An Electrochemical Evaluation of Nitrogen-doped Carbons as Anodes for Lithium Ion Batteries

1 AN ELECTROCHEMICAL EVALUATION OF NITROGEN-DOPED CARBONS AS ANODES FOR LITHIUM ION BATTERIES A. Gomez-Martina*, J. Martinez-Fernandeza, Mirco Ruttertb, Martin Winterb, c, Tobias Plackeb, J. Ramirez-Ricoa, * a Dpto. Física de la Materia Condensada and Instituto de Ciencia de Materiales de Sevilla, Universidad de Sevilla – CSIC, Avda. Reina Mercedes SN, 41012 Sevilla, Spain bUniversity of Münster, MEET Battery Research Center, Institute of Physical Chemistry, Corrensstraße 46, 48149 Münster, Germany c Helmholtz Institute Münster, IEK-12, Forschungszentrum Jülich GmbH, Corrensstraße 46, 48149 Münster, Germany Abstract New anode materials beyond graphite are needed to improve the performance of lithium ion batteries (LIBs). Chemical doping with nitrogen has emerged as a simple strategy for enhancing lithium storage in carbon-based anodes. While specific capacity and rate capability are improved by doping, little is known about other key electrochemical properties relevant to practical applications. This work presents a systematic evaluation of electrochemical characteristics of nitrogen-doped carbons derived from a biomass source and urea powder as anodes in LIB half- and full-cells. Results show that doped carbons suffer from a continuous loss in capacity upon cycling that is more severe for higher nitrogen contents. Nitrogen negatively impacts the voltage and energy efficiencies at low charge/discharge current densities. However, as the charge/discharge rate increases, the voltage and energy efficiencies of the doped carbons outperform the non-doped ones. We provide insights towards a fundamental understanding of the requirements needed for practical applications and reveal drawbacks to be overcome by novel doped carbon-based anode materials in LIB applications. With this work, we also want to encourage other researchers to evaluate electrochemical characteristics besides capacity and cycling stability which are mandatory to assess the practicality of novel materials. * Corresponding author. Email: [email protected], Tel: +34 9545 56423 *Corresponding author. Email: [email protected], Tel: +34 9545 50963 2 1. Introduction Lithium ion batteries (LIBs) are an essential technology for certain purposes such as electric automotive applications and renewable energy sources due to their superior gravimetric and volumetric energy and power densities, and low self-discharge when compared to other energy storage technologies [1-5]. However, a major challenge facing current LIBs is that state-of-the-art graphite anodes can no longer satisfy the ever-growing energy demands in terms of energy density and rate capability [1, 6]. Graphite represents so far the material of choice as negative electrode (=anode) for LIBs due to its combination of appropriate features such as low average discharge potential (≈0.2 V vs. Li|Li+), low cost and low voltage hysteresis yielding high energy efficiencies [7-10]. Unfortunately, graphite-based anode materials typically display a poor rate capability, especially at high charging rates (“fast charging”) and/or at low operating temperature, which results in a high risk for lithium metal plating at the graphite anode surface [8]. In turn, high charging rates might have a significant impact on the lifetime and safety of LIB cells. Despite intensive efforts to develop alternative anode materials [11], carbon-based electrodes are still at the forefront of research due to their low cost, tunable structure and possible eco-friendly synthesis route by direct carbonization of sustainable biomass resources [12, 13]. In previous works, alternative electrodes based on either nanostructured disordered carbons [14], graphene [15] or heteroatom-doped disordered carbons[16, 17] have shown promising rate capabilities as anode materials for LIBs owing to the shortened diffusion pathways of Li ions and minimized diffusive resistance. Doping of carbon precursors with surface heteroatoms, such as sulfur, boron, phosphorus and nitrogen, is a straightforward strategy to modify the physicochemical and electronic properties of a host carbon network [18]. In the case of LIBs, the addition of surface heteroatoms into carbon scaffolds typically results in additional Li ion adsorption sites and improved electrode wettability in comparison to non-doped carbons [19]. Among the different dopants, nitrogen has drawn the most attention because of the enhanced polarity and electronic conductivity, added active sites within the carbon lattice [20, 21] and stronger electronegativity of nitrogen (χ =3.04) compared to that of carbon atoms (χ = 2.55) according to Pauling scale [22]. Nitrogen atoms incorporated into a carbon network can strengthen the interaction between anchor sites doped with nitrogen and alkali metal ions due to the improved local density [18], which favors the insertion and reduces the charge transfer resistance and diffusion barriers [16]. The process of lithiation (charge) of nitrogen-doped carbons is believed to take place not only through a diffusion process entailing Li 3 insertion between adjacent graphene layers, but also through a capacitive process, i.e., by surface redox pseudo capacitance and/or double-layer capacitance, attributed to Li adsorption at the surface of nitrogen heteroatoms [23, 24]. Routes previously addressed in literature for doping carbons with nitrogen-heteroatoms can be distinguished as “in-situ” and “post” treatments [25]. Direct formation treatments include thermal annealing in ammonia flowing atmosphere [26], arc-discharge methods [27], liquid plasma methods [28] and chemical vapor deposition [29]. However, these processes are expensive, afford little control over doping homogeneity and have low yields. Post-treatment methods include hydrothermal treatment [30] and pyrolysis of nitrogen-rich precursors or of carbon resources mixed with nitrogenrich compounds [31]. One easy, inexpensive and eco-friendly method involves the mixing of carbon materials (or biomass resources [32]) with low-cost nitrogen-rich precursors such as urea and melamine followed by thermal annealing under inert atmosphere [33, 34]. During pyrolysis, the decomposition products of urea or melamine are incorporated into the resulting carbon, doping the surface with nitrogen. This process enables controlling the final nitrogen content by the mass ratio between nitrogen and carbon sources, and peak treatment temperature. Previous works have reported an increase in reversible capacity and an improvement in rate capability with increasing nitrogen content [17, 30, 35-40]sparking interest in its use as a dopant in silicon anodes [41] and other emerging post-LIB technologies such as sodium ion batteries [42, 43], lithium-sulfur batteries [44], lithium ion capacitors [45] and potassium ion batteries [46, 47]. Most of the works available in the literature only focus on improving reversible capacities and rate capability of these anode materials, disregarding other parameters of crucial importance – e.g. Coulombic efficiency (CEff), average de-lithiation (discharge) potential and voltage (VEff) and energy (EEff) efficiencies [10]. In particular, little attention is being paid to the large voltage hysteresis, high working de-lithiation potential and sloping-trend often observed in the potential profile of some nitrogen-doped carbons [48], in contrast to potential plateaus observed for lithiation/de-lithiation of graphite. While it is true that these characteristics could be suitable for supercapacitor applications [49], they also inevitably impact the cell voltage (= difference between positive and negative electrode potential) for LIB applications as a high negative electrode working potential, reduces the cell voltage and, thus, the energy density of a LIB cell. It is not clear to which extent these electrochemical features could limit the application of these materials for LIBs applications, making it necessary to perform a critical evaluation of this point as a function of the nitrogen doping level. 4 In this work, a comprehensive analysis of electrochemical properties of nitrogen-doped carbons as anode materials for LIB cells was performed. As representative and simple route, nitrogen-doped porous carbons were synthetized by a two-step process. First, a biomass resource as carbon precursor and urea powder as nitrogen precursor were homogenized by a ball-milling process. Afterwards, the mixtures were subsequently pyrolyzed under an inert atmosphere. Different carbon precursors and urea powder weight ratios were considered to study the effect of final nitrogen content on microstructural and textural features. The electrochemical properties of the resulting materials were critically evaluated as LIB anode materials in a half-cell [50] and pseudo full-cell setup with a capacity-overbalanced lithium nickel cobalt manganese oxide (NCM-111) cathode. Special attention was paid to the correlation between the N-doped carbon material properties and electrochemical characteristics, including the reversible capacity, rate capability, CEff as well as VEff and EEff. Post-mortem X-ray photoelectron spectroscopy (XPS) was used for investigating the surface chemistry of cycled electrodes. 2. Methods 2.1. Materials As-received olive stones (supplied by OleoMorón S.L) were first ground and sieved to a particle size lower than 75 µm. As a route to introduce homogeneous nitrogen-based surface functionality into the carbon material, olive stone powder was then mixed with urea powder (CO(NH2)2, purity ≥ 98%, Sigma Aldrich) by a planetary ball-mill (Pulverisette 6, Fritsch). The milling process was performed using an agate crucible and grinding agate balls (the ball-to-powder ratio was 3:1; four 13 mm and six 8.3 mm in diameter balls) at a rotation speed of 350 rpm for 2 h. Three different olive stone to urea weight ratios were considered in this study 1:0, 1:1 and 1:5. The resulting mixture was dried at 100 ºC for 48 h and then heated in a tube furnace under a flowing inert atmosphere of nitrogen (flow rate of ≈ 0.2 L·min-1,Alphagaz, 99.999 %) up to 800 ºC at a heating rate of 5 ºC·min-1, followed by a holding time of 60 min and then cooled down to room temperature. To avoid relevant microstructural changes in carbon microstructure and to facilitate investigating the direct influence of the amount of nitrogen dopant, only one peak heat-treatment temperature was considered for all samples. Resulting carbon materials will be referred throughout the text according to the corresponding weight ratio to as biomass:urea 1:0, 1:1 and 1:5. 5 2.2. Microstructural and surface characterization Scanning electron microscopy (SEM, TENEO, FEI) and transmission electron microscopy (TEM, Talos F200S,FEI) analyses were performed to characterize the surface morphology and microstructural features. Thermal analysis during heat-treatment was studied by thermogravimetric (TGA) and differential scanning calorimetry (DSC) experiments using a dual thermobalance (Thermal Advantage SDT-Q600, TA Instruments). Measurements were carried out applying a heating rate of 10 ºC·min-1 up to 800 ºC under a nitrogen flow rate of 100 ml·min-1. Calibration tests were periodically performed using a standard sapphire sample. Raman spectroscopy measurements (Bruker SENTERRA, Bruker Optics Inc.) were carried out using a green excitation wavelength of 532 nm and a laser power of 10 mW. In order to collect the spectra, 10 integrations were carried out with a time of 60 s. The spectra shown is the average of three measurements at different sample positions. Raman spectra were fitted to pseudo-Voigt functions using a least-square house-made code implemented in MATLAB considering all the relevant Raman bands of carbon materials. X-ray photoelectron spectroscopy (XPS) analyses of doped carbon powders were carried out on an Axis Ultra DLD spectrometer (Kratos) using a monochromatic Al Kα radiation (hv = 1486.6 eV; 10 mA emission current and 12 kV accelerating voltage) to analyze the surface composition of carbon materials. The binding energy (BE) was calibrated against the C1s peak of carbon (284.5 eV). To ensure the reproducibility of the results, three measurements per sample were performed. Data processing, curve fitting and peak deconvolution was performed using CasaXPS software and pseudo-Voigt line shapes. Atomic concentrations were calculated using integrated peak intensities. Additionally, postmortem XPS investigations of the cycled electrodes (after 100 cycles; biomass:urea 1:0 and 1:5 samples) were performed in order to investigate changes in solid electrolyte interphase (SEI) formation and active lithium losses. For that, electrodes were harvested from cycled pseudo full-cells in an argonfilled glovebox and washed with 200 µL of dimethyl carbonate (DMC; BASF, battery grade) prior to surface analysis, in order to remove electrolyte residues and LiPF6 salt. Textural properties were evaluated by nitrogen adsorption/desorption measurements at 77 K (ASAP 2420, Micromeritics). Prior to the measurement, samples were degassed at 200 ºC overnight under reduced pressure. Specific surface areas were determined by applying the Brunauer-Emmett-Teller (BET) theory to the adsorption isotherms. Pore size distributions were estimated by density functional (DFT) theory. 6 2.3. Electrode preparation Carbon samples obtained after heating at 800 °C were first ground with a mortar and pestle and sieved down to 45 µm. Working electrodes (WE) consisted of a mixture of 90 wt. % carbon as the active material, 5 wt. % sodium-carboxymethyl cellulose as the binder (Na-CMC, Walocel CRT 2000 PPA 12, Dow Wolff Cellulosics) and 5 wt. % Super C65 as conductive agent (Imerys Graphite & Carbon). The WE paste was prepared using deionized waster as solvent. The electrode paste was then completely homogenized by a high-energy T25 Ultra Turrax disperser (1 hour, 10.000 rpm) and coated on a dendritic copper foil (Carl Schlenk AG) using a doctor blade. Electrode sheets were punched into circular discs (12 mm in diameter (Ø)) and dried under reduced pressure (pressure level below 10-2 bar) using a BÜCHI® oven at 120 ºC overnight. The average active mass loading of the WE was about 2 - 2.5 mg·cm-2. 2.4. Cell assembly and electrochemical characterization Electrochemical investigations were first carried out in a half-cell configuration using three-electrode Swagelok-type T-cells and high-purity metallic lithium foil (Albemarle Corporation, battery grade) as counter (CE) and reference (RE) electrodes. Cells were assembled in a dry room with a dew point of at least -50 °C. 1 M lithium hexafluorophosphate (LiPF6) in ethylene carbonate (EC)/ethyl methyl carbonate (EMC) 3:7 (w:w) plus 2 wt. % vinylene carbonate (VC) as SEI forming additive [51] was used as the electrolyte (supplied by BASF SE). Six and three-layers Freudenberg FS2190 discs (Ø = 13 mm in diameter between WE and CE and Ø = 8 mm in diameter for RE) were used as separators. The separators for the WE/CE and RE were wetted with 120 µL and 80 µL of electrolyte, respectively. For the electrochemical investigations in carbon || Li metal cells, constant current charge/discharge experiments were carried out within cut-off potentials between 0.02 (lithiation) and 1.5 V (delithiation) vs. Li|Li+ using a Maccor Series 4000 automated test system (Maccor, Inc.). For comparison, electrochemical properties of a commercial synthetic graphite as anode material for LIBs (SMG A4; active mass loading 2.3 mg·cm-2, BET specific surface area 1.58 m2·g-1, average particle = 4.32 ± 0.04 µm, Hitachi) were also investigated. Rate experiments at different specific currents between 37.2 mA·g-1 to 7440 mA·g-1 were carried out as follows: first three formation cycles at 37.2 mA·g-1, followed by 5 cycles at 372 mA·g-1 and each 5 cycles at specific currents of 37.2, 74.4, 186, 372, 744, 1166, 1860, 3720, and finally, 10 cycles turning back to 372 mA·g-1. Biomass:urea 1:0 and 1:5 samples were also studied at specific currents of up to 7440 mA·g-1. Long-term cycling stability experiments (carbon || Li metal cells) were carried out by performing three formation cycles at 37.2 mA·g-1 and then 200 cycles at 186 mA·g-1 in a potential window between 0.02-1.5 V vs. Li|Li+. 7 In order to evaluate reaction mechanisms, cyclic voltammetry (CV) experiments were carried out in a potential range between 0.02 and 1.5 V vs. Li|Li+ by using a VMP multichannel potentiostatic– galvanostatic system (Biologic Science Instruments) connected to a climatic chamber at 20 °C (KBF 115, Binder). For estimation of capacitive reaction and diffusion reaction contributions, CV experiments were performed at different scan rates: first three SEI formation cycles at 0.1 mV·s-1, followed by 3 cycles at 0.2, 0.4, 0.6, 0.8, 1, 2, 5, 10 and 20 mV·s-1. Further electrochemical investigations were performed in a LIB pseudo full-cell setup. Three-electrode Swagelok-type T-cells consisting of Li metal as RE, nitrogen-doped biomass derived carbon as negative electrode and a capacity-oversized lithium nickel cobalt manganese oxide (LiNi1/3Mn1/3Co1/3O2, NCM-111) as positive electrode was considered in order ensure excess of active lithium within the cell.The capacity-oversized NCM-111 electrodes were produced by an in-house pilot line (95 wt.% active material, 3wt. % PVDF as binder and 2 wt.% Super C65 as conductive agent) with an areal capacity of ≈2.3 mAh·cm-2, resulting in an anode to cathode capacity balancing of ≈1:2.3 in the pseudo full-cell investigations. Three-layer Freudenberg FS2190 discs (Ø = 13 mm in diameter between WE and CE and Ø = 8 mm in diameter for RE) were used as separators. Long-term cycling stability experiments (NMC-111 || carbon full-cells) were carried out by performing three formation cycles at 37.2 mA·g-1 and then 100 cycles at 186 mA·g-1. During these pseudo full-cell experiments, the cut-off potentials of the negative electrodes were controlled and set to 0.02 V vs. Li|Li+ and 1.5 V anode vs. Li|Li+. For all experimental cycling experiments, three cells were evaluated for each sample to ensure a high reproducibility of the results. The associated standard deviation is represented as error bars in the corresponding figures. The CEff, (in %) of a lithium ion cell is an indicator of the reversibility of redox reactions and reflects the ratio of the delivered discharge capacity (=de-lithiation in this case) (Qd) to the capacity necessary to charge (=lithiation in this case) (Qc) the system over a full cycle [52]. This parameter should be close to 100 % and it is generally expressed in percentage as follows [53]: CEff (%)=100·Qd Qc (1) The EEff and V Eff reflect the ratio of the discharge energy/voltage to the charge energy/voltage, respectively [54]. When a constant current is applied in charge/discharge cycling experiments, EEff and VEff can be defined as follows: EEff (%)= 100 · ∫UdIddt td 0 ∫UcIcdt tc 0 (2) 8 VEff (%)= 100· Ud Uc (3) Where Ud and Uc represent the discharge and charge cell voltages, Id and Ic the discharge and charge current and td and tc represent the discharge and charge time, respectively. Ud and Uc refer to the average discharge voltage and average charge voltage, respectively. When a constant current is applied, the EEff can be defined as the product of CEff and VEff as follows: EEff = CEff ·∫Uddt td 0 td ∫UcIcdt tc 0 tc=CEff ·Ud Uc=CEff ·VEff (4) The VEff and EEff were estimated following the work of Meister et al.[10]. For the calculations, a virtual lithium iron phosphate (LFP) cathode with a constant potential of 3.4 V vs. Li|Li+ was considered as the positive electrode. The average de-lithiation working potential was defined at half of the discharge/de-lithiation capacity. 3. Results and discussion 3.1. Material characterization Nitrogen-doped carbons were synthetized by a pyrolysis process of a mixture of olive stones as carbon source and urea powder as nitrogen-rich source. The thermal behavior during pyrolysis was investigated by TGA/DSC analysis, and results are detailed in Figure S1 (Supplementary Information). The remaining yield decreased from 30 wt. % to 5 wt. % when increasing the biomass:urea weight ratio from 1:0 to 1:5. Figure 1-a, b shows scanning electron microscope (SEM) micrographs of representative samples. The carbon scaffold derived from the raw biomass precursor (Figure 1-a; biomass:urea 1:0) exhibits a porous structure with micron-sized slits and non-directional cylindrical pores. After the ball-milling homogenisation process with urea powder and subsequent pyrolysis process (Figure 1-b; biomass:urea 1:5), the resulting carbon preform shows a more irregular surface morphology with no evident porous channels. Close-up high-resolution TEM images (Figure 1-c) of doped carbon (biomass:urea 1:5 sample) shows a typical turbostratic disordered microstructure composed by a network of nanosized carbon crystallites arranged in randomly-oriented layers, according to the nongraphitizable nature of biomass [55]. No evident difference in such a microstructure was observed while varying the urea content ratio. The EDX elemental mappings of biomass:urea 1:5 sample shown in Figure 1-d-g prove the uniform distribution of nitrogen (green coloured area, Figure 1-f) on the carbon particle surface (red coloured area, Figure 1-e). The remaining oxygen present in the sample 9 (blue coloured area, Figure 1-g) comes from the uncompleted thermal decomposition of polysaccharide chains of lignocellulosic-based polymers at a pyrolysis temperature as low as 800 ºC. The surface chemical composition and nitrogen bonding configuration of doped carbons was further analysed by XPS. A summary of integrated atomic concentrations is listed in Table 1. The full-range XPS survey spectra (shown in the Supplementary Information, Figure S2) exhibits characteristic peaks centred at ≈285, 400, 533 eV, corresponding to C1s, N1s and O1s, respectively. Impurity phases of Si (< 4 at. %) and Na (< 2 at. %) were also detected, which may be inherent to the olive stone precursor and can be attributed to the inorganic fraction often incorporated during biomass growth [56]. The carbon content was ≈80 at. % irrespectively of the sample. With an increase in the starting urea content, nitrogen atomic concentration ranged between 0, 4.1 and 12.7 at. % for biomass:urea 1:0, 1:1 and 1:5 samples, respectively (Table 1). Thus, these results in combination with TEM elemental mapping confirm that the mixing of the biomass precursor with urea powder helped in homogeneously doping the resulting carbon scaffold with nitrogen. In contrast, the oxygen content decreased from 13 to 3 at. % from the sample with the highest concentration (biomass:urea 1:5) to that with the lowest concentration (biomass:urea 1:0) of nitrogen species. Figure 1. Representative SEM micrographs of a) biomass:urea 1:0 and b) biomass:urea 1:5 samples. c) Representative TEM micrograph of biomass:urea 1:5 sample. EDX elemental mappings of biomass:urea 1:5 sample (d) corresponding to e) C, f) N and g) O. h) Schematic structure of nitrogen-doped graphene monolayer. i) Results of N1s spectra fitting of XPS analysis. 16 proportionally to the content of nitrogen-functional groups and considerably exceeds to that of graphite, meaning that the Li insertion occurs at a more negative potential than the Li de-insertion. Although not investigated in detail for nitrogen-doped carbons, the voltage hysteresis in hydrogencontaining carbons is believed to be caused by the difference in energy involved in the change of the bonding of the nearby carbon atoms [71]. While the VEff is higher at low currents, this tendency is reversed at higher currents. In turn, at 1116 mA·g-1, the lower voltage hysteresis (Table S2) of biomass:urea 1:5 sample as compared to less doped carbons results in a higher VEff (and thus a higher EEff). At 37.2 mA·g-1, average de-lithiation potentials were ≈0.49 V (biomass:urea 1:0), ≈0.67 V (biomass:urea 1:1), ≈0.77 V (biomass:urea 1:5) and ≈0.18 V (graphite) vs. Li|Li+. The average delithiation potential is a key parameter for application as anode materials in LIB cells and it should be as low as possible to increase the cell voltage and thus to achieve the maximum energy density, while still being sufficiently high to avoid Li metal plating at the top of the anode, e.g., during fast-charging. The low working potential and almost constant potential profile (≈0.2 V vs. Li|Li+) of graphite makes it an ideal anode material for LIB cells. A raised average de-lithiation potential as a result of a large hysteresis negatively affects the cell voltage in a full-cell setup. The application of nitrogen-doped carbons for applications requiring high-energy density would be thus strongly hindered compared to graphite, as they turn out to be more suitable for high-power applications. As for conversion-type electrode materials, nitrogen-doped carbons can also enhance the reversible capacity at the expense of increasing the cut-off potential above 1.5 V vs Li|Li+. Figure 5-a displays the influence of increasing the cut-off potential from 1.5 to 3 V (vs. Li|Li+) at 186 mA·g-1, whereas Figure 5-b shows corresponding representative potential profiles of biomass:urea 1:5. While in lessdoped samples (biomass:urea 1:0 and biomass:urea 1:1) this change does not impact as much, the specific capacity for the biomass:urea 1:5 electrode can increase from ≈300 mAh·g-1 (0.02-1.5 V vs. Li|Li+) to ≈380 mAh·g-1 (0.02-3.0 V vs. Li|Li+) by increasing the cut-off potential due to an enhanced sloping contribution above 1.5 V (vs. Li|Li+). All the nitrogen content may not be contributing to storage process when setting an upper cut-off potential of 1.5 V (vs. Li|Li+). It is generally believed that the capacity below 0.5 V (vs. Li|Li+) is related to Li insertion into graphene layers, whereas the capacity above 0.5 (vs. Li|Li+) is related to surface-dominated reactions resulting from the adsorption of Li+ on pyridinic or pyrrolic defect structures [24, 72]. The work of Hu et al.[73] suggested that the pyridinic nitrogen dominated the capacity enhancement at a potential below 3 V (vs. Li|Li+), while the pyrrolic nitrogen contributed to the Li-ion storage below 1.5 V (vs. Li|Li+). However, although 17 commonly disregarded in the majority of works on doped carbons, it has to be noted that an enhanced upper cut-off potential could stress the electrode, speeding up ageing effects or resulting in the oxidation/dissolution of the SEI layer or even of the Cu current collector (above ≈3.5 V vs. Li|Li+)[74]. Figure 5. a) Specific discharge capacities and b) representative potential profiles at different upper cut-off potentials from 1.5 to 3.0 V vs. Li|Li+. Half-cell setup: carbon (WE)||Li metal (CE) cells using Li metal as RE. In this work, we achieved practical capacities in the range of ≈300-400 mAh·g-1 depending on the nitrogen content and upper cut-off potential, while previously reported capacity values of biomassderived nitrogen-doped carbon range between ≈400 and 820 mAh·g-1 [60, 75-79]. It must be noted that a fair comparison of electrochemical data is difficult as carbon source, processing temperatures, nitrogen content and electrochemical protocols vary from one paper to the next. In particular, a lower cut-off potential close to 0V can result in lithium plating at the carbon electrode and increased capacity, but such conditions are unrealistic for practical applications. Our work is focused on the effect of nitrogen functionality on electrochemical properties, instead of looking for the best material with the highest possible reversible capacity. 3.2.2. Electrochemical investigations in pseudo full-cells and post-mortem XPS analysis To avoid reactions occurring at the Li metal CE which may influence the performance in a half-cell setup, electrochemical investigations were also performed in pseudo full-cells, containing a capacityoversized NCM-111 cathode in order to ensure that there is an excess of active lithium within the cell. Figure 6-a shows the long-term cycling stability experiments for 100 charge/discharge cycles at 18 186 mA·g−1 for the different anode materials (NCM-111 || carbon full-cell setup). The capacity retention after 100 cycles, in comparison to the capacity at the 4th cycle (first cycle after formation) amounts to 98, 77 and 78% for biomass:urea 1:0, 1:1 and 1:5 samples, respectively. One of the often-disregarded limitations of doped carbon materials is the extremely poor 1st cycle CEff. Reported CEff values are far from those of commercial graphites which may achieve a first cycle CEff of ≥90-95% (according to graphite manufacturers). Values of only 39 ± 4% (biomass:urea 1:0), 40 ± 1% (biomass:urea 1:1), 41 ± 1% (biomass:urea 1:5) were achieved, likely as a result of the large BET specific surface area, surface heterogeneity and high chemical activity as a result of surface heteroatom groups [80]. A 1st cycle CEff of 50 % for the anode, using a realistic capacity balancing of ≈1.1:1 would lead to severe capacity fading after the first cycle due to significant consumption of active lithium from the cathode. Interestingly, an increase in the nitrogen content shows no evident influence on the 1st cycle CEff. The evolution of the accumulated irreversible capacity (QAIC)[54] upon the long-term cycling experiments is shown in Figure 6-b. The curve of the undoped carbon electrode (biomass:urea 1:0) shows a small slope after the SEI formation cycles (3rd cycle) and slightly increases upon cycling due to a decrease in decomposition reactions, with a maximum QAIC value of 549 mAh·g-1 after 100 cycles. The QAIC values add up to 890 mAh·g-1 (biomass:urea 1:1) and 907 mAh·g-1 (biomass:urea 1:5) after 100 charge/discharge cycles. These large values indicate severe parasitic side reactions that might stem from the ongoing SEI formation and decomposition of electrolyte but can also be related to lithium trapping within the anode. The trend of the accumulated values in half-cells (shown in Figure S5, Supplementary data) correlates well with that observed in the pseudo full-cells, as especially the 1:5 samples show steadily increasing QAIC values. Figure S9 shows representative potential profiles of different cycles during the long-term experiment (5th, 25th, 50th and 100th cycles) for the different anode materials. Previous investigations on the effect of nitrogen doping on Li diffusion by density functional theory simulations revealed that the strong Coulombic attraction between pyridinic/pyrrolic nitrogen and the adsorbed Li+ may hinder the delithiation process [69]. Furthermore, Machnikowski et al. [81] reported that the Qirr of nitrogen-doped carbon increased in a direct relation to the proportion of pyridinic groups. As pyridinic nitrogen atoms are located at the edge sites of graphene layers, they would act as active sites for electrolyte decomposition and solvated Li+ trapping, resulting in an enhanced QAIC for the more doped samples. 19 Figure 6. a) Cycling performance of carbonaceous anode materials at a specific current of 186 mA·g-1 for 100 cycles after three SEI formation cycles at 37.2 mA·g-1. b) Accumulated irreversible specific capacity vs. cycle number for constant current charge/discharge at 186 mA·g-1 during longterm cycling experiments. Pseudo full-cell setup: NCM-111 || carbon (N:P ratio 1:2.3) using Li metal as RE. Potential range of anode: 0.02-1.5 V vs. Li|Li+. c) F1s XPS core level spectra of the negative electrodes (biomass:urea 1:0 and 1:5) after cycling in NCM-111||carbon pseudo-full cells after 100 cycles. Post-mortem XPS investigations of the cycled electrodes (after 100 cycles; biomass:urea 1:0 and 1:5) were performed in order to investigate changes in SEI formation and active lithium losses. Detailed core spectra and summary of atomic concentrations are included in the Supplementary data (Figure S10). It should be noted that the results from the cycled electrodes should not be compared with the XPS analysis of pristine electrodes as XPS is a surface sensitive technique (information depth ≈10 nm) and thus predominantly probes the SEI and not only the active material. The presence of more nitrogen functionality severely increases the amount of fluorine species from 9.5 (biomass:urea 1:0) to 13.6 (biomass:urea 1:5) at. % after 100 cycles, indicating further decomposition of LiPF6. The relative amounts of carbon and oxygen do not significantly change between both electrodes. A slightly higher amount of Li is detected at the top layer of the biomass:urea 1:5 electrode, which indicates increased consumption of active Li. However, the relative concentration of Li estimated by XPS should not be directly correlated to the accumulated irreversible capacity (capacity loss) observed in the pseudo fullcells, because, as stated before, the analysis only elucidates ≈10 nm of the SEI. The F1s core level XPS spectra (Figure 6-c) shows two major components related to the formation of LiF and LixPOyFzand/or LixPFy phases resulting from the decomposition of the LiPF6salt [82]. For the electrode with a higher nitrogen content (biomass:urea 1:5; green curve), these two peaks exhibit a higher intensity in comparison to the undoped material (biomass:urea 1:0; red curve), indicating that 20 the addition of nitrogen enhances the decomposition of the conductive salt for the SEI formation. In the O1s spectra (Figure S10-b), carbonaceous electrodes exhibit two overlapping peaks centered at 531.4 eV (C=O) and 533.6 eV (C-O) related to lithium alkyl carbonates and Li2CO3[83], respectively. As can be seen, the ratio of C=O to C-O is enhanced for the most doped sample, which could be related to a further contribution of Li2CO3. Overall, further extensive electrochemical studies of capacity-balanced LIB full-cells and complementary analytical studies will be mandatory to clarify the mechanism of the enhanced irreversible capacities over cycling, e.g., in order to distinguish between parasitic reactions due to SEI formation and irreversible lithium trapping [84]. 4. Conclusions In this work, the electrochemical properties of nitrogen-doped biomass-derived carbons as anode materials in LIB cells were critically discussed in terms of the nitrogen content. Nitrogen-doped carbons were synthetized by a ball-milling homogenization process of biomass resources and urea with different weight ratios, followed by heat-treatment under inert atmosphere. This two-step process allows for the incorporation of nitrogen homogeneously on the carbon particle surface, with atomic concentrations ranging between 0 and 12.7 at. % for biomass:urea 1:0 and 1:5 samples, respectively. Based on our results, a higher nitrogen proportion results in an improved rate capability and enhanced maximum specific capacity as anode materials for LIB cells. However, these electrode materials suffer from disadvantages not frequently analyzed in the literature that may hinder their practical application. The energy efficiency (EEff), which means the ratio of input and output energies, decreased with increasing nitrogen content at the lowest current as a result of a large voltage hysteresis. EEff average values of ≈91% (biomass:urea 1:0), ≈87% (biomass:urea 1:1), ≈86% (biomass:urea 1:5) and ≈97 % (graphite) were reported. However, while non-doped carbons exhibited a declining trend in the EEff with increasing charge/discharge specific current, the EEff remained almost constant for the most doped sample up to 1860 mA·g-1. The working de-lithiation potential increased as a function of the nitrogen doping to almost 0.8 V vs. Li|Li+. Unfortunately, this would result in a decreased cell voltage in a fullcell setup and, therefore, a decreased energy density of the cell. Pseudo full-cell investigations using a capacity-oversized NCM-111 cathode with an excess of active Li revealed that the accumulated irreversible capacity continuously increased upon cycling for nitrogen-doped samples as a result of irreversible reactions. Post-mortem XPS analysis confirmed that 21 the high nitrogen content enhances further electrolyte decomposition and active Li losses in comparison to the undoped material. With this analysis we emphasize that researchers should focus on the practical aspects such as voltage hysteresis, energy efficiency and Coulombic efficiency when considering novel anode materials. The application of nitrogen-doped carbons for applications requiring high-energy density would be strongly hindered as electrochemical properties are not competitive to those of graphite anodes despite the better rate capability. According with our results, these materials will be more suitable for highpower applications. Nevertheless, the large irreversibility of the storage reactions remains as a major drawback and needs to be tackled by improved material designs, e.g., by reducing the large irreversible capacity in the first charge/discharge cycles. Acknowledgements The authors from University of Sevilla acknowledge funding by the Spanish Government Agency Ministerio de Economía y Competitividad under grant MAT2016-76526-R. SEM, TEM and textural characterization was performed at the CITIUS central services of the University of Seville (Spain). A. Gómez-Martín is grateful to the German Academic Exchange Service (DAAD) for funding her time at MEET under the “Short-term research grants 2018" program. The authors from University of Münster thank the MWIDE NRW for funding this work in the project “GrEEn” (funding code: 313- W044A). We are thankful to R. Nölle and L. Hanke for XPS measurements. 22 Captions Figure 1. Representative SEM micrographs of a) biomass:urea 1:0 and b) biomass:urea 1:5 samples. c) Representative TEM micrograph of biomass:urea 1:5 sample. EDX elemental mappings of biomass:urea 1:5 sample (d) corresponding to e) C, f) N and g) O. h) Schematic structure of nitrogendoped graphene monolayer. i) Results of N1s spectra fitting of XPS analysis. Figure 2. Rate capability of different anode materials: a) Specific discharge capacity vs. cycle number using specific charge/discharge currents between 37.2 and 7440 mA·g-1 and b) Capacity retention as a function of specific charge/discharge current, whereas the capacity at 37.2 mA·g-1 was defined as 100 % for each material. Error bars: standard deviation of three cells for each sample. Rate performance investigations: Cycles 1 - 3: 37.2 mA·g-1; Cycles 4 - 8: 372 mA·g-1; Cycles 9-59: specific currents of 37.2, 74.4, 186, 372, 744, 1166, 1860, 3720, 5580 and 7440 mA·g-1 for each step 5 cycles; Cycle 60 - onwards: 372 mA·g-1. Half-cell setup: carbon (WE)||Li metal (CE) cells using Li metal as RE. Potential range: 0.02 -1.5 V vs. Li|Li+. Figure 3. Evolution of a) EEff, b) CEff and c) VEff at different charge/discharge currents. Potential range: 0.02 V - 1.5 V vs. Li|Li+. Half-cell setup: carbon (WE)||Li metal (CE) cells using Li metal as RE. Error bars: standard deviation of three cells for each sample. The red dashed area points out the range where nitrogen-doped samples maintain constant EEff and VEff despite the increase in specific current. The dashed arrow indicates the evolution of the trend for the biomass:urea 1:5 sample. Figure 4. Representative potential as a function of the normalized capacity at a) 37.2 mA·g-1, b) 372 mA·g-1 and c) 1116 mA·g-1. The shadowed areas and arrows highlight the hysteresis between charge/discharge processes. Half-cell setup: carbon (WE)||Li metal (CE) cells using Li metal as RE. Figure 5. a) Specific discharge capacities and b) representative potential profiles at different upper cut-off potentials from 1.5 to 3.0 V vs. Li|Li+. Half-cell setup: carbon (WE)||Li metal (CE) cells using Li metal as RE. Figure 6. a) Cycling performance of carbonaceous anode materials at a specific current of 186 mA·g- 1 for 100 cycles after three SEI formation cycles at 37.2 mA·g-1. b) Accumulated irreversible specific capacity vs. cycle number for constant current charge/discharge at 186 mA·g-1 during long-term cycling experiments. Pseudo full-cell setup: NCM-111 || carbon (N:P ratio 1:2.3) using Li metal as RE. Potential range of anode: 0.02-1.5 V vs. Li|Li+. c) F1s XPS core level spectra of the negative electrodes (biomass:urea 1:0 and 1:5) after cycling in NCM-111||carbon pseudo-full cells after 100 cycles. 23 Tables Table 1. Summary of relevant structural properties, textural properties and surface chemistrya of nitrogen-doped carbon samples. [aThe concentration of pyridinic, pyrrolic, graphitic, pyridinic oxide nitrogen (at. %) is normalized to the total nitrogen content of each sample] Sample BET specific surface area (m2·g-1) 𝐈 𝐃 𝟏 / 𝐈 𝐆 Elemental composition (at. %) a C O N Total N pyridinic N pyrrolic N graphitic N pyr-oxide Biomass:urea 1:0 189.5 3.60 81.2 12.6 0 ---- ---- ---- ---- Biomass:urea 1:1 212.5 4.86 78.6 11.8 4.1 1.8 0.6 1.4 0.3 Biomass:urea 1:5 140.4 5.16 83.1 3.4 12.7 4.9 1.5 3.3 3.0 24 5. References [1] Schmuch R, Wagner R, Hörpel G, Placke T, Winter M, Performance and cost of materials for lithium-based rechargeable automotive batteries, Nature Energy 2018; 3(4):267-78. [2] Cano ZP, Banham D, Ye S, Hintennach A, Lu J, Fowler M, et al., Batteries and fuel cells for emerging electric vehicle markets, Nature Energy 2018; 3(4):279-89. [3] Blomgren GE, The Development and Future of Lithium Ion Batteries, J Electrochem Soc 2017; 164(1):A5019-A25. [4] Winter M, Barnett B, Xu K, Before Li Ion Batteries, Chem Rev 2018; 118(23):11433-56. [5] Placke T, Kloepsch R, Dühnen S, Winter M, Lithium ion, lithium metal, and alternative rechargeable battery technologies: the odyssey for high energy density, J Solid State Electrochem 2017; 21(7):1939-64. [6] Betz J, Bieker G, Meister P, Placke T, Winter M, Schmuch R, Theoretical versus Practical Energy: A Plea for More Transparency in the Energy Calculation of Different Rechargeable Battery Systems, Adv Energy Mater 2019; 9(6):1803170. [7] Wagner R, Preschitschek N, Passerini S, Leker J, Winter M, Current research trends and prospects among the various materials and designs used in lithium-based batteries, J Appl Electrochem 2013; 43(5):481-96. [8] Waldmann T, Hogg B-I, Wohlfahrt-Mehrens M, Li plating as unwanted side reaction in commercial Li-ion cells – A review, J Power Sources 2018; 384:107-24. [9] Winter M, Möller K, Besenhard J, Carbonaceous and Graphitic Anodes, in: G.A. Nazri, G. Pistoia (Eds.), Science and technology of advanced lithium batteries: basic aspects, New York: Kluwer Academic Publishers2003, p. 50. [10] Meister P, Jia H, Li J, Kloepsch R, Winter M, Placke T, Best Practice: Performance and Cost Evaluation of Lithium Ion Battery Active Materials with Special Emphasis on Energy Efficiency, Chem Mater 2016; 28(20):7203-17. [11] Zhu B, Wang X, Yao P, Li J, Zhu J, Towards high energy density lithium battery anodes: silicon and lithium, Chem Sci 2019; 10(30):7132-48. [12] Long W, Fang B, Ignaszak A, Wu Z, Wang Y-J, Wilkinson D, Biomass-derived nanostructured carbons and their composites as anode materials for lithium ion batteries, Chem Soc Rev 2017; 46(23):7176-90. [13] Gomez-Martin A, Martinez-Fernandez J, Ruttert M, Heckmann A, Winter M, Placke T, et al., Iron-Catalyzed Graphitic Carbon Materials from Biomass Resources as Anodes for Lithium-Ion Batteries, ChemSusChem 2018; 11(16):2776-87. [14] Holtstiege F, Koç T, Hundehege T, Siozios V, Winter M, Placke T, Toward High Power Batteries: Pre-lithiated Carbon Nanospheres as High Rate Anode Material for Lithium Ion Batteries, ACS Appl Energy Mater 2018; 1(8):4321- 31. [15] Al Hassan MR, Sen A, Zaman T, Mostari MS, Emergence of graphene as a promising anode material for rechargeable batteries: a review, Materials Today Chemistry 2019; 11:225-43. [16] Zheng F, Yang Y, Chen Q, High lithium anodic performance of highly nitrogen-doped porous carbon prepared from a metal-organic framework, Nature Communications 2014; 5:5261. [17] Wang J, Fan H, Shen Y, Li C, Wang G, Large-scale template-free synthesis of nitrogen-doped 3D carbon frameworks as low-cost ultra-long-life anodes for lithium-ion batteries, Chem Eng J 2019; 357:376-83. [18] Wang X, Sun G, Routh P, Kim D-H, Huang W, Chen P, Heteroatom-doped graphene materials: syntheses, properties and applications, Chem Soc Rev 2014; 43(20):7067-98. 25 [19] Tian W, Li H, Qin B, Xu Y, Hao Y, Li Y, et al., Tuning the wettability of carbon nanotube arrays for efficient bifunctional catalysts and Zn–air batteries, J Mater Chem A 2017; 5(15):7103-10. [20] Ma C, Shao X, Cao D, Nitrogen-doped graphene nanosheets as anode materials for lithium ion batteries: a firstprinciples study, J Mater Chem 2012; 22(18):8911-5. [21] Zhang X, Zhu G, Wang M, Li J, Lu T, Pan L, Covalent-organic-frameworks derived N-doped porous carbon materials as anode for superior long-life cycling lithium and sodium ion batteries, Carbon 2017; 116:686-94. [22] Yang Y, Jin S, Zhang Z, Du Z, Liu H, Yang J, et al., Nitrogen-Doped Hollow Carbon Nanospheres for High- Performance Li-Ion Batteries, ACS Appl Mater Interfaces 2017; 9(16):14180-6. [23] Zheng C, Hu X, Sun X, Yoo SJ, Li X, Large-scale synthesis of nitrogen-rich hierarchically porous carbon as anode for lithium-ion batteries with high capacity and rate capability, Electrochim Acta 2019; 306:339-49. [24] Tian L-L, Wei X-Y, Zhuang Q-C, Jiang C-H, Wu C, Ma G-Y, et al., Bottom-up synthesis of nitrogen-doped graphene sheets for ultrafast lithium storage, Nanoscale 2014; 6(11):6075-83. [25] Yen P-J, Ilango PR, Chiang Y-C, Wu C-W, Hsu Y-C, Chueh Y-L, et al., Tunable nitrogen-doped graphene sheets produced with in situ electrochemical cathodic plasma at room temperature for lithium-ion batteries, Materials Today Energy 2019; 12:336-47. [26] Bianco GV, Losurdo M, Giangregorio MM, Capezzuto P, Bruno G, Exploring and rationalising effective n-doping of large area CVD-graphene by NH3, PCCP 2014; 16(8):3632-9. [27] Glerup M, Steinmetz J, Samaille D, Stéphan O, Enouz S, Loiseau A, et al., Synthesis of N-doped SWNT using the arc-discharge procedure, Chem Phys Lett 2004; 387(1):193-7. [28] Senthilnathan J, Rao KS, Yoshimura M, Submerged liquid plasma – low energy synthesis of nitrogen-doped graphene for electrochemical applications, J Mater Chem A 2014; 2(10):3332-7. [29] Ito Y, Christodoulou C, Nardi MV, Koch N, Sachdev H, Müllen K, Chemical Vapor Deposition of N-Doped Graphene and Carbon Films: The Role of Precursors and Gas Phase, ACS Nano 2014; 8(4):3337-46. [30] Xing Z, Ju Z, Zhao Y, Wan J, Zhu Y, Qiang Y, et al., One-pot hydrothermal synthesis of Nitrogen-doped graphene as high-performance anode materials for lithium ion batteries, Scientific Reports 2016; 6:26146. [31] Arif AF, Chikuchi Y, Balgis R, Ogi T, Okuyama K, Synthesis of nitrogen-functionalized macroporous carbon particles via spray pyrolysis of melamine-resin, RSC Advances 2016; 6(86):83421-8. [32] Zheng F, Liu D, Xia G, Yang Y, Liu T, Wu M, et al., Biomass waste inspired nitrogen-doped porous carbon materials as high-performance anode for lithium-ion batteries, J Alloys Compd 2017; 693:1197-204. [33] Soares OSGP, Rocha RP, Gonçalves AG, Figueiredo JL, Órfão JJM, Pereira MFR, Easy method to prepare N-doped carbon nanotubes by ball milling, Carbon 2015; 91:114-21. [34] Schneidermann C, Jäckel N, Oswald S, Giebeler L, Presser V, Borchardt L, Solvent-Free Mechanochemical Synthesis of Nitrogen-Doped Nanoporous Carbon for Electrochemical Energy Storage, ChemSusChem 2017; 10(11):2416-24. [35] Mondal AK, Kretschmer K, Zhao Y, Liu H, Fan H, Wang G, Naturally nitrogen doped porous carbon derived from waste shrimp shells for high-performance lithium ion batteries and supercapacitors, Microporous Mesoporous Mater 2017; 246:72-80. [36] Jang J, Kim H-e, Kang S, Bang JH, Lee CS, Urea-assisted template-less synthesis of heavily nitrogen-doped hollow carbon fibers for the anode material of lithium-ion batteries, New J Chem 2019; 43(9):3821-8. [37] Wang H, Zhang C, Liu Z, Wang L, Han P, Xu H, et al., Nitrogen-doped graphene nanosheets with excellent lithium storage properties, J Mater Chem 2011; 21(14):5430-4.