Revealing the Impact of Different Iron-Based Precursors on the ‘Catalytic’ Graphitization for Synthesis of Anode Materials for Lithium Ion Batteries

Revealing the Impact of Different Iron-Based Precursors on the ‘Catalytic’ Graphitization for Synthesis of Anode Materials for Lithium Ion Batteries Lars Frankenstein,[a] Pascal Glomb,[a] Joaquin Ramirez-Rico,[b] Martin Winter,[a, c] Tobias Placke,*[a] and Aurora Gomez-Martin*[a] Low cost and environmentally friendly production of graphite anodes from naturally available biomass resources is of great importance to satisfy the increasing material demand for lithium ion batteries. Herein, graphitization of coffee ground was performed using four different iron-based activating additives, including iron (III) chloride, iron (III) nitrate, iron (III) oxide and pure iron, following either a wet or a dry mixing approach. The structural development regarding the type of activator used and the impact on the corresponding electrochemical performance are systematically investigated. A maximum degree of graphitization between 55 and 74% (as determined by Raman spectroscopy) is attained using iron (III) chloride and iron powder, respectively. The graphitic anode material synthesized using iron powder reached a maximum reversible capacity of �320 mAhg1at a rate of 0.1 C. This study provides significant insights into the impact of activators on the design of synthetic graphite from renewable sources. Introduction Since their first commercialization in the early 90 s, the demand for lithium ion batteries (LIBs) for electronic devices as well as electro mobility applications has increased steeply.[1] LIBs are considered as a key technology to decarbonize global transport and energy sectors.[1–2] Increasing consumer demands as well as recent concerns about process sustainability and shortages in material supply chains drive the development of advanced LIB active materials.[3] Graphite is the state-of-the-art anode active material for LIBs due its moderate gravimetric capacity (372 mAhg1, LiC6stoichiometry), long cycle life, low average (de-)lithiation potential (�0.2 V vs. Li jLi+) and a low voltage hysteresis which results in high voltage and energy efficiencies.[3–4] Despite silicon (Si) holds great promise as anode active material for future high-energy density LIBs,[5] its use in high proportions (�10–20 wt%) in commercial cells is still hampered by several challenges limiting cycle life.[6] Blending of Si with graphite particles is considered one of the most promising approaches to improve cycle life of future Si-based LIB full-cells.[7] Therefore, it is expected that graphite will remain a mandatory material for LIB anodes in the coming years. The demand for graphite is expected to experience an exponential growth over the coming years driven by the development of electric vehicles (EVs).[1b,8] Two types of graphite anode materials are typically used in commercial LIB cells, i.e., natural graphite (NG) and synthetic/artificial graphite (SG).[1b,8a] SG shows outstandingly high levels of purity and less fluctuating quality and characteristics, while NG exhibits lower costs. Therefore, the production of mixtures of SG and NG offers a good compromise between cost, energy and power densities and lifetime.[9] However, there are raising environmental and supply bottlenecks concerns for the production of both graphite types.[10] Currently, most of the global supply of NG and SG comes from China.[9,11] On the one hand, SG is produced from soft carbon (i.e., ‘graphitizable’ carbon) precursors such as petroleum coke, which is a non-sustainable waste by-product from the oil refinery industry but can be graphitized when heattreated at high temperatures.[1b,12] The processing of SG comprises first carbonization step up to 800–1200°C, a second heat-treatment up to temperatures higher than 2500°C for several days up to weeks and intermediate and final particle size refinement steps.[1b,13] Overall, the production of SG is expensive, energy-intensive, and environmentally harmful. On the other hand, NG has been classified as critical for Europe and the U.S. in terms of economic importance and supply risk due to extraction from graphite ores mainly present in China.[12a,14] Besides NG reserves are limited, the purification of NG has severe negative impacts on the environment, such as air and [a] L. Frankenstein, P. Glomb, Prof. Dr. M. Winter, Dr. T. Placke, Dr. A. Gomez- Martin University of Münster MEET Battery Research Center, Institute of Physical Chemistry Corrensstr. 46, 48149 Münster, Germany E-mail: [email protected] [email protected] [b] Prof. J. Ramirez-Rico Dpto. Física de la Materia Condensada and Instituto de Ciencia de Materiales de Sevilla, Universidad de Sevilla, Consejo Superior de Investigaciones Científicas Avda. Américo Vespucio 49, 41092 Sevilla, Spain [c] Prof. Dr. M. Winter Helmholtz Institute Münster, IEK-12 Forschungszentrum Jülich GmbH Corrensstr. 46, 48149 Münster, Germany Supporting information for this article is available on the WWW under https://doi.org/10.1002/celc.202201073 © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. ChemElectroChem www.chemelectrochem.org Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (1 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 16/28] 1 water pollution.[10] Therefore, further attention should be paid to the search of more sustainable processing of SG. The conversion of industrial waste by-products and naturally available biomass resources into valuable carbon materials for energy storage devices, i.e., LIBs, sodium ion batteries (SIBs) and supercapacitors, has been the subject of extensive research efforts recently.[15] The physicochemical properties of carbon materials (i.e., degree of graphitization (DoG), particle morphology, surface area, etc.) can be easily tailored to the final application.[16] Carbons obtained by heat-treatment of biomass resources are typically classified as ‘non-graphitizable’ or ‘hard carbons’ due to the strong cross-linking and random orientation between crystallites which precludes the development of a highly graphitic structure upon further heat-treatment.[17] While hard carbons are preferred as anode materials in SIBs as a result of the less favoured intercalation of Na+into graphite compared to that of Li+,[16b,18] graphitic carbons are desirable for LIBs. An effective approach to accelerate the development of ordered graphitic domains in hard carbons while reducing cost and energy consumption is the so-called ‘catalytic’ graphitization.[19] This process involves the wet or dry addition of transition metals (Fe, Co, Ni, etc.) acting as ’’activators’’ to the carbon precursor that, upon heating, can induce the precipitation of ordered carbon regions and thus a decrease in the activation energy for crystallization.[19a,d,20] Among transition metals, iron (Fe) is regarded as the most effective activator due to the number of free electron vacancies in the d-shell orbitals and thus the possibility to form strong covalent bonds between metal and carbon atoms.[19b,21] According to previous studies on the graphitization of biomass, activator concentration[22] and carbonization temperature[22–23] play a critical role in the DoG and the electrochemical performance as anode materials in LIB cells.[19a,22a] Gomez-Martin et al.[19a] showed an enhanced DoG with increased heat treatment temperatures (850–2000°C) when using an iron-based activator (iron (III) chloride) at a fixed concentration and observed improved specific capacities of the produced carbons. More recently, Hanhart et al. found an almost linear trend in achievable reversible capacities when increasing the activator content (iron (III) chloride) as well as graphitization temperature due to the increasing structural order within the carbon materials.[22a] However, our latest study also reported that the volatility of the iron (III) chloride precursor might drive continous evaporation from the carbon precursor causing a saturation point in the maximum DoG (�69%) and reversible capacity (�235 mAhg1) achievable.[24] Hoekstra et al. investigated the graphitization of cellulose with different iron salts and found a worse graphitization efficiency for iron (III) chloride. From those findings, it was pointed out that the extent of graphitization might strongly depend on the iron particle size and iron salt.[24] Banek et al. reported the conversion of hardwood sawdust to highly crystalline graphite flakes by laser pyrolysis using iron powder with similar maximum reversible capacities as known for commercial SG.[25] Therefore, further studies investigating the effect of different iron-based precursors are needed for further tailoring and optimization of the synthesis of more sustainable SG anodes for LIBs using biomass resources. In this work, a detailed study of the effect of different ironbased precursors on the graphitization of coffee grounds as biomass precursor is presented. High temperature graphitization was performed at 2000°C with four different iron-based precursors. Raman spectroscopy and total scattering X-ray diffraction experiments along with their pair distribution functions were performed to monitor the extent of graphitization in coffee-ground derived carbons. Finally, all graphitized carbon materials were electrochemically evaluated in carbon jjLi metal cells and the results are discussed in connection with their DoG, crystallite size and particle morphology. This study can pave the way for further tailoring the synthesis of graphitic materials from biomass resources. Results and discussion Synthesis procedure and morphology of the samples The graphitization of coffee ground was investigated using the same Fe:C mass ratio (1:1) and four different iron-based precursors, including iron (III) chloride (FeCl3·6H2O further referred to as ‘FeCl3’-sample), iron (III) nitrate (Fe(NO3)3·9H2O abbreviated as ‘Fe(NO3)3’-sample), iron oxide (Fe2O3) and pure iron (Fe). In Germany, more than 97,000 t of raw coffee is imported monthly resulting in a lot of waste given as coffee ground, leading to a possible sustainable graphite source when performing ‘catalytic’ graphitization.[26] As illustrated in Figure 1, two different approaches were considered for incorporation of the iron-based activator. For the first approach, FeCl3and Fe(NO3)3were dissolved in acetone and wet-mixed with coffee grounds. For the second approach, the biomass precursor and Fe2O3or Fe as activators were mechanically mixed. These mixtures were first carbonized up to 1000°C in an inert atmosphere to drive the thermal decomposition of the organic matter to amorphous carbon followed by a second heattreatment at 2000°C to promote graphitization. According to thermogravimetric analysis performed by Hahnart et al., the carbon yield from the coffee grounds carbonization can be estimated to be �20 wt% based on dry biomass weight, which seems to be low compared to other biomasses such as lignin.[22a,27] However, coffee ground is a waste product that one should take advantage of and modifications in the carbonization step could result in higher carbon yields.[28] The iron content after carbonization at 1000°C as well as after graphitization at 2000°C and acid washing was determined by inductively coupled plasma-optical emission spectrometry (ICP-OES). Results are listed in Table S1 for all samples. Despite the same initial Fe:C ratio (1:1), the total iron content slightly differs between samples. Values between �18 and 27% were achieved. The sample containing FeCl3showed the lowest iron proportion most likely due to dehydration and decomposition of the iron salt at temperatures >100°C, in agreement with previous studies.[22a] It has to be noted that the high amount of residual Fe for the Fe(NO3)3-treated sample might be ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (2 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 17/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License explained by the possible lower yield of carbon during carbonization, as Fe(NO3)3is highly oxidative. After graphitization at 2000 °C and acid washing, the residual iron content drastically fell below 1 wt%, leading to almost impurity-free graphitic carbon samples. Representative scanning electron microscopy (SEM) images of all samples graphitized at 2000°C shown in Figure 2 reveal significant morphological changes for different iron-containing activators. As can be seen in Figure 2g and 2h, the FeCl3-sample has a remarkably similar particle surface, morphology and porous structure to the starting biomass. SEM micrographs of the graphitized coffee grounds without iron activator are shown in Figure S1. In contrast, SEM images of the other samples (Figure 2a–f) reveal a flake-shaped morphology composed of agglomerates of graphite plates. In particular, for the Fe-sample, some particles showed a perfectly spherical morphology that resembles that of commercial SG materials and were not found for the other samples (Figure S2). One should note that Fe2O3 and Fe activator particles have initial sizes of <1μm and <5μm (Figure S3), respectively. Those particle sizes are expected to be significantly larger than those resulting from the decomposition of FeCl3and Fe(NO3)3and might have an influence on the graphitization mechanism.[29] Structural characteristics of carbon materials The structural properties as well as the DoG of carbonaceous samples were first examined via Raman spectroscopy due to its sensitivity towards structural disorder.[30] Characteristic Raman signals for carbons originate from various lattice vibrations from the graphene layers. Only one first-ordered band (G-band) would occur for perfectly stacked graphene layers at a Raman shift of 1580 cm1.[31] It can be attributed to the in-plane stretching motion of the C=C double bond with E2g- symmetry.[30] Theoretically, there are four additional firstordered defect bands (Dbands) occurring in graphitic carbons and carbonaceous materials stemming from structural disorder.[30] The most distinct defect band is the D1 mode observed at �1350 cm1, caused by lattice vibrations with A1g- symmetry mainly at the edges of graphene layers in amorphous carbon. Also, the presence of heteroatoms within the structure could lead to the occurrence of the D1 band.[32] A further band commonly observed is the D2 band at �1620 cm1as a shoulder of the Gband. It only occurs at graphene layers, which are not sandwiched by other layers.[32a,33] D3 and D4 defect bands are primarily observed for highly amorphous carbonaceous materials at 1500 cm1and 1200 cm1due to the sp2–sp3 bonds of CC and C=C stretching vibrations of polyene-like structures, respectively.[32a] An additional peak appears at �2700 cm1, which is the second order of the Dband (2D) and is influenced by the order of stacking in c-direction. The measured Raman spectra of carbonaceous materials exhibit all five D bands as well as the G band (Figure 3). Overall, the D1 band intensity decreases proportionally to the G band from FeCl3to Fe-powder (top to bottom, Figure 3-a). The intensity ratio between G and D1 bands (ID1/IG) is often used as an indicator of the extent of graphitization in carbonaceous samples.[32a] However, the values obtained using the integrated peak area intensities of the G and D1 bands are only an Figure 1. Graphical illustration of the synthesis conditions investigated in this work. Coffee ground was used as biomass-precursor. The activator was introduced either following a wet sol gel approach using iron (III) chloride or iron (III) nitrate in acetone or following a dry mixing approach using iron (III) oxide or iron powders. For further details, please check the experimental section. ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (3 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 18/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License estimation and should be interpreted with caution. A more accurate statement can be achieved by fitting G, D1, D2 and D4 bands with Lorentzian functions and the D3 band with a Gaussian function according to Sadezky et al.[32a] Figure 3-b shows an example of fitted spectrum for FeCl3-derived carbonaceous material. Detailed Raman fitting results and contributions of the different modes for all samples can be seen in Figure S4 and Table S2. The highest ID1/IG-ratio of 0.82 is obtained by the volatile activator FeCl3. The ID1/IG-ratio decreases to 0.49, 0.45, 0.36 when using Fe(NO3)3, Fe2O3and Fe, respectively (Figure S5a). Therefore, the sample synthetized using Fe-powder as activator has the highest development of the graphitic structure among all measured samples. The DoG (0�DoG�1) and the crystallite size in a-direction (La) can be calculated by the integrated intensity ratios of D1 and G bands according to the following equations (Eqs. 1 and 2):[34] DoG ¼IG ID1þIG(1) La¼2:4�1010 �l4ID1 IG � �1 (2) After fitting the Raman data as shown in Figure 3c, a decreasing DoG regarding the activator as follows was found: Fe (74%) >Fe2O3(69%) >Fe(NO3)3(67%) >FeCl3(55%). The highest DoG of 74% is thus achieved by using Fe as activator in the graphitization process, while the lowest value of 55% was obtained by the volatile activator FeCl3, as expected from our previous results and the estimated values based on the ID1/IG-ratio.[22a] The value reported here is a bit lower than that reported in our previous study for FeCl3(DoG of �69%). However, in that study only different spots within one particle were measured and might not be representative of the whole Figure 2. SEM micrographs of coffee-ground derived carbons at 2000°C using a,b) Fe, c,d) Fe2O3, e,f) Fe(NO3)3and g,h) FeCl3as activators. ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (4 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 19/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License sample. Interestingly, the DoG values of Fe2O3and Fe(NO3)3- derived carbonaceous materials are quite similar to each other. This could be explained by hydrolysis of Fe(NO3)3after addition of the solvent to Fe2O3. It might be that still some Fe(NO3)3 residues are left in the sample which explains the slightly different DoG. Furthermore, the relative peak area of the D2-band at �1620 cm1can be used to get information about the stacking mechanism of carbonaceous materials. The relative D2peak area decreases from 3.0% to 0.8% from FeCl3to Fe-powder (Figure S5b), indicating the higher extent of stacking of the Federived materials. Moreover the difference between Fe2O3and Fe(NO3)3is higher than for Fe and Fe2O3(Table S2) indicating the physical mixing being more beneficial towards the graphitization of carbonaceous materials. Similarly, the crystallite size in the a-direction (Figure S5c), La, was calculated to be �23.5 nm (FeCl3), 39.2 nm (Fe(NO3)3), 42.4 nm (Fe2O3) and 53.3 nm (Fe). Furthermore, I2D/IGas well as I2D/ID1 ratios were calculated to get an idea of the degree of stacking of graphene sheets.[20b] However, no clear trend that would be helpful to understand the effect of different activators was found (Table S2). The ID2/IGratio seems to decrease for samples with a higher DoG (i.e., Fe and Fe2O3following a dry mixing approach). According to these results, a simple physical mixing/addition of the activator (Fe) seems to have a more positive impact on the ‘catalytic’ graphitization mechanism, which we further investigated by X-ray diffraction measurements as well as electrochemical analysis. To gain furthers insights into the local ordering of the partially-graphitized samples synthetized at 2000°C using different iron-based activators, total scattering X-ray diffraction experiments were performed in MSPD beamline of the ALBA synchrotron. Figure 4 shows the normalized and corrected scattering factor, which is proportional to the diffracted intensity. There are slight differences between carbonaceous samples, especially with regards to the sharpening of diffraction reflections. The atomic pair distribution function (PDF) has emerged as a powerful tool for the study of hard carbon or partially graphitized materials with limited long-range ordering.[18,35] The PDF contains information regarding the interatomic distances in a material and was extracted from the scattered data according to the following equation (Eq. 3): G rð Þ ¼ 2 pZ∞ 0 Q S Qð Þ 1½ �sin Qrð ÞdQ � 2 pZQmax Qmin Q S Qð Þ 1½ �sin Qrð ÞdQ (3) Figure 3. a) Raman spectra of synthesized carbonaceous materials with different activators, b) fitting of Raman spectrum to determine the DoG (example for FeCl3-derived sample), c) calculated DoG values for each activator-derived carbonaceous material. ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (5 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 20/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License with Qbeing the modulus of the scattering vector, and rthe interatomic distance separating pair of atoms. The so-called small-box model using the crystalline structure of graphite was considered as a starting point to refine PDF data and obtain structural parameters, as previously reported to study hard carbons.[18,36] For this approach, a crystalline structure containing the basic short-range atomic unit (hexagonal carbon rings in this case) was initially considered and the long-range order was smoothed by introducing long-range disorder.[36] The disorder was introduced through anisotropic atomic displacement parameters (ADPs), which are allowed to take unphysical values describing static disorder. Lattice parameters aand c, anisotropic displacement parameters U11, average coherent crystalline distance or crystallite size, and a global scale factor were refined in the rrange 0.5–20 Å. U11 was set equal to U22 to maintain the in-plane symmetry of graphene layers in the graphitic structure. The observed and calculated PDFs from the refined structural model are shown in Figure 5a for all samples. As can be seen, this is a one-dimensional function which oscillates around zero and shows peaks centered on the average r interatomic distances separating pair of atoms, related to the increase of the local atomic density to the average value. For small interatomic distances, the PDF shows peaks at atom-atom distances well matching those in a basic graphene unit, while the lack of long-range ordering manifests as a decaying amplitude of the PDF with increasing atomic distance r. Detailed information of refinements, lattice parameters and ADPs can be found in Table S3. Agreement factors Rwp were generally below 30% suggesting that the refinements are reliable and there is a good agreement between measured and calculated data.[19f] The lattice parameter a(Figure 5b), which is related to the average in-plane interatomic distances in the structural graphene unit, remained constant and very similar to that of graphite (a=2.461 Å). The interlayer distances (c/2) are slightly higher than in ideal graphite (c=3.35 Å) and range between 3.35 and 3.38 Å (Figure 5c), corresponding to the Fe and FeCl3samples, respectively. A higher interlayer or interplanar spacing between graphene sheets can be ascribed to the presence of less-ordered turbostratic regions. U11 and U33 are indicative of in-plane and out-of-plane disorder in the graphitic structure, respectively. While U11 remains almost constant at �2×103Å2, a value that is physically compatible with thermal motion, the out-of-plane atomic displacement parameter U33 follows the same trend as the interplanar distance (Figure 5d). The sample synthesized with FeCl3as iron-source shows the highest value (�0.05 Å2) due to its higher disorder. The estimated crystallite size (Figure 5e) shows the following descending order for the investigated iron-based activators: Fe >Fe2O3>Fe(NO3)3>FeCl3, in agreement with the DoG estimated from Raman results. Dry mixing vs. wet-mixing approach using different activators Findings from SEM, Raman and PDF results indicate intriguing differences between samples. The carbonaceous sample prepared using Fe-powder showed a significant better development of the graphitic structure. The coherent crystallite size, or more specifically the coherent crystalline distance, is significantly larger (�115 Å) compared to the other samples. In contrast, the sample using FeCl3showed a limited long-range structural order which can stem from the volatility of the precursor itself.[24] According to previous works,[29] the micro- Figure 4. Normalized scattering factor for the coffee ground-derived carbonaceous samples. Left graph shows a magnified view of the (002) reflection in graphite. ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (6 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 21/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License structure of the resulting graphitic samples might not only intimately be linked to the carbonization temperature, additive concentration, and impregnation method[19a,23b,24,37] but also to the additive particle size. Even though samples wet-impreg- nated with Fe(NO3)3and FeCl3in an organic-based solution are expected to have a more homogenous dispersion of the additive and a better contact with the biomass-derived carbon, they unexpectedly show lower DoG values which might stem from a different reaction mechanism.[22a] According to some of the earliest works on catalytic graphitization by Oya et al. using Ni as additive, if the particles were finely divided in the order of a few nanometers in size (e.g., �20 nm) onion-like graphitic layers surrounding catalyst particles were precipitated.[29] However, larger particles formed by coarsening upon heating led to the precipitation of three-dimensional bulk graphite crystals. This effect was also observed in previous studies using FeCl3at temperatures ranging between 1000 and 2000°C by highresolution transmission electron microscopy (TEM).[19a,23b] Two reaction mechanisms have been proposed in literature for ‘catalytic’ graphitization of carbon/iron mixtures, i.e., dissolution-precipitation at temperatures higher than the melting point of pure iron (~1540°C)[38] and carbide formation/ decomposition at much lower temperatures.[39] In the case of wet-impregnation with FeCl3and Fe(NO3)3, the iron salt precursor is decomposed upon heat treatment resulting in the formation of iron oxide nanoparticles before forming a carbide that later decomposes into metallic iron and additional graphite from low temperatures below 1000°C.[19f] Conversely, for the dry mixing approach using Fe powder with a relatively large particle size, a dissolution-precipitation mechanism within molten iron ‘catalyst’ particles might be responsible for the formation of three-dimensional graphite platelets.[39c] Li et al.[40] reported the onset of graphitization to occur at ~1200°C when mixing powders of milled coke carbon and Fe (particle sizes <74 and <5μm, respectively). It might be expected that large particle sizes of Fe in that study probably limited the formation of iron carbide (Fe3C) due to the sluggish diffusion kinetics, and thus, graphitization only started when a liquid phase formed via a solution/re-precipitation mechanism. However, the (002) reflection of graphite is present in all our samples at 1000°C (Figure S6) regardless of the initial iron-based precursor. The absence of iron carbide (Fe3C) reflections in the XRD pattern of Fe and Fe2O3samples (Figure S6) might indicate that the most dominant graphitization mechanism for these samples is the dissolution-precipitation mechanism. Nevertheless, future studies with precise control of the additive particle size are needed to investigate the actual impact on the DoG. Electrochemical performance in carbon jjLi metal cells The impact of the development of graphitic structures, observed by both Raman spectroscopy and XRD, on the electrochemical performance was investigated by constant current cycling (CCC) experiments in carbon jjLi metal cells (half-cell setup using a three-electrode cell configuration with lithium metal as counter (CE) and reference electrodes (RE)).[41] The rate-capability investigations, to check the fundamental properties of the materials, at rates between 0.1 C and 10 C (1 C equals to 372 mAg1) are shown in Figure 6a. The maximum specific de-lithiation capacities attained follow the same trend as the DoG and crystallite size estimated by Raman analysis and PDF results: Fe >Fe2O3>Fe(NO3)3> FeCl3. During the whole rate capability experiment, Fe-derived carbonaceous material exhibits outstanding performance and the highest capacities with little deviation at even high rates of Figure 5. Results obtained from the fitting of the PDFs to a disordered graphite structure: a) Observed (coloured hollow circles) and calculated (red lines) values of the PDF, along with the difference curves (green ones). b) Lattice parameter a, c) interplanar spacing (c/2), d) in-plane (U11) and out-of-plane (U33) ADPs, and e) average crystallite size. U11 is multiplied by a factor of 100. ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (7 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 22/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 5 C and 10 C. During (de-)lithiation at high C-rates, the capacities significantly fade due to slower lithiation kinetics due to hindrance of Li+ion intercalation into the graphite layered structures compared to the adsorption/insertion of Li+ions into hard carbon structures.[42] This capacity drop becomes more evident for the more graphitic samples (i.e., Fe, Fe(NO3)3and Fe2O3). Once the rate returns back to 1 C (after 59 cycles), the delithiation capacities are nearly maintained or increased compared to the capacities at the beginning of the C-rate experiments, reflecting good electrochemical reversibility. Initial Coulombic efficiencies (CEff) of �63% (FeCl3), 66% (Fe(NO3)3), 72% (Fe2O3) and 76% (Fe) were achieved and differences can be due to solid electrolyte interphase (SEI) formation in the first cycle. In order to present the evolution of the CEff upon cycling more qualitatively, accumulated Coulombic inefficiencies (ACI) were plotted in Figure 6b. ACI can reflect parasitic reactions within the LIB cell, including but not limited to reductive electrolyte decomposition and loss of active lithium for interphase formation (SEI[43]).[44] As can be seen, all cells show a similar increase of ACI except for FeCl3-derived materials which exhibit a steady increase, which can be caused by distinct active lithium loss or more severe electrolyte degradation. This can be most likely related to the more amorphous character, and thus higher porosity and higher amount of surface groups in that sample.[42] According to the 1st cycle potential profiles and specific differential capacity (dQ/dV)vs. potential plots in Figure 6c–d, the highly graphitic character of the materials can be reflected in the characteristic peaks of staging phenomena of Li+ion intercalation into graphite at potentials <0.2 V vs. LijLi+.[45] By further examination of the potential profiles, the ascending trend regarding the DoG is also supported by increasing peak intensities, i.e., the highest intensity is achieved for the Fecarbonaceous sample and the lowest for FeCl3. Also the quite similar capacity for both Fe2O3and Fe(NO3)3derived carbonaceous materials supports the above mentioned positive effect of undissolved addition of the activator due to hydrolysis of Figure 6. a) Charge/discharge rate performance studies and b) accumulated Coulombic efficiencies at rates between 0.1 and 10 C of carbon jjLi metal cells for carbonaceous materials synthesized at 2000°C. c) 1st cycle potential profiles and d) differential capacity (dQ/dV) vs. potential profiles of carbon jjLi metal cells (three-electrode configuration, half-cell setup with Li metal as CE and RE). Electrolyte: 1 M LiPF6in EC/EMC (3:7) +2 wt% VC; potential range: 0.02 to 1.5 V vs. Li jLi+. ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (8 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 23/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Fe(NO3)3to Fe2O3resembling a ‘quasi in-situ’ physical addition process. To better elucidate the effect of different iron sources on the electrochemical performance, Figure 7 shows the 1st cycle CEff as well as specific delithiation capacities of coffee groundderived carbons at 0.1 C (averaged specific delithiation capacities of first three cycles) as a function of the DoG estimated from Raman measurements and the crystallite size estimated from PDF measurements. There seems to be a strong dependence between both structural parameters and the electrochemical results. As can be seen, maximum reversible capacities of �240 mAhg1(FeCl3), 290 mAhg1(Fe(NO3)3), 294 mAhg1 (Fe2O3), 320 mAhg1(Fe) were achieved at a rate of 0.1 C. Therefore, the higher the DoG and crystallite size, the higher the achievable capacity and CEff in the first cycle. Moreover, the more graphitic character of some samples result in lower nonbasal surfaces, as seen by SEM analysis, that might help with the formation of a more effective SEI, less active lithium losses and thus higher CEff values in the first cycles.[46] Although the electrochemical performance of Fe-derived materials is superior to that of other samples and to even that of petroleum-coke derived carbons reported in previous works using similar experimental conditions for cell testing,[45c] it is important to note that initial CEff values and initial capacities achieved herein are lower than those of commercial graphites.[4b,19c] The theoretical specific capacity of graphite is 372 mAhg1and values about 350–360 mAhg1are often achieved for commercial SG anodes. To meet the requirements for industrial application it is mandatory to achieve >99.9% CEff within the first few cycles. Furthermore, the materials need purities >99.95%, high crystallinity, appropriate cycle as well as calendar life and a fitting morphology. Further particle sizing and shaping, surface refinements or carbon coating approaches typically done in graphite anode material production industrially could also be performed to improve such values. However, further modification steps are beyond the goal of this work and can only be performed in kg-scale.[1b] The goal of this work was rather to provide a systematic study on the effect of different iron-based precursors on the graphitization efficiency. Future works will focus not only on further material optimization but also on recycling of the used iron-based activator during the synthesis which could be used in further ‘catalytic graphitization’ processes. When comparing the results reported here with previous literature (Table S4), Banek et al. investigated hardwood sawdust using 62.5 wt% of Fe-powder as activator resulting in a maximum capacity of 358 mAhg1.[25a] Gomez-Martin et al. also investigated ‘catalytic graphitization’ behavior of wood-based biomass precursors using FeCl3as activator. Reversible capacities of 307 mAhg1were achieved at 2000°C using ~35 wt% FeCl3.[19a] Our previous results from Hanhart et al. showed a saturation point regarding both, achievable DoGs and reversible capacities, between 40 and 50 wt% when using FeCl3as activator.[22a] After graphitization at 2000°C, a specific delithiation capacity of 243 mAhg1was reached from coffee ground with FeCl3which is ~60 mAhg1less than wood-based materials. Comparing results from coffee ground treated with Fe-powder to Banek’s wood materials, a difference of only 41 mAhg1in capacity is observed, also supporting the higher impact of Fe-powder regarding the graphitizing mechanism than volatile activators such as FeCl3. However, the highest achieved specific delithiation capacity here of 320 mAhg1is already close to capacities often achieved by commercial graphites, and might be increased by further optimization.[47] In light of these results, one should keep in mind that parameters such as starting biomass material and synthesis procedure itself (e.g., conventional heat-treatment vs. laser pyrolysis) differ for various reports and those differences can directly impact the graphitization behavior and might not be directly compared. However, our results suggest that it is worth further investigating the role of Fe-powder content to reveal whether there is also a saturation point in the degree of crystallinity and whether the particle size of the activator has also a key role on the graphitization.[25a] Figure 7. a) Averaged specific delithiation capacity of first three cycles and 1st cycle CEff as a function of the DoG and iron-based activator. b) Averaged specific delithiation capacity as a function of crystallite size as estimated from PDF results. ChemElectroChem Research Article doi.org/10.1002/celc.202201073 ChemElectroChem 2023,10, e202201073 (9 of 13) © 2023 The Authors. ChemElectroChem published by Wiley-VCH GmbH Wiley VCH Donnerstag, 23.02.2023 2305 / 280401 [S. 24/28] 1 21960216, 2023, 5, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/celc.202201073 by Universidad De Sevilla, Wiley Online Library on [19/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License