Ultrastable Surface‐Dominated Pseudocapacitive Potassium Storage Enabled by Edge‐Enriched N‐Doped Porous Carbon Nanosheets

Abstract The development of ultrastable carbon materials for potassium storage poses key limitations caused by the huge volume variation and sluggish kinetics. Nitrogen‐enriched porous carbons have recently emerged as promising candidates for this application; however, rational control over nitrogen doping is needed to further suppress the long‐term capacity fading. Here we propose a strategy based on pyrolysis–etching of a pyridine‐coordinated polymer for deliberate manipulation of edge‐nitrogen doping and specific spatial distribution in amorphous high‐surface‐area carbons; the obtained material shows an edge‐nitrogen content of up to 9.34 at %, richer N distribution inside the material, and high surface area of 616 m2 g−1 under a cost‐effective low‐temperature carbonization. The optimized carbon delivers unprecedented K‐storage stability over 6000 cycles with negligible capacity decay (252 mA h g−1 after 4 months at 1 A g−1), rarely reported for potassium storage.


Introduction
Limited Li resources in the earthscrust (20 ppm) and the ever-increasing demands for large-scale electrical energy storage have stimulated numerous efforts for alternative energy storage systems beyond the state-of-the-art Li-ion batteries. [1] K-ion based energy storage systems,such as K-ion batteries or related hybrid capacitors,have recently emerged as promising candidates based on abundant Kr esources (17 000 ppm), low cost, and relatively low redox potential (À2.93 Vv s. standard hydrogen electrode (SHE)) compared to Li (À3.04 Vv s. SHE). [1b,2] Ak ey requirement to realize high performances for K-ion based energy storage lies in developing stable anode materials,a sh uge anode volume variation and sluggish kinetics are typically associated with large-sized Ki ons. [2c,d] Thus,r ecent years have witnessed an explosive growth of exploration. Va rious alloying and intercalating anodes,s uch as metals, [3] transition-metal selenides/ sulfides, [4] and carbonaceous materials have been proposed. [2d, 5] Carbonaceous materials have been demonstrated to be among the most promising candidates due to their high surface area, abundant active sites,a nd large interlayer spacing for storing Ki ons.
An obvious choice is graphite as electrode,the successful material applied in Li ion batteries with formation of stage I LiC 6 via intercalation. Graphite is electrochemically active to Ki on intercalation, leading to the formation of fully potassiated stage IK C 8 and at heoretical capacity of 278 mA hg À1 . [2d, 5] However,r ate performance and cycling stability are still poor, and graphite suffers from large volume expansion (58 %) and al ow Ki on diffusion rate during the intercalation/extraction process. [5,6] Progress has been made to extend the stability to 500 cycles by expanding interlayer space and by nanostructure design in graphitic carbons. [7] To pursue al ong-cycling lifespan with high rate capability, considerable efforts have been devoted to the exploration of novel carbonaceous materials.Progressive enhancement in cyclability reveals several critical structural parameters must be controlled for achieving as urface-induced K-adsorption mechanism beyond the intercalation chemistry,i ncluding developed pore nanostructures with high surface area and defects of heteroatom-doped active sites. [8] In this regard, N doping, particularly edge N( pyridinic/pyrrolic configurations) has been proved experimentally and theoretically to be efficient for adsorbing Ki ons with abundant electronegative defect sites and higher adsorption energy. [8a, 9] Despite massive investigation, Ns pecies in carbons cannot be rationally tailored in edge-enriched configurations,o wing to trial-anderror strategies with uncontrolled Ns pecies involved in doping. [8a,c, 10] Designing precursors with definite edge-type N seems possible to promote edge-enriched Ni nr esulting carbons,s uch as using polypyrrole-derived carbons with enhanced capacity and cyclability. [9,11] However, there is still enough room to further push forward the cycling performance with less decay rate (< 0.002 %p er cycle). [9a, 12] Thep ossible reasons for restricted cycling stability could be,onone hand, the incapability of rationally tailoring Ns pecies into edgeenriched configurations.O nt he other hand, it is difficult to simultaneously achieve edge-enriched Na nd high surface area, because of the well-known trade-off for both structural features. [13] Moreover,l ittle attention has been paid to the spatial Ndistribution or doping depth, which could also affect Kstorage performances.Thus,elaborate engineering of edgeenriched Nincarbonaceous materials is still highly desirable for achieving ultrastable cycling performance while maintaining high capacity and rate capability.
Herein, we demonstrate anew strategy to produce edgeenriched N-doped porous carbon nanosheets (ENPCS) for ultrastable Ki on storage.T he ENPCSs were fabricated by pyrolysis-etching of apyridine-coordinated polymer network, taking into account the predefined pyridinic Ni nt his precursor.B yl owering the pyrolysis temperature to 500 8 8C, we obtained ENPCS bearing both ah igh edge-N content of 9.34 at %and ahigh surface area of 616 m 2 g À1 ,aswell aricher Nd istribution in the inner part of material. TheE NPCS delivers ah igh reversible capacity (313 mA hg À1 ), high rate capability,a nd ultralong cycling stability with almost no degradation (0.0009 %d ecay rate and remaining capacity of 252 mA hg À1 )over 6000 cycles at 1Ag À1 ,representing one of the best cycling stabilities.The high edge-N content, richer N distribution inside the material, and high surface area are together responsible for its impressive ultrastable performance through full utilization of surface adsorption capacitive behavior. Our work constitutes an important step in deploying carbonaceous materials for stable K-ion based energy storage systems.

Results and Discussion
Thef abrication of ENPCS ( Figure 1a)i nvolves the fast coordination of bipyridine with copper ions for producing pyridine-coordinated polymer network and subsequent pyrolysis and etching. [14] Thep yrolysis transforms aromatic pyridine moieties into edge-N heterocyclic carbon rings fusing and stacking around copper species,while the etching enables the creation of microporosity by leaching the copper species, and meanwhile introduces some Of unctional groups.T he polymer network itself shows aw ell-defined cuboid-like nanosheet morphology with as mooth surface (Figure 1b). After carbonization at 500 8 8C, the resulting carbon ENPCS-500 largely retains the precursor-immanent morphology (Figure 1c)w ith randomly aggregated cuboids.H owever,d ue to the shrinkage and pore creation effect, the surface becomes rough with many holes (Figure 1c,d), indicative of the presence of structural defects that could be beneficial for K + storage.T he high-resolution transmission electron microscopy (TEM) image of ENPCS-500 reveals an amorphous structure with short turbostratic nanodomains,w hich are randomly packed and interconnected with rich microporosity (Figure 1e). Some lattice fringes are observed with an interlayer distance of 0.38-0.41 nm (Figure 1e), greater than that of graphite (0.335 nm). According to the HAADF-STEM and element mappings,E NPCS-500 exhibits an even distribution of Na nd Oo ver the cuboids (Figure 1f-i).
With increasing pyrolysis temperature,E NPCS-650 and ENPCS-800 also reveal similar morphology features (Figure S1). However,t he intrinsic crystallinity,h eteroatom doping, and microporosity change.X -ray diffraction (XRD) patterns with hump peaks of (002) and (100) diffractions suggest amorphous structures (Figure 2a). Thec rystallinity degree was further assessed using an empirical R value. [9b, 15] Higher pyrolysis temperatures give rise to increased R values from 1.91 for ENPCS-500 to 2.03 and 2.32 for ENPCS-650 and ENPCS-800, respectively,manifesting the gradual development of ag raphite-like microcrystal structure.T he intensity ratio of I D (disorder-induced D-band) to I G (in-plane vibrational G-band) in the Raman spectra for ENPCS-500 has av alue of 2.71, implying ad isordered structure with many defects.Increased pyrolysis temperatures lead to higher I D /I G values ( Figure 2b and Table S1). This is probably because at relatively lower temperature,ENPCS-500 possesses acharacteristic sp 2 -hybridized hexagon structure inherited from the pyridinic precursor;higher temperatures promote the decomposition and conversion to amorphous carbons with defect nanopores/nanovoids. [9b] From N 2 adsorption/desorption curves,E NPCSs exhibit predominately micropores with an arrow pore size around 0.5 nm and al ess intense peak at 1.2 nm (Figure 2c). ABrunauer-Emmett-Teller (BET) surface area up to 616 m 2 g À1 was obtained for ENPCS-500. Higher temperature results in enhanced microporosity and surface area, which is probably ascribed to the gradual loss of carbon and doped heteroatoms at elevated temperatures, forming more defects in the carbon structure.
Thes urface content of Oa lso decreases with increasing carbonization temperature (Table S2), which can be deconvoluted into three peaks for C = O, C À OH or C À O À C, and COOH (Table S5). C1ss pectra with C=C/CÀC, CÀO/CÀN, C=N, and C=Owere recorded (Table S6). Thehigh N-and Odoping is also responsible for the enlarged lattice spacing (Figure 1e), and could make the surface more polar and hydrophilic. [19] Thed ynamic contact angle (DCA) measurements can reflect the surface characteristics to alarge extent for porous materials,once the water drops start to contact the surface.E NPCS-500 always shows the lowest DCAa t intervals between 0.1 and 0.4 s ( Figure 2f,F igures S3 and S4), demonstrating its high N-and O-decorated surface. Likewise,the superior low-pressure water capture of ENPCS-500 also indicates the higher N-and O-content and enhanced hydrophilicity (Figure 2g). In contrast, the total capacity of ENPCS-500 is lower ( Figure S5), ascribed to the larger surface area and pore volume of ENPCS-650 and ENPCS-800.
Theb ulk elemental composition was measured by combustion elemental analysis.The bulk N:Catom ratio decreases with increasing pyrolysis temperature,c onsistent with the conclusion from XPS (Table S2). Intriguingly,itisfound that ENPCS-500 bears abulk N:Catom ratio of 1:4.7, higher than that (1:6.8) determined by XPS and even the value of pyridine rings in the precursor (1:5) (Figure 2h). Considering XPS is as urface technique with ap enetration depth of several nanometers,t he higher bulk N:Cr atio indicates that the distribution of Ni sm ore concentrated at the interior of the cuboids.T his probably originates from the thermally induced contraction and compression at the nanointerface between the organic pyridine moieties and the copper species template,t hus retarding the decomposition of Ni nside. [20] The removal of the template could expose more edge Natoms at defect sites like nanopore walls at inner materials.F or the other two carbons obtained at higher temperatures,similarly higher bulk N:Cr atios were also observed, but with significant decreased values.S uch ag radient-like distribution of N could act as ar eservoir for exposing renewed active Ns ites upon continuous cycling. Taken together,these results clearly suggest that the use of an aromatic pyridine-coordinated polymer network and alower pyrolysis temperature simultaneously achieved edge-enriched Ncontent of 9.34 at %, richer Nd istribution inside the defect sites of the materials,a nd al arge surface area (616 m 2 g À1 ), which are expected to facilitate the surface-dominated capacitive storage. [21] Figure 3a presents the initial four cyclic voltammetry (CV) curves of ENPCS-500. Acathodic peak at 0.41 Vinthe initial cycle disappears in subsequent cycles,probably because of the formation of as olid electrolyte interface (SEI) film. Another small peak near the cutoff voltage (0.01 V) is associated with the K + intercalation. Upon charging,ahump peak around 0.87 Vc orresponds to the depotassiation process.The curves overlap well after the second scan, suggesting ah ighly reversible electrochemical behavior. In spite of similar CV characteristics for ENPCS-650 and ENPCS-800, ap rogressive reduction in areas of the CV curves was observed ( Figure S6), revealing ad ecreased capacity.T his is mainly caused by the elimination of Ns pecies at higher temperature.G alvanostatic discharge-charge curves show behaviors in keeping with the CV results,a nd an initial discharge capacity of 628 mA hg À1 and ar eversible charge capacity of 313 mA hg À1 were obtained for ENPCS-500, corresponding to an initial Coulombic efficiency (ICE) of 50 %( Figure S7a). Thel ow ICE is frequently observed for amorphous carbonaceous materials.T he ICE of ENPCS-500 is higher than that of most reported N-doped carbons. [8a,d,h, 16a, 17, 18, 22] However,further work should be focused on reducing the irreversible capacity such as design of novel structures,p repotassiation, and optimization of electrolyte.
After two cycles at 0.05 Ag À1 ,c ells were subjected to further cycling at 0.1 Ag À1 .E NPCS-500 shows as light capacity loss in the initial 15 cycles and then ag radual increase to 246 mA hg À1 after 150 cycles (Figure 3c). However,E NPCS-650 and ENPCS-800 show continuous fading with residual capacities of 206 and 166 mA hg À1 after 150 cycles,r espectively.T he capacity decay in the initial cycles could be attributed to the incomplete SEI formation as Figure 3. a) CV curves of ENPCS-500a t0 .1 mVs À1 ;b )the second galvanostatic discharge-charge curve of ENPCS-500 at 0.05 Ag À1 ; c) cycle performances and Coulombic efficiencya t0.1 Ag À1 and d) rate performances with 10 cycles at each rate;e)long-term cycling performance and Coulombic efficiency of ENPCS-500 at high rate of 1Ag À1 . The cell was restarted from 5000 cycle;f)comparisonsofcycle number and corresponding average capacity decay rate per cycle of various heteroatom-doped carbons. Note:f or (c)-(e), the initial two cycles were performed at 0.05 Ag À1 .
well as the irreversible decomposition of electrolyte on the high-surface-area carbon. [24] Thegradual increase of ENPCS-500 afterwards is probably caused by the progressive utilization of N-rich domains located inside the particles and deeper penetration during cycling.I nc ontrast, ENPCS-650 and ENPCS-800 show significantly reduced Nd oping, giving rise to little increase or even decrease in capacity.E NPCS-500 shows the best rate capabilities (Figure 3d), delivering reversible capacities of 276, 255, 229, 206, 184, 157, and 110 mA hg À1 at current densities of 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, and 4.0 Ag À1 ,r espectively.W hent he rate turns back to 0.05 Ag À1 ,t he capacity is entirely recovered, indicative of high reversibility and excellent structural integrity.I nc ontrast, ENPCS-650 and ENPCS-800 retain lower capacities of 32 and 73 mA hg À1 ,r espectively at ah igh rate of 4.0 Ag À1 . Ther ate performance of ENPCS-500 is comparable to or even better than that of some reported carbons ( Figure S8). Generally,t he rate performance is governed by both the electronic and ionic conductivity.T he former is mainly influenced by the carbonization temperature,whereby ahigher temperature will increase the conductivity. [18] Thel atter, associated with ion transport/diffusion, is governed by doped active sites and porous structure.C onsidering that both the lower carbonization temperature and surface area are thought to be negative to electronic and ionic conductivity, respectively,E NPCS-500 still delivers superior rate performance,again highlighting the critical role of edge-enriched N for facilitating the ion transport/diffusion.
Thelong-term cycling of ENPCS-500 was tested at 1Ag À1 (Figure 3e). After the first two cycles at 0.05 Ag À1 ,E NPCS-500 was subjected to cycling at 1Ag À1 and shows ac apacity fading from 266 to 183 mA hg À1 at 137 cycles at 1Ag À1 .Then the capacity stabilized and ac apacity of 192 mA hg À1 was obtained after about 2000 cycles ( Figure S9). Subsequently, ag radual rise was observed and the capacity reached 247 mA hg À1 after 5000 cycles,very close to its initial capacity at 1Ag À1 .T he stabilization during the 2000 cycles indicates the wetting of the surface or near surface,e nhancing the reversibility of the charge storage.T he continuous rise afterwards is mainly related to further utilization of the richer Nd istribution inside the material, which serves as an N reservoir to release more active Ns ites for capturing Ki ons. Ther esult suggests the importance of constructing gradientlike Nd istribution for ultrastable cycling. This phenomenon can also be reflected by electrochemical impedance spectra ( Figures S10 and S11), in which the calculated charge transfer resistance (R ct )gradually decreases and then stabilizes (R ct is 1902, 1522, 1581 Ohm at 100, 1500, 2000 cycles,r espectively, Table S7). When the cell was restarted after 5000 cycles,there was still no decay (252 mA hg À1 to 6000 cycles), further revealing the robust structure.Acapacity retention of 94.5 %c an be obtained during the whole 6000 cycles at 1Ag À1 ,c orresponding to ac apacity decay rate of only 0.0009 %p er cycle (Figure 3f). Such al ow attenuation rate is rarely observed for carbonaceous materials including heteroatom-doped carbons (Figure 3f), other porous carbons, and even some carbon composites with metallic compounds (Table S8), validating the advantage of such an architecture design on alleviating structural degradation. Thec uboid-like structure is well retained after 1000 cycles (Figures S12 and S13a). Elemental mapping still displays ah omogeneous distribution of C, N, O, and Ko ver the cuboids (Fig-ure S13b-f), confirming the remarkable structural and Ndoping stability for effective Ks torage.I nc ontrast to intercalation chemistry,t he surface-driven pseudocapacitive behavior mainly occurs at edges,defect sites,orpore surface, without causing damage to the electrode materials.E NPCS with high edge-enriched Ndoping, defect structure,and high surface area could enable the full utilization of pseudocapacitive behavior,a sd iscussed below with ENPCS-500 and ENPCS-800.
Thek inetics were investigated by conducting CV tests with different scan rates (Figures S14a and S15a). According to the equation i(V) = k 1 v + k 2 v 1/2 , i is separated into as urface-driven pseudocapacitive process (k 1 v)a nd ad iffusion-controlled contribution (k 2 v 1/2 ). Figure 4a,b show the contributions from pseudocapacitive and diffusion processes at 0.6 mV s À1 .Apredominately capacitive contribution of 77.1 %i sa chieved for ENPCS-500, significantly higher than 54.7 %f or ENPCS-800. By analysis of other scan rates (Figures S14b-d and S15b-d), the capacitive contribution improves with higher scan rates.E NPCS-500 always shows higher capacitive contribution than ENPCS-800 ( Figure 4c). Even at avery low scan rate of 0.1 mV s À1 ,ahigher capacitivedominated contribution of 59.7 %isobtained for ENPCS-500, whereas ENPCS-800 is characterized by ad iffusion process with alow capacitive contribution of 22.9 %. Such pseudocapacitive behavior can be enhanced by increasing the surface area or introducing defects such as N-doping sites.T he simultaneous achievement of edge-enriched Na nd high surface area in ENPCS-500 (Table S4) is beneficial for the highly efficient pseudocapacitive process.Surface O-containing groups could contribute to the performance, [8g, 9a, 25] generally around 1.7 Vf or adsorbing Ki ons.T he much lower capacity contribution around 1.7 Vand the relatively low O content result in much less capacity from O. Thehigher level of edge-N sites and developed porosity induce more defects, leading to fast Ki on diffusivity kinetics.T he diffusion coefficient, D k was calculated by the Randles-Sevcik equation, showing linear plots with higher slope for ENPCS-500 (Figure 4d,e). D k is 1.47 10 À11 and 4.56 10 À12 cm 2 s À1 for reduction and oxidation peaks,r espectively,g reater than those for ENPCS-800 ( Figure 4f). Likewise,E NPCS-500 demonstrates smaller R ct (Figure 4g and Table S7). Consequently,the superior Kion diffusivity kinetics in ENPCS-500 is responsible for its high capacity and rate performance.
Galvanostatic intermittent titration (GITT) measurements were further performed to elucidate the diffusion kinetics of Ki ons during the dynamic potassiation/depotassiation process (Figure 5a). ENPCS-500 exhibited smaller overpotentials compared to ENPCS-800, indicating ah igher D k .T he resulting D k as afunction of the depth of dischargecharge was calculated, and ENPCS-500 shows higher D k than ENPCS-800 during the whole potassiation and depotassiation process (Figure 5b), manifesting its faster Ki on diffusivity.
Considering that the surface area of ENPCS-500 is slightly lower than that of ENPCS-800 (616 vs.8 42 m 2 g À1 ), one may suggest that favorable Ki on diffusivity is mainly dominated by the accessible edge-enriched Nsites.Aprogressive decline of D k was observed as potassiation proceeds,mainly because Ki ons have to overcome the repulsive gradient from previously adsorbed Ki ons. [22c] In situ Raman spectra were  recorded to validate the pseudocapacitive dominated charge storage property (Figure 5c). Previous studies indicate that the G-band or the intensity ratio of I D /I G could be employed to monitor the staging behavior. [8a,f,26] Thei ntercalation of K ions into graphitic layer generally results in achange of I D /I G , weakening and even disappearance of the Ga nd Db ands at the fully discharged state due to the formation of KC x . [10a, 27] However,i tc an be seen that the Da nd Gb ands remain almost unaffected (Figure 5c,d), demonstrating that the storage of Ki ons does not induce structural changes,w hich further supports the surface-adsorption mechanism. Asimilar observation was reported for non-graphitic carbon even at higher temperature. [10a] Ther esult can also explain the nearsloped line in the discharge-charge curve (Figure 3b)w ith predominate surface-driven process and minor potassium intercalation, [8d, 24] in contrast to graphite which displays ap lateau region below % 0.2 V. Although such as lopedominated process could avoid the possible plating of Kmetal for safety purpose,t he relatively high potential without ap lateau will decrease the energy density of the full cell. Studies are in progress to construct some local expanded graphitic structures in such defect edge-enriched Nc arbon layers for practical applications.
To theoretically reveal the superior Ki on storage capability by edge-enriched N, the relative Ka dsorption abilities on carbon lattice with various Nconfigurations were simulated (Figure 5e). Thea dsorption energy (E ads )o fK atoms on an idealized graphene layer is calculated to be À0.59 eV,and E ads on an N-Q site is À0.55 eV.For the edge N doping, we can envisage,i deally but reasonably,t hat two situations can be considered:o ne is the location at the material outer surface (edge plane of carbon rings) and the other is interior of the materials such as defects of carbon basal planes (e.g., pore walls and nanovoids). Forthe former, the E ads for N-6 is À1.16 and À1.01 eV with two possible sites, and the E ads for N-5 is À1.55 eV ( Figure S16). Ther esult indicates that N-5 shows higher adsorption ability than N-6 in this case,but both are superior to the N-Q site.The higher E ads of the N-Q site originates from the electron-rich structure with anegative effect on Kadsorption. Forthe latter, the E ads for N-5 and N-6 at defect sites inside materials is determined to be À2.62 and À3.61 eV,r espectively,s ignificantly lower than that of N-Q and those Na toms at the edge plane of carbon rings on the material outer surface.E NPCSs exhibit similar pores with small sizes at 0.5 nm, enabling concentrated edge-located No nt he defect pore surface.T he much higher content of edge-enriched No fE NPCS-500 thus exhibits superior capacity and high rate capability than other ENPCSs. More interestingly,m uch richer Nd istribution (N:C > 1:4.7) within ENPCS-500 also makes continuous utilization of N-6/ N-5 at defect sites inside possible with astronger tendencyfor Kcapturing, resulting in agradual increase of capacity with an outstanding cycling performance.T hese results vividly show that it is vital to increase the edge-N doping level, especially at the defect carbon layer sites to maximize the Ka dsorption and cycle stability.

Conclusion
In summary,w ed emonstrated the preparation and utilization of edge-enriched N-doped porous carbons for fast and stable Kion storage.The pyrolysis-etching of apyridinecoordinated polymer enables the conversion of pyridine moieties into edge-enriched Na ctive sites at lower carbonization temperature,a nd the generation of micropores with rich defects via etching.T he optimized sample with superior edge No btained at 500 8 8Cd isplays high capacity,e xcellent rate capability,a nd robust cycle stability,t hanks to the surface-dominated pseudocapacitive storage characteristic with fast ion diffusivity.Most noteworthy is the extraordinary cyclability with 0.0009 %capacity decay rate for 6000 cycles at 1Ag À1 .This is also related to the unique structure with richer Ndistribution inside the defect sites of the materials,acting as ar eservoir to expose successively renewed Ns ites upon cycling. Theoretical simulations show the much higher adsorption energy for edge Na tt he defect sites inside the materials.T he utilization of defect edge-N active sites with controlled spatial distribution could be valuable for various carbon-based electrode systems in the future.