Journal of Power Sources Advances 1 (2020) 100002

Contents lists available at ScienceDirect

Journal of Power Sources Advances

journal homepage: www.journals.elsevier.com/journal-of-power-sources-advances

Catalytically graphitized freestanding carbon foams for 3DLi-ion microbatteries

Antonia Kotronia a, Habtom Desta Asfaw a, Cheuk-Wai Tai b, Kristina Edstr€om a,Daniel Brandell a,*

a Department of Chemistry-Ångstr€om Laboratory, Uppsala University, Box 538, SE-75121, Uppsala, Swedenb Department of Materials and Environmental Chemistry, Arrhenius Laboratory, Stockholm University, SE-10691, Stockholm, Sweden

A R T I C L E I N F O

Keywords:EmulsionPolymerCarbonGraphitic foamThree-dimensionalLi-ion battery

* Corresponding author.E-mail address: daniel.brandell@kemi.uu.se (D.

https://doi.org/10.1016/j.powera.2020.100002Received 10 November 2019; Received in revised f2666-2485/© 2020 The Author(s). Published by Elsnc-nd/4.0/).

A B S T R A C T

A long-range graphitic ordering in carbon anodes is desirable since it facilitates Liþ transport within the structureand minimizes irreversible capacity loss. This is of vital concern in porous carbon electrodes that exhibit highsurface areas and porosity, and are used in 3D microbatteries. To date, it remains a challenge to graphitize carbonstructures with extensive microporosity, since the two properties are considered to be mutually exclusive. In thisarticle, carbon foams with enhanced graphitic ordering are successfully synthesized, while maintaining theirbicontinuous porous microstructures. The carbon foams are synthesized from high internal phase emulsion-templated polymers, carbonized at 1000 �C and subsequently graphitized at 2200 �C. The key to enhancing thegraphitization of the bespoke carbon foams is the incorporation of Ca- and Mg-based salts at early stages in thesynthesis. The carbon foams graphitized in the presence of these salts exhibit higher gravimetric capacities whencycled at a specific current of 10mA g�1 (140 mAh g�1) compared to a reference foam (105 mAh g�1), whichamounts to 33% increase.

1. Introduction

Lithium-ion batteries (LIBs) are currently among the most popularenergy storage technologies and have found use in applications rangingfrom consumer electronics to large-scale grid energy storage [1–3].Another emerging area of application for LIBs is microelectronics, such asmedical implants and internet-of-things (IoT) devices wherein autono-mous energy storage is sought. Because the miniaturization of electronicdevices limits the space available for the batteries, it is essential toconstruct suitable three-dimensional (3D) electrode architectures thatcan pack as much energy as required without affecting power perfor-mance [4–6]. To that end, considerable research efforts have beendirected towards the creation of micro- and nanostructured electrodedesigns that can maximize the surface area available for Li-ion interca-lation, shorten the diffusion paths for Li-ion transport and render thecurrent distribution inside the cell more homogeneous [2,7–9].Three-dimensional electrode architectures that have been considered sofar include micro- and nanopillar arrays [10–14], interconnected gra-phene sheets [15–18] and foams [19–29]. Carbon-based monolithic 3Delectrodes have been suggested for potential use inmicrobatteries as both

Brandell).

orm 23 January 2020; Acceptedevier Ltd. This is an open access a

active materials and current collectors [10,24,30,31]. This multifunc-tional role of carbon is associated to its capability to reversibly hostlithium-ions, its high electronic conductivity and the high interfacialsurface area it possesses [32]. In contrast to graphite, disorderedcarbon-based 3D electrodes suffer from significant irreversible capacityloss during the first discharge and this hampers their eventual use incommercial cells. By increasing the degree of graphitization of carbon,the coulombic efficiency and reversible capacity of the 3D electrode canbe improved [33–35].

The synthesis of carbons with an optimal porosity and a highgraphitic content is, however, challenging as the presence of pores cre-ates defects in the carbon structure and inhibits complete graphitization[36,37]. The difficulty to substantially graphitize microporous carbonswas initially attributed to the presence of diamond-like sp3-crosslinks[38] and later to ribbon- [39,40] or fullerene-like structures [36,37,41].The latter hypothesis was supported by evidence from X-ray, EELS [42]and aberration-corrected HRTEM data [43]. More specifically, the pres-ence of such structures in the precursor prevents the development oflarge, well-aligned graphitic crystallites in the resulting material, therebyreducing the reversible lithiation capacity based on intercalation. The

23 January 2020rticle under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-

Fig. 1. Schematic summary of the procedure for the synthesis of polyHIPEpolymers: high internal phase emulsion (HIPE) preparation, polymerization at65oC (1), evaporation of the internal phase (water droplets) to form highlyporous and networked polyHIPE (2), and subsequently chemical functionaliza-tion, pyrolysis and catalytic graphitization (3) which yield monolithic carbonfoams that are suited for 3D micro-battery applications. Photos of the polyHIPEpolymer and carboHIPE foam are given on top of the schematic.

A. Kotronia et al. Journal of Power Sources Advances 1 (2020) 100002

effect of grain boundaries and crystallite misalignment has been previ-ously studied and found to result in a significantly lower diffusion coef-ficient for the Li-ion. For instance, the diffusion coefficient obtained inartificial graphite, where the Li-ion moves along the graphene planes, issuperior to that of hard carbons, where crossing of grain boundaries withdifferent orientations becomes a necessity [44]. Carbonaceous materialswith a large fraction of misaligned graphitic crystallites are hence likelyto exhibit accessibility issues, especially for the crystallites residing in thebulk material. Further, another substantial issue is that extensivelymicroporous, carbonaceous precursors do not normally result in theformation of perfectly ordered hexagonal “h-graphite”. Domains of tur-bostratically disordered carbon, henceforth referred to as “t-carbon”, areinstead formed, in which a fraction of the ABAB structure of h-graphitecan be both translated and rotated with respect to the normal. Thistranslation/rotation of the graphene layers implies changes in theinterlayer spacing and the curvature of the domains. As regards theelectrochemical performance, t-carbon is known to deliver lowerreversible lithiation capacities than h-graphite. The stacking faults pre-sent in t-carbon lead to the blocked galleries, which in turn impair thedevelopment of staged phases with long-range order [45,46]. Increasingthe fraction of h-graphite in the carbonaceous material and improvingthe alignment between adjacent crystallites can thus lead to a significantenhancement of the lithiation capacity.

In a recent publication, it was demonstrated that carbon foamsderived from high internal phase emulsions exhibit significantimprovement in their electrochemical performance as a result of heattreatment at 2200 �C [47]. The partially graphitized foam deliveredreversible gravimetric capacities of about 100 mAh g�1 for a specificcurrent of 10mA g�1. A key aspect of that work was that the monolithic,porous structure of the carbon foam was preserved after the graphitiza-tion step, with promising electrochemical performance for use in 3Dmicrobatteries. The degree of graphitization depends on several param-eters like the type of carbon precursor [48–50], heat treatment [49],applied pressure [51], stress [52–55] and inclusion of catalysts [50,56].Regarding the latter, alkaline earth metals bound in minerals such as CaOand CaCO3 [57,58] have a graphitizing effect and are reported to initiategraphitic growth at temperatures as low as 1500 �C in N2 at atmosphericpressure, and at temperatures close to 1000 �C at pressures of 0.3MPa.Inclusion of Ca(OH)2 in the carbon matrix lowers the temperature limitfor graphite formation to a mere 800 �C [59].

This study explores the graphitization of emulsion-templated carbonfoams with the help of CaCl2 and MgCl2 salts, and their potential use aselectrodes in 3D-microbatteries. These salts are initially introduced tostabilize the high internal phase emulsion (HIPE) and prevent dropletcoalescence during the polymerization step, and are removed during thewashing step to prepare clean polymer foams. In this particular study,however, they are purposefully left in the polymers and employed aspotential graphitizing agents during pyrolysis. The full synthesis pro-cedure is summarized in Fig. 1.

Thorough characterizations are conducted to understand whether theincorporation of these salts leads to an enhanced graphitic content andgraphitic order. To probe the structural characteristics, X-ray diffraction(XRD), Raman spectroscopy and transmission electronmicroscopy (TEM)are employed, while the electrochemical performance is evaluated in Lihalf-cells using voltammetric and chronopotentiometric techniques.Changes in the sample morphology, porosity and surface area are trackedby the means of scanning electron microscopy (SEM) and nitrogen gassorption. The purity of the graphitized samples is examined using ther-mogravimetric analysis (TGA), X-ray photoelectron spectroscopy (XPS)and TEM.

2. Experimental

Synthesis of graphitic carbon foams: The polystyrene foam (polyHIPE)used as precursor for the fabrication of the graphitic carbon foam issynthesized from a water-in-oil high internal phase emulsion (HIPE). An

2

HIPE forms when a hydrophobic organic phase is blended with anaqueous phase of much larger volume (>80%) in the presence of suitablesurfactants. The organic phase in this case consisted of styrene (3mL,Aldrich®, �99%), divinylbenzene (1.5mL, Aldrich®, 80%), 4-vinyl-benzyl chloride (0.5mL, Fluka®, �90%) and the surfactant Span 80(1.5 g, Aldrich®). All monomers and reagents were used as-received. Theaqueous phase contained a stabilizing salt, which was either CaCl2⋅6H2O(0.825 g, Aldrich®, �99%) or an equimolar amount of MgCl2⋅6H2O(Bioxtra, Aldrich®, �99%), and the initiator K2S2O8 (0.075 g, Merck99%) dissolved in deionized water (30mL). The aqueous phase wasadded drop-wise into the organic phase under constant stirring. Thestirring speed was gradually increased to a maximum of 2000 rpm. Afterstirring for 30min, the emulsion was transferred to an oven and under-went polymerization for 48 h at 65 �C. Samples of the synthesized poly-mers were washed in order to extract the stabilizing salts. The stabilizingsalt CaCl2⋅6H2O was leached out by immersing pieces of the polymersfirst in distilled water (100 �C, 24 h) and then in ethanol (85 �C by meansof a Soxhlet apparatus, 48 h). The remaining samples containingCaCl2⋅6H2O and MgCl2⋅6H2O, respectively, were left intact until thefunctionalization step. All polymers were functionalized by immersion inconcentrated sulfuric acid (95–97%, Emsure®, Merck) at 100 �C and for24 h. This was followed by washing with distilled water, drying at 80 �Covernight and finally under vacuum (100 �C, 24 h) [24,31]. The sulfo-nated foams were first carbonized by step-wise pyrolysis reaching amaximum temperature of 1000 �C under an Ar-atmosphere. The finalgraphitization step involved heating the samples to 2200 �C in a graphitefurnace (Thermal Technology) (in N2) for 1 h.

Materials characterization: TGA was performed using a DQ 500 in-strument (TA Instruments). The carbon foams were ground to finepowders and a quantity of 2–4mg was distributed evenly in the aluminacrucible. The heating took place in air, between RT and 800 �C at a rate of5 �C min�1. SEM images of the carbon foams were acquired with a ZeissMerlin microscope. The in-lens detector provided surface sensitive im-aging at a 3.0 kV acceleration voltage and 100 pA probe current. Furtherinsight into the extent of graphitization was provided using high reso-lution TEM. A JEOL JEM-2100 field-emission microscope equipped witha Gatan Ultrascan 1000 CCD camera was operated at 200 kV to collectTEM images of the three samples. Powders of each sample were dispersedand ground in ethanol. Certain drops of the dispersion were cast ontocarbon-coated Cu-grids, which were mounted on the TEM holder and

A. Kotronia et al. Journal of Power Sources Advances 1 (2020) 100002

introduced into the microscope for analysis. The porosity and Brunauer-Emmett-Teller (BET) surface area were estimated using N2 gas sorptionwith a Micromeritics ASAP 2020 analyzer at 77 K. The amount of sampleused varied in the range 0.1–0.3 g. The samples were degassed prior toanalysis under vacuum (250 �C, 4 h). Raman spectra of the carbon foamswere obtained with a Renishaw Ramanscope equipped with a Lieca LMoptical microscope, a CCD camera and a 532 nm laser. A focusingmagnification of 50� was used and internal calibration was performedon a Si standard prior to the measurement. During the run, 40 sweepsbetween 400 and 3200 cm�1 were collected with an exposure time of 20 sand at 0.5% laser power. Diffraction patterns were collected in the 2θrange 10–90�, using a step size of 0.012� (1.5 s per step) with a Bruker D8Advance Powder diffractometer in reflection mode using Cu Kα radia-tion. Prior to the XRD measurements, the carbonized and graphitizedfoams were ground to fine powders in ethanol and dispersed onto a flat Sisample holder. XPS measurements were conducted with an in-house PHI5500 spectrometer usingmonochromatic Al Kα radiation (1487 eV) and a45� electron emission angle. The acquired spectra were energy-calibratedwith respect to the hydrocarbon peak, fixed at the reference value of284.8 eV. Cycled samples were washed with dimethyl carbonate (DMC)in an Ar-filled glove-box and brought to the spectrometer with an inerttransfer shuttle.

Electrochemical measurements: The carbon foams were cut into regularshapes and polished to a thickness ranging from 200 to 300 μm in orderto ensure comparable electrolyte penetration and active material utili-zation. A washing step with 1M HCl (Analar Normapur®, VWR) andsubsequent sonication in distilled water and ethanol were used to cleanthe foams and ensure that residual impurities were not clogging thepores. All electrodes were dried under vacuum inside a Büchi oven in anAr-filled glove-box (120 �C, 12 h). Pouch cells were assembled usingmetallic lithium counter and reference electrodes (Cyprus Foote Min-eral), glass-fiber separators (Whatman) soaked in LP40 (1M LiPF6 in 1:1ratio of 1:1 v/v EC:DEC) and graphitized carbon foams as workingelectrodes. The carbon foam electrodes were soaked in LP40 for 10minprior to assembly, to ensure that the electrolyte permeates the pores. ABiologic VMP2 potentiostat was used for cyclic voltammetry and an

Fig. 2. (a) From bottom to top: XPS of carboHIPE, carboHIPE-Ca and carboHIPE-M

3

Arbin BT2000 battery tester for galvanostatic cycling.

3. Results and discussion

3.1. Structure and morphology

To shed light on the impact of heat treatment and inclusion ofgraphitizing salts on the carbon foam properties, powders of each samplewere studied using XPS, XRD and Raman scattering analyses. Thesetechniques can provide adequate evidence as regards which allotrope ofcarbon and chemical bonding exist in the samples. The three sets ofsamples will be denoted as carboHIPE, carboHIPE-Ca, and carboHIPE-Mgreferring to carbon foams with (i) no graphitizing additives, (ii) Ca-, and(iii) Mg-based graphitizing salt. The XPS and Raman spectra provided inFig. 2 along with the XRD patterns, show that considerable physico-chemical changes occur depending on the choice of graphitizing agent.

The graphitic component of the carbon foams is clearly visible in allsamples in the C 1s X-ray photoelectron spectra (Fig. 2a), which was fixedto a binding energy of 284.8 eV. The main difference between the threesamples is that the use of the Mg- and Ca-based graphitizing salts leads toa more oxidized surface. This is supported by the increase in intensity ofthe peaks corresponding to -C-OH/-C-O- and -CO2/-OCO2 moieties in thecarboHIPE-Ca and carboHIPE-Mg spectra.

The Raman spectra shown in Fig. 2b display first- order and doubleresonance scattering bands that are well-separated and sharper thanthose of heavily disordered carbon materials [47,60]. These bands alongwith the G-band position at about 1586 cm�1 are indicative of the pres-ence of graphitic carbon [61]. An estimate of the degree of graphitizationcan be expressed by the intensity ratio between the peak heights of the Dand the G bands [61–65]. The calculated ID/IG ratios are 1.53, 1.22 and1.52 for carboHIPE-Mg, carboHIPE-Ca and carboHIPE, respectively,suggesting that carboHIPE-Ca is the most graphitic sample. The 2D bandat ca. 2684 cm�1 is also slightly more pronounced for the carboHIPE-Casample, which is indicative of exfoliation and/or formation of smallergraphitic domains.

The XRD of the carboHIPE sample (shown in Fig. 2c and Fig. S1)

g (b) equivalent Raman spectra and (c) corresponding X-ray diffractograms.

Fig. 3. In column (a) SEM and TEM imaging of carboHIPE. In columns (b) and (c) corresponding SEM and TEM images from carboHIPE-Mg and carboHIPE-Ca,respectively.

A. Kotronia et al. Journal of Power Sources Advances 1 (2020) 100002

exhibits a single, sharp diffraction peak at approximately 26.0� due to the(002) reflection for t-carbon (T-component). This corresponds to aninterlayer spacing of 0.343 nm along the c-axis, which is slightlyexpanded compared to h-graphite (0.335 nm). The T-component isvisible at 26.0� for carboHIPE-Ca and carboHIPE-Mg as well. Addition-ally, a second sharp (002) reflection appears around 26.5�–26.6� forcarboHIPE-Ca and carboHIPE-Mg. This peak corresponds to h-graphitewith smaller interlayer spacing (0.335 nm) and is referred to as “G-component” [66]. The G-component is indicative of heterogeneous,catalytic graphitization [59]. The G-component is absent in the carbo-HIPE sample, but increases substantially for carboHIPE-Ca andcarboHIPE-Mg. Apart from the T and G-components, a broad peak isobserved for all samples, denoted as the “A-component” which is alsocharacteristic of disordered carbon materials [38,66]. The intensity ofthis feature decreases in the following order: carboHIPE, carboHIPE-Mgand carboHIPE-Ca, which in turn indicates that carboHIPE-Ca is themost graphitized sample. In addition, the increased degree of graphiti-zation is further confirmed by the appearance of the (010) reflection at43.5�, which can be evidenced in the full diffraction pattern (Fig. S1).Further insight into the impact of graphitization on microstructure andmorphology of these samples was acquired through SEM, TEM and N2gas-sorption measurements.

The SEM micrographs for all three graphitized carbon foams areprovided in Fig. 3, with further imaging being supplied in Fig. S10 and

4

Fig. S11. Notably, all carbon foams retain the bicontinuous architectureeven after treatment at 2200 oC, with open voids on the microscale range.This architecture allows good electrolyte percolation throughout theporous electrode and provides large surface area for Li-ion intercalation.The BET specific surface areas of the graphitized foams (10, 12 and8m2 g�1 for the graphitized carboHIPE, carboHIPE-Ca and carboHIPE-Mg, respectively) do not differ significantly from each other and aremuch lower than those of the carbon foams synthesized at 1000 �C (545,725 and 399m2 g�1, respectively), as can be seen in Table S2.

In agreement with the XRD data discussed above, the TEM imagesshown in Fig. 3 suggest that all three samples are partially graphitic.Accordingly, it can be concluded that the graphitic stacking in the car-boHIPE sample does not exceed more than 5–10 graphene layers perdomain. In the case of carboHIPE-Ca and carboHIPE-Mg, ribbons or shellsof graphitic layers (G-component) are observed around what appears tobe the salt seed crystal. This is confirmed both by the detection of saltresidues in the XRD pattern of carboHIPE-Ca (Fig. S1) and through EELSmeasurements performed on the salt crystal (Fig. S16). The graphiticordering in the case of carboHIPE-Ca and carboHIPE-Mg is of the long-range, ordered type, which is in good agreement with the appearanceof the G-component in the XRD pattern.

The mechanism of graphitization for these samples has not beeninvestigated in detail, but most probably involves the formation of theintermediary species CaC2 (or MgC2) and its reaction with the N2-

A. Kotronia et al. Journal of Power Sources Advances 1 (2020) 100002

atmosphere to produce gaseous calcium cyanamide and free carbon. Theanticipated reactions are designated by Equations (1)–(5).

CaCl2ðaqÞþH2SO4ðaqÞ→CaSO4ðsÞ þ 2HCl ðaqÞ (1)

CaSO4ðsÞþ 2C ðsÞ ↔ CaSðsÞ þ 2CO2ðgÞ (2)

2CaS ðsÞþ 5C ðsÞ ↔ 2CaC2ðsÞ þ CS2 ðgÞ (3)

CaC2ðsÞþN2ðgÞ → CaðCNÞ2ðsÞ (4)

CaðCNÞ2ðsÞ→ CaCN2 ðgÞ þ C ðsÞ (5)

The CaCl2 in the polymer foams is precipitated in the form of CaSO4,as shown in Equation (1), during the sulfonation step that is required toimpart thermal stability to the polymers. During the carbonization step(between 850 and 950 �C), CaSO4 is reduced to CaS, which is confirmedwith XRD (Fig. S2). Although no conclusive evidence for the conversionof CaS to CaC2 is presented here, its occurrence is highly probable, in asimilar way as for the well-studied reaction between CaO and carbon atelevated temperatures [67]. In the presence of the N2 atmosphere, theunstable carbide intermediate decomposes into gaseous calcium

Fig. 4. (a) CV of the 1st cycle for all carboHIPEs (b) galvanostatic charge-dischargecapacity and coulombic efficiency as a function of cycle number and (d) rate capabiliin the literature.

5

cyanamide and free carbon [58,68]. The free carbon bonds to the existinggraphitic layers and, since ordered graphite is the most thermodynami-cally stable configuration, it is preferentially formed [69]. The reactionmechanism is presumed to be similar for MgCl2. CaCl2 was initiallychosen since it already constituted a part of the polyHIPE synthesis.MgCl2 is chemically close to CaCl2, and was therefore anticipated to be agood candidate for emulsion stabilization. The lower thermal stability ofMgS as compared to CaS is likely to facilitate its conversion to the car-bide. MgS decomposes at 2000 �C in an inert atmosphere, while CaSdecomposes only after 2525 �C. Hence, in case the conversion of thesulfide to carbide would not go to completion, the Mg-containing carbonwould still be free from impurities. This appears to be the case based onthe observations in the TEM micrographs, since carboHIPE-Ca still con-tains salt traces while the carboHIPE-Mg does not exhibit any residues. Asignificant disadvantage of the Mg-based graphitization route, however,proved to be the high solubility of the salt in the MgSO4 form present inthe sulfonated polymer (36.4 g/100 g of water at 25 �C in comparison to0.21 g/100 g for CaSO4) [70]. Since a washing step is necessary aftersulfonation to remove the excess sulfuric acid before carbonization(Fig. 1, step 3), it is probable that a higher amount of the MgSO4 salt waswashed away in comparison to CaSO4. This would then leave theMg-based foam with a lesser amount of catalyst, which is hence most

voltage profiles of the 1st (solid line) and 15th (dotted line) cycle (c) dischargety test and performance comparison to notable examples of 3D carbon electrodes

A. Kotronia et al. Journal of Power Sources Advances 1 (2020) 100002

probably the reason why the resulting carbon did not graphitize to thesame extent as its Ca-based counterpart.

3.2. Electrochemical performance

The graphitized foams were investigated by the means of CV, recor-ded at a rate of 0.050mV s�1 and within a potential window of0.001–2.5 V vs Li/Liþ. In these scans, the staging nature of the Liþ

intercalation in ordered graphite becomes visible as indicated in Fig. 4a.While the CV of carboHIPE exhibits merely a sloping profile for the po-tentials at which Liþ intercalation takes place in graphite (0–0.25 V vs Li/Liþ), carboHIPE-Ca immediately shows four peaks. This is in goodagreement with the growth of well-graphitized domains observed withXRD, Raman and TEM. The CV of carboHIPE-Mg exhibits likewise peaksat the potentials for Liþ intercalation. These are, however, not equallydistinct, which supports the observation that this sample has fewer long-range ordered graphitic domains. Moreover, a portion of the observedcapacity originates from the A- and T-components, as these also have acertain lithiation capacity [33,45].

In Fig. 4b, the galvanostatic voltage-capacity profiles recorded at aspecific current of 10mA g�1 are given for the 1st and 15th cycle. Incontrast to the subsequent cycles, the initial discharge curves show aplateau around 0.8–1.0 V, which corresponds to electrolyte decomposi-tion and the formation of the solid electrolyte interphase (SEI). The SEIbuild-up limits the initial coulombic efficiency to 67% for carboHIPE-Mgand carboHIPE-Ca and to 77% for carboHIPE. The higher Li-ion con-sumption in the first formation cycle of the Mg- and Ca-based samples ismost likely due to the increased amount of surfaces functionalities inthese samples. The more oxidized nature of the two heterogeneouslycatalyzed foams in comparison to carboHIPE was observed with XPS, asdiscussed previously (Fig. 2a). The SEI appears however to be sufficientlystable upon repetitive cycling (Fig. S8), even though a change in itscomposition is observed over time, favoring the formation of LiF.Moreover, the absence of parasitic side-reactions is best seen from thefact that the coulombic efficiency for both samples increases quickly to100% in the subsequent cycles. Nevertheless, the carboHIPE-Ca andcarboHIPE-Mg samples exhibit higher gravimetric capacities than thecarboHIPE sample, which may be attributed to the appearance of largercrystalline domains, consisting of h-graphite. The carboHIPE sampleexhibits a reversible specific gravimetric capacity of approximately 105mAh g�1, a value which is in good agreement with previous works [47,60]. CarboHIPE-Ca and carboHIPE-Mg reach 140 and 130 mAh g�1

respectively, which corresponds to a 33% increase of the reversiblegravimetric capacity. It is important to note here that the capacity ob-tained for all samples is much lower than that expected for graphiteelectrodes (350–370 mAh g�1). As discussed in the introduction, this canbe explained by the complicated structure of the carbon foam, which iscomposed of domains with distinctly different characteristics. Forinstance, the domains corresponding to the A- and T-components areexpected to have a lower lithiation capacity than those corresponding tothe G-component. In addition, the nanocrystalline nature of these sam-ples increases the risk of misalignment between adjacent crystallites.Hence, a substantial part of the bulk of the carbon foam may be inac-cessible to the Li-ions during battery operation, thus reducing the prac-tically achievable capacity [28,29,71]. The rate capability of theelectrodes is shown in Fig. 4d along with notable examples of 3D carbonelectrodes in the literature [10,27–29,47]. As it may be deduced fromFig. 4d, the graphitic foams perform well both in terms of lithiation ca-pacity and rate capability in comparison to similar, 3D alternatives. It isalso worthwhile to highlight again the fact that the graphitic foam pre-sented in this work exhibits a high coulombic efficiency (close to 100%)and stable cycling after the first formation cycle. This is not necessarilythe case for other 3D carbonaceous electrodes, as many of the literatureexamples are not sufficiently graphitized and optimized with respect tothe available surface area and porosity. The latter type of carbonaceouselectrodes may achieve impressive initial capacities due to their large

6

surface areas and numerous heteroatom terminations; but they willinevitably suffer from rapid capacity fading.

4. Conclusions

This work has demonstrated that the graphitic content and order inemulsion-templated carbon foams can be enhanced through catalyticgraphitization based on the use of CaCl2 and MgCl2 additives. The resultsindicate that long-range, graphitic stacking occurred around the Ca- andMg-rich particles used as catalysts. X-ray diffraction, Raman scatteringand TEM analyses provided evidence of that carboHIPE-Ca andcarboHIPE-Mg had graphitized to a larger extent than carboHIPE.Nevertheless, both of the catalytically graphitized samples exhibitedcomplex structures, which were composed of varying fractions ofcompletely disordered carbon, t-carbon and h-graphite. These findingsreceived further support from the electrochemical characterization, asthe catalytically graphitized samples exhibited approximately 33%higher lithiation capacities in comparison to the carboHIPE sample. Thiswas in good agreement with the initial hypothesis, since h-graphiteshould lead to an increased reversible lithiation capacity, based on theprocess of intercalation. Nevertheless, the highest reversible gravimetriccapacity observed (140 mAh g�1, for carboHIPE-Ca) was still not com-parable with the theoretical capacity for pure graphite (372 mAh g�1), asthe random spatial orientation of the nanographitic crystallites canseverely impair the accessibility of the crystallites to the Li-ions, henceresulting in a lower practical capacity. Apart from the capacity increase,another noteworthy property of the graphitized carbon foams proved tobe their coulombic efficiency, which, in contrast to similar literaturealternatives, rose to approximately 100% within the course of the firstcycles. This is an additional advantage of the graphitization, since itensures that the energy storage mechanism will be mainly based on Li-ion intercalation, which in the case of graphite is a highly reversibleprocess. Also, it is worth noting that the monolithic, 3D network withmacroscale porosity remained intact upon graphitization, which rendersthe graphitized foams highly promising as electrode materials formicrobattery applications.

Declaration of competing interest

The authors declare no conflict of interest.

Acknowledgements

Financial support for the project was provided by the Swedish EnergyAgency (2015-009549) and the StandUp for Energy consortium. Theauthors acknowledge technical assistance and fruitful discussions fromVictoria Sternhagen, Henrik Eriksson, Pedro Berastegui, Maria Hahlinand Ashok S. Menon. The microstructural analyses included in this paperwere carried out using the electron microscopy facilities at myfab-LIMSof Uppsala University and the Electron Microscopy Center of Stock-holm University (through a grant from Knut and Alice WallenbergFoundation).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.powera.2020.100002.

References

[1] B.N. Loeffler, D. Bresser, S. Passerini, M. Copley, Johnson Matthey TechnologyReview 59 (2015) 34–44.

[2] Y. Wang, B. Liu, Q. Li, S. Cartmell, S. Ferrara, Z.D. Deng, J. Xiao, J. Power Sources286 (2015) 330–345.

[3] J.-K. Park, Principles and Applications of Lithium Secondary Batteries, John Wiley& Sons, 2012, pp. 3–6.

A. Kotronia et al. Journal of Power Sources Advances 1 (2020) 100002

[4] M. Valvo, M. Roberts, G. Oltean, B. Sun, D. Rehnlund, D. Brandell, L. Nyholm,T. Gustafsson, K. Edstr€om, J. Mater. Chem. 1 (2013) 9281–9293.

[5] K. Edstr€om, D. Brandell, T. Gustafsson, L. Nyholm, The Electrochemical SocietyInterface 20 (2011) 41–46.

[6] M. Roberts, P. Johns, J. Owen, D. Brandell, K. Edstrom, G. El Enany, C. Guery,D. Golodnitsky, M. Lacey, C. Lecoeur, J. Mater. Chem. 21 (2011) 9876–9890.

[7] S. Ferrari, M. Loveridge, S.D. Beattie, M. Jahn, R.J. Dashwood, R. Bhagat, J. PowerSources 286 (2015) 25–46.

[8] T.S. Arthur, D.J. Bates, N. Cirigliano, D.C. Johnson, P. Malati, J.M. Mosby, E. Perre,M.T. Rawls, A.L. Prieto, B. Dunn, MRS Bull. 36 (2011) 523–531.

[9] J.F. Oudenhoven, L. Baggetto, P.H. Notten, Advanced Energy Materials 1 (2011)10–33.

[10] N. Cirigliano, G. Sun, D. Membreno, P. Malati, C. Kim, B. Dunn, Energy Technol. 2(2014) 362–369.

[11] C. Wang, L. Taherabadi, G. Jia, M. Madou, Y. Yeh, B. Dunn, Electrochem. Solid StateLett. 7 (2004) A435–A438.

[12] S.K. Cheah, E. Perre, M. Rooth, M. Fondell, A. Hårsta, L. Nyholm, M. Boman,T.r. Gustafsson, J. Lu, P. Simon, Nano Lett. 9 (2009) 3230–3233.

[13] G. Oltean, L. Nyholm, K. Edstr€om, Electrochim. Acta 56 (2011) 3203–3208.[14] J.I. Hur, L.C. Smith, B. Dunn, Joule 2 (2018) 1187–1201.[15] Y. Tang, F. Huang, H. Bi, Z. Liu, D. Wan, J. Power Sources 203 (2012) 130–134.[16] Y. Ding, H. Ren, Y. Huang, F. Chang, P. Zhang, Mater. Res. Bull. 48 (2013)

3713–3716.[17] X. Cao, Y. Shi, W. Shi, X. Rui, Q. Yan, J. Kong, H. Zhang, Small 9 (2013) 3433–3438.[18] Z. Wang, X. Yu, W. He, Y.V. Kaneti, D. Han, Q. Sun, Y.-B. He, B. Xiang, ACS Appl.

Mater. Interfaces 8 (2016) 33399–33404.[19] A.P. Cohn, L. Oakes, R. Carter, S. Chatterjee, A.S. Westover, K. Share, C.L. Pint,

Nanoscale 6 (2014) 4669–4675.[20] Z. Zhou, H. Zhang, Y. Zhou, H. Qiao, A. Gurung, R. Naderi, H. Elbohy,

A.L. Smirnova, H. Lu, S. Chen, Sci. Rep. 7 (2017) 1440.[21] R. Paul, V. Etacheri, V.G. Pol, J. Hu, T.S. Fisher, RSC Adv. 6 (2016) 79734–79744.[22] J. Dong, Y. Xue, C. Zhang, Q. Weng, P. Dai, Y. Yang, M. Zhou, C. Li, Q. Cui, X. Kang,

Adv. Mater. 29 (2017) 1603692.[23] J. Mao, J. Iocozzia, J. Huang, K. Meng, Y. Lai, Z. Lin, Energy Environ. Sci. 11 (2018)

772–799.[24] H.D. Asfaw, M. Roberts, R. Younesi, K. Edstr€om, J. Mater. Chem. 1 (2013)

13750–13758.[25] H.D. Asfaw, M.R. Roberts, C.-W. Tai, R. Younesi, M. Valvo, L. Nyholm, K. Edstr€om,

Nanoscale 6 (2014) 8804–8813.[26] P. Johns, M. Roberts, J. Owen, J. Mater. Chem. 21 (2011) 10153–10159.[27] Z. Wang, F. Li, N.S. Ergang, A. Stein, ECS Transactions 6 (2008) 199–204.[28] N. Brun, S. Prabaharan, M. Morcrette, C. Sanchez, G. P�ecastaings, A. Derr�e,

A. Soum, H. Deleuze, M. Birot, R. Backov, Adv. Funct. Mater. 19 (2009) 3136–3145.[29] N. Brun, S.R.S. Prabaharan, C. Surcin, M. Morcrette, H. Deleuze, M. Birot, O. Babot,

M.-F. Achard, R. Backov, J. Phys. Chem. C 116 (2012) 1408–1421.[30] J.W. Long, B. Dunn, D.R. Rolison, H.S. White, Chem. Rev. 104 (2004) 4463–4492.[31] H.D. Asfaw, R. Younesi, M. Valvo, J. Maibach, J. Ångstr€om, C.W. Tai, Z. Bacsik,

M. Sahlberg, L. Nyholm, P.K. Edstr€om, ChemistrySelect 1 (2016) 784–792.[32] N. Nitta, F. Wu, J.T. Lee, G. Yushin, Mater. Today 18 (2015) 252–264.[33] K. Guerin, A. Fevrier-Bouvier, S. Flandrois, B. Simon, P. Biensan, Electrochim. Acta

45 (2000) 1607–1615.[34] H. Fujimoto, A. Mabuchi, K. Tokumitsu, T. Kasuh, J. Power Sources 54 (1995)

440–443.

7

[35] A. Sleigh, U. Von Sacken, Solid State Ionics 57 (1992) 99–102.[36] P.J. Harris, Int. Mater. Rev. 42 (1997) 206–218.[37] P.J. Harris, J. Mater. Sci. 48 (2013) 565–577.[38] R.E. Franklin, Proc. Roy. Soc. Lond. Math. Phys. Sci. 209 (1951) 196–218.[39] L. Ban, D. Crawford, H. Marsh, J. Appl. Crystallogr. 8 (1975) 415–420.[40] G. Jenkins, K. Kawamura, Nature 231 (1971) 175.[41] P. Harris, Phil. Mag. 84 (2004) 3159–3167.[42] Z.-l. Zhang, R. Brydson, Z. Aslam, S. Reddy, A. Brown, A. Westwood, B. Rand,

Carbon 49 (2011) 5049–5063.[43] P.J. Harris, Z. Liu, K. Suenaga, Journal of Physics: Conference Series, IOP

Publishing, 2010, p. 012050.[44] H.-j. Guo, X.-h. Li, X.-m. Zhang, H.-q. Wang, Z.-x. Wang, W.-j. Peng, N. Carbon

Mater. 22 (2007) 7–10.[45] T. Zheng, J. Dahn, Synth. Met. 73 (1995) 1–7.[46] T. Zheng, J.N. Reimers, J.R. Dahn, Phys. Rev. B 51 (1995) 734–741.[47] H.D. Asfaw, A. Kotronia, C.-W. Tai, L. Nyholm, K. Edstr€om, Energy Technol. 7

(2019) 1900797.[48] K.P. Gierszal, M. Jaroniec, T.-W. Kim, J. Kim, R. Ryoo, New J. Chem. 32 (2008)

981–993.[49] O. Fromm, A. Heckmann, U.C. Rodehorst, J. Frerichs, D. Becker, M. Winter,

T. Placke, Carbon 128 (2018) 147–163.[50] M. Johnson, K. Faber, J. Mater. Res. 26 (2011) 18–25.[51] O. Beyssac, J.-N. Rouzaud, B. Goffe, F. Brunet, C. Chopin, Contrib. Mineral. Petrol.

143 (2002) 19–31.[52] K. Matsui, L.J. Lanticse, Y. Tanabe, E. Yasuda, M. Endo, Carbon 7 (2005)

1577–1579.[53] L.J. Lanticse-Diaz, Y. Tanabe, T. Enami, K. Nakamura, M. Endo, E. Yasuda, Carbon

47 (2009) 974–980.[54] R.J. Zaldivar, G.S. Rellick, Carbon 29 (1991) 1155–1163.[55] M. Inagaki, F. Kang, M. Toyoda, H. Konno, 5. Stress graphitization, in: Advanced

Materials Science and Engineering of Carbon, Elsevier, pp. 88-109.[56] A. �Oya, S. �Otani, Carbon 17 (1979) 131–137.[57] C. Zhao, W. Wang, Z. Yu, H. Zhang, A. Wang, Y. Yang, J. Mater. Chem. 20 (2010)

976–980.[58] G. Yang, H. Han, T. Li, C. Du, Carbon 50 (2012) 3753–3765.[59] M. Inagaki, F. Kang, M. Toyoda, H. Konno, 5.2.3 mechanism for acceleration of

graphitization, in: Advanced Materials Science and Engineering of Carbon, Elsevier,2014, pp. 96–100.

[60] H.D. Asfaw, A. Kotronia, C.-W. Tai, L. Nyholm, Energy Technology, 2019.[61] A.C. Ferrari, J. Robertson, Philos. Trans. R. Soc. London, Ser. A: Mathematical,

Physical and Engineering Sciences 362 (2004) 2477–2512.[62] A.C. Ferrari, D.M. Basko, Nat. Nanotechnol. 8 (2013) 235.[63] C. Sole, N.E. Drewett, L.J. Hardwick, Faraday Discuss 172 (2014) 223–237.[64] R.J. Nemanich, S. Solin, Phys. Rev. B 20 (1979) 392.[65] Y. Kawashima, G. Katagiri, Phys. Rev. B 52 (1995) 10053.[66] S. �Otani, A. �Oya, J. Akagami, Carbon 13 (1975) 353–356.[67] G. Li, Q. Liu, Z. Liu, Ind. Eng. Chem. Res. 52 (2013) 5587–5592.[68] C.A. Pickles, J.M. Toguri, High Temperature Materials and Processes, 2004, p. 405.[69] H.P. Boehm, Carbon 16 (1978) 385–387.[70] G.H. Aylward, T.J.V. Findlay, SI Chemical Data, Wiley, 2002.[71] J.R. Dahn, T. Zheng, Y. Liu, J. Xue, Science 270 (1995) 590–593.