Green Precursors and Soft Templating for Printing Porous Carbon‐Based Micro‐supercapacitors

Abstract A combination of soft lithographic printing and soft templating has been used to fabricate high‐resolution interdigitated micro‐supercapacitors (MSC). Surfactant‐assisted self‐assembly produces high surface area ordered mesoporous carbons (490 m2 g−1). For the first time, such precursors have been printed by nano‐imprint lithography as microdevices with a line width of only 250 nm and a spacing of only 1 μm. The devices are crack‐free with low specific resistance (1.2×10−5 Ωm) and show good device capacitance up to 0.21 F cm−3.


Introduction
An increasing demandf or flexible and portable energy-storage systemsp owering micro-electromechanicals ystems (MEMS), implantable medicald evices, and the internet of things requires extensive developments in on-chip energy storage systems. [1,2] In the field of rechargeable micro-power devices micro-batteries are being extensively investigated. [3] Their lack of cycling stabilitya nd low power densities can be overcome by using electrochemical double-layerc apacitors (EDLC). [4][5][6] Their energy-storage capacity is based on the electrostatic adsorptiono fe lectrolyte ions on the inner surfaceo fh ighly porousc arbons. This physical ion-storage mechanismw ithout faradaic reactions enables fast charge-discharge rates and remarkable cycling stabilities over millions of cycles. Hence, systems providing high power densities are commercialized. [7][8][9] Ak ey feature of micro-supercapacitors is their two-dimensional geometrical design, which is based on interdigitated structures integrated on af lats urface, rendering such systems ideal for microelectronic device integration. These structures enable an increase in surfacea rea and foster ag ood electrode accessibility and shorter diffusion pathways. With the fixed electrode setup, the use of separators may be dispensed with; this further facilitates the ion transport. [2,4] Direct structuring and deposition methods enable binder-free electrodes, further improving the performance. [1,10] Solvent-assisted nanoimprint lithography (SA-NIL) is an elegant method to directly create an interdigitated pattern on a flat surface. Micro-and nanostructured electrodes can be fabricated by using this low-cost, high-throughput method. [11] NIL typicallya chieves higher resolution (down to nanoscale) than simple inkjet printing or laser micro-structuring.M oreover,t he liquid precursor will not cause particlec ontamination, as subtractive laser structuring is accompanied by massive dusting causing shunts.
Soft lithography is based on patterning ap recursor with an elastomeric stamp. [12,13] Scalability has been demonstrated by integration into roll-to-roll processes. [14,15] In the case of SA-NIL, al iquid and curable precursor is customizedf or printing, hardening and thermal transformation into crack-free line patterns consisting of ah igh surface area material with defined porosity.
In the field of MSCs, aw ide range of materials and various structuring approaches have been investigatedi nr ecent years. Appropriate electrode materials supplyingh ighly active surface areas and good electrical conductivities are activated carbons, [16][17][18][19] conducting polymers, [20] MXenes, [21,22] or metal oxides( RuO 2 ,M nO 2 ). [23] The use of various composite materials is also widely established. [24] In the area of porousc arbons several materials such as carbide-derived carbon (CDC), [25][26][27][28] graphene, [29,30] or CNTs [31][32][33] with high specific surfacea reas (SSAs; > 1000 m 2 g À1 )h ave been investigated for MSCsi nt he past. [34] Customized liquid carbon precursors such as phenolic resins made of phenol or resorcinol in combination with formaldehyde fulfill the demands for SA-NIL. [35][36][37] However,c arcinogenicity and low environmental compatibilitya re disadvantageous for sustainable production and safe workspaces. From this pointo fv iew,G himbeu et al. have presented promising precursors based on so called "green resoles" that contain phloroglucinol and glyoxylic acid as monomers and react in a catalyst-free polymerization. [38] In the following, we report the development and excellent performance of "green resoles"c ombined with soft templating for the SA-NILp rocesses by using an UV-assisted evaporation induced self-assembly (UV-EISA) for micelle assembly that enables one-step curing of the precursor during the NIL-process.

Results and Discussion
The SA-NIL process is ideal to produce in-plane micro-supercapacitors with interdigitated geometry. [12] Precursor development is crucial to enhancet he capacitance or tailor the frequencyresponse. The precursor needs to fulfill alist of requirements with respectt op atterning, resolution, structurals tability,and the resulting porous carbon electrical conductivity,pore size distribution and surfacea rea. The requirements for the application of al iquid precursor in the SA-NIL-process are, that the precursor completely cures under thermalo rU Vt reatment and resultsi nh omogeneous structures with good dimensional stability. In this work we customize the green carbonp recursor based on phloroglucinol and glyoxylic acid. The phenolic resin is suitable for the pattering process because of its fast polymerization during temperature and UV treatment. Furthermore, the addition of as urfactant like Pluronic F127 enables soft templating in order to induce ad efined porosity. [39] In the following we compareanon-templated and as oft templated carbon precursor for SA-NILa pplications ( Figure 1). The precursors are printed in twod ifferent interdigitated patterns with line widthso f250 (IDE250) and5 00 nm (IDE500).

Carbonprecursor and thinfilm characteristics
The pore size and surface properties of the material were estimated using carbon powders from the resol precursor (carbonized at 900 8C) via nitrogen physisorption measurements at 77 K. The non-templatedc arbon has no appreciable SSA (Figure 2g). With Pluronic F127 as surfactant am esoporousc arbon materialw ith aS SA of 491 m 2 g À1 and ap ore volume of 0.5 cm 3 g À1 was produced. The carbon shows at ype IV(a) isotherm characteristicf or mesoporouss ystems. [40] The pore-size distributions were calculated using quenched solid density functional theory for slit-like pore geometry (Figure 2g). With Pluronic F127, mainly mesopores with as ize of around 10 nm were formed. The surfactant generated ordered micelles in the resol. [38] The TEM images showa no rdered hexagonal pore structure in the carbon powder ( Figure 2a). Using SAXS measurements the ordered pore structure also was confirmed (Supporting Information S.1.). In order to obtain more detailedi nformation of the precursor characteristics at micro-and nanoscales, carbon thin films were formedb ys pin coating. The polymerized resin was either formed using temperature treatment (non-templated) or by UV-EISA (templated). The additional UV treatment supports the self-assembly of surfactant and precursor molecules and enables ar apid hardening,r equired for the NIL-process.
Te mplated thin films ( 2 mm, 900 8C) were characterized by TEM measurements after scraping off sections from the substrate (Figure 2c-f) confirming the ordered hexagonal pore structure in the EISA-generated films. Interestingly, within the films ordered pores with different structureso ro rientations occur,s or egions with pores arranged verticallya nd horizontally are found. We conclude that the templating also works at microscale. The electrochemical properties of the carbon films were investigated with respect to carbonization temperatures varying from7 00 to 1000 8C. The specific resistances are studied via four-point measurements on carbon thin films (Table 1). Typically,e lectrical conductivity is expected to increasew ith increasing carbonization temperatures due to ah igherg raphitization, however in af ilm, macroscopic cracks may cause a growing resistance. [41] We measured the lowest electrical resistance for the 900 8Cc arbonized sample for both carbon materials. At 1000 8Cc racking in the film starts to affect the films. Overall the templated sample showst he lowest resistance with 1.2 10 À5 Wmt he surfactant additionally stabilizes the precursor and prevents crystallization and cracking.

Nanoimprintl ithography
The basic requirements for interdigitated electrodes are crackfree electrode fingers which are perfectly separatedf rom each other.T herefore the lines must have an adequate and homogeneoush eight to avoid dimension-variation related resistance fluctuations. Furthermore, short spacings in between the fingers facilitate the ion transport. [30,42] Finally,c arbonization should lead to high surface area carbons with controlled pore size in the given structure on the substrate. The nanoimprint lithographyi si deally suited to produce patterned surfaces at micro-and nanoscale. [43] In the following, we show the printing of interdigitated micro-structures with line widthsd own to 250 nm. The line widths and spacings of the structures are shown in Ta ble 2.
The "green resol" precursor is excellentf or the application in the SA-NIL processb ecause of its adjustable viscosity and the moderate curing conditions. The used solvente thanoli sc onvenientf or the NIL process because of low evaporation temperaturesa nd the inertness of the polydimethylsiloxane (PDMS)s tamps against it. The ethanol easily diffuses through the membrane withouti nducings welling of the PDMS. For the pure resin, at emperature treatment at 120 8Ci sn ecessary for the complete curing of the precursor duringt he NIL-process. Using this precursor,h omogeneous and defect-free electrodes with negligiblep recursor residue in between the lines were printed (Figure 3). The used PDMS stamps have channels with ad epth of 500 nm. The amount of solvent in the precursor is 84 wt/% which is completely removedd uring printing. This fact causes am assive shrinkage of the polymer and lines after the NIL process. After printing IDE500 remains with line heights of around2 00 nm andI DE250 with 120 nm. After carbonization stable interdigitated structuresw ere formed with   typical line heights of 100 (IDE500) and 60 nm (IDE250), thus indicating avolumereduction of 50 %. The surfactant in the precursor supports the formation of homogeneous and defect free lines. For the smalli nterdigitated structure IDE250 the amount of Pluronic F127 was reduced in order to form completely separated electrodes without residual films in between. Possible residues shrink and vanish during pyrolysis and will not shortcut the electrodes. For IDE250 line heights up to 100 nm and for IDE500 250 nm were reached. Nevertheless,t he NIL process wass atisfying for this system.I nterdigitated structures with completely contacted and homogeneous fingers were printed. Also, after pyrolysis at 900 8Ct he structuress tayed defect-free with moderate line heights fort he envisioned electrochemical applications.T he non-templated IDE500 structures resulted in line heights of 100 nm. Compared to those,t he templated IDE500 structures again showedadrastic shrinkage, because of the surfactant removal. The lines remained smaller than the non-templated lines (50 nm). Due to the smallc arbon volumea nd mass it is difficult to investigate the porosity and thea rrangemento f pores in the interdigitated structures. The lines in IDE250 were between 30 and5 0nmi nh eight. All generated patterns showedp romisingc haracteristics for further micro-supercapacitor applications.

Electrochemical characterization
The basic electrochemical characteristicso ft he carbon materials were investigated in thin filmE DLCs. Thereforf reestanding carbon films (thickness = 2 mm) were attached with the PVA/ H 2 SO 4 -hydrogele lectrolyte and ap olypropylene separator.T he non-porous (non-templated) carbon material reached as pecific areal capacitance of 5.4 mF cm À2 (Supporting Information S.2.). Compared to that, the soft templated carbon film reached a slightly higherc apacitance (6.0 mF cm À2 ). Depending on the orientation of the hexagonal pores the accessibility fort he electrolyte ions varies. As ar esult the actual capacitance was smaller than expected based on SSA data.
The carbonized interdigitated structuresw ere firstly contacted with at hin gold layer at the outer contacts. After isolation with aP MMA ring the structures were activated in ac old Ar plasma in order to form hydrophilic surface groups.A fter that at hin PVA/H 2 SO 4 film was attached. The electrochemical characterization by cyclic voltammetry,g alvanostatic charge-discharge( GCD) and impedance spectroscopy is shown in Figure4(Supporting Information S.3.). CV-curves of the porous carbon micro-supercapacitors show the typical rectangular shape even at higher scan rates. The calculated capacitances are listed in Table 3.
The Nyquist plots of the MSCs display the expected shape for capacitive behavior (Figure 4c). An equivalents eries resist- ance (ESR) of around 15 kW was observedt hat was mainly influenced by the small line width in the interdigitated structure. Typically,t he capacitance increases with decreasing electrode spacings. [44] In our case we observed the reverse trend. For the IDE500 higher device capacitances were reached. As ignificant influenceo nt he performance of the MSC originates from the line width relating to the electrode resistance. With smaller lines also the overall line heightd ecreases and higher resistance values effectively reduce the performance. Smaller lines also increaset he possibility of small cracks in the electrode fingers. Nevertheless also for the small IDE250 structure promising capacitances were reached.
Furthermore astark influence of the surfactant in the precursor is evident. Both structures showedacapacitance increase with additional soft templating, indicating, that even in these small structures the surfacea rea can be enhanced. For the larger structure the increasew as higher than for the smaller structures. In the IDE250 the higher probability of pore collapse is disadvantageous. Regarding shorterd iffusion length the IDE250 had as ignificant advantage. The smaller structure shows abetter rate capability than the IDE500 (Figure 5a).
The cycling stability of the MSCs was shown exemplarily for the IDE250 structure. The MSC was charged and discharged over 10 000 cycles demonstratinga ne xcellentc ycling stability over the numerous cycles with ac apacitance retention of 120 %f or the template and 180 %f or the non-templated MSC. The slighti ncrease originates from improved wetting by the hydrophilic electrolyte on the carbon surfaceo ver the first cycles.
The leakage current of these micro-supercapacitors was approximately 10 À3 mAf or the templated and around1 0 À4 mAf or the non-templated samples (Supporting Information S.4.). As already mentioned, the surfactant also supports the formation of homogeneous structures. During the printing process the self-assembly occurs within af ew seconds.T herefore, residues could remaini nb etween the electrode fingers which were not completely displaced by the stamp. These residual shunts cause the increased self-discharge.I nF igure 4c energy and power density of the different MSCs were compared in a Ragone plot. In correlation witht he calculated capacitances, higher energyd ensities were reached by the templated MSCs in comparison to the non-templated ones. Also, the power densities were increased for the templateds ystems. The power density was influencedb yt wo important electrode properties. Firstly,t he serial electrode resistance was ac rucial factor.A s discussed before, larger lines decrease the resistance enabling higher power and current density.A tt he same time the length of the diffusion pathways is important. Faster diffusion also  generates enhanced power capability.Based on these compensating effectst he powerd ensities of the templateds tructures cover as imilar range. Corresponding to these measurements relaxation time constants were calculated (Supporting Information S.5.). The IDE500 shows the best relaxation time constants of 50 (non-templated) and 70 ms (templated). In comparison with other solid-state carbon based MSCs for the presented cells the capacitances and energy densities are in the same ranges (Supporting Information S.6.) despite of the much smaller lines and higher resolutions.I nt erms of powerd ensities and chargingt imes, the high electrode resistance caused by the small interdigitated fingersl imits the performance. An additional heteroatom doping or geometry optimizationc ould furtherimprovethe performance.

Conclusion
As oft templated "green resol" based precursor system was successfully used as an ink in the SA-NIL-process. High resolution interdigitated patterns down to 250 nm were printed and transferred into stable highly porous carbon electrodes. For the first time, mesoporous carbon electrodes were processed at such tiny line width for aq uasi-solid-state micro-supercapacitor using PVA/H 2 SO 4 -hydrogele lectrolytes. The devices show excellent supercapacitor characteristics with cycling stabilities over 10 000 cycles. The solvent assisted nanoimprint lithography was performed with the Micro-Contact-Printing System m-CP 3.0 from GeSiM mbH and patterned PDMS stamps (Sylgard 184 elastomer kit, Dow Chemicals). The process of stamp preparation and molding is published elsewhere. [45] For the printing process ad roplet (4 mL) of the precursor solution was deposited on the substrate and later the stamp was pressed into it. Thereby the solution was displaced into the spaces of the structured stamp. The substrate was then heated to 120 8Cf or 15 min in order to induce the polymerization and evaporate the solvent. The precursor containing surfactant was treated at 80 8Cf or 30 min with an additional exposure of UV light (Delolux-04, Delo, 8mWcm À2 )t hrough the stamp membrane. After treatment the stamp was peeled off and the precursor structure remains. Interdigitated structures with line width of 500 (IDE500) and 250 nm (IDE250) were printed.

Experimental Section
TEM measurements were carried out using aJ EM 1400plus Microscope (120 kV). The samples were applied on ac opper grid using a suspension of the carbon material in acetone. For the measurement of thin films, they were scraped off from the substrate. AFM measurements were performed via aD imension 3000 (Digital Instruments) in tapping mode. Small-angle X-ray diffraction (SAXS) experiments were carried out on aB ruker Nanostar diffractometer in transmission mode with Cu Ka1 radiation (0.15405 nm) coupled with ap osition sensitive HiStar detector.N 2 -Physisorption measurements were carried out in aQ uadrasorp EVO/SI (3P instruments) at 77 K. Previously the samples were degassed at 150 8Cf or 12 h. The SSAs of the materials were calculated using the multipoint BET method. The pore size distributions were calculated assuming slitlike pore geometry using quenched solid density functional theory (QSDFT) method.
Carbonization:C arbon materials were produced in three different ways:1 )Carbon films and 2) carbon interdigitated structures on substrates, furthermore, 3) carbon powders were formed for further characterization. All carbonizations were carried out at 900 8Cf or 2h with ah eating rate of 150 Kh À1 under argon. For powder synthesis the precursor was hardened by treatment with UV for 20 min.
Preparation of micro-supercapacitors:Apolymer hydrogel based on polyvinyl alcohol (PVA; Merck;m olecular weight 145 000) and sulfuric acid was used as electrolyte. Therefor 0.5 go fP VA was dissolved in 7mLo fd eionized water at 90 8Cwith vigorous stirring. In the next step 0.5 go fconc. sulfuric acid was added to the solution.
With that, the electrolyte was ready for coating and drying at room temperature.
For the preparation of the symmetric full cell setup, free-standing carbon films with at hickness about 1 mmw ere produced. The films can be detached from the substrates by soaking in acetone. After drying at 70 8Ct he films were placed on two titanium pistons (12 mm diameter) with electro-DAG. As for the structured micro-supercapacitors the carbon surface was activated with Ar plasma and the hydrogel electrolyte was deposited on the electrodes. The electrodes were assembled with apolypropylene separator in aSwagelok setup.
The carbonized interdigitated structures were cleaned from extra carbon residuals. The electrode area was masked with silicone during the deposition of ac hrome (10 nm) and gold (100 nm) current collector by physical vapor deposition (B39, Malz &S chmidt) with ad eposition rate around 15-20 k s À1 .A fter that, the mask was removed and ar ing of PMMA was deposited around the interdigitated area as an isolation of the current collector.M oreover, the ring serves as ar eservoir for the electrolyte. Finally,t he electrode surface was activated using an Ar plasma treatment (KINpen, neoplas tools) in order to generate am ore hydrophilic surface and 10 mLo ft he hydrogel electrolyte were added. After evaporating water excess at room temperature, the quasi-solid-state electrolyte remains.
Electrochemical testing:A ll cells were tested in aV MP3 Potentiostat from Bio-Logic and cyclic voltammetry (CV) electrical impedance spectroscopy (EIS) and galvanostatic cycling were performed. CVs were recorded with the hydrogel electrolyte PVA/ H 2 SO 4 as electrolyte between 5a nd 100 mV s À1 in ap otential range between AE 1V .E IS was carried out in af requency range between 5mHz and 100 kHz. Galvanostatic cycling was executed in the voltage range of 0-1 Vatacurrent density of 0.6-2.5 mA cm À2 .
The specific capacitances are calculated using CV data and GCDdata. For the capacitance calculation out of CV-data, the area of the charge-and discharge curve is integrated and normalized to the voltage window DU and the scan rate u [Eq. (1)]: The GCD capacitance is calculated using the discharge curve and the charging current [Eq. (2)]: For normalization the device area and device volume (including electrolyte volume and area) are used. Furthermore, an areal capacitance based on the effective carbon electrode area is calculated.
Energy and power density are calculated by integration of the galvanostatic discharge curves using Equations. (3) and (4): The area of the integrated discharge curve A Dis ,t he applied current I, t diss discharge time and the device volume of interdigitated electrodes V D are used as factors for the calculation.