Szymon
Sollami Delekta
a,
Karin H.
Adolfsson
b,
Nejla
Benyahia Erdal
b,
Minna
Hakkarainen
b,
Mikael
Östling
a and
Jiantong
Li
*a
aKTH Royal Institute of Technology, School of Electrical Engineering and Computer Science, Division of Electronics, Electrum 229, SE-164 40 Kista, Sweden. E-mail: jiantong@kth.se
bKTH Royal Institute of Technology, Department of Fibre and Polymer Technology, Teknikringen 58, SE 11428 Stockholm, Sweden
First published on 2nd May 2019
The advance of miniaturized and low-power electronics has a striking impact on the development of energy storage devices with constantly tougher constraints in terms of form factor and performance. Microsupercapacitors (MSCs) are considered a potential solution to this problem, thanks to their compact device structure. Great efforts have been made to maximize their performance with new materials like graphene and to minimize their production cost with scalable fabrication processes. In this regard, we developed a full inkjet printing process for the production of all-graphene microsupercapacitors with electrodes based on electrochemically exfoliated graphene and an ultrathin solid-state electrolyte based on nano-graphene oxide. The devices exploit the high ionic conductivity of nano-graphene oxide coupled with the high electrical conductivity of graphene films, yielding areal capacitances of up to 313 μF cm−2 at 5 mV s−1 and high power densities of up to ∼4 mW cm−3 with an overall device thickness of only ∼1 μm.
In this work, we present a full inkjet printing process for the fabrication of all-graphene microsupercapacitors with an ultrathin solid-state electrolyte based on nano-graphene oxide (nGO), a form of graphene oxide with lateral dimensions at the nanoscale.24 Moreover, thanks to its excellent dispersibility resulting from small particle size, nGO is remarkably suitable for inkjet printing, enabling accurate thickness control of the printed thin film. As a result, the volumetric performance can be maximized by minimizing the thickness and volume of the electrolyte, producing ultrathin (overall device thickness ∼1 μm) devices with high energy and power densities.
The all-graphene MSCs consist of interdigitated electrodes based on electrochemically exfoliated graphene (EG) and a nGO-based electrolyte. The ink formulation process and the MSC fabrication are described in Experimental methods, ESI, and in the schematic in Fig. S1.† Briefly, to fabricate the electrodes, the EG inks were formulated as previously reported21 and printed on flexible Kapton substrates to form the interdigitated electrodes, followed by a brief annealing step to remove the binder (ethyl cellulose). Next, the electrolyte was produced by inkjet printing the nGO inks. The nanodispersions of nGO were derived from cellulose through a microwave-assisted process as reported previously,25,26 exhibiting a C/O ratio of ∼2.1 (see X-ray photoelectron spectroscopy analysis, ESI, and Fig. S3†). Because the so-obtained nGO has a small particle size and the oxygen-containing groups (C–OH, C–O, COO–H and CO) are located on the edges and surface of the nanosheets,27 it has a very high dispersibility and colloidal stability in a range of solvents, especially water.28,29 As a result, we were able to obtain stable inks containing highly concentrated nGO (20 mg ml−1) in a mixture of deionized water, ethylene glycol and concentrated phosphoric acid (H3PO4). The inks are stable (without any sedimentation) for at least several months. Their stability was also confirmed by zeta potential measurements where, even after the addition of ethylene glycol and H3PO4, the nGO exhibited negative charge which increases the colloidal stability by preventing flake coagulation (Fig. S4†).30 Thanks to their colloidal stability, the nGO inks exhibited ideal jetting performance (Fig. S5†) which enabled reliable inkjet printing at a resolution down to around 50 μm. The deposited nGO thin film exhibits high ionic conductivity due to the water molecules adsorbing onto its surface30–32 (Fig. S6a†) and becomes solid-state after drying, indicating that the devices do not suffer from leakage. By adding H3PO4 into the material, we were able to increase its ionic conductivity to ∼3 mS cm−1 (Fig. S6b†).
The inkjet printing technique enables the fabrication of devices with arbitrary geometry and thickness. The devices were printed with a finger length of 1.9 mm, a finger width of 550 μm, an 8 finger and inter-finger gap of 350 μm, with a geometrical area of 0.16 cm2 (excluding finger gaps) and a footprint area (area covered by the electrolyte) of 0.24 cm2 (Fig. 1a). Scanning electron microscopy (SEM) was used to investigate the morphology of the printed films as shown in Fig. 1b and c. The surface of the bare graphene electrode (Fig. 1b) exhibits the sharp protruding edges of the graphene flakes, a typical trait of printed graphene films which contributes to maximizing their surface area. On the other hand, the printed nGO (Fig. 1c) appears as a uniform and conformal coating fully covering the underlying graphene layer. Even the edges of the graphene flakes seem to be covered by nGO, indicating a large contact area between the electrode and electrolyte. The nGO can be observed in the TEM image (Fig. 1d), where the nanoparticles appear to be mostly spherical with an average diameter of 60 nm although somewhat agglomerated. To investigate the influence of the volume on the device performance, the thicknesses of both the electrode and electrolyte were varied by controlling the number of printing passes of the inkjet printed layers. The devices were named according to the number of printing passes, so e.g. the EG20L/nGO20L device consists of 20 printing passes of EG and 20 printing passes of nGO. The corresponding thicknesses of the electrodes were found to be about 30, 50, and 100 nm for 5L, 10L and 20L, respectively, while the electrolyte thickness was around 0.5, 1.0, 1.3 and 1.6 μm for 5L, 10L, 20L and 30L, respectively (Fig. S7†). We could observe that for both EG and GO patterns, the thickness increases almost linearly with the number of printing passes (Fig. 2a) and that the profiles of the printed films are uniform with negligible coffee-ring effects, which causes non-uniform thickness between the center and perimeter of the printed patterns.33 Cyclic voltammetry (CV), galvanostatic charge–discharge (GCD) and electrochemical impedance spectroscopy (EIS) were performed to characterize the performance of the devices. Fig. 2b shows the CV curves of the EG20L/nGO20L device at scan rates of up to 50 mV s−1. The device exhibits rectangular CV curves indicating ideal electric double layer capacitive behavior. As shown in Fig. S8†, this behavior continues at higher scan rates albeit with elongated CV curves caused, as expected, by a less efficient ion migration to the electrodes. The thickness of both the electrolyte (Fig. 2c) and electrode (Fig. 2d) plays a significant role in the CV performance of the devices and a summary of the extracted areal capacitances can be found in Fig. 2e and f, respectively. Although the device with the thinnest electrolyte (5L) showed a slightly resistive and lens-shaped CV curve with an areal capacitance (CA) of around 200 μF cm−2 (at 5 mV s−1), the behavior improves significantly for thicker nGO films. It is found that the optimal amount of the printing passes of the electrolyte is 20L (CA = 313 μF cm−2 at 5 mV s−1) because thicker films yield similar capacitances within the studied scan rate range, indicating an overabundance of the electrolyte. The observations from the CV curves are confirmed by the results of the GCD measurements. Fig. 3a and b show the GCD performance of the EG20L/nGO20L device at various current densities, exhibiting a triangular and symmetric response to charging and discharging, respectively. This confirms that the energy storage mechanism is based on the electrical double layer capacitance with an ideal capacitive behavior. The GCD curves of the devices (Fig. 3c and d) yield areal capacitances (Fig. 3e and f) in agreement with the CV measurements. Finally, we also performed GCD cycling for 11000 cycles on the EG20L/nGO20L device, showing a good capacitance retention of >65% (Fig. S9a†), although some fluctuation can be observed after the 6000th cycle due to slight variations of humidity.
To confirm our previous remarks on the dependence of electrolyte thickness on ionic conductivity, we performed EIS measurements on all of the devices (Fig. S9b†) and measured their equivalent series resistance (ESR) corresponding to the intercept of the semicircle from EIS. The results confirm that although the ionic conductance generally increases with the electrolyte thickness, the increase is not linear for thicker films and 30 printing passes of electrolyte end up with similar ESRs to 20 passes. Fig. 4a summarizes the areal energy and power densities of our devices, showing that the EG20L/nGO20L and EG20L/nGO30L devices have the highest performance, especially when compared to a reference device with EG20L as the electrode and drop-cast polyvinyl alcohol (PVA)/H3PO4 as the electrolyte.
Fig. 4 Ragone plots of microsupercapacitors (MSCs). (a) Areal energy and power densities of our MSCs. (b) Volumetric energy and power densities of our MSCs, showing the difference in performance depending on whether just the volume of the electrodes is considered or the full volume of the device. These values are also reported in Table S1.† (c) Volumetric performance comparison between our best MSC (green line, circular markers) and recently reported MSCs in the literature7,8,21,36–38 with their electrode material and electrolyte being shown. In all plots, hollow markers denote that the whole device volume was accounted for when extracting the metrics, while filled markers refer to metrics extracted from the electrode volume only. |
Volumetric figures-of-merit are often overlooked but essential for reporting the true performance of SCs in a reliable way,34 especially when envisioning the device for practical applications. In this respect, the footprint of all the components of the device should be considered (e.g. electrodes, electrolyte, separator, current collectors and packaging) as they significantly affect the overall device performance.35 In the case of microdevices, the weight of a thin film of the electrode material is negligible and their gravimetric performances do not scale up linearly with the thickness of the electrode, making these metrics potentially misleading.10 Volumetric energy and power densities are generally less vulnerable to these uncertainties, but they also depend on what component is included, as can also be seen in the case of our devices. We illustrate how the volumetric performance depends on the device components in Fig. 4b. The hollow markers in the Ragone plot refer to the power (Pd,full) and energy densities (Ed,full) of the full volume of the device (including the volume of the electrolyte) and filled markers only include the volume of the two electrodes (Ed,el and Pd,el – excluding finger gaps). It can be seen that the EG20L/nGO20L MSC exhibits an Ed,full value of ∼0.2 mW h cm−3 with a corresponding Pd,full value of ∼4 mW cm−3. Both of these figures increase by a factor of more than 10 when just the volume of the electrodes is considered, a trend which is common in all of our devices. For comparison, we also studied a MSC with a typical drop-cast PVA/H3PO4 electrolyte. Because drop-casting yields poor control on the uniformity and thickness of the electrolyte layers, the PVA-based electrolyte exhibits a thickness higher than 20 μm. However, the corresponding device can barely attain comparable Ed,el and Pd,el to our nGO-based devices, whereas Ed,full and Pd,full are significantly (more than one order of magnitude) lower (Fig. 4b), indicating that our fully printed solution is the most space efficient. The volumetric performances of our devices are in line with MSCs reported recently in the literature (Fig. 4c), indicating that our technique has the potential for fabricating high-performance energy storage devices.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9nr01427f |
This journal is © The Royal Society of Chemistry 2019 |