Boosting the Supercapacitance of Nitrogen‐Doped Carbon by Tuning Surface Functionalities

Abstract The specific capacitance of a highly porous, nitrogen‐doped carbon is nearly tripled by orthogonal optimization of the microstructure and surface chemistry. First, the carbons’ hierarchical pore structure and specific surface area were tweaked by controlling the temperature and sequence of the thermal treatments. The best process (pyrolysis at 900 °C, washing, and subsequent annealing at 1000 °C) yielded a carbon with a specific capacitance of 117 F g−1—nearly double that of a carbon made by a typical single‐step synthesis at 700 °C. Following the structural optimization, the surface chemistry of the carbons was enriched by applying an oxidation routine based on a mixture of nitric and sulfuric acid in a 1:4 ratio at two different treatment temperatures (0 and 20 °C) and different treatment times. The optimal treatment times were 4 h at 0 °C and only 1 h at 20 °C. Overall, the specific capacitance nearly tripled relative to the original carbon, reaching 168 F g−1. The inherent nitrogen doping of the carbon comes into interplay with the acid‐induced surface functionalization, creating a mixture of oxygen‐ and nitrogen‐oxygen functionalities. The evolution of the surface chemistry was carefully followed by X‐ray photoelectron spectroscopy and by N2 sorption porosimetry, revealing stepwise surface functionalization and simultaneous carbon etching. Overall, these processes are responsible for the peak‐shaped capacitance trends in the carbons.


Introduction
Movingt os ustainable energy sources requires efficient energy storages olutions. [1] Electrochemical devices such as batteries, fuel cells, and supercapacitors play ac entralr olei nt his field. [2][3][4] Each renewable energy source (e.g.,w ind, solar,g eothermal) sets its own requirements for energy density,p ower density,l ife-time, cost, and size. Supercapacitors (also called electrochemical capacitors or ultracapacitors) are important power sourcesfor applications that require fast chargeand discharge. [5][6][7][8][9] For example, the transport industry uses supercapacitors for regenerative braking and for boosting power delivery alongside batteries. [10] In addition to their high power density, supercapacitors offer excellent reversibility (typically % 10 6 cycles) and safer use and production than batteries. However, the energy density of supercapacitors is 2-3 orders of magnitude lower than that of batteries.
Supercapacitors operate through two fundamentalm echanisms. The first is electric double layer capacitance (EDLC), in which charging the electrode leads to adsorption and desorption of electrolyte counter-ionsa ti ts surface. This sorptioni s fast and reversible, determining the high-power density of the device and its longevity.T he second mechanism is pseudocapacitance (PC), wherein fast Faradaic reactions occur at the surface, storing charge in chemical bonds and boosting the energy density. [11,12] These redox reactions are fast enoughs o that diffusion limitations are small and power density remains high. PC was reported for oxidess uch as MnO 2 [13] and Nb 2 O 5 , [14] as well as in conducting polymers (e.g.,p olyaniline). [15] However,t hesem aterials suffer from limitations such as low conductivity and stability as well as high cost.
Porous carbon, with its high surfacea rea, high conductivity, and low cost, is the mostpopular materialfor EDLC supercapacitors. [16][17][18] Furthermore, doping carbon with Oo rNatoms creates active sites for pseudocapacitive reactions (Figure 1). Dopants can be introduced either during synthesis (e.g.,bypyrolysis of biomass [24] or of nitrogen-containing salts) [7,[25][26][27][28] or by postpyrolysis surfacet reatment. Surface functionalization decouples the carbon bulk structuref rom surfacec hemistry.T his allows for orthogonal optimization and for ab etter understanding of pseudocapacitive active sites. Surfacet reatments vary in complexity from brute-force boiling in acid [29] to grafting of predesigned functional groups. [30] The simpler treatmentsa re more practical;however,their effect on supercapaci- The specific capacitance of ah ighly porous, nitrogen-doped carbon is nearly tripled by orthogonal optimization of the microstructure and surface chemistry.F irst, the carbons' hierarchical pore structure and specific surfacea rea were tweaked by controlling the temperature and sequence of the thermal treatments. The best process (pyrolysis at 900 8C, washing, and subsequenta nnealing at 1000 8C) yielded ac arbon with as pecific capacitance of 117Fg À1 -nearly doublet hat of ac arbon made by at ypical single-step synthesis at 700 8C. Following the structuralo ptimization,t he surfacec hemistry of the carbons was enriched by applying an oxidation routine based on am ixture of nitric and sulfuric acid in a1 :4 ratio at two differ-ent treatment temperatures (0 and 20 8C) and different treatment times. The optimal treatment times were 4h at 0 8Ca nd only 1hat 20 8C. Overall, the specific capacitance nearly tripled relative to theo riginal carbon, reaching1 68 Fg À1 .T he inherent nitrogen doping of the carbon comes into interplay with the acid-induced surfacef unctionalization, creatingamixture of oxygen-and nitrogen-oxygen functionalities. The evolution of the surfacec hemistry was carefully followed by X-ray photoelectron spectroscopy and by N 2 sorptionp orosimetry,r evealing stepwise surfacef unctionalization and simultaneous carbon etching. Overall, these processes are responsible for the peak-shapedc apacitance trends in the carbons.
We recently reported an ew family of carbons [25][26][27][28] for whichh ierarchical porosity,h igh nitrogen doping, and conductivity suggested they may also be useful as supercapacitors. [33] These carbons were prepared by pyrolyzing highly crystalline magnesium-nitrilotriacetate metal-organic frameworks (MOFs). The robust and sustainable synthesis methodu ses safe and abundant materials and is performed on am ultigram scale.
We now report that the capacitance of such carbons is nearly tripled by orthogonally optimizing the carbon structure and the surface chemistry.F irst, we studied the microstructurea nd doping of the bulk carbon by tweaking the temperature and sequence of the thermalt reatments. Second, we functionalized the surfaceb yasimple yet powerful treatment in mineral acids. The surfacef unctionalities evolve with time and temperature of the oxidative treatments. We found that the pre-existing nitrogen dopants play an important role in the formation and pseudocapacitive contribution of the new surface functionalities.

Results and Discussion
Capacitance and microstructure Supercapacitor carbons are often produced by pyrolysis at temperatures of approximately 700 8C. This temperature is high enough to carbonize the precursorsy et low enough to conserve the heterodopants. Pyrolysis of magnesium nitrilotriacetic acid (MgNTA) at 700 8Cyields acarbon with high specific surfacea rea (SSA;8 30 m 2 g À1 )a nd high nitrogen content (11at%), yet moderate specific capacitance (62 Fg À1 in 1 m H 2 SO 4 at 5mVs À1 ). The low capacitance may result from insufficient micropore volume( 0.27 mL g À1 ,o ut of at otal pore volumeo f1 .05 mL g À1 ), from low conductivity,o rf rom as mall number of pseudocapacitive sites. The pyrolysis temperature was increased up to 1000 8C( samples denoted NC-700 to NC-1000, Table 1). Both the micropore volume andS SA increased to 0.54 mL g À1 and 1606 m 2 g À1 ,r espectively,w hen the samples were pyrolyzed at 900 8C. This increase was attributed to the evaporation of tar from nanometric interstices between graphene sheets. [27] As ar esult,t he carbon's specific capacitance was boosted by 55 %upt o9 6Fg À1 .
The pyrolysis temperature has ac omplex effect on the SSA, micropore volume, nitrogen content, and specific capacity ( Table 1). For example, raising the pyrolysis temperature to 1000 8Cc auses micropore shrinkage, yet an increase in capacitance. However,h eatingt he carbon to 1000 8Ca fter the MgO had been washed out yielded the bestp orosity:S SA of 1830 m 2 g À1 and micropore volume of 0.61 mL g À1 (3.10 mL g À1 total pore volume). Carbons heated in this mannera re denoted with an asterisk. Possibly,t he removal of MgO eases the clearing of amorphous goo and anneals the surface by rearranging the more labile kinks (a similar effect is seen with mixed oxides). [34] However,M gO may also stabilize the presence of Nc ontent;s ample NC-1000 has 11.0 at %n itrogen, whereas NC-900* has only 4.1 at %nitrogen.
The highest capacitance (117-118 Fg À1 )i nt he heat-treatment series was obtained from the two carbons heatedt o 1000 8C( NC-1000 and NC-900*). The shape of the voltammograms of these samples is closest to as quare shape ( Figure 2 and FigureS1i nt he Supporting Information). This indicates a fast and homogeneous charging of the carbon electrodes. The fact that these were not necessarilyt he carbons with the highest SSA emphasizes the importance of other factors. These include conductivity (as determined by degree of graphitization) and surfacer edox chemistry introducing slower pseudocapacitive reactions.

Capacitancea nd surfacechemistry
Next, we set out to improve the PC of our carbons. Perd efinition, PC depends on fast redox reactions between incoming  (ionic) speciesa nd surface functional groups. Pristine (i.e.,u ndoped) carbon doesn'ts how any PC, but nitrogen-doped carbon has PC activity. [29,35] In our carbons, ah int of redox activity is evident in the broad, reversible "hump"i nt he cyclic voltammogram (see for example Figure 2a;NC-1000). In our acidic aqueous environment, the most commoni onic speciesa re hydronium( H 3 O + )i ons. Therefore, we tried to create surface sites that would facilitatep roton/electron transfer reactions with thesei ons. To do this, we used a1:4 mixture of concentrated nitric acid and sulfuric acid. Nitric acid is a strong oxidizing and nitratinga gent at low temperatures, [36,37] especially in the presence of sulfuric acid. [31] Recently,L in et al. treated an itrogen-doped carbon in hot nitric acid, boosting its PC through an acid-bases urface reaction. [29] This is the only example of such at reatment for an itrogen-dopedc arbon that we know.
We then ran two sets of experiments,a t08Ca nd at 20 8C, for up to 18 h. Control experiments showed that longer and hotterr uns led to ac ollapse of the structure. Samples are denoted as NC-xx-yy,w here xx is the treatment temperature (0 or 20 8C) and yy is the treatment time (e.g.,1h).
Cyclic voltammetry of the carbonsi n1m H 2 SO 4 showed that the oxidative treatments considerably increased the specific capacitance (Figure3). First, the voltammogram shapes changed from square to more round. This indicates as lower charge transfer during electrode polarization reversal, which may be related to surface functionalization through severalm echanisms. These include the introduction of slow-charging pseudocapacitive sites, the blocking of micropores, and increased electron localization (lower conductivity). Ab road peak appeared at 0.5 Vv s. saturated calomel electrode (SCE) after the treatment (Figure 3a and S3). This reversible peak corresponds  to an ew redox process occurring at the carbon surface. The width of the peak-which is probably relatedt oabroad distribution of pseudocapacitive surfaces ites [9,29] -does not allow precise integration. [38] However,i fw er ule out contributionsb y EDLC, we can assign the entire increaset oP Ca nd estimate its contribution to total capacitance (see below).
The temperature and time of the treatment have ap rofound effect on the capacitance (Figure 3b and ca nd Ta ble 2). At 0 8C, the capacitance increases with treatment time until it peaks at 4h.A t2 08C, even 1his sufficient to boost the capacitance to its maximum value. The maximum capacitances were 158-168Fg À1 at 5mVs À1 ,a ni ncrease of approximately 40 % over the untreated nitrogen-doped carbon. This carbon retained its activity over several cycles and faster scan rates (Figure S2), including at 10 (133 Fg À1 )a nd 25 mV s À1 (81 Fg À1 ). The lower response at faster scan rates corresponds to slower charge-discharge processes similart ot he change in the shape of the voltammogram (see above).
The rise and fall of capacitance versus treatment time indicates that two opposing factors are at play.T ou nderstand these factors, we must first understand how microstructure and surface chemistry change witht he surface treatments. First, let us consider EDLC, which depends strongly on surface area and pore structure. These were studied using N 2 sorption porosimetry at 77 K( Ta ble 2), combined with two-parameter Brunauer-Emmett-Teller (BET2)a nalysis of the specific surface area and Dubinin-Radushkevitch analysis of the micropore volume. Interestingly,t he surfacet reatments decreased the carbon's SSA (by as much as 30 %w hen treated at 0 8C). The main effect of the treatment was ad ecrease in the micropore volume. This could reflect the blocking of micropores by newly grafteds urface functionalities [36,39] and/or the etching of micropore walls, which turns them into mesopores. [40] Thes econd explanation, however,i sh ard to reconcile with the N 2 sorption isotherm; the hysteresis in the isotherm,a ssociated with mesopore content, does not increase ( Figure S4). Thus, the most probable explanation for the decrease in SSA and in micropore volumei st he blocking of pores by new surface groups. Importantly,t he improvement in capacitance cannotb ea ttributed to an increase in EDLC charge storage because the SSA actually decreases. Therefore, PC contributes at least 45 %o ft he capacitance of NC-0C-4h-thee ntire rise effectedb yt he treatments.
To explain this boost in capacitance, we examined the changes to the surfacec hemistryd uring the treatment. The oxygen/carbon ratio increased with treatment time (Table 2), indicating surfaceo xidation. The oxygen content reached saturation without any change to the shape of the O1sX -ray photoelectron spectroscopy (XPS) peak ( Figure S5). This suggested that the oxygen-based functional groups saturated the surface. Beyondt his point, further treatment led to carbon deterioration as observedf rom the total disappearance of carbon after prolonged/hot oxidations.
To study how the acidic treatment affects the distribution of nitrogen functionalities in the carbon, XPS spectra of the N1sr egion were mathematically deconvoluted and their areas were compared ( Table 2). The deconvoluted spectra for the untreated carbon and the best-treated carbon are presented in Figure 4. The nitrogen dopantsi nt hese carbons include graphitic (401.1 eV), pyridinic (398.3 eV), and pyridonic/pyrrolic nitrogen atoms  (399.7 eV);v arious oxidized nitrogen atoms (402-404 eV);a nd nitro groups (405.7 eV). [41] The evolution of nitrogen contentw ith treatment time is shown in Figure 5. The most common nitrogen functionalities were graphitic and pyridinic,w hich were presenti na ll pyrolysis-derived carbons. Both of these groups decrease in relative intensity throughout the treatments (Figure 5a and b), from approximately 2.0 at %t o1 .0 at %. This decrease was the most pronounced and smooth in the 20 8Ct reatments. In contrast, only pyridinic nitrogen was removed in the 0 8Ct reatment, whereas the graphitic nitrogen fraction remained roughlyc onstant (2.0-2.6 at %). This process was attributed to the etching, as discussed earlier. Aw armer and/or prolongedt reatment is more effective in etching, as seen earlier in the saturation values of the oxygen/carbon ratio. Pyridinic nitrogen atoms at plane edges (Figure 1) were the first to leave. Hence, even the colder treatment was sufficient for their removal.
Etching was not the only process driving the decrease in the pyridinic nitrogen fraction.O xidation of the vicinal carbon atoms formed ap yridonic group ( Figure 6) with higher N1s binding energies. The increasei na mount of this group throughout the 20 8Ct reatment is evident in Figure 5c.I nthe 0 8Ct reatment, the pyridonic nitrogen fraction also increases and finally drops. This drop coincides with ad rop in the total nitrogen fraction and in specific capacitance after prolonged treatment. The assignment of the 399.7 eV peak is often challenging [41] as this is the binding energy of both pyridonic and pyrrolic nitrogen atoms. However,p yrrolic groups cannot form by treatment at room temperature. [42] Thus, the increasei nt his peak wasa ssigned to the formation of pyridonicn itrogen atoms by the oxidation of ac arbon atom vicinal to ap yridinic nitrogen atom.
Interestingly,t he pyridonic groups were not the only new groups resulting from the acid treatment.
The new peak at 405.7 eV revealed the formation of nitro groups ( Figure 6). This is in agreement with previous studies, in which nitric/sulfuric acid treatments led to pronouncedn itro peaks in the XPS spectra. [43,44] These groups grew steadily in relative fraction duringt he 0 8Ct reatment and passed through am aximum during the 20 8Ct reatment (Figure 5d). This indicates that the nitro groups were not the most stable oxidationp roductsa nd suggested that optimization of the treatmentt ime is crucial for obtaining the best conductivity.

Conclusions
In this work, we demonstratedt hat nitrogen-doped, hierarchically porousc arbonsd erived from Mg-based metal-organic frameworks (MOFs) can be used as electrochemical capacitors. The structure ands urface chemistry of these carbonsc an be optimized orthogonally.S electing an optimal pyrolysist emperature, or combining ap yrolysis processw ith as econd thermal process, is crucial for identifying an optimal combinationo f surfacea rea, pore structure, and dopant concentration.R elative to at ypical one-step pyrolysis at 700 8C( specific capacitance 62 Fg À1 ), the two-step heat treatment (900 8C/washing/ 1000 8C) nearly doubled the specific capacitance (117 Fg À1 ). Next, treating the carbon with a1:4 mixture of nitric ands ulfuric acid introduced av ariety of oxygen-and nitrogen-oxygen surfacef unctionalities. These groups boosted the pseudocapa-  citive charge storagem echanisms, increasing the capacitance up to 168 Fg À1 -over 40 %i mprovement just from the treatments and nearly 200 %r elative to the original carbon.T he effect of the surfacet reatment was complex, includinge tching out pyridinic andg raphitic nitrogen atoms while introducing new functional groups such as pyridonic and nitro groups.T he density of the pyridonic groups reached saturation as the treatment progressed, whereas the specific capacitance peakeda nd then dropped. The positive effect of the surface functionalities waseventually outweighed, and capacitance declined. This may reflect the blocking of micropores by functional groups [39] (less electrochemical double-layerc apacitance) or ad ecrease in conductivity caused by oxygenation. [45,46] Experimental Section

Materials and instrumentation
Unless stated otherwise, chemicals were purchased from either Sigma-Aldrich or Alfa Aesar,a nd used without further purification. N 2 adsorption-desorption isotherms were measured on aT hermo Scientific Surfer instrument at 77 K, using vacuum-dried samples (200 8C/3 h). Isotherms were analyzed by the ThermoFischer Advanced Data Processing 6.0 software, using the BET2 model for specific surface area, the Dubinin-Radushkevitch model for micropore volume. The XPS measurements were performed using aP HI VersaProbe II scanning XPS microprobe (Physical Instruments AG, Germany). Analysis was performed using am onochromatic AlK a X-ray source with ap ower of 24.8 Wa nd ab eam size of 100 mm. The spherical capacitor analyzer was set at a4 5 8 take-off angle with respect to the sample surface. The pass energy was 46.95 eV,y ielding af ull width at half maximum of 0.91 eV for the Ag 3d 5/2 peak. Peaks were calibrated using the C1sp osition. Curve fitting was performed using XPSPeak 4.1.

Procedure for carbon synthesis
Nitrogen-doped carbon was prepared using the method reported previously. [25][26][27] Briefly,n itrilotriacetic acid (N(CH 2 COOH) 3 ;N TA)w as mixed with basic magnesium carbonate ((MgCO 3 ) 4 Mg(OH) 2 )a t1 :1 Mg/NTAr atio in deionized (DI) water at 85 8C. The solid was precipitated by slowly adding an excess of ethanol and then chilling in an ice bath for 2h.T he resulting white paste was scraped out, vacuum dried at 40 8Cf or 48 h, and ground in an automated grinder.T he resulting hard white powder was pyrolyzed in argon at temperatures of 700-1000 8C( samples denoted herein as NC-700 to NC-1000, respectively). The MgO nanoparticles were washed by stirring overnight in 0.5 m citric acid. The carbons were dried at 130 8C. In some cases, the carbons were further heat-treated in argon to 1000 8C/1 h.

Procedure for surfacetreatment
Surface treatments were performed by adding 10 mL of H 2 SO 4 (96 %, 178 mmol) to 80 mg of NC-900* in a2 5mLr ound-bottom flask. This mixture was stabilized at the desired temperature (0 or 20 8C). Subsequently,2 .5 mL of HNO 3 (65 %a q.,3 7mmol) was added in ten portions while stirring over 15 min. The resulting suspension was stirred for the indicated time (1-18 h). The reaction was quenched by pouring the reaction mixture in 300 mL ice water and washed on am obile phase filtration setup using an ylon filter (0.45 mmp ore size) with 1L of DI water.T he sample was dried at 80 8Cf or 24 h.

Electrochemical experiments
Electrodes were prepared using a8 0:10:10 mass ratio of carbon sample/PTFE/carbon black (PTFE = polytetrafluoroethylene). In a typical preparation, the carbon sample (16 mg) was mixed with carbon black (2 mg, 50 %c ompressed) in am ortar.P TFE (2.2 mL, 60 %s uspension in water,2mg PTFE) was then added with water (75 mL). This slurry was ground in am ortar and pressed to af ilm. The films were then cut into squares of roughly 1cm 2 .The working electrode consisted of carbon films pressed between a0 .125 mm tantalum plate and Whatman filter paper using the electrode press. The layers were pressed together in as pecialized testing clamp, 3D-printed from high impact polystyrene ( Figure S6). Cyclic voltammetry experiments were performed on aG amry Instruments Reference 600 potentiostat using at hree-electrode setup. AS CE was used as ar eference in at hree-electrode cell, with ag raphite rod serving as the counter electrode.