Fe3O4/Graphene Composite Anode Material for Fast-Charging Li-Ion Batteries

Composite anode material based on Fe3O4 and reduced graphene oxide is prepared by base-catalysed co-precipitation and sonochemical dispersion. Structural and morphological characterizations demonstrate an effective and homogeneous embedding of Fe3O4 nanoparticles in the carbonaceous matrix. Electrochemical characterization highlights specific capacities higher than 1000 mAh g−1 at 1C, while a capacity of 980 mAhg−1 is retained at 4C, with outstanding cycling stability. These results demonstrate a synergistic effect by nanosize morphology of Fe3O4 and inter-particle conductivity of graphene nanosheets, which also contribute to enhancing the mechanical and cycling stability of the electrode. The outstanding capacity delivered at high rates suggests a possible application of the anode material for high-power systems.


Introduction
Current concerns about limited energy resources, coupled with the need to decrease greenhouse gas emissions, are leading worldwide research and development efforts to enhance efficient utilization of renewable energies as main energy sources.
Due to the fitful nature of renewable sources' availability and energy needs, efficient energy storage for stationary and mobile applications is mandatory. Among the available technologies, Li-ion batteries have played a main role for the last three decades, due to their current performances and development prospects. However, to fulfil the needs of emerging and consolidating technologies, such as electrified vehicles, the research activity has had to focus efforts on new materials with improved power and energy densities.
On the anode side, insertion materials represent the state-of-art, especially graphite [1], thanks to its unique features, such as flat and low working potential vs. lithium, low cost, and good cycle life. However, graphite allows the intercalation of a limited number of lithium ions dictated by its layered structure, leading to a stoichiometry of LiC 6 as the end-term of intercalation process and a specific capacity limited to 372 mAh g −1 [2].
In order to overcome these drawbacks, research has focused on different reactivity concepts, such as the conversion process, entailing the reversible electrochemical reaction of lithium with transition metal oxide (TMO). The conversion reaction can be generalized as in Equation (1): where M is a transition metal such as Ni, Mn, Fe, etc. Several TMOs have been studied as possible candidate anodes for LIBs [3], such as Co 3 O 4 [4], Fe 2 O 3 [5] and Fe 3 O 4 [6]. Among them, the very large capacities associated O bonds [25,26]. The D-band located at 1288 cm −1 can be indexed to the defects present on the surface of the nanoparticles. In addition, the presence of Fe2O3 additional phase is revealed by the peaks at 215 and 276 cm −1 . Even if Fe2O3 is expected only in minor amounts, the relatively strong signal is probably due to an enhancement, which can be ascribed to a partial oxidation of Fe 2+ to Fe 3+ occurring by laser irradiation during the Raman measurement [27].  Figure 1b shows the Raman response of bare rGO. Three characteristic peaks of carbonaceous materials are observed [28]. The D band located at 1338 cm −1 and the G band located at 1581 cm −1 shows a recovery of the hexagonal pattern of carbon atoms with defects. The high ID/IG ratio (≈1.31) evidence a high number of structural defects on rGO surface. Figure 1c shows the Raman response of the Fe3O4/rGO nanocomposite. All the signals evidenced in Figure 1a,b are retained, suggesting that the Fe3O4 embedding by the rGO matrix relies on electrostatic interactions, without any chemical modifications.
The diffractogram of the pristine magnetite powder ( Figure 1d) presents a series of reflections consistent with Fe3O4 as indexed in JCPDS, card no. 19-0629. The crystallite size was estimated to be ~5.65 nm, by applying the Scherrer's equation (Equation (2) Figure 1b shows the Raman response of bare rGO. Three characteristic peaks of carbonaceous materials are observed [28]. The D band located at 1338 cm −1 and the G band located at 1581 cm −1 shows a recovery of the hexagonal pattern of carbon atoms with defects. The high ID/IG ratio (≈1.31) evidence a high number of structural defects on rGO surface. Figure 1c shows the Raman response of the Fe 3 O 4 /rGO nanocomposite. All the signals evidenced in Figure 1a,b are retained, suggesting that the Fe 3 O 4 embedding by the rGO matrix relies on electrostatic interactions, without any chemical modifications.
The diffractogram of the pristine magnetite powder ( Figure 1d) presents a series of reflections consistent with Fe 3 O 4 as indexed in JCPDS, card no. 19-0629. The crystallite size was estimated to be~5.65 nm, by applying the Scherrer's equation (Equation (2)) [29]: with k = 0.94, λ = 0.709319 Å (Mo Kα source) and β = FWHM (full-width at half-maximum), the 311 reflection has been considered for the calculation. The X-ray diffractogram of the magnetite-carbon composite (Figure 1e) micrographs of Fe 3 O 4 nanoparticles at 40,000× and 275,000× magnification levels, respectively. Despite the formation of aggregates, probably due to agglomeration by magnetic stirring, the particle size can be visually estimated to 5-10 nm, consistent with the crystallite size estimation. hump around 2θ = 26° is the signature of the amorphous rGO carbon phase. Figure 1f-g show the theoretical reflections of the Fe3O4 and Fe2O3 phases. Figure 2a,b show the SEM micrographs of Fe3O4 nanoparticles at 40,000× and 275,000× magnification levels, respectively. Despite the formation of aggregates, probably due to agglomeration by magnetic stirring, the particle size can be visually estimated to 5-10 nm, consistent with the crystallite size estimation.  Figure 2c,d show the SEM micrographs of Fe3O4/rGO nanocomposite at 40,000× and 275,000× magnification levels, respectively. In both SEM images, carbon sheets embedding nanoparticles bigger than the pristine oxide are clearly evidenced, suggesting a probable agglomeration during the composite synthesis. At the higher magnification, it is possible to see iron oxide nanoparticles embedded into the amorphous carbon sheets. Figure 3 depicts the thermogravimetric analysis of Fe3O4/rGO. The composite was heated with a heating rate of 10 °C/min in air atmosphere. After some low-T weight loss due to water evaporation, the analysis shows a weight drop from around 96% to 76% (about 20%) at T ≈ 450 °C, due to the oxidation of the rGO to carbon dioxide. This corresponds to an approximate Fe3O4:rGO mass ratio of 79:21, which allows to estimate specific capacity for the composite material of the order of 885 mAh g −1 . This section may be divided by subheadings.  In both SEM images, carbon sheets embedding nanoparticles bigger than the pristine oxide are clearly evidenced, suggesting a probable agglomeration during the composite synthesis. At the higher magnification, it is possible to see iron oxide nanoparticles embedded into the amorphous carbon sheets. Figure 3 depicts the thermogravimetric analysis of Fe 3 O 4 /rGO. The composite was heated with a heating rate of 10 • C/min in air atmosphere. After some low-T weight loss due to water evaporation, the analysis shows a weight drop from around 96% to 76% (about 20%) at T ≈ 450 • C, due to the oxidation of the rGO to carbon dioxide. This corresponds to an approximate Fe 3 O 4 :rGO mass ratio of 79:21, which allows to estimate specific capacity for the composite material of the order of 885 mAh g −1 . This section may be divided by subheadings.

Electrochemical Characterization
As shown in Figure 4, the voltametric response of Fe3O4/rGO composite reveals, during the first cathodic scan, three main features: a very broad, and lo around 1.5 V (*), and two peaks at 1.02 V (B) and 0.75 V (A). The feature around has been observed also for other transition metal oxides [30], and describes irre interfacial processes only occurring during the first discharge. As regards the peak at 1.02 V (B), Thackeray [31] proposes a mechanism i intercalation of Lithium into the spinel structure of Fe3O4 occurs according to E (3):

Electrochemical Characterization
As shown in Figure 4, the voltametric response of Fe 3 O 4 /rGO composite material reveals, during the first cathodic scan, three main features: a very broad, and low hump around 1.5 V (*), and two peaks at 1.02 V (B) and 0.75 V (A). The feature around 1.5 V (*) has been observed also for other transition metal oxides [30], and describes irreversible interfacial processes only occurring during the first discharge.

Electrochemical Characterization
As shown in Figure 4, the voltametric response of Fe3O4/rGO composite material reveals, during the first cathodic scan, three main features: a very broad, and low hump around 1.5 V (*), and two peaks at 1.02 V (B) and 0.75 V (A). The feature around 1.5 V (*) has been observed also for other transition metal oxides [30], and describes irreversible interfacial processes only occurring during the first discharge.  As regards the peak at 1.02 V (B), Thackeray [31] proposes a mechanism in which intercalation of Lithium into the spinel structure of Fe3O4 occurs according to Equation The subsequent reduction of Fe 3+ to Fe 0 (described with the sharp peak at 0.75 V (A)) occurs by the conversion reaction leading to the Fe 0 nanoparticles dispersed in a Li2O matrix (Equation (4)): As regards the peak at 1.02 V (B), Thackeray [31] proposes a mechanism in which intercalation of Lithium into the spinel structure of Fe 3 O 4 occurs according to Equation (3): Molecules 2021, 26, 4316 6 of 14 The subsequent reduction of Fe 3+ to Fe 0 (described with the sharp peak at 0.75 V (A)) occurs by the conversion reaction leading to the Fe 0 nanoparticles dispersed in a Li 2 O matrix (Equation (4)): In this potential region, the decomposition of electrolyte towards the carbon surface, forming the passivation layer, take place. During the anodic scan, a couple of broad peaks (C) and (D) are visible at 1.57 V and 1.89 V and can attributed to the oxidation of Fe 0 to Fe 2+ and Fe 3+ respectively.
During the cathodic scans of the subsequent cycles, peak (B) disappears and peak (A) is shifted to 0.8 V, while the behavior during anodic scans is retained. Figure 5 shows the galvanostatic cycling behavior of Fe 3 O 4 /rGO composite anode at the estimated 1C rate. In this potential region, the decomposition of electrolyte towards the carbon surface, forming the passivation layer, take place. During the anodic scan, a couple of broad peaks (C) and (D) are visible at 1.57 V and 1.89 V and can attributed to the oxidation of Fe 0 to Fe 2+ and Fe 3+ respectively.
During the cathodic scans of the subsequent cycles, peak (B) disappears and peak (A) is shifted to 0.8 V, while the behavior during anodic scans is retained. Figure 5 shows the galvanostatic cycling behavior of Fe3O4/rGO composite anode at the estimated 1C rate. The specific capacity of the first discharge reaches a value of 1634 mAh g −1 , while the subsequent charge exhibits a value of 1186 mAh g −1 , with an efficiency of 72%. During the following 40 cycles, both lithiation and delithiation specific capacities remain higher than 1000 mAh g −1 , with an efficiency close to 100%. Capacity values higher than the theoretical ones have already been reported for conversion materials by several authors and could be associated with several mechanisms. Among those, interfacial lithium storage [32] and reversible Li storage processes by the external layers of SEI, mainly given by the partly reversible formation/dissolution of carbonates and semicarbonates [33] and commonly described as a 'gel-type layer' [34], can provide extra capacity. The specific capacity of the first discharge reaches a value of 1634 mAh g −1 , while the subsequent charge exhibits a value of 1186 mAh g −1 , with an efficiency of 72%. During the following 40 cycles, both lithiation and delithiation specific capacities remain higher than 1000 mAh g −1 , with an efficiency close to 100%. Capacity values higher than the theoretical ones have already been reported for conversion materials by several authors and could be associated with several mechanisms. Among those, interfacial lithium storage [32] and reversible Li storage processes by the external layers of SEI, mainly given by the partly reversible formation/dissolution of carbonates and semicarbonates [33] and commonly described as a 'gel-type layer' [34], can provide extra capacity.
Capacity subsequently fades upon long-term cycling, with a drop to around 700 mAh g −1 at the 100th cycle, still with a coulombic efficiency close to 100%.
The galvanostatic E vs. Q profiles (Figure 5b) reveal during the first discharge a shoulder around 1.5 V, a small plateau at 1.01 V, and a larger one at 0.81 V for the lithium insertion and conversion reactions, while the charge step shows a slopping plateau from 1.54 V to 1.83 V due to the oxidation of Fe. These plateaus are consistent with the differential analysis dQ dE −1 vs. E (Figure 5c), which reveals, during the first lithiation, a broad hump around 1.5 V (*), a peak (A) at 0.80 V and a peak (B) at 1.01 V, while peaks (C) and (D) at 1.54 V and 1.80 V, corresponding to the subsequent Fe oxidation steps, are observed during delithiation. During the second lithiation/delithiation cycle, any sign of the previous plateaus disappears in favor of a more pronounced plateau at 0.89 V. The curves are characterized by a large voltage hysteresis, which is a typical feature of conversion materials. This phenomenon has been explained by several authors by concurring processes, such as a pseudocapacitive behavior due to the increase of surface area upon cycling [35], or different reaction pathways for conversion and deconversion reactions [36]. Figure 6 reports the performances of the Fe 3 O 4 /rGO composite material at 2C (1848 mA g −1 ) and 4C rate (3696 mA g −1 ). Capacity subsequently fades upon long-term cycling, with a drop to around 700 mAh g −1 at the 100th cycle, still with a coulombic efficiency close to 100%.
The galvanostatic E vs. Q profiles (Figure 5b) reveal during the first discharge a shoulder around 1.5 V, a small plateau at 1.01 V, and a larger one at 0.81 V for the lithium insertion and conversion reactions, while the charge step shows a slopping plateau from 1.54 V to 1.83 V due to the oxidation of Fe. These plateaus are consistent with the differential analysis dQ dE −1 vs. E (Figure 5c), which reveals, during the first lithiation, a broad hump around 1.5 V (*), a peak (A) at 0.80 V and a peak (B) at 1.01 V, while peaks (C) and (D) at 1.54 V and 1.80 V, corresponding to the subsequent Fe oxidation steps, are observed during delithiation. During the second lithiation/delithiation cycle, any sign of the previous plateaus disappears in favor of a more pronounced plateau at 0.89 V. The curves are characterized by a large voltage hysteresis, which is a typical feature of conversion materials. This phenomenon has been explained by several authors by concurring processes, such as a pseudocapacitive behavior due to the increase of surface area upon cycling [35], or different reaction pathways for conversion and deconversion reactions [36]. Figure 6 reports the performances of the Fe3O4/rGO composite material at 2C (1848 mA g −1 ) and 4C rate (3696 mA g −1 ). The material shows very stable performances at high rates, thanks to graphene and its high electronic conductivity. The first discharge at 2C gives a capacity of 1529 mAh g −1 , and the subsequent charge delivers 1103 mAh g −1 , with an initial coulombic efficiency of 72%. The capacity remains quite stable until cycle 55, giving an average value of 1050 mAh g −1 . At 4C the material exhibits a first-cycle discharge capacity of 1323 mAh g −1 and a charge capacity of 993 mAh g −1 , with an initial coulombic efficiency of 75%. Also at this high C-rate, the charge/discharge capacity remains quite stable during the cycling, with an average value of 980 mAh g −1 , suggesting that the Fe3O4/rGO composite works better at high rates. The difference in performances and stability between the lower (1C) and higher (2C, 4C) rates may be rooted in a possible decomposition of the electrode material at lower rates, due to irreversible chemical processes concurrent with the electrochemical The material shows very stable performances at high rates, thanks to graphene and its high electronic conductivity. The first discharge at 2C gives a capacity of 1529 mAh g −1 , and the subsequent charge delivers 1103 mAh g −1 , with an initial coulombic efficiency of 72%. The capacity remains quite stable until cycle 55, giving an average value of 1050 mAh g −1 . At 4C the material exhibits a first-cycle discharge capacity of 1323 mAh g −1 and a charge capacity of 993 mAh g −1 , with an initial coulombic efficiency of 75%. Also at this high C-rate, the charge/discharge capacity remains quite stable during the cycling, with an average value of 980 mAh g −1 , suggesting that the Fe 3 O 4 /rGO composite works better at high rates. The difference in performances and stability between the lower (1C) and higher (2C, 4C) rates may be rooted in a possible decomposition of the electrode material at lower rates, due to irreversible chemical processes concurrent with the electrochemical Li storage. Furthermore, the SEI formation and irreversible processes at slow rate are enhanced by the high surface area of graphene sheet. On the contrary, as the rate increases, the reversible electrochemical processes become predominant and the performances of the electrode are very stable.
These findings have been validated by rate capability measurements in an extended C-range, from C/10 (88 mA g −1 ) up to 10C (8800 mA g −1 ). The rate capability results of Fe 3 O 4 /rGO are presented in Figure 7 and detailed in Table 1. Li storage. Furthermore, the SEI formation and irreversible processes at slow rate are enhanced by the high surface area of graphene sheet. On the contrary, as the rate increases, the reversible electrochemical processes become predominant and the performances of the electrode are very stable. These findings have been validated by rate capability measurements in an extended C-range, from C/10 (88 mA g −1 ) up to 10C (8800 mA g −1 ). The rate capability results of Fe3O4/rGO are presented in Figure 7 and detailed in Table 1. The rate capability test confirms that the charge/discharge behavior of Fe3O4/rGO (Figure 7a) is more stable at higher than at lower currents, supporting the hypothesis that the low-current high irreversible capacity is rooted in irreversible processes that are enhanced at low C-rates. In fact, up to C/2 some capacity fade is evidenced, which can be ascribed to irreversible, slow-kinetics interfacial phenomena, as confirmed by relatively low coulombic efficiencies.
As the rate increases, the specific capacity decreases, reaching the minimum of 484 mAh g −1 at 10C with a coulombic efficiency of 99.2%. When the 1C charge/discharge rate conditions are restored, the electrode is able to retain a steady reversible capacity of ~900 mAh g −1 . The increase of capacity during the initial cycles at this regime can be ascribed to the progressive electrode morphological rearrangement and wetting by electrolyte, which makes more active sites accessible by Li + ions.   The rate capability test confirms that the charge/discharge behavior of Fe 3 O 4 /rGO (Figure 7a) is more stable at higher than at lower currents, supporting the hypothesis that the low-current high irreversible capacity is rooted in irreversible processes that are enhanced at low C-rates. In fact, up to C/2 some capacity fade is evidenced, which can be ascribed to irreversible, slow-kinetics interfacial phenomena, as confirmed by relatively low coulombic efficiencies.
As the rate increases, the specific capacity decreases, reaching the minimum of 484 mAh g −1 at 10C with a coulombic efficiency of 99.2%. When the 1C charge/discharge rate conditions are restored, the electrode is able to retain a steady reversible capacity of~900 mAh g −1 . The increase of capacity during the initial cycles at this regime can be ascribed to the progressive electrode morphological rearrangement and wetting by electrolyte, which makes more active sites accessible by Li + ions.
When the Fe 3 O 4 /rGO results here shown are compared with those reported in Ref. [6] by a pristine Fe 3 O 4 anode prepared in the same conditions, some similarities are evidenced. In fact, at the lower C-rate values (C/10, C/5, C/2), irreversibility and capacity fades are evidenced for both electrodes by the relatively low coulombic efficiencies (91.4% vs. 88.6% for Fe 3 O 4 and Fe 3 O 4 /rGO at C/10, respectively). However, the current composite Fe 3 O 4 /rGO electrode constantly shows capacity values, which are about 150 mAh g −1 higher than pristine Fe 3 O 4 at C-rate > C/2. Also at higher rates, when both electrodes exhibit enhanced capacity retention, the present Fe 3 O 4 /rGO composite exhibits both better coulombic efficiency and higher capacity than the pristine Fe 3 O 4 . The difference is strongly evidenced at the highest rate (10C) with the Fe 3 O 4 /rGO composite exhibiting 484 mAh g −1 capacity vs. 293 mAh g −1 of the benchmark Fe 3 O 4 . This behavior may be explained by considering that the rGO matrix, in addition to better buffering and confining the volume changes and structure rearrangements that take place upon reversible Li storage, also strongly enhance electrode conductivity, possibly resulting in lower overall electrode polarization and better high-rate tolerance. The profiles of E vs. Q galvanostatic cycles selected at every C-rate Figure 7b are consistent with the evidenced charge/discharge performances: after the 1st cycle, where most of the irreversible processes take place, charge and discharge capacities are always matching, confirming the high coulombic efficiency in every of the cycling conditions investigated. In addition, the low polarization is confirmed by the fact that the lithiation/delithiation plateaus, around 1 V and 1.5 V, respectively, are still visible even at the highest current applied (10C). Comparison of lowand high-C-rate performances with some literature findings is briefly evidenced in Table 2. With the aim to shed light onto Li exchange kinetics, with particular regard to the role played by the rGO matrix, the impedance response of both electrodes, based on bare Fe 3 O 4 and Fe 3 O 4 /rGO composite, has been characterized. Figure 8a,b shows the Nyquist plots acquired at selected cycles (first cycle, then every tenth cycles) during lithiation of Fe 3 O 4 and Fe 3 O 4 /rGO-based electrodes, respectively. The applied bias potentials correspond to the voltametric peak (A) in Figure 4, i.e., E = 0.75 V for the first cycle and 0.9 V for the following cycles. All the impedance dispersions reveal common features, which are typical for a lithium-ion battery anode [6,41,42], namely: (i) an intercept on the real axis corresponding to electrolyte resistance; (ii) a high-frequency arc corresponding to charge accumulation and migration through the passivation layer, partially overlapped by (iii) a medium-frequency arc corresponding to interfacial charge transfer and electrical double layer formation; (iv) a 45 • dispersion, bending toward a vertical line, corresponding to diffusion to a blocking electrode.   When the impedance responses of bare Fe3O4 and Fe3O4/rGO are compared, it appears that for the rGO-modified anode, the overall increase is more limited, but the growth of the high-frequency arc is more marked, probably because of a larger electrode/electrolyte interface provided by the carbonaceous rGO matrix. At the same time, for the rGO-modified anode, the medium-frequency arc exhibits a limited growth upon cycling, confirming the dual role of graphene matrix in stabilizing the electrode morphology and in enhancing electrode conductivity and rate of charge-transfer process. The ac-dispersions have been modeled, by using Boukamp's EQVCRT software [43], to an equivalent circuit Rel(RSEICSEI)(RctCdl)W in Boukamp's notation. Rel, RSEI, CSEI, Rct, Cdl and W represent the pure ohmic resistance of the electrolyte, the resistance and capacitance associated with the passivation layer, the resistance of the charge-transfer process, the capacitance of the electrical double layer, and the Warburg diffusion element, For both electrodes, as expected, the overall impedance increases with cycling. The most relevant contribution to the growth is given by the medium-frequency arc related to the charge-transfer, which can be explained either by possible particle aggregation or by increase of inter-particle resistance. A minor increase is evidenced by the high-frequency arc, which may be associated with a limited growth of passivation layer upon cycling.
When the impedance responses of bare Fe 3 O 4 and Fe 3 O 4 /rGO are compared, it appears that for the rGO-modified anode, the overall increase is more limited, but the growth of the high-frequency arc is more marked, probably because of a larger electrode/electrolyte interface provided by the carbonaceous rGO matrix. At the same time, for the rGO-modified anode, the medium-frequency arc exhibits a limited growth upon cycling, confirming the dual role of graphene matrix in stabilizing the electrode morphology and in enhancing electrode conductivity and rate of charge-transfer process. The ac-dispersions have been modeled, by using Boukamp's EQVCRT software [43], to an equivalent circuit R el (R SEI C SEI )(R ct C dl )W in Boukamp's notation. R el , R SEI , C SEI , R ct , C dl and W represent the pure ohmic resistance of the electrolyte, the resistance and capacitance associated with the passivation layer, the resistance of the charge-transfer process, the capacitance of the electrical double layer, and the Warburg diffusion element, respectively. During the fit procedure, constant phase elements Q substituted pure capacitors C in order to take into account electrode inhomogeneities and roughness [44]. The evolutions of R ct and R SEI upon cycling are reported in Figure 8c,d, respectively. For both materials, the calculated values of both resistances increase upon cycling. Specifically, the Fe 3 O 4 /rGO nanocomposite exhibits a lower R ct than the pristine Fe 3 O 4 , which confirms the enhanced electronic conductivity of the composite, given by the rGO matrix. On the contrary, the R SEI is higher in the composite, which confirms the role played by the large surface area of rGO nanosheets in the buildup of a more extended passivation layer.

Synthesis
A base promoted coprecipitation of FeCl 2 and FeCl 3 was pursued for the synthesis of pristine Fe 3 O 4 nanoparticles [6]. The synthesis of Fe 3 O 4 in basic media without any thermal annealing treatment have been widely investigated [45]. The formation mechanism has been explained by the following set of reactions (Equations (5)-(9)): With the overall reaction: In this regard, stoichiometric amounts of FeCl 2 ·4H 2 O and FeCl 3 ·6H 2 O were dissolved in 40 mL of distilled water. After the complete dissolution of the salts, 100 mL of 10% NH 4 OH was added to the solution giving a black precipitate. The solution was stirred and heated until reaching 70 • C and then further 30 mL of concentrated NH 4 OH were added and left to react for 8 h. The obtained black powder was easily washed first with acetone and then with ethanol. After that, it was vacuum dried at 50 • C.
200 mg of graphene oxide was added to 150 mL of distilled water and sonicated for 1 in order to obtain a homogeneous suspension. After the sonication, 500 mg of pristine Fe 3 O 4 nanoparticles were added to the solution and sonicated again for 1 h. After the second sonication, 10 mL of hydrazine hydrate was added to the solution and an ice bath was used to dissipate reaction heat. The solution underwent a third sonication for 2 h. The obtained powder was filtered with millipore (0.2 µm GTTP), thoroughly washed with ethanol, and vacuum dried at 50 • C.

Electrode Processing
Electrodes based both on the Fe 3 O 4 /rGO composite were prepared by using the same procedure. The anode slurries were obtained by mixing active material, SuperC65 (TIMCAL C-ENERGY TM) as a conductive agent and Polyacrilic Acid (Mw = 450,000-Aldrich) as a binder in the ratio 70:20:10, respectively. The active material and the conductive agent were finely grounded in an agata mortar and then added to the binder solution. The slurries were prepared using ethanol as solvent and stirred by a magnetic stirrer overnight. The well-mixed slurries were first casted on copper foil by Doctor Blade at 100 µm wet thickness and dried on a heating plate at 70 • C for 2 h. Circular electrodes (9 mm diameter) were cut using an electrode puncher (EL-CELL). Electrodes were further pressed using a hydraulic press at 4.7 tons cm −2 . The average loading of active material was of ≈1 mg cm −2 . Eventually, the electrodes were dried at 120 • C under vacuum for 12 h and stored in an Ar-filled glovebox.

Structural and Electrochemical Characterizaiton
The structure of the prepared materials was characterized by Raman spectroscopy by using Horiba iH320 spectrometer with a 532 nm laser source, X-ray diffraction (XRD) was acquired by using a Mo-Kα source (λ = 0.709319 Å) coupled with a curved multichannel detector INEL CPSD 180 (Curved Position Sensitive Detector). For the sake of comparison with literature findings, the reflection angle values were converted into 2θ/Cu Kα. Thermal analysis of Fe 3 O 4 composite was carried out by employing a TGA-DTA Perkin-Elmer STA6000. The morphological characterization was performed with a ZEISS Sigma Series 300 field emission scanning electron microscope (FE-SEM).

Cell Assembly and Electrochemical Test
Three-electrode Swagelok-type cells with metallic lithium foil as counter and reference electrode, stainless steel current collectors were employed for characterization of electrochemical performances. Circular Whatman GF/A glass fiber 12 mm diameter disks were used as separator and 1 M LiPF 6 in Ethylene carbonate (EC):Dimetheyl carbonate (DMC) 1:1 (v:v) (Solvionic, Tolouse, France) was used as electrolyte.
The electrochemical behavior of the electrodes was characterized by a VMP-3 galvanostat/potentiostat (Bio-Logic). Cyclic voltammetry (CV) was acquired in the potential range within 0.001-3.000 V at 0.1 mV s −1 scan rate. GCPL cycles were performed in the same potential range and at C-rates between C/10 (calculated as 88.1 mA g −1 ) and 10 C (8810 mA g −1 ). Electrochemical impedance spectroscopy (EIS) measurements were performed at bias potential E = 0.75 V for the first cycle and E = 0.90 V for the subsequent cycles, by applying a sinusoidal perturbation of amplitude ∆E = ±5 mV in the frequency range 101 kHz < f < 9 mHz. All potential values are given vs. Li + /Li redox couple (E • Li+/Li = −3.04 V vs. SHE).

Conclusions
Fe 3 O 4 nanoparticles were synthesized by a base-promoted method and characterized, resulting in an average nanoparticle size of about 6 nm computed by Scherrer's equation and confirmed by SEM micrograph.
A composite material based on Fe 3 O 4 nanoparticles and reduced graphene oxide was synthesized by ultrasonication of graphene oxide and subsequent reduction with hydrazine. The synthesized composite was characterized, giving an average nanoparticle size of 6 nm and confirming the embedding of magnetite by graphene matrix. The electrochemical tests revealed a quite stable charge/discharge behavior of the composite (~900 mAh g −1 at 1Crate after the rate stresses), and impressive rate capability (980 mAh g −1 at 4C; 484 mAh g −1 at 10C). The EIS measurements confirmed that the outstanding cycling performances are rooted in stable morphology and low charge-transfer polarization, given by the embedding of nanoparticles into reduced graphene oxide sheet.
The combination of low-environmental-impact and facility of the synthesis of the composite material and electrode processing, together with the high performances herein reported, make the composite a promising anode material for high-energy and especially high-power applications.