Mesoporous graphite nanoflakes via ionothermal carbonization of fructose and their use in dye removal

The large-scale green synthesis of graphene-type two-dimensional materials is still challenging. Herein, we describe the ionothermal synthesis of carbon-based composites from fructose in the iron-containing ionic liquid 1-butyl-3-methylimidazolium tetrachloridoferrate(III), [Bmim][FeCl4] serving as solvent, catalyst, and template for product formation. The resulting composites consist of oligo-layer graphite nanoflakes and iron carbide particles. The mesoporosity, strong magnetic moment, and high specific surface area of the composites make them attractive for water purification with facile magnetic separation. Moreover, Fe3Cfree graphite can be obtained via acid etching, providing access to fairly large amounts of graphite material. The current approach is versatile and scalable, and thus opens the door to ionothermal synthesis towards the larger-scale synthesis of materials that are, although not made via a sustainable process, useful for water treatment such as the removal of organic molecules.


Introduction
Graphene, currently one of the most popular two-dimensional materials, is a well-known carbon aIlotrope consisting of a single layer of sp 2 carbon atoms. 1,2The exceptional electronic and mechanical properties of graphene lend themselves to applications in electronics, 3,4 chemical sensing, 5 and catalysis. 6[9][10] Indeed high-quality graphene lms have been prepared by chemical vapor deposition of hydrocarbons on metal substrates, [11][12][13][14][15][16] but the accessible amounts are far from what is needed for large-scale applications.
One viable synthesis is chemical oxidation and exfoliation of graphite.It is amenable to large-scale production and therefore remains the most important pathway for bulk graphene synthesis. 9,17,18However, graphite exfoliation is a multistep and rather energy-intensive process.In view of the growing demand for facile chemical processes, new and efficient approaches towards larger amounts of graphene-type materials are clearly required.A sustainable process towards graphene-like materials from simple organic molecules such as glucose has recently been put forward; the reaction either proceeds through an intermediate carbon nitride (C 3 N 4 ) or uses molten metal chloride salts as the solvent and template. 7,19ther molten salts, ionic liquids (ILs), are now established solvents, 20,21 templates, 22 and precursors for inorganic materials, 8,[23][24][25][26][27] and as stabilizers for exfoliated graphite. 9,2824]29 As of now, ironcontaining ILs have been among the most actively studied ILs for the carbonization of biomass and other carbon precursors. 8,29For instance, hierarchical porous carbon nanoparticles can be obtained by simply heating biomass in iron-containing ILs.The process takes advantage of the multiple roles ILs can play as catalysts, templates and structure directing agents for carbon formation.Especially iron is an efficient promoter of the polymerization and dehydration of carbohydrates and the subsequent formation of graphitic structures. 25Iron-containing ILs possess unique properties: i.e. the cation can serve as a template, 29 whereas its anion is catalytically active 26,30 and enables the synthesis of graphene-type materials. 25Indeed, Dai et al. reported that graphene-type powders can be prepared with metal halide-ionic liquids as the carbon source and magadiite (a layered sodium silicate) as the template. 8Layered templates like magadiite play a crucial role that enable intercalation of metal-halide ionic liquids between layers prior to carbonization and thus promote the formation of 2D instead of 3D materials.This approach, however, requires the removal of the silicate template with toxic uorine-containing reagents such as HF or ammonium uorides. 8Therefore, a facile, large scale, and nonhazardous towards graphene-type materials is highly desirable.
Herein, fructose and 1-butyl-3-methylimidazolium tetrachloridoferrate(III), [Bmim][FeCl 4 ] are selected as an example to study the carbonization of biomass in the presence of an ironcontaining IL, Scheme 1.The resulting composite is mesoporous, magnetic, and rich in sp 2 carbons.These properties enable the removal of organic dyes such as rhodamine B from contaminated water using adsorption and magnetic separation.

Results
As depicted in Scheme 1, low temperature (LT, 180 C) ionothermal carbonization of fructose results in monolithic carbon ionogels.Carbon ionogels are versatile and fascinating hybrid materials comprised of a carbon framework and ILs incorporated within the carbon network. 31The high homogeneity of all monoliths suggests that they are stabilized by strong interactions between the carbon network and the IL constituents such as Coulombic interactions and physical adsorption. 31irect carbonization of carbon ionogels at 750 C under nitrogen results in high temperature (HT, 750 C) ionothermal carbons.X-ray diffraction (XRD) patterns of HT ionothermal carbons clearly show the presence of graphitic carbon and iron carbide (Fig. 1A).The reection at 26 could be related to a reasonably ordered (002) interlayer carbon packing. 32The reection is well resolved and sharp, suggesting that the materials are highly oriented graphitic structures. 33Additionally, there are a number of diffraction peaks centered at $45 (2q) which can be assigned to Fe 3 C (JCPDS 65-0393) within the graphitic matrix. 25The strong and sharp diffraction peaks corresponding to Fe 3 C suggest the formation of fairly large and well crystallized particles.
Raman spectroscopy, Fig. 1B, supports the assignment of turbostratic graphitic carbon.Spectra obtained from the ionothermal carbons show three characteristic peaks at 1340, 1574, and 2672 cm À1 , respectively.Narrow D and G bands along with a strong 2D band indicate partial ordering of the graphite sheets.Moreover, the intensity ratio of I D /I G (1.04) indicates that numerous sp 3 defect sites are also present. 8,16,34,35he N 2 sorption isotherm and pore size distribution of the nal materials are shown in Fig. 1C.Barret-Joyner-Halenda (BJH) analysis yields a surface area of 122 m 2 g À1 .The characteristic type IV isotherm indicates the presence of mesopores without a signicant fraction of micropores. 29,36,37The latter is conrmed by the fairly broad hysteresis loop, which closes upon desorption.This indicates capillary condensation and swelling effects. 29The hysteresis loop indicates unrestricted adsorption in the high pressure regime and the increased gas uptake at p/p 0 > 0.9 indicates the presence of slit-like pores. 38The mesopores  are emptied by a cavitation mechanism, as indicated by the spontaneous desorption at p/p 0 $ 0.45. 38Non-linear density functional theory (NLDFT) analysis yields a pore distribution with relatively monodisperse pore diameters around 5 nm.
Consistent with XRD and Raman data, TEM and high angle annular dark eld scanning TEM (HAADF-STEM) images reveal the presence of a graphite matrix and agglomerated iron carbide particles (Fig. 2).The presence of Fe 3 C is conrmed by energy dispersive X-ray spectroscopy (EDXS) on individual nanoparticles, which clearly shows the presence of iron and carbon.The large particles observed in (S)TEM are also consistent with the fairly intense and narrow reection in XRD.
High resolution TEM (HRTEM) and TEM-selected area fast Fourier transform (FFT) again support the formation of Fe 3 C and suggest that the Fe 3 C particles in fact are mesocrystals 39 or particles formed by oriented attachment. 40,41The lattice fringes of the Fe 3 C (1 0 0) crystalline plane, with d-spacing of 0.5 nm, can clearly be seen from the HRTEM image.Moreover, all TEM data show that, besides the large Fe 3 C particles within the samples, other domains mainly consist of carbon nanoakes, indicating a relatively heterogeneous distribution of the Fe 3 C nanoparticles in a homogeneous carbon matrix.
The surface morphology of an as-synthesized HT ionothermal carbon is shown in Fig. 3A.The carbon surfaces exhibit unique petal-like and interconnected nanoakes with a mean size in the micrometer range.Complementary transmission electron microscopy (TEM) nds a ake-like nanomorphology composed of stacked graphene-type sheets.The interconnected nature of the nanoakes likely generates the mesopores within the carbon frameworks.The nanoakes have highly homogenous morphologies, but vary in their thickness.
High resolution TEM (HRTEM) clearly shows lattice fringes in the carbon, conrming the carbon's graphitic nature.The interplane spacing measured from different nanoakes is in the range between 0.33-0.38nm, similar to that of carbon nanosheets (0.34 nm) reported previously. 42The typical honeycomblike molecular structure of hexagonally connected carbon atoms is observed in many unfolded carbon sheets.These domains are fairly small on the order of a few nm.Besides the single carbon nanosheets mostly discussed so far, all samples also contain multilayer domains; we hence call the materials obtained via our approach "oligo-layer graphite" (OLG) nanoakes.
The iron carbide particles within the mesoporous graphite nanoakes are accessible through the pores.Treatment of the materials with dilute hydrochloric acid results in the dissolution of the iron carbide nanoparticles.XRD patterns of the acidtreated materials exclusively show reections from carbon and all reections corresponding to iron carbide are gone aer 48 hours of acid etching (Fig. S1 †).This is also conrmed by TEM, which shows no signicant changes in the graphite nanoake morphologies, but no more dark speckles indicative of the Fe 3 C nanoparticles.This indicates that the acid etching does not affect the carbon sheets, but selectively removes the iron carbide from the samples.
Additional X-ray elemental mapping experiments were performed for further chemical analysis of the acid-etched samples (Fig. 4).X-ray maps show that the carbon is homogeneously distributed, and oxygen is also present in the sample.In spite of the fact that XRD (Fig. S1 †) does not show any indication of iron carbide or iron chloride, X-ray analysis still nds iron and chlorine signals homogeneously distributed throughout all samples.The apparent contradiction between the fact that XRD does not detect any iron-based mineral phases anymore (Fig. S1 †) while EDXS (Fig. 4D) still detects Fe can be assigned to (i) rather low concentrations of Fe aer the etching process (which can still be detected by X-ray mapping, which is a local method, but not by XRD, which is a global method) or (ii) small fractions of residual anhydrous iron chloride, which has a poor solubility in dilute aqueous hydrochloric acid and does not generate an XRD signal if poorly ordered.EDX spectrum aer etching again support that Fe signal is much lower than that of as-prepared Fe 3 C@OLG materials shown in Fig. 2D (Fig. S2 †).With the stripping of iron carbide by HCl treatment, the surface areas of ionothermal carbon increase from 122 to 180 m 2 g À1 , probably due to the change of materials density aer removing of large Fe 3 C particles (Fig. S3 †).
As stated in the introduction, magnetic carbon materials are candidates for water purication. 43,44In addition, graphite materials with high specic surface areas are useful for adsorption of (mostly unpolar) organic molecules. 45The OLGs prepared here are clearly attractive for water treatment because they are relatively hydrophobic and, as demonstrated by XRD and Raman spectroscopy (Fig. 1), contain a high fraction of aromatic sp 2 carbon atoms.Indeed, the OLG can be dispersed in toluene but precipitate in water (Fig. S4 †).
This implies a high affinity for organic molecules, such as dyes or aromatics which oen are hydrophobic, but still present in aqueous media in concentrations posing a risk to the environment. 46To evaluate the OLG materials for the removal of organic molecules from water, we have studied the adsorption of rhodamine B (RB) from aqueous solution (Fig. 5) as a model reaction.Adsorption kinetics was studied at an initial RB concentration of 25 mg L À1 in water.Adsorption was monitored by UV-vis spectroscopy vs. adsorption time.Fig. 5A shows that the RB absorbance gradually decreases with increasing contact time.Visual inspection of the respective solutions conrms that a large fraction of RB is removed from the aqueous phase leaving behind a clear solution aer 5 hours.The adsorbent can be easily extracted for reuse by magnetic ltration.As shown in Fig. 5B, equilibrium is reached aer 120 min of contact time.
Fig. 5C shows that the solution pH only has a weak inuence on the adsorption performance between pH 3 and 9.These results suggest that the OLGs could effectively adsorb RB over a wide pH range without signicantly altering the adsorption characteristics.The key advantage of the OLG material over other materials used for dye removal is that the adsorption capacity of the current OLGs reaches up to 150 mg g À1 , as illustrated in Fig. 5D and Table 1.This is signicantly better both in terms of equilibrium adsorption time (which translates into speed or rate of a technical process) and capacity (which  The adsorption isotherm was calculated using the equation: q e ¼ (C 0 À C e )V/m, where q e is the concentration of dye adsorbed (mg g À1 ), C 0 and C e are the initial and equilibrium concentrations of dye (in mg L À1 ), m is the mass of OLG nanoflake (g), and V is the volume of solution (L).
translates into amount of RB removed by the material per adsorbent weight unit).We currently surmise that the RB adsorption onto the OLG material is due to strong van der Waals, hydrophobic, and p-p interactions between OLG nanoakes and the dye molecules.Also, the mesoporosity and high surface areas of the current material lend it to having the satisfactory RB adsorption performance.

Discussion
Ionic liquids are promising active components for the creation of (multi)functional materials. 23,52Among others, there is an increasing research and technology effort in making use of ILs for the preparation of carbon-based materials. 8,26,29One major advantage for using of ILs as precursors for carbons is that heteroatom doping (such as N, B, S) and metal doping can be accomplished fairly easily. 53Moreover, the almost unlimited exibility of IL structure and chemical composition makes them highly suitable for tuning carbon structure with desirable electronic properties or specic surface area. 54As stated in the introduction, however, there is a lack of protocols for the largescale synthesis of graphene-like materials.The current article provides a viable approach towards nano-and mesotructured graphitic carbon using a sustainable raw material (although the OLG synthesis as such is clearly not sustainable at the moment), fructose, and an IL that is easily accessible on a large scale.
Iron-based ionic liquids are interesting components not only because they are magnetic 55 or catalytically active. 30,56,57They can also serve as powerful solvents for biomass conversion. 29Bmim][FeCl 4 ] has been used as precursor for iron/iron carbide and graphene-type materials synthesis in the presence of an inorganic template.For instance, Göbel et al. synthesized Fe/ Fe 3 C/C hybrid materials with rather high surface areas using functionalized SiO 2 as template and [Bmim][FeCl 4 ] as precursor.25 Dai et al. synthesized graphene-type carbon powders by pyrolysis of iron-based ILs and magadiite, a layered sodium silicate.8 Moreover, we have previously shown that the IL [Bmim][FeCl 4 ] is an effective medium for conversion of biomass to porous carbons at relatively low temperature.29 This process results in a monolithic carbon ionogel.In spite of focusing on different nal materials, all studies used the same IL, showing its potential for a wide variety of chemical transformations.Especially given the fact that both biomass and iron chloride are relative low-cost, the large-scale synthesis of such carbon ionogels can be economically achieved, possibly by adapting the chemistry of the IL from imidazolium-based (which still constitute more expensive ILs) to ILs based on other cations such as ammonium, which may be cheaper.
Carbonizing of as-synthesized carbon ionogels leads to graphite nanoakes.For structure assignments, we have used various techniques.SEM and TEM (Fig. 2 and 3) indicate an "oligo-layer-graphite" (OLG) morphology with characteristic thin graphite nanoakes, thus justifying the term "oligo-layergraphite".Remarkably, the honeycomblike molecular structure corresponding to single-layer graphene is clearly seen in HRTEM (Fig. 3E), conrming the formation of graphene-like materials at low carbonization temperature of only 750 C. Our approach thus introduces a new strategy towards graphite composites containing highly ordered graphitic structures termed OLG.In addition, agglomerated rather than homogeneously distributed Fe 3 C particles are is found within the graphitic matrix.This may be assigned to migration/ coalescence of particles (sintering) or Ostwald ripening of iron species at high temperature.
Raman spectroscopy is highly sensitive to the electronic structure of carbon-based materials. 34The characteristic peaks of carbon-based materials are the G band at 1580 cm À1 corresponding to the rst-order scattering of the E 2g phonons of sp 2 carbon atoms, and the D band at 1350 cm À1 corresponding to breathing mode of k-point phonons of A 1g symmetry. 35Our results show D and G bands at 1340 and 1574 cm À1 , respectively.Moreover, a strong 2D band is observed at 2672 cm À1 , which supports the assignment of OLG found in the current study. 8,28,35otably, we have demonstrated previously that ironcontaining ILs are a powerful solvent for carbonization. 29The signicant difference to the current study, however, is the fact that in the previous case, the IL was recovered by Soxhlet extraction (in fact demonstrating a somewhat sustainable process).Moreover, the process was chemically different yielding spherical carbon particles of approximate 50-100 nm in diameter. 29y contrast, the current case introduces an adapted process allowing for the synthesis of OLG nanoakes via carbonization of carbon ionogels (Scheme 1 and Fig. 6).The formation of graphitic carbons from such iron catalysts usually occurs through a dissolution-precipitation mechanism that involves the dissolution of amorphous carbon into iron particles followed by the precipitation and segregation from the Fe to form graphene layers. 26Indeed, the formation of the OLG nanoakes is clearly different to the previously described materials. 29he combination of magnetic moment and large p-system with acceptable porosity suggests that the current materials could be attractive for water purication, particularly for the removal of organics. 45,58Many organic molecules such as dyes or aromatics are toxic, persistent in the environment and nonbiodegradable. 59Physical adsorption is an efficient and attractive way to overcome this issue for wastewater treatment. 60ndeed, many carbon materials that have shown effective dye adsorption 43,45,59 and Fig. 5 and Table 1 clearly show that the current material is comparable to (and in fact better than some) other carbon materials used for dye removal both in terms of capacity and equilibration time.Although clearly not made via a "green" or "sustainable" process as such, the material is interesting from an environmental application perspective.The OLG nanocomposites combine the advantages of mesoporosity, graphitic surface structure, and magnetic moment making it a comparatively cheap (although still far too expensive for the developing countries) and simple material for water remediation.Upscaling of the materials synthesis, replacement of the still rather const-intensive imidazolium IL with a cheaper, e.g.ammonium-based, IL, and optimization of the water treatment process as such will, however, be necessary prior to commercialization.

Conclusion
The current report describes a general approach towards mesoporous oligo-layer Fe 3 C/graphite nanoakes from a renewable monomer, fructose, under ionothermal conditions using the IL [Bmim][FeCl 4 ] as "all-in-one" solvent-reactant-catalyst. The composites are efficient adsorbents for the removal of RB from aqueous solution and can be isolated by magnetic separation with a permanent magnet.Via acid etching, the iron carbide particles can be removed leaving pure mesoporous OLG materials behind.These can be dispersed in hydrophobic solvents such as toluene.Overall, the ionothermal carbonization of renewable raw materials such as the approach demonstrated here may in the future allow the large-scale synthesis of tailormade adsorbents or dispersible (and thus solution processable) oligo-layer graphite with potential application in electronics or sensing.

Nanoake synthesis
Oligo-layer graphite/iron carbide nanoakes were synthesized by direct carbonization of the carbon ionogels by heating the carbon ionogels in an oven under N 2 ow at a heating rate of 10 K min À1 to 750 C and hold 4 h.The iron/iron carbide nanoparticles within the carbon matrix were removed via soaking in 1 M HCl (aq) solution for 48 h under stirring.

Characterization
Scanning electron microscopy (SEM) images were acquired on a LEO 1550 with Everhard-Thornley secondary electron and inlens detectors.Transmission Electron Microscopy (TEM) and HRTEM were done on a FEI C s -corrected Titan 80-300 microscope operated at 300 kV with a Gatan energy lter.The Raman spectrometer (Kaiser Optical) was equipped with a frequencydoubled Nd:YAG 532 nm laser.Nitrogen sorption isotherms were measured at 77 K on a Quadrachrome Adsorption Instrument (Quantachrome Instruments).Sample was dried at 120 C for 12 h prior to nitrogen sorption analysis.UV-vis spectra were recorded on a Perkin-Elmer Lambda 25 UV-vis spectrometer at room temperature.

Scheme 1 Fig. 1
Scheme 1 Synthesis of Fe 3 C-graphite composites from fructose and IL and their use in dye removal.

Fig. 2
Fig. 2 TEM and HAADF-STEM images (A and B) and HRTEM and corresponding EDXS analysis (C and D).Inset is a selected area FFT pattern of the sample shown in (C).

Fig. 3
Fig. 3 SEM (A) and TEM images at increasing magnifications (B-E) of a graphite composite (D and E are HRTEM images).

Fig. 4 (
Fig. 4 (A, B) TEM images, (C) HAADF-STEM image and (D) X-ray maps of the OLG after acid etching.Scale bars in panels (C) and (D) are 200 nm.

Fig. 5 (
Fig. 5 (A) Absorption spectra of RB solutions treated with the ionothermal carbon hybrid material.(B) Adsorbed RB vs. time.(C) The effect of pH on the adsorption of RB onto nanoflakes, C 0 ¼ 25 mg L À1 .(D) Adsorption isotherm.The adsorption isotherm was calculated using the equation: q e ¼ (C 0 À C e )V/m, where q e is the concentration of dye adsorbed (mg g À1 ), C 0 and C e are the initial and equilibrium concentrations of dye (in mg L À1 ), m is the mass of OLG nanoflake (g), and V is the volume of solution (L).
Carbon ionogel synthesisD-Fructose was purchased from Sigma-Aldrich® and used without further purication.3.0 g of fructose was dispersed in 8 mL of [Bmim][FeCl 4 ] at room temperature (fructose is not soluble in the IL at r.t.)The mixtures were loaded into PTFE lined autoclaves and treated at 180 C for 5 hours.The resulting carbon ionogels were dried overnight at 60 C.

Fig. 6
Fig. 6 SEM and TEM images of carbon structures prepared via second annealing process in the presence of iron catalyst (A and C), and in the absence of iron catalyst (B and D).