Hierarchical cellulose-derived CNF/CNT composites for electrostatic energy storage

The composite CNF/CNT electrodes were produced via chemical vapor deposition of CNTs on top of cellulose-derived CNFs. Initially, CNF sheets were fabricated by executing the following three consecutive steps: cellulose acetate electrospinning (17 wt% solution of the polymer in 2:1 solvent ratio of acetone and dimethylacetamide), cellulose regeneration (in 0.1 M water solution of NaOH) and carbonization (in a quartz tube furnace with N₂ flow by heating up to 800 °C with the heating rate of 5 °C min⁻¹) according to [9]. Subsequently, CNTs were thermally grown on CNF substrates at 700 °C for 10 min using acetylene as a carbon source, 2 nm thick iron layer as a catalyst and hydrogen as a carrier gas (in AIXTRON Nanoinstruments Black Magic 2 inch machine). Finally, prior to electrochemical measurements, the CNF/CNT composites were dipped into a mixture with equal volumetric proportions of 1M nitric and sulphuric acids to remove catalytic particles, and thoroughly washed afterwards with DI water to avoid contamination with dissolved ions.

The morphology of the composite materials was observed using high resolution scanning electron microscopy (SEM) (Leo Ultra 55 FEG SEM, Zeiss) in a secondary electron mode at an acceleration voltage of 3 kV.

For transmission electron microscopy (TEM), as-grown CNF/CNT samples sandwiched between two Cu-grids (Athene G204, 400 mesh) were placed in a single tilt sample holder and further examined in a JEOL (JEM 2100) TEM equipped with a LaB₆ cathode and a Gatan (SC1000 Orius) digital cyclic charge–discharge (CCD) camera. Imaging was done using an electron acceleration voltage of 80 kV, in order to stay below the threshold for knock-on damage and to minimize electron-beam induced damages in the CNTs [22].

The surface area of the materials was measured using the Brunauer–Emmett–Teller nitrogen adsorption method, and mesopore size distribution was quantified by the Barett–Joyner–Halenda method using an adsorption isotherm (TriStar 3000 V6.04 A surface area and pore analyzer). The samples were degassed under vacuum at 225 °C for 4 h prior to the measurements.

The x-ray photoelectron spectroscopy (XPS) was performed on a Quantum 2000 scanning ESCA microprobe (Physical Electronics) to assess the presence of catalyst residues in the samples. An Al Kα x-ray source (1486.6 eV) with a beam size of 100 μm was used. The area analyzed was about 400  ×  500 µm² with a depth of 4–5 nm. The results were evaluated using MultiPAK 6.1A software.

X-ray diffraction (XRD) analysis was made using a Philips X'Pert Materials Research Dif-fractometer with an x-ray tube with Cu anode (Kα radiation, λ  =  1.541 84 Å) as a radiation generator at 45 kV and 40 mA. Phase analysis was carried out with X'Pert HighScore 3.0 (PANalytical BV).

Raman spectroscopy was implemented to evaluate the microstructure of vapor-grown CNTs. Raman spectra were obtained using a 638 nm laser source on a Horiba XploRA spectrometer equipped with an Olympus BX41 microscope. Spectra were analyzed with LabSpec software.

The electrical conductivity of the materials was evaluated using a four-point probe system (CMT-SR2000N, AIT). Five different measurement spots were taken for analysis of each sample.

Electrochemical performance was measured in a Swagelok supercapacitor cell consisting of a symmetrical two-electrode system with CNF-based carbon nanomaterials as working electrodes, electrospun cellulose as a separator, and 6 M aqueous solution of KOH as an electrolyte. The working electrodes and separators were cut to a circular area of 0.5 cm² to fit current collectors. Before starting the measurements, the electrodes were immersed in the electrolyte solution for 24 h.

Electrochemical measurements were performed with Gamry Reference 3000 potentiostat/galvanostat/ZRA and data were analyzed with Gamry Echem Analyst. A voltage range between 0 and 1 V was used for CV (cyclic voltammetry) measurements at five different scan rates (5, 10, 20, 100 and 200 mV s⁻¹), while GCD (galvanostatic charge–discharge) tests were performed at six different current densities (0.5, 1, 1.5, 2, 5 and 10 A g⁻¹). EIS (electrochemical impedance spectroscopy) was completed at an open circuit potential with an AC amplitude of 5 mV over a frequency range from 100 kHz–10 mHz. Electrochemical stability tests were performed by CCD for 2000 cycles with a current density of 1 A g⁻¹.

Freestanding CNF mats with a thickness of 25–40 µm were obtained via carbonization of electrospun cellulosic precursors. The mats consist of fibers with 50–250 nm diameters. The continuous fibers are randomly oriented and have a smooth topography (figure 1, bottom right corner). The morphology of the composite material is rather different (figure 1, top left corner). After chemical vapor deposition, the bigger CNFs were densely covered with much smaller uniform CNTs (5–10 nm tube diameters), thus forming a hierarchical nanocomposite material. A straight boundary line between the pristine CNF region and the region with deposited CNTs validates the reliable space controllable deposition of iron catalytic particles.

Zoom In Zoom Out Reset image size

Figure 2(A) reveals slightly uneven growth of CNTs on different sides of the CNF, which could be due to the heterogeneous distribution of heat or catalytic particles within a CNF substrate during CNT growth. A closer look at the CNTs (figure 2(B)) exposes catalytic particles which are commonly seen inside the tip of the CNTs but occasionally also remote from the tip, still inside the tube walls but further towards the tube base. In such cases, the tip could sometimes be curled up, indicating that the direction of growth is less straight once a tube grows past a catalyst particle. The number of tube walls typically varied between three and eight, with rare occurrences of single- and double walled tubes (figure 2(C)). Measurements were taken in several different distances from the CNT base, both near the CNF on which they were grown, as well as close to the CNT tips, and the distribution of outer and inner diameters was found to be fairly homogeneous. The measurements revealed outer tube diameters d_o in the span of 5–10 nm and inner diameters d_i—3–5 nm, resulting in a normalized thickness ${{t}_{\text{n}}}=\left(\frac{{{d}_{\text{o}}}-{{d}_{\text{i}}}}{{{d}_{\text{i}}}}\right)$ of 0.4–0.55. The range of t_n is in the lower region of typical values for CVD-grown CNTs and considerably lower than for CNTs grown by arc-discharge (where typically t_n is much larger than 0.5) [23].

Zoom In Zoom Out Reset image size

Figure 3(A) shows a nitrogen adsorption isotherm and pore size distribution for the CNF/CNT material. The shape of the isotherm is characteristic for meso- and macro-porous (pores with a size range of 2–50 and  >50 nm, respectively) materials without a developed microporous structure [24], which is in accordance with figures 1 and 2. Meso- and macro-porosity of the composite are very important for electrostatic energy storage as they provide sufficient access of electrolyte ions to the surface, support charge propagation and accumulation, and, as a result, increase the power capability of a supercapacitor [25]. Besides this, pore size distribution points to the presence of pores with a diameter below 10 nm, which makes a substantial contribution to the total surface area of the synthesized composites (table 1). Pores of this size should be attributed to the CNTs which are much smaller than the CNFs and thus have much higher specific surface area [14].

Zoom In Zoom Out Reset image size

The XPS analysis (figure 3(B)) confirmed that both the pure CNF material and the CNF/CNT composite almost completely consist of carbon with predominant C–C bonds (a peak at 284.5 eV). The pure CNFs contain slightly more oxygen which can be due to their more amorphous nature in comparison to the CNTs. It should lead to faster oxidation of CNFs when they are exposed to air.

The XRD analysis was used to evaluate the crystalline structure of the composite CNF/CNT material (figure 3(C). It is hard to differentiate a graphitic peak around 25°, which is usually attributed to CNTs [12], as it overlaps with a broad amorphous peak around 18° which is a characteristic peak for cellulose-derived carbons obtained after carbonization at temperatures below 3000 °C [24].

The Raman spectra (figure 3(D)) shows two broad bands at ~1320 cm⁻¹ and ~1590 cm⁻¹ that can be respectively assigned to amorphous sp²-bonded carbons with structural defects (D-band) and crystalline sp²-bonded carbons (G-band) [26]. I_D/I_G ratio for the composite material was calculated to estimate the level of disorder in the carbonaceous structure. This ratio is 1.07, which confirms the prevalence of an amorphous CNF part.

The conductivity values for the CNF/CNT composite are more than 15 times higher than for pure CNFs (table 1). This is due to the dense coverage of the CNF material with the highly conductive CNTs, which is enough to reach a critical electrical percolation threshold [27]. Effective combination of high electrical conductivity and appropriate morphology, i.e. the intrinsic properties of the synthesized CNF/CNT composite, is necessary for effective performance of carbon electrode materials used in supercapacitors [16].

Various electrochemical measurements were used to evaluate the performance of the CNF/CNT composites as electrode materials in a supercapacitor device.

Figure 4(A) shows the dependence of capacitance on a scan rate for the composite CNF/CNT electrode. The CV curves have a moderately rectangular shape, which indicates an EDL capacitive behavior [28]. Distortion of the rectangular shape at higher scan rates (100–200 mV s⁻¹) happens because of the lack of time for electrolyte ions to penetrate completely inside the electrode, whereas for lower scan rates (5–20 mV s⁻¹) the ions reach the inner surface of the electrode providing higher accumulative charge [5]. For each electrode, the evaluation of the specific capacitance C_s was made according to equation (1),

$$\begin{eqnarray}&&{{C}_{\text{s}}}=4~\frac{{{{\int}^{}}^{}}{{I}_{\left(\text{V}\right)}}\text{d}V}{m\left(\text{d}V/\text{d}t\right) \Delta V}\end{eqnarray} \tag{ 1 }$$

where I_(V) is the voltammetric current obtained from the integrated area of the CV curves, m is the total mass of carbon materials in two electrodes, dV/dt is the scan rate, and ΔV is the voltage range [10, 11]. The comparison of the CV curves of two different CNF-based nanostructured materials is presented in figure 4(B). The composite electrodes show about two times higher values of capacitance in comparison to the pure CNF electrodes. Addition of CNTs has a big influence on the performance of the CNF-based electrodes as CNTs significantly increase their surface area and electrical conductivity (table 1), which are key factors in making electrostatic charge accumulation process more efficient [14].

Zoom In Zoom Out Reset image size

The GCD curves show an almost triangular shape specifying EDL behavior as well (figure 5(A)). From the GCD test, power and energy density values were calculated using equations (2) and (3) respectively,

$$\begin{eqnarray}&&{{E}_{\text{d}}}=\frac{{{C}_{\text{s}}}{{\left(\Delta V\right)}^{2}}}{2}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{{P}_{\text{d}}}=~\frac{{{E}_{\text{d}}}}{t}\end{eqnarray} \tag{ 3 }$$

where C_s indicates the specific capacitance, ΔV is the potential difference excluding the IR drop and t is the discharge time [29]. From the GCD test power and energy density values were found to be reasonably high for the composite electrode materials. A fast current–voltage response proves that an electrode material has a high power density [24]. Figure 5(B) shows the dependence of the specific capacitance on current density. For the composite CNF/CNT electrode, at 0.5 A g⁻¹ the capacitance is about 1.5 times higher than at 2 A g⁻¹ and about 10 times higher than at 10 A g⁻¹. At the higher current density many electrolyte ions cannot diffuse fast enough to an electrode surface and occupy available sites, which negatively effects accumulation of charges, i.e. decreases the capacitance [30].

Zoom In Zoom Out Reset image size

According to EIS the equivalent series resistance (ESR) of a system is defined as the initial intercept of a plot with the X-axis in the high frequency region (figure 6(A)). ESR mostly comprises of bulk electrolyte resistance and interfacial resistance between an electrolyte and an electrode material. The composite electrode and the pure CNF electrode have low ESR values (0.57 and 0.38 Ω, respectively), which is good for effective electrochemical performance. A slightly lower value for the pure CNF material can be explained by its better wettability with the aqueous KOH electrolyte due to the higher amount of hydrophilic oxygen-containing groups (according to the XPS analysis). In contrast, charge transfer resistance, expressed as the intercept of a semicircle in the mid-higher frequency region, is lower for the CNF/CNT electrode compared to the pure CNF electrode (≈0.5 and  ≈1 Ω, respectively), which verifies enhanced conductivity of the composite [12]. Moreover, the verticality of the Warburg line, the slope of the 45° segment of the curve in the low and medium frequency regions, validates sufficient pore accessibility for electrolyte ion diffusion [31].

Zoom In Zoom Out Reset image size

The long term exploitation of electrode materials is validated through their ability to withstand a large number of charge–discharge cycles and retain capacitance. The CNF/CNT electrodes retained 96.6% of the initial capacitance after 2000 cycles (figure 6(B)), which is a very good stability for an energy storage device such as a supercapacitor, as it has to deliver the harvested energy through quick charging and discharging many times [32]. High capacitance retention indicates excellent stability of the composite electrode material and confirms strong adhesion of vapor-grown CNTs to CNFs.