Natural Polymers as Green Binders for High‐Loading Supercapacitor Electrodes

Abstract The state‐of‐the‐art aqueous binder for supercapacitors is carboxymethyl cellulose (CMC). However, it limits the mass loading of the coatings owing to shrinkage upon drying. In this work, natural polymers, that is, guar gum (GG), wheat starch (WS), and potato starch (PS), were studied as alternatives. The flexibility and adhesion of the resulting coatings and electrochemical performance was tested. The combination of 75:25 (w/w) ratio PS/GG showed a promising performance. Electrodes were characterized by SEM, thermal, adhesion, and bending tests. Their electrochemical properties were determined by cyclic voltammetry, electrochemical impedance spectroscopy, and cycling experiments. The PS/GG mixture conformed well to criteria for industrial production, enabling mass loadings higher than CMC (7.0 mg cm−2) while granting the same specific capacitance (26 F g−1) and power performance (20 F g−1 at 10 A g−1). Including the mass of the current collector, this represents a +45 % increase in specific energy at the electrode level.


Introduction
Electrochemical double-layerc apacitors (EDLCs) are highpower energy-storage deviceso ffering long cycle life.N umerous unique applications already exist for EDLCs, and their use in the expanding ecosystem of renewable energy is growing steadily as more and more effort is put into fighting climate change. Their energy-storage mechanism is based on the fast and reversible adsorptiono fi ons from an electrolyte onto a poroush igh-surface-area material. Commonly,c ommercial EDLCs include activated carbon (AC) electrodes and organic electrolytes, which reacha pproximately 3V operating voltage. [1] Production of these devices involves coating large sheets of metal (Al) currentc ollectors by using fast, low-cost,r oll-to-roll processes.C oatings are applied as slurries consisting of the active material, conductive additive(s) (carbonb lack, CB), and polymericb inder(s). This latter must fulfill several requirements, [1][2][3][4] such as: 1) gives uniform coatings on the Al current collector; 2) endures drying at high temperature in air without degrading; 3) provides mechanical stability( the resulting coatings have to survivem echanical stresses such as bending and rolling against steel rolls, pressing in ac alender and, finally,t ight winding in cylindrical cells); 4) grantsa no ptimal microscopic structure of the electrodes, with well dispersed conductive additives to reduce resistance; 5) is electrochemically inert to not jeopardize the operation of the EDLC; 6) requires an environmentally friendly solvent( such as water) for slurry processing or,i deally,n osolventatall.
Despite some alternatives having been recently proposed, the state of the art in aqueous, non-fluorinated bindersf or EDLC is carboxymethyl cellulose( CMC). [2,5,6] Although this material conforms to the requirements outlined above,i th as one major drawback, which is that the maximum thickness (or mass loading) possible for coatings is relativelyl ow.The reason for this lies in the shrinking of the CMC coatings upon drying, which can result in cracking of the coating. Additionally,t he low flexibility of the coating at high mass loadings causes cracking upon, for example, rolling after drying.A lso, electrodes based on alternative water-soluble polymers, such as poly-(acrylic acid) (PAA) or sodium alginate, have been only reported with relatively low mass loadings (< 5mgcm À2 ), [7] suggesting that such bindersm ay suffer from the same issues as CMC. Achieving higher mass loadings is extremely important because it would allow higher active material/currentc ollector mass ratios,c onsequently improving the specific energy of EDLCs. [1] In this work, new bindersb ased on water-soluble natural polymers were comprehensivelys tudied. The study allowed the identification of apromising new binder mixture ful- The state-of-the-arta queous binder for supercapacitors is carboxymethyl cellulose (CMC). However, it limits the mass loading of the coatings owing to shrinkage upon drying. In this work, natural polymers, that is, guar gum (GG), wheat starch (WS), andp otato starch (PS), were studied as alternatives. The flexibility and adhesion of the resulting coatings and electrochemicalp erformance was tested. The combinationo f7 5:25 (w/w)r atio PS/GG showed ap romising performance. Electrodes were characterized by SEM, thermal, adhesion, and bend-ing tests. Their electrochemical properties were determined by cyclic voltammetry,e lectrochemical impedance spectroscopy, and cycling experiments. The PS/GG mixture conformed well to criteria for industrial production, enabling mass loadings higher than CMC (7.0 mg cm À2 )w hile granting the same specific capacitance (26 Fg À1 )a nd power performance (20 Fg À1 at 10 Ag À1 ). Includingt he mass of the current collector,t his represents a + 45 %i ncreasei ns pecific energy at the electrode level.
filling the above-mentioned criteria and substantially outperforming CMC in terms of flexibility and maximum mass loadings of the resulting electrodes.

Coating tests
Af ew binder candidates [cold-water-soluble wheat starch (WS), potato starch (PS), and guar gum (GG)] were investigated for the preliminary coating tests. This served to check the basic suitability of the materials as possible binders. Starch is an atural polymer consisting of two types of polysaccharides,t hat is, amylosea nd amylopectin.T he first is constituted by a(1-4) glycosidic-bonded a-d-glucose units forming al inear chain. Amylopectin is similar, but with many additional side chains connected by a(1-6) glycosidic bonds. Their mass ratios, chain lengths, branching, and other factorsd epend on the botanical source.T hese polysaccharides are produced by plants in the form of so-called "granules", that is, particles of 2-100 mmi n size. They also contain varying small amounts of protein and minerals. [8] In the case of cold-water-soluble starch, these granules are shreddedo rp redissolved to increaset heir solubility in water without need for elevated temperatures. [8,9] GG is also a polysaccharide produced fromg uar beans. It differs from starches in that the backbone is made from b(1-4)-bonded dmannopyranose units carrying short side chains on every second unit. These side chains are single (1-6)-bonded a-d-galactopyranose units. GG is known to strongly increaset he viscosity of aqueous solutions even at low concentration,f or which it is used in aw ide range of applications from food additives to fracking. [10] Each binder candidate was mixed with water,activematerial, and conductive additive to yield 5wt% binder content with respect to all solids in the slurry.T hree qualitative criteria were appliedfor the initial screening: 1) Processability of the slurry:t he slurry should not show signs of sedimentation within the timescales needed for the coating process (1-3 h). 2) Uniformity of the coating:t he slurriesn eededt op roduce homogeneous coatings without bubbles, pinholes, stripes, or similar defects;t he coating quality could typicallyb e modified by the addition or removal of water. 3) Solid content of the slurry:b indersn ot enabling sufficiently high solid content (> 25 wt %) but fulfilling the previous two points were excluded.
These criteria result from the needs of cell production.I n fact, homogeneous coatings are indispensable to ensure tight wrapping of the electrode and separator rolls, which in turn ensuresp recise balancing and low performance variation between cells. High solid contents are desirable because any excess solventn eeds to be evaporated before further processing. The more water is in the slurry,t he longer (slower)t he drying line hastobe(run), which results in higher costs and reduced productivity,r espectively. [1] Settling of solids in the slurry can cause density gradients in the slurry container.T his would cause the properties of the slurry to change during coating and therefore causei nhomogeneous coatings. [7] The resultso ft he binder screening are summarized in Ta ble 1, including those of the bending test discussed in the next section. Additionalp hotographs can be found in Figure S1 in the Supporting Information. It should be noted that none of the bindersa re found to be entirely suitable on their own. The typical high viscosity induced by GG reduces the maximum possible solid content for coatable slurries to only 20 wt %( Figure S1 Ai nt he SupportingI nformation). WS-based coatings are brittle ands how poor adhesion at all mass loadings ( Figure S1 Bi nt he Supporting Information). Homogeneous WS and PS slurriest ypically have low solid contentsa nd therefore result in low mass loadings even when applied with large doctor blade slit heights. Using higher solid contents is not possible because mixing does not result in homogeneous slurries(Figure S1 Cint he Supporting Information).
However,m ixtures of both starches (WS and PS)w ith GG allow up to 31.6 wt %s olidc ontents, suggesting synergistic effects of the two binders. GG servest oe nables lurries with highers olid contents, acting as an emulsifier and stabilizer,a llowingf or am uch easier slurry mixing. In fact, the advantageous interaction of GG with starches is well known in the food industry,i nw hich it is used, amongo ther things, to reduce the starchyt exture of soups anda void syneresis (the expulsion of water from the polymer network). [11][12][13] This interaction is also evidenced by the tendency of slurriest og el reversibly when left standing. However, the perturbation owing to stirring and pumping in industrial coating lines would easily disrupt the slurry gelation because simples tirring before coating was alwayss ufficientt or e-homogenizet he slurries. The two most promising mixingr atios, also with respect to the coating flexibility (see below), were 1:1f or WS/GG and 3:1f or PS/GG. They are denoted as WS50/GG50 and PS75/GG25.

Bending tests
Besides the rheological properties of the slurry,w hich were here only qualitatively monitored, the most important criterion for the binder selection is the morphological and mechanical quality of the resulting coating. In the industrial environment, This means that the coating must not crack during bending and must be able to endurew ithout too much abrasion, which is particularly criticalf or high mass electrode loadings. Achieving high bending flexibility is the key here because thick coatings easily break at small curvatureradii. [1,14] To test the flexibility of the high-mass-loadingc oatings based on the new binders, the electrodes were cut into approximately 2cmw ides trips and wrappeda round 12 mm diameters teel pins. It should be noted that the flexibility test used here was quite conservative, considering that rolls in industrial coating lines are typically severalc mi nd iameter. However,t his allowedu st oc omparet he relative bendability of the electrodes featuring different binders. [15] For the sake of comparisonw ith the state of the art, CMC-based electrodes were also subjectedt ot he same procedure. As noticeable in Figure 1A,h igh-mass-loading CMC-basede lectrodes show large cracks. In fact, small-to-medium-size cracks are already noticeable at loadings of approximately 3.5-4.0 mg cm À2 .H owever, Figure 1B shows that the WS50/GG50 mixture also tends to crack andd elaminate at approximately 6mgcm À2 ,precludingits use as abinder.Incontrast, PS75/GG25 only shows minor cracks even up to 7mgcm À2 mass loading, demonstrating much better flexibility.T his is as trong indication that PS75/GG25 represents as ignificant improvement over the state of the art, enabling the high mass loadingsr equested at the industrial scale. Thus, the followingi nvestigationsw eref ocusedo nt he PS75/GG25 mixture. Further details aboutt he coating thicknesses and mass loadings of this mixture can be found in Figure S2 in the SupportingI nformation.
As an ext step, the adhesion strengthf or both PS75/GG25 and CMC electrodes was determined. For PS75/GG25-based electrodes the value is 400 kPa, but the limiting factor is the inner electrode cohe-sion rather than adhesion to the current collector.F or CMC,i n contrast,t he mechanicals tabilityo ft he CMC-based electrodes is 1100 kPa, limited by the adhesion of the layer to the current collector.N onetheless,t he coating stabilityo ft he PS75/GG25 electrodes is deemed sufficient.

Micro-morphology
The morphologyo ft he coated electrodes was investigated by SEM. These electrodes were not subjected to any mechanical stress but simply taken after drying. Figure 2s hows as eries of SEM images comparing CMC- (Figure 2A)a nd PS75/GG25based electrodes ( Figure 2B)w ith increasing mass loadings. It is clear that the number and size of cracks in the CMC-based electrodes are larger than in those employing the PS75/GG25 binder,e specially at high mass loadings. Indeed, the first cracks in PS75/GG25-basede lectrodes only appear at very high mass loadings (9.7 mg cm À2 ).
It is well knownt hat CMC tends to contract strongly during drying. This is clearlyt he reason for the greater extent of microscopic cracks compared with PS75/GG25. [3] Microscopic cracking may promotea ccelerated disintegration of the electrode because parts of the active material layer can lose contact with the bulk of the electrode and the current collector.I nf act, CMC-based layers at higherm ass loadings ( % 7mgcm À2 )c ontract so strongly that macroscopic cracks appear when the layers are not allowed to relax and roll up under the tension. In the absence of constraints, as during the drying in industrial processes in which the foil is only supported by ah ot air stream, the foil would strongly deform.H owever,t his limitation does not seem to occur when using the mixture of PS and GG as binder.
Another important aspect in the electrode morphology is the microscopic homogeneity.Awell-dispersed conductive additive is essential to achieve low electrode resistance. [7] If the interaction of the binder with either the active material or the conduc-  tive additive is too strong there is ar isk of agglomeration. The surfaceo fP S75/GG25-based electrodes at high magnification shows well-dispersede lectrode components (see Figure S3 in the Supporting Information). These electrodes also show a rather good homogeneity along their thickness as demonstrated in Figure S4 in the Supporting Information, depicting the cross-section of the electrode made by focused ion beam (FIB) milling.

Thermal stability
High temperatures are necessary for cost-effective electrode production in short and fast drying lines, ensuring high productivity. [1] However,i ft he binder degrades during the drying process,t he integrity of the coating is compromised. Therefore, the thermals tability of the binder components is rather important and needs to be evaluated. For this purpose,t hermogravimetric analysis( TGA) experimentsw ere performed on the GG and PS separately,a sw ell as CMC as ac ontrol. Because the binders adsorb significant amounts of water at room temperature, [16] which might hide the weightl oss associated with their decomposition, all samples weref irst subjected to a3h heatings tep at 100 8Ct oe nsure the complete removal of water.T or eliably determine the thermals tability,i sothermal steps were used during TGA rather than ac onstant heating rate. The temperature was increased stepwise and held at each value for 2h.A sc an be seen in Figure 3A,e ven at 180 8Ct he mass loss for both PS and GG is only 0.7 %h À1 .I na ni ndustrial production line, the electrodes would typically pass the drying stage within af ew minutes. [15] It followst hat PS and GG are sufficiently stable in air at high temperatures to enable electrode production.A ctually,t hese binders releasem ost of the water already in the isothermal step at 100 8C, whereas CMC does not, meaning that they are much easier to dry than CMC.

Electrochemical stability
With the production-related hurdles for the PS75/GG25-based electrodes having been passed, the electrochemical properties were investigated.F irst, the electrochemical inertness of the binder mixture itself at the relevant potentials employedi n EDLCs had to be demonstrated. In fact, the binder(s) must not limit the usable operating voltage of the deviceo wing to side reactions. Furthermore, side reactions might compromise the bindinga bility of the binder(s) and result in pulverization or delamination of the electrode, which in turn would cause capacity loss. [17,18] Twocommon state-of-the-art electrolytes, tetraethylammonium tetrafluoroborate (TEA BF 4 )i na cetonitrile (MeCN) or propylenec arbonate (PC), weres elected along with three novel ones developed by Balducci and co-workers. [19][20][21] These latter electrolytes wered eveloped to improvec onductivities and electrochemical stabilityw indows beyond the state of the art, together with reduced toxicity, cost, and flammability.F or example, N-methyl-N-butylpyrrolidiniumt etrafluoroborate (Pyr 14 BF 4 )a nd 3-cyanopropionic acid methyl ester (CPAME) combined with conventionals alts and solvents are promisingn ew electrolyte components.
The intrinsic stabilityo ft he binder was evaluated by cyclic voltammetry.T he binder was first coated on the current collectors (mass loading % 0.15 mg cm À2 ,s ee Figure S5 Ai nt he Supporting Information) without any carbonaceous material to assesst he faradaic currents originating from the binder decomposition, which would be hidden by the large capacitive current from af ull carbon electrode. [22,23] The "binder"e lectrodes were used as workinge lectrodes in three-electrode cells by using al eaklessr eference Ag/AgCl3 .5 m KCl electrode for accurate determinationo ft he onset potential of significant faradaic currents. [24] Separate cells were assembled for cathodic and anodic sweeps. The sweepss tarted at À0.250 Vv ersus Ag/ AgCl 3.5 m KCl, whichr oughlyc orresponds to the initial potential of carbon, [25] to ensure equal scope of probingo ft he electrochemical window in both the cathodic anda nodic directions. Because at ypical EDLC will go to the extreme potentials in its operating life many times, it is not reasonable to conclude limits from the first cycle currento nset, in whicho nly minor reactions associated with transient phenomena are typically observed. Hence, the sweepsw ere repeated ten times to show the development of reactivity at the edges of the potential window.
The top curve in Figure 4s hows the current response for a completely blank Al current collector.I nthis exemplary measurement,t he anodic sweep exhibits an irreversible oxidation peak, resulting from Al dissolution, whichd ecreases upon cycling as the result of passivation phenomena. This anodic dissolution peak is also found in the anodic sweeps of binder-coated Al electrodes;h owever,t he onset occurs at much higherp otentials. Moreover, the anodic current recorded during the following cycles is lower,i ndicatingapassivation occurs even with the coated binder covering the Al surface. Finally,t he binder does not reactt oa ny appreciable extent with any of the employede lectrolytes. Even at the highestp robedp otentials, the current response decayst oaf ew tens of mAcm À2 after ten cycles, whereas the binder shows no sign of decomposition or delamination( see Figure S5 A-C in the SupportingI nformation). These results demonstrate that the PS75/GG25 binder does not limit the electrochemical stabilityo ft he electrodes. Indeed, by using conventional binders and full carbon electrodes, Balducci and co-workersf ound smaller potentialw indows,a nd smaller values are also reported for TEA BF 4 in PC and MeCN in the literature. [19,21,26] Electrochemical impedance spectroscopy To characterize PS75/GG25-based AC electrodes, in particular, the influence of the binder on the electrochemical behavior, electrochemical impedance spectroscopy (EIS) measurements were performed on symmetrical cells. Different cells were investigated employing CMC-and PS75/GG25-based electrodes with 3.5 mg cm À2 active materiall oading (designated as "standard loading") and 7.0 mg cm À2 (designated as "high loading"; these electrodes were achieved only with PS75/GG25). The corresponding EIS resultsa re shown in Figure 5. Although the standard loading electrodes employing the twob inders show differentt otal impedance ( Figure 5A,B), their impedance spectra could be described with the same equivalent circuit (see Figure 5C). Besides the expected electrolyte resistance at high frequencies, modeled by R1, and the diffusion-limited/capacitive responsed escribed by the open Warburg element Wo1 (at low frequencies), the semicircle at medium frequencies is indicative of charge-transfer processes at an interface. This was modeled by ap arallel RQ element comprisingaresistance( R2) and constant phase element (CPE2). In the case of CMC, the value for R2 is approximately 1 W,w hereasR 2e quals 3 W for PS75/GG25-based electrodes. Interestingly,i nh igh-loading-type cells (see Figure 5D), two distinguishable semicircles could be resolved in the same frequency range. Therefore, two separate RQ elements were employed to fit this spectrum (see model EC in Figure 5E). Here, the values are 2 W for R2 and 5 W for R3. These semicircles are attributed to the transfer of electrons betweent he active materialp articles and the electrode coatinga nd the current collector. [27,28] In the case of CMC, the interfacial resistances are relatively low,b ut PS75/GG25 exhibits slightly higher resistance contributions from both sources. Because the interfacial resistancea tt he current collector should be independentf rom the mass loading, it is reasonable to attribute the second RQ element (R3 j CPE3, resolved only for high loading atl ower frequencies) to the particle-particlei nterfaces.
In summary,t he EIS resultsi ndicate as light drawback of PS75/GG25 versus CMC in terms of electrode resistance, well matching the FIB cross-section observation (see Figure S4 in the Supporting Information), indicating that the electrode layer/current collector contact still needs to be improved.

Long-term cycling
Although the three-electrode measurements excluded evident detrimental reactions of the binder with any of the tested electrolytes, it is also important to assess the performance of full electrodes in two-electrode cells over longer timescales. This helps to rule out other possible issues such as swelling of the binder or slow degradation. [5] In addition, as the EIS measurements illustrated, the binder has am ajor impact on the electrode resistance, which could be detrimental for cycling or rate performance. [7] Very often in the literature, EDLCs are tested with cycling protocols that chargea nd discharge the cell repeatedly,usually more than 10 000 times, but without any constantv oltage step at full charge. Even with slow charging rates,t his resultsi nt he EDLC spending only asmall fraction of time at high voltages. [29] Consequently,t he long-term stabilityi so ften overestimated. In industry,h owever,t esting protocols typically involve long steps at high voltage. The protocol used here was directly adaptedf rom aE uropean manufacturer of EDLCs and based on IEC standard 6239/1. [30] Further details can be found in the Supporting Information.
Here, symmetric coin cells with 1 m TEA BF 4 in PC were assembled with electrodes based on CMC or PS75/GG25. Once more, cells employing differentb inders (CMC and PS75/GG25) and different electrode loadings (3.5 and 7.0 mg cm À2 ," standard" and "high loading", respectively) were tested. The results of the voltage hold test can be seen in Figure 6A.I ti se vident that the PS75/GG25 enables ap erformance equivalentt oC MC in terms of capacitance retention. This is even true for the high-loading electrodes, which only show as lightly increased equivalent series resistance (ESR). Therefore, such high-loading electrodes can significantly increase the activem aterial/current collector ratio withoutp enalizing the performance. If one includes the current collector mass into the calculation (5.18 mg cm À2 at the thickness of 20 mme mployed here), this represents as pecific energy gain of approximately + 45 %a t the electrode level.
The cycling performance of the electrodes was also tested by using the same type of coin cells but employing fast constant current charge/discharge cycles (i.e.,w ithout voltage hold steps). The results, illustrated in Figure 5B,s how that the rate performanceo fP S75/GG25-based electrodes is mostly equivalent to CMC-based ones. The capacitance for all cells is only moderately affected, and only at very high currents. Overall, the cycling data strongly suggests that PS75/GG25 is basically identical to CMC but allowsd oubling the mass loading.

Conclusions
Three naturalb inder materials for electrochemical double-layer capacitors (EDLCs), guar gum (GG), potato starch (PS), and wheat starch were investigated and compareda gainst the widely used aqueous binder carboxymethyl cellulose (CMC). The 3:1m ixture of PS and GG resulted in the mostf avorable rheology for coating, enabling high solid contents and flexible high-mass-loading electrodes. PS75/GG25 coulds urviveb ending tests at up to 7mgcm À2 ,t wice as much as CMC. SEM microscopy showedg ood dispersion of the active materiala nd conductive additive, as well as reduced crack formation during drying. Both PS and GG were shown to be stable at the typical drying temperatures in air,enabling fast drying. Binder-covered current collectors were used as workinge lectrodes in cyclic voltammetry experiments to determine their electrochemical stabilityi nb oths tate-of-the-art and novel electrolytes. They showede xcellent stability windows that were larger than that of activated carbone lectrodes, proving that they would not limit the devices' operating voltage. Electrochemical impedance spectroscopyw as employed to further characterize the PS75/GG25 electrodes, and as lightly increased contact resistance compared with CMC was observed. Symmetric full cells were subjected to long-term voltage-hold experiments. PS75/ GG25-and CMC-based electrodes showedi dentical and excellent capacitance retention with only minor equivalentseries resistancei ncrease. Doubling the mass loading for PS75/GG25 had only as mall impact on performance but represents al arge increasei ns pecific capacitance on the full electrode level,

Experimental Section
Electrodepreparationand characterization Potato starch (PS) was purchased from Sigma-Aldrich. Wheat starch (WS) and guar gum (GG) were kindly supplied by Krçner Stärke GmbH and Lamberti S.p.a, respectively.C MC was purchased from Dow WolffC ellulosics (CMC Walocell 2000 PA). Activated carbon (AC, YP50 from Kuraray) and etched aluminum foil (20 mm, 5.18 mg cm À2 )w ere kindly supplied by Skeleton Te chnologies GmbH. Carbon black (CB) was purchased from Imerys Graphite & Carbon (C-ENERGY Super C45). All slurries were prepared with weight ratios of AC/CB/binder of 90:5:5. The solid content was adjusted as needed to achieve homogeneous coatings. The binder was stirred in ultrapure water (milli-Q) until fully dispersed and, in the case of PS, heated for 30 min at 70 8C. CB was added next, and the slurry was stirred for at least 1h.W ater lost during stirring was added again. AC was mixed in by manual mixing and kneaded until the slurry was well homogenized. The slurries were coated onto Al foils with an adjustable doctor blade at 50 mm s À1 .C oatings were allowed to dry at room temperature for approximately 30 min and then dried overnight at 80 8C. Bending tests were done with 12 mm diameter steel pins and 2cm wide strips cut from dried electrodes. Mass loadings were determined by cutting 12 mm disks from the strips and weighing them. Counter electrodes were made with weight ratios of AC/CB/polytetrafluoroethylene (PTFE) of 80:10:10. Using the 60 wt %P TFE solution in water (Sigma-Aldrich), CB and AC were dispersed by stirring until the water evaporated and ad ough-like slurry was formed. The slurry was rolled out to obtain at hick layer (about 0.5 mm), which was cut into 12 mm diameter disks. TGA was done in synthetic air (20 %O 2 ,8 0% N 2 )b yu sing the Discovery TGA by TA Instruments. SEM images were taken with a Zeiss LEO 1550 microscope at 3kV, using combined backscattered electron and secondary electron imaging.

Electrochemical characterization
Te traethylammonium tetrafluoroborate (TEA BF 4 ), acetonitrile (MeCN), and propylene carbonate (PC) were purchased from Sigma-Aldrich. N-Methyl-N-butylpyrrolidinium tetrafluoroborate (Pyr 14 BF 4 )a nd 3-cyanopropionic acid methyl ester (CPAME) based electrolytes were kindly supplied by the group of Prof. Andrea Balducci (Friedrich-Schiller-University Jena). Electrodes were cut into 12 mm disks and dried at 110 8Cu nder vacuum overnight and transferred to an argon-filled glovebox (LabMaster,M braun GmbH) with < 0.1 ppm O 2 and < 0.1 ppm H 2 Of or all cell assembly.S wagelok-type three-electrode cells were used for the CV experiments. AC-overloaded (in terms of capacitance) counter electrodes were employed. As reference electrodes, Ag/AgCl 3.5 m KCl leakless minielectrodes (edaq) were used. Approximately 300 mLo fe lectrolyte was impregnated into glass fiber disks (GF/D, thickness:6 70 mm, diameter:1 3mmd iameter;W hatman). Ap otentiostat/galvanostat (VMP3, Biologic Science Instruments) was employed to record the CVs. Two-electrode EDLCs were assembled in coin cells with symmetric mass loadings of either 3.5 mg cm À2 (standard loading) or 7.0 mg cm À2 (high loading), by using 150 mLo fe lectrolyte impregnated in the glass fiber separator.T hese were cycled in aM accor Battery Tester 4300. EIS experiments were performed in symmetric two-electrode Swagelok cells with an Impedance/Gain-Phase Analyzer 1260 (Solartron Analytical). All electrochemical tests were performed in climatic chambers at T = (20 AE 2) 8C( KB115, Binder GmbH).
collectors heets and activated carbon, as well as many helpful comments on industrial requirements;A.Balducci and C. Schütter (FSU Jena) for providing electrolytes;M .M arinaro (ZSW Ulm) for measuring adhesion strengths;M .K ünzel( HIU-KIT) for help with the FIB cross section.T he authors would like to acknowledge the Helmholtz Institute Ulm (Karlsruhe Institute of Technology), and the Bundesministerium fürW irtschaft und Energie( BMWi)f or fundingt his work within the project "ULTIMATE-Ultrakondensatoren auf Basis innovativer Materialien füre rhçhte Energiespeicherfähigkeit" (contract number03ET6131E).