20 electrochemical analysis system (Eco Chemie B.V., The Netherlands), equipped with a Frequency Response Analysis (FRA2) module, which allows to record impedance and capacitance measurements. This system was computer controlled with the GPES and FRA software packages (version 4.9 in both cases). The electrochemical measurements were done over a single carbon electrode immersed in a deoxygenated H2SO4 (1 M) aqueous solution, at ambient temperature. A standard three-electrode cell configuration was employed. The carbon material was used as working and as a counter electrode. The samples weight used as working electrodes was between 50 and 80 mg. The weight of the counter electrode material was two to three times greater. Both electrode materials were connected to a stainless steel wire, with a hook configuration. An Ag/AgCl/NaCl (3 M) electrode, mounted in a saline bridge, was used as the reference electrode. Cyclic voltammetric measurements were carried out within a potential range of 0.1–1.0 V, at a potential scan rate that was ranged from 0.1 to 50 mV/s. These measurements were used in order to access more promptly the capacitive behavior of samples within the working potential region. The chronopotentiometric (charge/ discharge) measurements were performed at constant current of 0.1 and 1 mA, while the impedance measurements were carried out within a potential range of 0.5–0.6 V, using the ac frequency range of 10 kHz to 10 mHz and an excitation amplitude of 5 mV. 3. Results and discussion 3.1. Materials textural, microstrutural and chemical characterisation The porous characteristics of the carbon materials, shown in Table 2, were evaluated by N2 adsorption/desorption isotherms at 77 K and analysis of the isotherms by suitable methods such as Brunauer–Emmett–Teller (BET), Dubinin–Radushkevich and aS. The isotherms, not shown here for the sake of simplicity, were all type I according the IUPAC classification (Rouquerol et al., 1994),

2783

J.M. Valente Nabais et al. / Bioresource Technology 102 (2011) 2781–2787 Table 2 Textural and chemical characterisation. Sample

CC C823 C840 C863 AQ605 AQ62 F920 F932 F932oxi F993

Porosity

pzc

BET

aS

ABET (m2 g1)

Vs (cm3 g1)

Aext (m2 g1)

Vo (cm3 g1)

a

a 0.33 0.34 0.46 0.17 0.11 0.04 0.23 0.26 0.30

a

a 0.34 0.34 0.44 0.16 0.11 0.04 0.23 0.24 0.21

695 709 1038 361 245 89 464 540 633

Elemental composition (wt%)

DR

3 7 60 9 8 4 42 38 83

9.90 10.0 9.96 10.0 9.25 9.09 10.1 11.0 2.72 10.3

O

C

H

N

10.10 11.52 11.59 11.43 10.47 15.60 7.31 8.74 13.54 8.76

64.9 63.12 60.24 56.97 62.24 61.96 82.73 80.38 71.51 73.01

1.00 0.58 1.16 0.83 1.00 1.02 0.92 1.60 1.06 0.95

2.40 1.50 1.89 1.76 2.51 2.25 7.94 5.27 4.62 4.73

a, null N2 adsorption at 77 K.

which indicates microporous solids with low external area. It can be seen in Table 2 that all samples all samples have very low external area in reference with their apparent surface, with values between 3 and 83 m2 g1. The porosity development is dependent on the precursor and production method used. Therefore, the samples have apparent BET surface area values between 89 and 1038 m2 g1. ACFs samples F920, F932, F932oxi and F993 have lower BET area than the activated carbon samples. In all cases, the value increases with samples’ burn-off. The ACs produced by chemical activation with KOH (AQ605 and AQ62) show slightly less developed porosity structure when compared with ACs produced from the same precursor, coffee endocarp, but with CO2 activation. The same trend can be observed for the total micropore volume evaluated by as method (Vs) and primary micropore volume indicated by DR method (Vo). The difference between the pore volumes given by Vs and Vo, can provide an estimate of the primary and secondary micropore volume considering the as volume as the total pore volume and the DR pore volume as an indication of the primary micropores. As shown in Table 2, all ACs samples have very similar values for both volumes, which indicate that most of the pores present in the AC samples are narrow micropores, the pore volume in the range of mesopores and secondary micropores is almost zero. The volumes of the primary micropores are the Vo volumes in Table 2. For the ACFs samples we can observe the same behavior with the exception of the highly activated sample, F993, where some broader micropores were formed. In this sample the volume of primary and secondary micropores are 0.21 and 0.09 cm3 g1, respectively. All pristine samples, ACs and ACFs, have basic properties as indicated by the pH values correspondent to the point of zero charge (pzc) always superior to 9. In relation to the elemental composition shown in Table 2 the main focus should be the oxygen and nitrogen content as both heteroatoms has a possible impact on the electrochemical characteristics of the materials. As can be observed in Table 2 both type of materials have different features in this regard. The pristine ACFs samples have less oxygen content, 7–8 wt%, and higher nitrogen quantity, 5–8 wt%, than ACs samples, which have oxygen content between 10 and 15 wt% and nitrogen in the range 1.5–2.5 wt%. In order to study the influence of the surface chemistry on the electrochemical properties the sample F932 was oxidised with nitric acid. This treatment had produced drastic alterations to the surface chemistry of the sample, as indicated by the pzc values that changed from 11.0 to 2.72. The impact of the nitric acid treatment on the porosity was much less noticeable. As can be seen in Table 2, the apparent BET surface area and micropore volume only showed a slight increase.

The evaluation of the surface chemistry by FTIR, spectra not shown here, has shown that different materials have also different surface properties. The spectra interpretation was done using the following nomenclature: m, d, c, qw for stretching, deformation, bending and waging mode of vibration, respectively. The main functional groups indentified in the activated carbon samples produced by CO2 activation are quinone, lactone, Si–H, phenol, hydroxyl, carbonyl and ether. The bands situated in the range 3400–3200 cm1 can be attributed to –OH groups (phenol and free). This groups are also responsible for the bands at 1400– 1300 cm1 (d(O–H)), and 680–620 cm1 (m(C–O) and c(O–H)). The presence of Si is indicated by the band at 900–820 cm1 (qw(Si–H)), which intensity increases with activation. The presence of C@O bond is clearly visible in the spectra integrated in several functional groups, specifically ketones (1800–1760, 1700– 1675 cm1) and quinones (1675–1616 cm1). The presence of cpyrone is indicated by the band situated at 1450–1420 cm1. The principal functional groups on the activated carbon samples produced by chemical activation, samples AQ605 and AQ62, were lactones, quinones, ketones and carbonyls in several structures. We can also indentify c-pyrone, phenol and alcohol. The band situated at 900–820 cm1 demonstrates the existence of Si–H bonds (qw(Si–H)). The bands in the AQ605 spectrum are less intense than in the case of AQ62 sample, which indicates a poorer surface chemistry (Guo and Lua, 2003; Machnikowski et al., 2005). The activated carbon fibres have a different surface chemistry mainly due the presence of nitrogen. The bands situated at 3900–3600 cm1 can be attributed a –OH and NH groups in amide. The bands at 1400 (dCH, dNH), 1340 (mCN) and 668 cm1 (mNH, qwNH) are directly related to the presence of amines. Another functional group relevant to the basicity of the samples is pyrones, which presence is indicated by the bands at 1874–1830 cm1 (m(C@O)). The presence of lactones is also indentified. In the oxidised sample, F932oxi, the main difference is the presence of more groups with oxygen, namely carboxylic acid. The microstructural characterisation was carried out by scanning electron microscopy (SEM) and X-ray diffraction (XRD). The monoliths have cylindrical shape with similar dimensions, approximately 9 mm length and 3 mm diameter, medium values for the activated carbon monoliths. The activated carbon fibers monoliths show a slightly bigger diameter, 5 mm. Despite the similar macroscopic shape the materials have different morphologies that are linked to the type of precursor. The acrylic fibres monoliths, such as F920, are made of activated carbon fibres with kidney/bean cross section. On other side, ACs monoliths are much more compact without the filaments visible in the ACFs case. We can also notice a more heterogeneous surface for sample AQ62, produced by KOH activation, when compared with C840, produced by CO2 acti-

2784

J.M. Valente Nabais et al. / Bioresource Technology 102 (2011) 2781–2787

vation. From the macroscopic and SEM observation is clearly visible that after the activation process the raw materials maintain the monolithic shape without the need for any other process such as the use of a chemical binder. The dimension of the microcrystaline domains in each sample can be evaluated from the XRD patterns, namely from the two common peaks for carbon materials due to reflections from the (0 0 2) and (10) planes, which are clearly visible in all XRD patterns, not shown here. The interplanar spacing, d002, can be evaluated by the application of Bragg’s Law to the position of the (0 0 2) peak. Estimates of the mean microcrystallite dimensions can be obtained by application of the Debye–Scherrer Equation. When applied to carbon materials, the equation takes the forms (Carrott et al., 2001):

Lc ¼ 0:90k=b cos h002

ð1Þ

La ¼ 1:94k=b cos h10

ð2Þ

where b is equal to the peak width at half height corrected for instrumental broadening, Lc and La are estimates for the height and width of the microcrystallites. The mean number of layer planes in the microcrystallites (Np) can be estimate using the ratio Lc/d002. Also, the relative density of edge and basal planes in microcrystallites can be estimated by the ratio Lc/La, materials with a higher edge orientation have also higher values of Lc/La. The results of characterisation by XRD measurements are shown in Table 3. The samples show dissimilar dimensions for the microcrystalline, as can be seen in Table 3. The AQ series have the lowest La values of all samples, which indicate the presence of narrow microcrystallites. The wider microcrystallites were found for F series. Different precursors show different trend, in one hand C series show a decrease in La value with activation but, in another hand, for F series we can observe the opposite behavior. The ACFs show an almost constant Lc value (1.35 nm) independently of the sample’s burn-off while C and AQ series show samples with variable microcrystallites height. We can also conclude that the liquid phase oxidation did not have a significant impact in the microcrystallites characteristics. The evaluation of the microcrystallite size and orientation is relevant as both parameters can have influence on the formation of the electric double layer and on the capacitance value for each sample. 3.2. Electrochemical performance

Representative cyclic voltammograms in 1 M H2SO4 solution are shown in Fig. 1. In the middle potential region, the cyclic voltammograms shows almost rectangular shape with no distinct peaks, which indicates the absence of strong pseudocapacitance (faradaic) phenomena. At more extreme potentials (close to 0.100 and 1.000 V) the electron transfer reactions involving the solvent and some surface groups of the electrode material occurs, which contributes extensively to the measured current. It was also observed that this behavior is independent of the number of charge/discharge cycles. At the middle polarization conditions the charge–discharge response of the tested sample/solution interfaces seems to be purely electrostatic. For AQ605 and F932 samples the positive branch, of the respective cyclic voltammogram, is more horizontal than the corresponding negative branch, and also than the cyclic voltammogram branches of C863 sample. This suggests that the charging of AQ605 and F932 electrodes with anions produces a steady state that is more independent of the electrical potential than its charging with cations. The time taken to achieve a positive steady state current is greater for the AQ605 electrode than for the F932. With respect to the C863 electrode, both anionic and cationic charging/ discharging processes are more potential dependent. The voltammograms also show the better performance of ACs when compared with ACFs samples. The specific capacitance values (Cm), which are proportional to the measured current intensity according to the relation (Fuertes et al., 2005; Pell and Conway, 2001; Niu et al., 2006):

C m ¼ iðmmÞ1

ð3Þ

where m is the scan rate value and m is the weight of the carbon electrode, are greater for C863 and A605 samples, than for F932 sample. The rate capability to store charge, as expressed by Eq. (3) was also studied. Fig. 2 shows the scan rate dependence of specific capacitance for C863 in 1 M H2SO4. The results indicate that the charge–discharge performance of this porous electrode is markedly time dependent. The electrode shows a good capacitive behavior when is charged/discharged over the range of scan rate of 0.25–5 mV s1, decreasing its specific capacitance with the increasing of scan rate. For a scan rate of 5 mV s1, the specific capacitance at 0.500 V is 50% of the value measured at 1 mV s1 (37.6 F g1). For the AQ605 electrode, the drop in capacitance with increasing scan rate is even higher (the value obtained at 5 mV s1 is 25% of the value obtained at 1 mV s1). The porous electrode with the best performance is F932. When higher scan rates are

The electrochemical performance of all carbon samples, as electrode materials, was tested using the most common techniques, which gives complementary results regarding the different phenomena that could occur at the electrode/solution interface, namely cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and chronopotentiometry (CP).

Table 3 Results of characterisation by XRD measurements. Sample

d002 (nm)

La (nm)

Lc (nm)

Lc/La

Np

CC C823 C840 C863 AQ605 AQ62 F920 F932 F993 F932oxi

0.350 0.384 0.348 0.304 0.342 0.341 0.345 0.344 0.352 0.352

2.33 3.82 3.21 2.91 2.39 2.43 3.12 3.49 4.13 4.23

1.03 1.19 1.15 1.41 1.60 1.11 1.35 1.36 1.34 1.42

0.44 0.31 0.36 0.48 0.67 0.46 0.43 0.39 0.32 0.34

2.3 3.1 3.3 4.6 4.7 3.3 3.9 4.0 3.8 4.0

Fig. 1. Cyclic voltammograms of C863, AQ605 and F932 samples in 1 M H2SO4 at m = 1 mV s1. For convenience, the current intensity was directly converted in specific capacitance values (Cm).

J.M. Valente Nabais et al. / Bioresource Technology 102 (2011) 2781–2787

Fig. 2. Specific capacitance of C863 sample in 1 M H2SO4 vs. potential, at different scan rates: (a) 0.25; (b) 1; (c) 2 and (d) 5 mV s1. Arrows shows the trend of specific capacitance values with the increasing scan rate.

used, the voltammograms tend to become distorted from the ideal rectangular shape, indicating that the capacitive behavior of electrode material becomes very weak. This behavior is typical for porous materials with almost ideal polarizable electrodes characteristics. Considering that the charge/discharge current of the non faradaic systems, at time t is given by (Doménech-Carbó, 2010):

h i i ¼ mC þ Ein R1  mC eðt=RCÞ

ð4Þ

where C and R represent the capacitance and resistance of the system, respectively, the product of these two variables (RC) is the time constant of this system, Ein is the start potential and the potential scan rate, it is expected that with very low scan rates or for systems with moderate values for RC, the electric current may reach a steady state value (i.e., ic = mC), since its transient component (second term in Eq. (4)) becomes negligible. In this case, the measured current, within a given potential window, may be less dependent or even independent, of the value of the electrical resistance of the porous capacitor system. Consequently, a linear dependence between the electric current and the scan rate can be observed, indicating that the charge storage is purely capacitive with an insignificant contribution of electrical resistive effects (known as the equivalent series resistance, ESR (Takeuchi et al., 2006)). In these cases, it can be shown that the specific capacitance can be almost independent of the potential scan rate. As was already mentioned, in particular for the electrodes AQ605 and F932, the capacitive behavior predominates at low scan rates, suggesting that the RC time constant of these systems can have a strong effect at higher charge/discharge rates. For example, the AQ605 system despite having a capacitance greater than F932, it takes more time to be fully charged, as suggested by the more sloping branch at the beginning of the voltammogram (see Fig. 1). The use of small speed values creates a more homogeneous electric potential along the material with minimal variations of potential and dispersion of capacitance in the porous electrode (Pell and Conway, 2001). On the contrary, when the sweep rate is increased the transient component becomes greater and more dependent of the RC time constant of the capacitor system. As can be seen in Fig. 2 the specific capacitance is scan rate dependent for all electrodes. Additionally, at very high sweep rates, the increase of current increases the ohmic drop (iR). Consequently, the electric potential in the porous electrode is not constant leading to a capacitance dispersion of the system and to a degradation of the electrical properties, like the decrease of charge and discharge rate.

2785

The electrochemical impedance spectroscopy (EIS) was also used to monitor the electrochemical behavior of the porous electrodes. Typical Nyquist plots for the F932, AQ605 and C863 electrodes in 1 M H2SO4 solution at 0.500 V are given in Fig. 3. As shown in Fig. 3, the Nyquist plots for all samples consist of a relatively large semicircle in the high-frequency region, followed by a discontinuity at intermediate frequency values and by a linear segment with increasing slope, in the low-frequency region. The semicircle represents the combined response of a not significant capacitive effect with resistive effects at the electrode/solution interface. The capacitive response is attributed to the faradic process of charge/discharge that occurs in the interfacial regions more exposed and accessible to the ions of the electrolyte. Since this capacitance is only related to the more external surface area of the material, it does not represent the maximum allowable capacitance of the porous material. The resistive effects, are the result of several aspects, such as electronic resistance of the electrode, interfacial resistance between the electrode and the current-collector, electrolyte solution composition, ionic diffusion of ions moving in small pores and, when redox species/groups are present, to the resistance of charge transfer of faradaic phenomena (Pandolfo and Hollenkamp, 2006; Doménech-Carbó, 2010). As the cyclic voltammograms did not reveal the presence of any redox peak at this potential region, the contribution of faradaic processes with a considerable resistance of charge transfer appears to be negligible, which indicates the resistive effects associated with the ionic and electronic conduction trough the system as the principal factor. The magnitude of these resistive effects, which are collectively quantified by the equivalent series resistance (ESR) (Niu et al., 2006), can be correlated to the size of the semicircle, being proportional to their diameter. The value of ESR, as indicated in the literature (Niu et al., 2006; Kötz and Carlen, 2000), could be obtained from the value of Z0 at the frequency of 1 kHz. As can be seen in Fig. 3, the diameter of the semicircle of the impedance spectra of each sample increases in the following sequence F932 > C863 > AQ605. Therefore, it can be assumed that the resistive behavior of these porous electrodes increases in the same sequence. The series equivalent resistance values obtained for F932, C863, and AQ605 were 7.4, 11.7 and 18.9 X, respectively. The size and orientation of the microcrystallites can also be relevant to the resistance of the material. It is very difficult to analyze the influence of the material characteristics onto the electrochemical properties because of the complexity of the process and to the dissimilar impact of different parameters on the performance of

Fig. 3. Nyquist plots of F932, AQ605 and C863 electrodes in 1 M H2SO4; applied potential: 0.500 V vs Ag/AgCl/3 M NaCl. The mass of AQ605, C863 and F932 was 0.0780, 0.0491 and 0.0610 g, respectively.

2786

J.M. Valente Nabais et al. / Bioresource Technology 102 (2011) 2781–2787

the materials. The overall influence is the sum of the interaction between those parameters. Nevertheless, it can be seen that the resistance of the material is not related to the porosity but to the heteroatom content and to the edge orientated microcrystallites. The resistance has smaller values for samples with higher content of nitrogen, lower content of oxygen and lower amount of edge orientated microcrystallites evaluated by the ratio Lc/La. The discontinuity, at intermediate frequency values, commonly observed with microporous carbon electrodes, is attributed to the resistance of ion migration inside the carbon micropores, combined with the increasing capacitive phenomena that can occur in the pores. When this discontinuity is very small, which is the case of the impedance spectra of F932 and C863 electrodes, it can be assumed that the capacitive phenomena on the micropores become increasingly important. On the contrary, when this discontinuity is large, which is the case of the impedance spectra of AQ605 electrode, it is suggested that the resistive effects are still more important than the capacitive, since the real component of impedance continues to increase while the imaginary component remains practically invariable. Consequently, the data indicates that AQ605 is the electrode material with performance capacitive more dependent on the loading rate. This characteristic of the AQ605 system, as already noted, can be also seen on the voltammogram in Fig. 1. The linear segment, in the low-frequency region, represents the increasingly capacitive behavior of the electrically charged interface. For such frequencies, a greater amount of ionic charge can penetrate in the porous structure and be accumulated in the available internal surface of the porous material, contributing to its maximum electrical capacitance. The slope of this segment indicates how time dependent is the penetration of ionic charge and its accumulation in the porous material, and consequently the capacitance of the electrode/solution interface. This dependence of the capacitance with the frequency is shown in Fig. 4 for all samples, plotting the specific capacitance (Cm) of the active material vs. log f. The specific capacitance was calculated according to Eq. (5) (Fuertes et al. 2005; Yuan et al., 2008):

    00 C m ¼ ð2pfZ mÞ1 

ð5Þ

where f, Z00 and m are the frequency, the imaginary impedance and the mass of the active material, respectively. From Fig. 4 it can be seen that the specific capacitance of the samples, at the lower frequency (f = 10 mHz), follows the sequence C863 > AQ605 > F932. We can also observe that the frequency

Fig. 4. Plot of specific capacitance vs. log f of C863, AQ605 and F932 electrodes in 1 M H2SO4; applied potential: 0.500 V vs. Ag/AgCl/3 M NaCl.

increases with the diminution of the capacitance. The biggest variation of the specific capacitance value with the frequency is obtained for AQ605. These observations are in agreement with the results obtained from the cyclic voltammetric studies. The time constant (RC) estimated by the formula RC = ESR1kHz  m  Cm, where ESR is the equivalent series resistance measured at 1 kHz (=Z0 1kHz) and m the mass of active material, have the values of 56.3, 52.3 and 6.4 s for AQ605, C863 and F932, respectively. The results show that C863 and AQ605 have superior capacitance values than F932 despite having slower response times. The results were confirmed by chronopotentiometry. From the galvanostatic charge–discharge curves, E vs. Q (curves not shown here for simplicity), the specific capacitance of each sample was calculated according to Eq. (6) (Chen et al. 2004):

C m ¼ ½ðDE=DQ Þ  m1

ð6Þ

where E is the potential, Q the charge that was passed during a period t, when a constant intensity current (i) charge/discharge the electrode, and (DE/DQ) represents the slope of the linear part of the curve. Considering a charging/discharging current of 1 mA the values of Cm obtained for F932, AQ605 and C863 were 9.3, 20.4 and 94.4 F g1, respectively. These values are consistent with the values obtained by cyclic voltammetry and electrochemical impedance spectroscopy. For larger values of i the performance of the electrodes becomes poor because their response times are relatively large, as already discussed. The specific capacitance values measured by the galvanostatic method (chronopotentiometry) at 10 mA between 0.2 and 0.6 V were for samples C823, C840 and C863, respectively, 149, 176 and 167 F g1. For samples chemically activated were 69 and 16 F g1, respectively for samples AQ605 and AQ62. In the case of the activated carbon samples the value measured were 2, 7 and 85 F g1 for F920, F932 and F993, respectively. The capacitance of oxidised activated carbon fiber, F932oxi, is very low (0.07 F g1). The observed trend for ACs in series C can only be justified by the differences on the porosity and microcrystallite dimensions and orientation as all samples have similar surface chemistry, as can be seen in Table 2. The capacitance value increases with the porous development and edge orientated microcrystalites, indicated by bigger values of Lc/La. Regarding the chemically activated carbons, series AQ, both samples have similar porous structure but dissimilar elemental composition and microcrystallite organization. It can be observed that the presence of oxygen is detrimental for the capacitance value obtained. Sample AQ62 has higher oxygen content and smaller capacitance. The same trend can be observed for the activated carbon fibres, namely samples F932 and F932oxi, for which the oxidation and the correspondent oxygen content increase lead to a diminution of the capacitance in about two orders of magnitude. The oxidation did not introduced pseudocapacitance behavior, as observed by the non-appearance of the respective peaks on the cyclic voltamogramms. The oxidation, and the increment of oxygen content, besides having a direct impact on the mobility of the delocalized electrons at the aromatic planes of the materials, also has changed the pzc value (see Table 2), which is relevant for the formation of the electric double layer. The value of pzc for sample F932oxi is very similar to the pH of electrolyte used thus the surface of the material do not have a highly developed surface charge which is detrimental for the formation of the electric double layer. For ACFs samples the porous development namely the enlargement of the pores is crucial for achieving a higher capacitance value. The porous enlargement can be evaluated from the difference between the values Vs and Vo, higher differences means that the percentage of wider microporous is also bigger. The sample with better electrochemical performance is the sample with higher sec-

J.M. Valente Nabais et al. / Bioresource Technology 102 (2011) 2781–2787

ondary micropore volume, which proves that bigger pores give better electrodes. The ACFs have a worst performance than activated carbons because they also have higher content of nitrogen, despite having smaller content in oxygen, and because the porous structure is also different, as already mentioned in Section 1. This characteristic of the ACFs involves a bigger resistance to the ion mobility of the electrolyte and thus poorer capacitance values. Additionally, the surface chemistry is different between the activated carbon samples and activated carbon fibres used as found by FTIR, the major differences are the presence of amine, amide and pyrone surface functional groups for ACFs samples and quinone, Si–H and ether groups for ACs samples. The oxygen content has a different influence on the capacitance values depending on the sample type. For the F and AQ series the capacitance decreases with the increase of oxygen content, which indicates that the pseudocapacitance effect is not relevant. However, for C series it appears that oxygen content does not influence the electrochemical performance of the materials. It seems that higher pore volume or surface area lead to better performance. The large surface area and suitable porous structure are crucial for charge accumulation, easy electrolyte wetting and rapid ionic motion. The comparison of the results obtained in different laboratories must be carefully done as different conditions were used, namely electrochemical cell construction and operational conditions. So, we must have this in mind when compare our results with other already published papers. Nevertheless, we can note that the specific capacitance obtained in some of our samples can be considered very good. It is also evident that the electrochemical properties are not controlled by a single parameter, such porosity or surface chemistry, but by a set of factors which influence are many times overlapped being the final results the sum of such overlapping impact. Also, for different type of carbon materials the parameters that constitute the driving force for better performance as supercapacitor can be also different.

4. Conclusion The innovative aspect of producing suitable materials to be used as electrochemical supercapacitors in monolithic shape but without the use of a binder was achieved, which can be considered advantageous over other materials from the economical and environmental point of view because the electrode pos-production step and the use of binders are not needed. The electrochemical characterisation has lead to the following conclusions: – Materials more effective if used as anodes; – Pseudocapacitance phenomenon is not relevant for the performance; – Main charge process is the formation of the electric double layer; – Properties controlled by a set of factors.

Acknowledgements The authors are grateful to the Fundação para a Ciência e Tecnologia (Portugal), COMPETE, QREN e European Union (European Regional Development Fund, FEDER) for financial support through Project FCOMP-01-0124-FEDER-007142) and to Novadelta-Comércio e Indústria de Cafés, S.A. and FISIPE for the provision of the coffee endocarp and acrylic fibres used as precursors. The authors are

2787

also grateful to Peter Carrott and Manuela Carrott for the discussions and suggestions that have improved the paper. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.biortech.2010.11.083. References Amaya, A., Medero, N., Tancredi, N., Silva, H., Deiana, C., 2007. Activated carbon briquettes from biomass materials. Bioresource Technology 98, 1635–1641. Carrott, P.J.M., Nabais, J.M.V., Ribeiro Carrott, M.M.L., Pajares, J.A., 2001. Preparation of activated carbon fibres from acrylic textiles fibres. Carbon 39, 1543–1555. Carrott, P.J.M., Nabais, J.M.V., Ribeiro Carrott, M.M.L., Pajares, J.A., 2002. Scanning electron microscopy of activated carbons prepared from commercial acrylic textile fibres. Fuel Processing Technology 77–78, 381–387. Chen, Q.-L., Xue, K.-H., Shen, W., et al., 2004. Fabrication and electrochemical properties of carbon nanotube array for supercapacitors. Electrochimica Acta 49, 4157–4161. Crini, G., 2006. Non-conventional low-cost adsorbents for dye removal: a review. Bioresource Technology 97, 1061–1085. Doménech-Carbó, A., 2010. Electrochemistry of Porous Materials. CRC Press, Boca Raton. Frackowiak, E., Béguin, F., 2001. Carbon materials for the electrochemical storage of energy in capacitors. Carbon 39, 937–950. Fuertes, A.B., Lota, G., Centeno, T.A., Frackowiak, E., 2005. Templated mesoporous carbons for supercapacitor application. Electrochimica Acta 50, 2799–2805. Gamby, J., Taberna, P.L., Simon, P., Fauvarque, J.F., Chesneau, M., 2001. Studies and characterisations of various activated carbons used for carbon/carbon supercapacitors. Journal of Power Sources 101, 109–116. Guo, J., Lua, A.C., 2003. Textural and chemical properties of adsorbent prepared from palm shell by phosphoric acid activation. Materials Chemistry and Physics 80, 114–119. Ismanto, A.E., Wang, S., Soetarejo, F.E., Ismadji, S., 2010. Preparation of capacitor’s electrode from cassava peel waste. Bioresource Technology 101, 3534–3540. Kötz, R., Carlen, M., 2000. Principles and applications of electrochemical capacitors. Electrochimica Acta 45, 2483–2498. Machnikowski, J., Grzyb, B., Machnikowska, H., Weber, J.V., 2005. Surface chemistry of porous carbons from N-polymers and their blends with pitch. Microporous and Mesoporous Materials 82, 113–120. Niu, J., Pell, W.G., Conway, B.E., 2006. Requirements for performance characterization of C double-layer supercapacitors: applications to a high specific-area C-cloth material. Journal of Power Sources 156, 725–740. Pandolfo, A.G., Hollenkamp, A.F., 2006. Carbon properties and their role in supercapacitors. Journal of Power Sources 157, 11–27. Pell, W.G., Conway, B.E., 2001. Voltammetry at a de Levie brush electrode as a model for electrochemical supercapacitor behaviour. Journal of Electroanalytical Chemistry 500, 121–133. Rouquerol, J., Avnir, D., Fairbridge, C.W., Everett, D.H., Haynes, J.H., Pernicone, N., Ramsay, J.D.F., Sing, K.S.W., Unger, K.K., 1994. Recommendations for the characterization of porous solids. Pure and Applied Chemistry 66 (8), 1739– 1758. Suhas Carrott, P.J.M., Ribeiro Carrott, M.M.L., 2006. Lignin – from natural adsorbent to activated carbon: a review. Bioresource Technology 97, 2301–2312. Obreja, V.V., 2008. On the performance of supercapacitors with electrodes based on carbon nanotubes and carbon activated material – a review. Physica E: Lowdimensional Systems and Nanostructures 40 (7), 2596–2605. Takeuchi, K.J., Marschilok, A.C., Lau, G.C., Leising, R.A., Takeuchi, E.S., 2006. Carbon structure/function relationships: characterization and electrochemistry of carbon nanofibers. Journal of Power Sources 157, 543–549. Valente Nabais, J., Carrott, P.J.M., Ribeiro Carrott, M.M.L., Luz, V., Ortiz, A.L., 2008a. Influence of preparation conditions in the textural and chemical properties of activated carbons from a novel biomass precursor: the coffee endocarp. Bioresource Technology 99, 7224–7231. Valente Nabais, J.M., Carrott, P.J.M., 2006. A project for chemical characterisation of activated carbon fibres and activated carbons. Journal of Chemical Education 83 (3), 436–438. Valente Nabais, J.M., Carrott, P.J.M., Ribeiro Carrott, M.M.L., Padre-Eterno, A.M., Menéndez, J.A., Dominguez, A., Ortiz, A.L., 2006. TTT new acrylic monolithic carbon molecular sieves for O2/N2 and CO2/CH4 separations. Carbon 44, 1158– 1165. Valente Nabais, J.M., Carrott, P.J.M., Ribeiro Carrott, M.M.L., Menendez, J.A., 2004. Preparation and modification of activated carbon fibres by microwave heating. Carbon 42 (7), 1315–1320. Valente Nabais, J.M., Nunes, P., Carrott, P.J.M., Ribeiro Carrott, M.M.L., MaciasGarcia, A., Diaz-Diez, M.A., 2008b. Production of activated carbons from coffee endocarp by CO2 and steam activation. Fuel Processing Technology 89, 262– 268. Yuan, C.Z., Gao, B., Su, L., Znag, X., 2008. Interface synthesis of mesoporous MnO2 and its electrochemical capacitive behaviors. Journal Colloid Interface Science 322, 545–550.