Low-cost micro-supercapacitors using porous Ni/MnO2 entangled pillars and Na-based ionic liquids

The enhanced areal energy of three-dimensional (3D) micro-supercapacitors has made these miniaturized energy-storage components increasingly important at the dawn of the Internet of Things. Although ultrahigh-capacitances have been obtained with Ru-based pseudocapacitive materials, their substitution with abundant non-noble transition metals is a key requirement to reduce the price of electrochemical micro-storage systems and enable long-term sustainability. Here we report a cost-effective and industrially feasible approach to realize 3D micro-supercapacitors based on highly porous scaffolds of Ni/MnO 2 . These low-price electrodes exhibit a huge areal capacitance exceeding 4 F cm -2 and excellent cycling stability. In addition, extended cell voltages up to 2.6 V with areal energy of 1159 mJ cm -2 ( i.e. 0.3 mWh cm -2 ) and high power of 11.1 mW cm -2 were achieved using innovative Na-based ionogel electrolytes. We also show a novel micro-supercapacitor design based on entangled porous Ni/MnO 2 pillars, combining both energy and power ability on a small footprint area.


Introduction
The recent development of 3D electrodes based on pseudocapacitive materials [1][2][3][4] has greatly reduced the difference between the characteristics of micro-supercapacitors and micro-batteries for embedded systems and on-chip electronics.Small dimension electrochemical energy storage is especially crucial for the upcoming Internet of Things (IoT) based applications, which involve extensive utilization of miniature sensor nodes for health, environmental or industrial monitoring [5,6].Compared to microbatteries, micro-supercapacitors can provide high-power delivery and, more importantly, have a much longer operating lifetime.This comes, however, at the expense of a much lower energy density.
Pseudocapacitive micro-supercapacitors are a category of electrochemical capacitors whose charge storage mechanism relies on rapid and reversible faradaic surface reactions [7].When the pseudocapacitive material is deposited as a thin-layer on 3D architectures, high areal energy density is achieved within the limited available space of the electronic circuitry.Using this strategy, we recently reported unprecedented high capacitances per surface area using ruthenium dioxide RuO2 deposited on porous Au or Pt current collectors [8,9].These Ru-based micro-supercapacitors deliver an energy density per unit footprint area very close to that of lithium-based microbatteries, but with superior power and better cycling stability [10].
However, the high and volatile price of platinum and ruthenium, as well as the environmental issues related to the mining and refining of these rare and precious metals, have made these micro-devices relegated to niche applications.Replacing ruthenium with alternative transition metals with higher abundance and lower cost is, therefore, fundamental for commercial scaleup.
Among pseudocapacitive materials, manganese dioxide MnO2 stands out for its high theoretical specific capacitance, wide availability and environmental friendliness [11,12].Although its capacitance is lower than that of RuO2, manganese is the fifth most abundant metal in the Earth's crust making its price derisory compared to the overpriced ruthenium oxide in trace amount.However, MnO2 suffers from low conductivity, inefficient ion diffusion and a low structural stability, which results in low electrochemical utilization and poor cycling life [13].
We show in this study that the electrodeposition of MnO2 nanomaterials onto conductive and high surface area Ni foams overcome these issues.The porous metallic structure facilitates electrolyte penetration and enables efficient electronic and ionic pathways resulting in high electrode capacitance.Moreover, the open structure and the nanoscale deposition of MnO2 in the form of nanoflakes accommodate volume change and mitigate stress upon charge/discharge cycling over long periods of time.Using these inexpensive electrodes along with a costeffective and scalable manufacturing process, we obtain one of the highest cell capacitances for a micro-supercapacitor device.

Elaboration of the electrodes
A thin metallic sublayer (100/200 nm of Ti/Ni) was deposited by evaporation on an oxidized silicon substrate followed by DHBT electrodeposition of porous Ni as current collector with thickness ranging from 144 to 257 µm, using a constant current density of 2 A cm -2 in 0.1 M NiCl / 2 M NH4Cl solution.Pulsed electrodeposition of MnO2 was afterwards carried out on Ni DHBT in 0.14 M MnSO4 using a potentiostatic pulse of 1.25 V vs. SCE (saturated calomel electrode) for 2 s followed by a rest period of 3 s at open circuit potential.The ionic liquids and sodium salts used in this study were purchased from TCI Europe.The electrolytes were prepared in an argon-filled glovebox at room temperature.The electrodes were soaked in the electrolyte overnight prior to electrochemical characterizations.

Material characterizations
Scanning electron microscopy (SEM) was performed using a Hitachi S-4800 field-emission microscope.Transmission electron microscopy (TEM) imaging was carried out using a JEOL 2100F microscope equipped with a Schottky field emission gun (FEG) operated at 200 kV and a Gatan RIO16IS CMOS camera.The TEM sample was prepared using an epoxy resin for hardening before mechanical thinning (SiC polishing cloth and concave polishing using a diamond suspension) and final finishing by ion beam.The crystallographic structures were analyzed by grazing incidence X-ray diffraction (GI-XRD) measurements on a Bruker D8 advanced X-ray diffractometer with Cu Ka radiation (1.54184 Å) operating at 40 kV and 40 mA.The surface chemical composition of manganese oxide was estimated via X-ray photoelectron spectroscopy (XPS) using a ThermoScientific K-Alpha system operating with a monochromatic Al Ka X-ray source (1486.6 eV).The spectrometer energy was calibrated using the Au 4f7/2 (83.9 ± 0.1 eV) and Cu 2p3/2 (932.8 ± 0.1 eV) photoelectron lines.

Electrochemical characterizations
The electrochemical synthesis and characterizations were performed with a VMP-3 Biologic potentiostat connected to an external 10 A booster channel.The three-electrode system was made up of the working electrode, a Pt mesh counter electrode, a saturated calomel electrode (SCE) for synthesis and a KCl saturated Ag/AgCl reference electrode for characterizations.
Electrochemical impedance spectroscopy (EIS) measurements were carried out at open circuit potential and frequencies ranging from 100 kHz to 10 mHz.
The accessibility of charges has been investigated by calculating the voltammetric charge, q * , as a function of the sweep rate, v.The outer charge, qo, is obtained from the extrapolation of q * when v → ∞ from the q * = f(1/v ½ ) plot.The total charge, qt, is determined using the extrapolation of the inverse of the voltammetric charge, 1/q * , when v → 0 from the 1/q * =f(v ½ ) plot.

AEF factor calculation of Ni scaffolds
The Area Enlargement Factor (AEF) of the metallic current collector was defined as follows: The EASA of porous Ni DHBT coatings was calculated by integrating the charge associated to the b-NiOOH à b-Ni(OH)2 reduction peak using cyclic voltammetry at 5 mV s -1 in 1 M KOH, considering a cathodic charge value of 2.1 mC cm -2 as a reference [14,15].

Porous Ni/MnO2 electrodes
Highly porous Ni current collectors were synthesized by a facile and scalable electrodeposition process on oxidized Si wafers: the dynamic hydrogen bubble template (DHBT) method [16].
In this process, the electrodeposition of the metallic material from aqueous solutions is performed at high cathodic overpotentials where H + is reduced to H2.The generation of hydrogen gas bubbles acts as a dynamic template for the growth of highly porous metal films interspersed with nanostructured pore walls.This synthesis method is very simple and provide porous structures with good mechanical stability and strong adherence to the substrate.
Moreover, compared to other procedures based on stationary physical templates [17,18], DHBT process does not require any subsequent etching or template removal step, thus reducing postprocessing and associated difficulties or compatibility.
An Area Enlargement Factor (AEF), defined as the ratio between the developed area of the Ni porous structure over its projected area, can be deduced from cyclic voltammograms (CVs) in alkaline media (see Materials and Methods section).Fig. 1a shows the evolution of the CVs and corresponding calculated AEF of different porous Ni foams obtained with increasing DHBT deposition times.As expected, the AEF increases steadily with the deposition duration and hence the film thickness, from 342 cm 2 cm -2 to a colossal AEF value of about 1000 cm 2 cm -2 for 30 min of deposition.The formed metal foams are composed primarily of polycrystalline Ni as confirmed from their X-ray diffraction (XRD) pattern (Fig. 1b).The scanning electron micrographs (SEM) (Fig. 1b inset) show porous surface structure with different pore sizes, with diameters varying between 10 and 50 µm for the largest.Individual pores are constructed with converging nanoscale spicule-like structures, thus providing enormous surface area.
We then electrodeposited the MnO2 as active material using a pulsed current technique on the highly structured nickel current collector (Figure S1).From cross-sectional SEM observations (Figure S2), the electrodes display a pine tree type morphology with uniformly spread active material.High resolution transmission electron microscopy (TEM) imaging (Figure S3) and fast Fourier transform (FFT) analysis (Figure S4) show continuous and uniform deposition of ultra-small nanoflakes with ~77 nm thickness, with interplanar spacings characteristic of ramsdellite phase of the 1D allotropic group of MnO2 [19].The elemental mapping images (Fig. 2a) further illustrate the homogeneous distribution and coverage of Mn on the Ni metal support with unfilled open pores/space for electrolyte penetration.The composition of deposited MnO2 was also confirmed through X-ray photoelectron spectroscopy (XPS) studies by tracking the peak splitting difference of Mn-3s states at electrochemically reduced (0 V vs. Ag/AgCl) and oxidized (0.9 V vs. Ag/AgCl) forms (Fig. 2b).The exchange-splitting of the 3s core levels results in a binding energy separation directly related to the valence of manganese and its average oxidation state in the oxide, with a negative linear relationship [20].The energy splitting of the Mn-3s core level peaks is 4.9 eV for the as-deposited material, and 4.7 eV after oxidation, which is consistent with the occurrence of a MnO2 phase in a Mn(IV) oxidation state [21].This high Mn(IV) content is also in agreement with the quantitative determination computed from the Mn-2p3/2 and O-1s core level spectra (Figure S5).In its fully reduced state (Mn 3+ ), the binding energy difference of the Mn-3s core level peaks should be ~5.4 eV, but in the current case, the peak separation in the reduced state was ~5.1 eV, suggesting incomplete reduction at 0 V.These XPS results are nevertheless in accordance with the established pseudocapacitance reaction based on the change of oxidation state varying between Mn(IV) and Mn(III) with one-electron transfer in neutral aqueous medium [22].indicating small internal electrical resistance and energy loss.The cyclic voltammograms, recorded at different scan rates, show an almost ideal symmetrical rectangular shape (Fig. 2d), revealing easy and efficient access of electrons and cations to the pseudocapacitive 3D electrode to afford fast and highly reversible redox reactions.More importantly, the electrode exhibits a huge areal capacitance of 1.8 F cm -2 when cycled at 0.1 mV s -1 and 2.9 F cm -2 at 10 mHz using EIS (Fig. 2c), which is by far higher than all reported 2D thin-film micro-supercapacitor electrodes [5,[23][24][25].

Electrochemical characterizations of the porous
The accessibility of the electrolyte to the electroactive material has been further investigated by calculating the voltammetric charge, q * , as a function of the sweep rates, v, from 0.1 to 20 mV s -1 (Fig. 2e) according to a procedure developed by Trasatti and co-workers [26].This advanced electrochemical data analysis has been performed for Na + storage on different samples with increasing number of electrodeposited cycles of MnO2 (Fig. 2f).The difference between the outer capacitance, Co, related to the most accessible Mn 3+/4+ sites at the surface of the MnO2 deposits, and the total capacitance, Ct, gradually increases with deposition cycles indicating that accessibility of active surface area becomes progressively more and more difficult.Indeed, the reaction mechanism involves an adsorption/desorption process of Na + at the surface of MnO2 flakes as well as an insertion/extraction of the alkali ion into the bulk of the material.However, even for the thicker MnO2 coating obtained after 2500 cycles of deposition, the main contribution of the total capacitance Ct comes from the outer capacitance Co, which accounts for 69 % of the total capacitance, depicting the effectiveness of the porous matrix in enabling a large part of the nanostructured MnO2 film to be electrochemically active.
In order to verify whether the DHBT structure can provide a reliable electrical connection to the inherently poorly conductive MnO2, the long-term behavior of the electrode was tested by repeated galvanostatic charge/discharge cycles at high current of 4 mA cm -2 (Fig. 2g).The cycling experiment indicates excellent stability of the composite porous electrodes with more than 80 % retention of the initial capacitance after 10 000 cycles, and SEM images after cycling tests show that morphology of the MnO2 is preserved (Figure S6).This impressive stability of nanocrystalline MnO2 when extra-long measurements are performed can be explained by the conductive and porous structure of the metallic current collector as well as the beneficial effect of nanosizing on transport properties which govern the electrode reversibility [32].Moreover, unlike 2D thick-film electrodes, delamination issue is not a concern for these binder and additive-free structures.

Cell performances
Motivated by the excellent capacitive performance of porous Ni/MnO2 architectures in threeelectrode configuration, their integration and functioning in full-cell microdevices were subsequently examined.The realization of a 3D configuration at the device-level integrating nanostructured electrodes is an important milestone to further improve the surface-to-volume ratio and obtain the desired combination of energy, power, lifetime on small footprint areas [33].While the interdigitated planar configuration can shorten the diffusion length between the positive and negative electrode [34], a better areal capacitance and energy can be achieved for a micro-supercapacitor in stack configuration [5].We provide here an innovative framework to shown in Figure S10) to maximize their density per unit area (see Table S1 for example).
Important to note here is that this general scheme of construction can be easily extended to other class of substrates since the fabrication process is facile and needs thin metallic films, from which porous Ni foams can be grown through aqueous electrodeposition.
Moreover, the CV profiles did not show resistive behavior even at high scan rate of 50 mV s -1 .

MnO2 micro-supercapacitor based on Na-based ionic liquids
The use of aqueous electrolyte restricted the operating voltage range because of water decomposition and partial dissolution of MnO2 into soluble MnOOH species above ca. 1 volt.
The development of advanced solid-state electrolytes suitable for Na + charge storage with extended electrochemical stability window (ESW) is therefore an important challenge as energy stored is directly proportional to the square of the cell voltage.We have demonstrated in previous studies that a high pseudocapacitive effect of RuO2 with extended ESW could be obtained using protic ionic liquids [39,40].Herein, we have synthesized and tested various innovative electrolytes using Na salts diluted in aprotic pyrrolidinium-and imidazolium-based ionic liquids solvents for MnO2. Figure S11 shows the structure of the different ionic liquids explored and Figure S12 their electrochemical behaviour with Ni/MnO2 electrodes.Ionic liquids are considered safe alternatives to conventional aqueous and organic electrolytes due to their non-flammability, low volatility, high ionic conductivity and widened potential windows.
Moreover, ionic liquids can be easily turned into quasi-solid state by the use of ionogels [41,42] and address the packaging issues of micro-supercapacitors.
We tested the micro-device using 1 M sodium bis(fluorosulfonyl)imide (NaFSI) salt diluted in aprotic 1-methyl-1-propylpyrrolidinium bis(fluorosulfonyl)imide (Pyr13FSI) ionic liquid (Fig. 3d), where Na + ion can be expected to be the dominant contributor to faradaic pseudocapacitance [43] and greatly enhances charge storage [44,45].Ionic liquid based on FSI - anions are known to be chemically stable and to provide relative low viscosity [46,47], which often correlates with good ionic conduction.CVs were performed at 2 mV s -1 with increasing cell voltage windows.To our surprise, the micro-supercapacitor was able to cycle up to 2.6 V with no oxidation peaks that would be related to active material degradation.The voltammetric charge/discharge ratio of Figure S13 also reveals an appreciable level of stability over this voltage range.This wide extension of cell voltage could suggest that manganese oxide pseudocapacitance may rely on oxidation states above +IV using stable ionic liquids, with more than one-electron transfer, according to the following reaction: Mn n+ On/2 + (n-3)Na + + (n-3)e -↔ Na + (n-3)Mn 3+ On/2 where n ³ 4 is the maximum oxidation number of Mn.
A significant portion of the stored energy could also come from the electrochemical doublelayer capacitance (EDLC) of aprotic Pyr13FSI solvent with the highly porous structure.
Moreover, earlier studies on MnO2 supercapacitors using neat ionic liquids without added salts have shown that small anions can intercalate into the structure to support faradaic charge storage [50].Although this high voltage comes at the expense of a lower cell capacitance (343 mF cm -2 ), the specific energy reaches in this case an outstanding value of 1159 mJ cm -2 (i.e.0.3 mWh cm -2 ).
Fig. 3e and Figure S14 show the cycling performance of micro-supercapacitor prototypes up to 2000 cycles in various ionic liquid-based media.Using NaFSI/Pyr13FSI, a slight increase of capacitance is observed at the beginning of cycling ascribed to a gradual penetration of the electrolyte within the porosity, followed by a very high stability, pointing out the reliability of the microdevices on long cycles.The overall performances of the cells are finally compared with state-of-the-art reports on micro-supercapacitors in an area-normalized Ragone plot (Fig. 4 and Table S2) [17,23,37,[51][52][53][54].Our economical aqueous MnO2 microdevice rivals the best reported RuO2 micro-supercapacitor while the ionic liquid-based one outperforms it.

Conclusion
We have demonstrated the possibility of realizing low-cost micro-supercapacitor MnO2-based electrodes having ultra-long lifetimes and areal capacitances competing with the archetypal pseudocapacitive RuO2.Furthermore, these performances can be easily improved by increasing the depth of the 3D scaffold and tuning the MnOx thickness accordingly.We also lay down general scheme of microfabrication process to construct full encapsulated devices that could be easily extended to other substrates and geometric shapes/forms.The entire technological process is cost-effective, environmentally friendly, with all processing steps performed at room temperature, from the electrodeposition of Ni DHBT and MnO2 active material to wafer-level integration of the micro-devices.To enable extended operational potential window, we have screened various ionic liquid-based media and show prototype cells with optimized electrolyte having superlative performance.Each of the sub-components of the final device is tailored to be economically feasible for scale up, thus opening avenues for successful commercialization.) Areal power density (mW cm -2 ) 3D MnO 2 (EMIM-TFSI) 17  Activated C (NET 4 BF 4 / PC) 51   CDC (in H 2 SO 4 ) 23   Carbon Onions (NET 4 BF 4 /PC) 51  Graphene (in H 2 SO 4 ) 52   VN (in KOH) 53   3D PEDOT (LiCl/PVA) 37      The size, shape, height and density of the pillars on a given surface can be easily adjusted to optimize 3D Gain, defined as the ratio between the exposed surface including all the pillars, S3D, and the exposed surface without pillars, S2D.Table S1 shows some examples of pillar topology on a S2D = 4 mm 2 surface.The comparison between NaFSI and NaTFSI was to evaluate if the transport of ions (and viscosity) has an impact on the charge storage properties.FSI -anions provides higher conductivities from the better mobility of the smaller ion.Regarding the EMIM-FSI and Pyr13-FSI solvents, they were chosen for their relatively low viscosity (25 and 53 cP, respectively).EMIM + cation has better transport properties but is less stable at lower potential values because of the acidity of the proton at the C2 position (the carbon between both N on the cycle).Pyr13 + cation has been applied with more success as battery electrolyte † since the absence of acidic protons and of conjugation on the cycle decreases the reactivity.† X. Wang No significant difference is observed between the different investigated ionic liquids on the performances (power rating and cell voltage) of the MnO2 micro-supercapacitor.The limitations seem therefore to come from interfacial reactions and/or ion transport within the active material, rather than ion transport within the electrolyte itself.Moreover, the potential limits do not appear to be too stringent for EMIM + cation (i.e. the negative limit is not negative enough for EMIM + to react) for a difference to be observed with the Pyr13 + .
Ni/MnO2 electrode (25 min of Ni DHBT and 1500 pulses of MnO2 electrodeposition) were performed in a three-electrode configuration in a nitrogen-purged 1 M Na2SO4 aqueous electrolyte.Fig. 2c shows the obtained electrochemical impedance spectroscopy (EIS) measurements along with the simulated fit using a classical equivalent circuit model to distinguish pseudocapacitance from double-layer capacitance.The composite electrode is characterized by a low equivalent series resistance (ESR) of 7.0 W cm 2 realize three-dimensional micro-supercapacitors consisting of entangled pillars of porous Ni/MnO2, as schematically shown in Fig.3a,b.The microfabrication steps are detailed in FigureS7and assembling details in FigureS8.The principle is based on the electrodeposition of Ni pillars inside photoresist molds, followed by DHBT electrodeposition of porous Ni and electrodeposition of MnO2.This new concept of 3D device design combines both the advantages of the interdigitated and the ones of the stack configuration, with reduced interelectrode distance and increased active material loading over limited surface area.This arranged matrix of 3D porous electrodes also facilitates better electrolyte percolation and higher electronic conductivity, thus resulting in superior interfacial kinetics.The entire process of fabrication is easily scalable (FigureS9) thanks to the simplicity of the DHBT template, and can be used with pillars of different geometries (square, hexagonal, parallelogram or spinner as

Fig. 1 .Figure 2 Fig. 2 .
Fig. 1.Characterizations of Ni porous current collectors.a) Evolution of the Area Enlargement Factor (AEF), deduced from CVs performed at 5 mV s -1 in 1 M KOH, with DHBT deposition time.Large AEF up to ca. 1000 cm 2 cm -2 were obtained.b) XRD pattern showing the diffraction

Figure S1 .
Figure S1.Potentiostatic pulsed electrodeposition of MnO2 on porous Ni current collectors.Each cycle consists of a potentiostatic pulse of 1.25 vs. SCE for 2 s followed by a rest period of 3 s at open circuit to stabilize the porous/electrolyte interface.

Figure S6 .
Figure S6.SEM image of a Ni DHBT/MnO2 electrode after long cycling tests.a) Low magnification.b) High magnification showing the preservation of the MnO2 nanoflakes.

Figure S7 .
Figure S7.Microfabrication of micro-supercapacitors based on 3D entangled pillars: topview and cross-section schematics of the different processing steps.a) Deposition and patterning using conventional photolithography techniques of a thick WBR photoresist on an oxidized silicon wafer coated with a thin Ti/Ni sublayer.b) Electrodeposition of Ni inside photoresist molds to obtain Ni pillars after WBR stripping.c) Design of square electrode area using a 400 µm-thick WBR photoresist.d) Uniform electrodeposition of porous Ni using the DHBT technique on the exposed surface (flat surface and pillars).e) Pulsed electrodeposition of MnO2 on DHBT Ni. f) Protection of electrode area and electrical contact pads by a sacrificial WBR photoresist.g) Wet etching of the exposed Ni flat surface.h) Application of a doublesided self-adhesive tape (Tesa ® chemical resistant filmic tape) on the first electrode to form a cavity.i) Filling of the cavity with electrolyte and final assembly of the two electrodes by controlled adjustment of the interspacing.

Figure S8 .
Figure S8.Assembling details.The FC150 -FlipChip Bonder equipment was used to assemble the interdigitated device with a precision of ± 1 µm by following the steps below:

Figure S9 .
Figure S9.Electrochemical cell dedicated for 4-inch substrates for large-scale electrolytic deposition of the porous current collector and active material.a) Front view.b) Side view.c) Immersed in the deposition bath.The custom-made cell was designed in collaboration with Yamamoto-MS.

Figure S11 .
Figure S11.Structures of the cation and anion groups of ionic liquid candidates explored for MnO2 micro-supercapacitors.a) Na salts.b) Ionic liquids solvents.

Figure S13 .
Figure S13.Voltammetric charge/discharge ratio Qc/Qd with increasing cell voltage in 1 M NaFSI /Pyr13FSI computed from Fig. 3d.The charge/discharge ratio is constant from 1 to 2.6 V, indicating appreciable level of stability of the micro-supercapacitor within this voltage range ‡, §