Effect of Germanium Incorporation on the Electrochemical Performance of Electrospun Fe 2 O 3 Nanoﬁbers-Based Anodes in Sodium-Ion Batteries

Featured Application: Authors are encouraged to provide a concise description of the speciﬁc application or a potential application of the work. This section is not mandatory. Abstract: Fe 2 O 3 and Fe 2 O 3 :Ge nanoﬁbers (NFs) were prepared via electrospinning and thoroughly characterized via several techniques in order to investigate the effects produced by germanium incorporation in the nanostructure and crystalline phase of the oxide. The results indicate that reference Fe 2 O 3 NFs consist of interconnected hematite grains, whereas in Fe 2 O 3 :Ge NFs, constituted by ﬁner and elongated nanostructures developing mainly along their axis, an amorphous component coexists with the dominant α -Fe 2 O 3 and γ -Fe 2 O 3 phases. Ge 4+ ions, mostly dispersed as dopant impurities, are accommodated in the tetrahedral sites of the maghemite lattice and probably in the defective hematite surface sites. When tested as anode active material for sodium ion batteries, Fe 2 O 3 :Ge NFs show good speciﬁc capacity (320 mAh g − 1 at 50 mA g − 1 ) and excellent rate capability (still delivering 140 mAh g − 1 at 2 A g − 1 ). This behavior derives from the synergistic combination of the nanostructured morphology, the electronic transport properties of the complex material, and the pseudo-capacitive nature of the charge storage mechanism. Simultaneous structure, composition, and morphology transformations during the creation of hollow iron oxide nanocapsules.


Introduction
Affordable and clean energy represents one of the seventeen sustainable development goals (SDGs) adopted by the 193 members of the United Nations General Assembly five years ago [1]. To ensure the achievement of the SDGs by 2030, countries have introduced their implementation in the own national development programs. These joint actions should allow facing the global energy crisis through a worldwide approach, simultaneously reaching other strictly related SDGs, e.g., the "climate action" (the thirteenth SDG). 2

of 15
Although a recent report has revealed that the "affordable and clean energy" goal (seventh SDG) will not be met by the agreed date [2], there are encouraging signs that energy is becoming more sustainable and available to the global population. Actually, thanks to the impressive advances of renewable energy sources (RESs) in the electricity sector [3], access to electricity in poorer countries is accelerating.
In the scenery of post-lithium rechargeable batteries, the ones exploiting sodium-ion storage, even if not commercialized at the large scale yet, probably represent the technology with the highest level of maturity and sustainability [19,32]. The development of highly performing anode materials is one of the barriers to still be overcome [33]. Transition metal oxides, exhibiting high theoretical reversible capacities and low cost, have broadly been evaluated as anode materials.
This work presents the results of a comprehensive investigation of the effects produced by germanium incorporation in the nanostructure and crystalline phase of electrospun Fe 2 O 3 nanofibers (NFs). The electrochemical activity of Fe 2 O 3 :Ge NFs as active anode material in SIBs is evaluated.

Preparation of the Electrospun NFs
The synthetic route followed to produce the electrospun NFs ( Figure S1) was described in detail in previous papers [36,39,42]. Concisely, the spinnable solutions were prepared via sol-gel method. In a typical procedure, 0.325 g polyacrylonitrile (PAN) was dissolved in 4.550 g N,N-dimethylformamide (DMF) and a homogeneous solution was obtained after magnetic stirring at room temperature (RT) for 2 h. Then, 0.389 g iron (II) acetate (FeAc 2 ), corresponding to a Fe concentration of 38.5% relative to PAN, was added to prepare the (reference) iron oxide NFs. The Ge-doped iron oxide NFs were prepared by the addition of 0.350 g FeAc 2 and 0.0531 g germanium (IV) isopropoxide (GeIPO), corresponding to a (Fe + Ge) concentration of 38.5% relative to the polymer and to a Ge:Fe mass ratio of 1:9 (corresponding to 0.085 Ge:Fe molar ratio). After 2 h magnetic stirring at RT, the as-prepared solution was loaded into a 20 mL glass syringe equipped with 40 mm long 0.8 gauge stainless-steel needle and electrospun at a DC voltage of 17 kV via a CH-01 Electro-spinner 2.0 (Linari Engineering s.r.l.). Solution was fed at a rate of 23.5 µL min −1 . The electrospun NFs were collected on a grounded aluminum foil placed at a distance of 11 cm from the tip of the needle. After drying at RT overnight to remove the DMF residuals, the as-spun fibrous membranes were peeled from the collector and calcined in air for 2 h through a single-step process at 600 • C (heating rate: 5 • C min −1 ). Upon calcination, the organic components were completely removed from the FeAc 2 -PAN and GeIPO-FeAc 2 -PAN NFs and the iron oxide was generated. For further details on the syntheses and experimental setup utilized see refs [36,39,42].
SEM images were acquired on a Zeiss microscope (15 kV), equipped with software for the automated measurement of the NF diameters from their SEM images. A Philips CM 200 microscope (200 kV), equipped with energy-dispersive X-ray (EDX) spectrometer, was utilized for the TEM analysis. XRD patterns were recorded with a Bruker D2 diffractometer using Ni β-filtered Cu-K α radiation source (λ = 0.1541 nm). Raman scattering was measured in air at RT by means of a NTEGRA-Spectra SPM NT-MDT confocal microscope coupled to a solid-state laser operating at 633 nm. A variable ND filter set the laser power at 250 µW at the sample surface to prevent local heating. The scattered light from the sample, collected by means of a 100X Mitutoyo objective, was detected by a cooled ANDOR iDus CCD Camera. Spectra from several random positions on each specimen were recorded to investigate its spatial homogeneity. The collected spectra were averaged to have a reliable picture of the entire sample. XPS analysis was conducted with a Physical Electronics GMBH PHI 5800-01 spectrometer, equipped with a monochromatic Al-K α source (1486.6 eV) with a power beam of 300 W. The C 1s peak at 284.5 eV was assumed as reference for binding energies. Further details on the instrumentation and experimental data processing can be found elsewhere [36,37,39,42].
The XAS spectra were collected at the Fe and Ge K-edges (7112 and 11,103 eV, respectively) at the XAFS beamline [43] operating at the Elettra synchrotron radiation facility in Trieste, Italy. The spectra were recorded at RT in transmission mode; for the measurements, a proper amount of sample (as to give a unit jump in the absorption coefficient) was mixed with cellulose and pressed into a pellet. Energy calibration was performed measuring the absorption spectrum of either metallic Fe pellet and Ge foil, placed in a second experimental chamber after the sample and after the I1 ionization chamber. The ring current and the energy were 200 mA and 2.4 GeV, respectively. A Si(111) double-crystal monochromator was used, ensuring high-order harmonic rejection by detuning of the second crystal. A water-cooled Pt-coated silicon mirror was used to obtain vertical collimation of the beam. The X-ray signal extraction and analysis, including the linear combination fit, were performed by means of Athena, belonging to the set of interactive programs IFEFFIT [44,45]. For the X-ray absorption near edge structure (XANES) analysis, the raw spectra were first background subtracted using a straight line and normalized to unit absorption above 800 eV, where the extended X-ray absorption fine structure (EXAFS) oscillations are no more visible. The EXAFS data analysis at the Ge K-edge was performed employing the Excurve code using a k2 weighing scheme and isolating the χ(k) signal given by the first shell through Fourier filtering, using a Gaussian window between 0.7 and 2.4 Å. A PND pattern of the Fe 2 O 3 :Ge NFs was collected at RT in a vanadium sample holder on the high intensity D20 diffractometer at the Institut Laue-Langevin (ILL), Grenoble, France, with the use of 1.54 Å wavelength, in the 0-150 • angular range, using the highresolution option with a takeoff angle of 90 • . Data were analyzed using the FullProf Suite to perform LeBail fitting and traditional Rietveld refinement.

Electrochemical Measurements
Electrodes based on Fe 2 O 3 :Ge NFs were obtained by mixing the active material with SuperP carbon black and an organic binder mixture in the weight ratio 6:3:1. The binder consisted of a 1:1 weight mixture of polyacrylic acid and carboxymethylcellulose. The mechanical mixture was dispersed in water and homogenized in an Ultra-turrax batch dispersers. The obtained slurry was then deposited by doctor blade with a wet film thickness of 75 µm on an Al foil. An electrode load of 1.5 mg cm −2 was obtained after drying at 140 • C in vacuum for two hours. Round disks of 16 mm in diameter were cut to fabricate 2032 coin cell equipped with a metallic sodium counter electrode and a Whatman glass fiber separator soaked with a solution of 1 M NaClO 4 in propylene carbonate. The cell assembling was made in an Ar filled glove box. Potentiodynamic cycling with galvanostatic acceleration (PCGA) and galvanostatic cycling with potential limitation (GCPL) were performed using a Biologic VMP3 potentiostat-galvanostat.

Physicochemical Properties of the NFs
Morphological and textural properties of the electrospun NFs were investigated by SEM and TEM analyses (Figure 1). In both types of produced NFs (Fe 2 O 3 NFs and Fe 2 O 3 :Ge NFs), no beads were observed. Nonetheless, the results revealed that the addition of GeIPO to the FeAc 2 -PAN-DMF precursor solution translated into evident differences in the morphology and texture of the calcined NFs. SEM images showed that the NFs prepared without adding GeIPO had rough and porous surface (Figure 1a,b). Their diameters, as calculated by the image analysis software, varied mainly between 120 and 500 nm, although largely thicker NFs (>0.8 µm) were also present. The center of the diameter distribution was located at 230 nm. Differently, the surface of the germanium-containing NFs was smoother (Figure 1c,d) and their size varied more uniformly in the 65-1180 nm range. The center of the distribution moved at higher values (360 nm). Many factors, such as feeding rate of the solution, applied voltage, collection distance and ambient humidity, affect the morphology and diameters of the fibers. In the present study, feeding rate of the solution, applied voltage and collection distance were kept constant. In addition, the relative air humidity (40%) was the same in the two syntheses. Nonetheless, the well-known extreme moisture sensitivity of the germanium isopropoxide and miscibility of water with DMF might have played a role in giving rise to a more uniform fiber diameter distribution since humidity causes changes in the nanofibers diameter by controlling the solidification process of the charged jet. the fibers that is due to the large temperature gradient along the radial direction experienced by the FeAc2-PAN NFs upon calcination [54]. Different from the pure iron oxide NFs, the germanium-containing ones consisted of finer and elongated nanostructures that developed mainly along the longitudinal axis of the fiber (Figure 1h-j). A hierarchical structure with similar characteristics had been previously reported for electrospun Si-doped iron oxide NFs [36]. Indeed, both silicon and germanium are believed to drive the growth along preferential crystallographic directions to a limited extent [41,55,56].  [39]. The resulting coral-like architecture seems to be peculiar to electrospun iron oxide NFs, as it has been frequently observed regardless of the precursors utilized to prepare the spinnable solution and the heat treatment conditions [46][47][48][49][50][51][52][53]. Previous studies on NFs prepared under the same conditions [42] had revealed the single-crystal or quasi-single-crystal nature of the oxide grains, as well as the hollow section of the fibers that is due to the large temperature gradient along the radial direction experienced by the FeAc 2 -PAN NFs upon calcination [54]. Different from the pure iron oxide NFs, the germanium-containing ones consisted of finer and elongated nanostructures that developed mainly along the longitudinal axis of the fiber (Figure 1h-j). A hierarchical structure with similar characteristics had been previously reported for electrospun Si-doped iron oxide NFs [36]. Indeed, both silicon and germanium are believed to drive the growth along preferential crystallographic directions to a limited extent [41,55,56].
The spatial distribution of the NF components was investigated via TEM/EDX ( Figure S3). The elemental maps revealed a homogeneous distribution of iron, oxygen and germanium within the NFs, with no organic residuals from the pristine NFs.
The oxide phase in the NFs and its crystallinity were investigated by X-ray and neutron diffraction and MRS analyses. Figure S4 compares the diffraction patterns of the NFs produced from the FeAc 2 -PAN-DMF and GeIPO-FeAc 2 -PAN-DMF solutions. In the former case, all detected peaks were ascribable to the reflections from crystallographic planes of rhombohedral hematite (α-Fe 2 O 3 , JCPDS card no. 33-0664.), indicating the formation of a single crystalline phase, as expected on the basis of previous studies [36,39,42] and literature [46,47,49,[51][52][53]57] as well. The average crystallite size (30.4 nm) was estimated from the FWHM (full width at half-maximum, β) of the most intense (104) peak, via the Scherrer formula, d = Kλβcosθ, with K = 0.9 and λ = 0.1541 nm. In the X-ray diffractogram of the germanium-containing NFs, the peaks indexed to α-Fe 2 O 3 were weaker and broader. In addition, the diffraction peaks typical of γ-phase of the iron (III) oxide (maghemite, JCPDS No.39-1346) were detected [58][59][60]. The estimation of the average crystallite size from the (104) peak FWHM gave a smaller value (18.1 nm) with respect to the pure oxide, in good agreement with the TEM analysis and literature reports on doped Fe 2 O 3 [61][62][63] as well. Finally, the detection of a very broad background in the pattern of Ge-containing NFs suggested the presence of an amorphous component. Figure 2 compares the micro-Raman spectra of Fe 2 O 3 and Fe 2 O 3 :Ge NFs, as resulting from the average on different random locations in each specimen. The group theory predicts seven normal phonon modes for α-Fe 2 O 3 [64][65][66], namely two "external" modes involving the translations and rotations of entire Fe 2 O 3 units (E g (1) and E g (2) at 245 and 294 cm −1 , respectively) and five "internal" modes associated to the motion within a single Fe(O) 6 octahedral unit (A 1g (1), E g (3), E g (4), A 1g (2) and E g (5) at 225, 298, 412, 500, and 612 cm −1 , respectively) [67][68][69]. In addition, the peaks originating from these modes, a strong asymmetrical band peaking at~1320 cm −1 and two weak features at~660 and 830 cm −1 were detected in the spectrum of the Fe 2 O 3 NFs. The peak at~660 cm −1 is associated to the IR-active longitudinal optical (LO) E u mode, activated by disorder (grain size, defects, heteroatoms) in the hematite lattice [36,42,55,64,66,69,70], while the asymmetrical band has been recently ascribed to its overtone [66,67,69]. The peak at 830 cm −1 is assigned to the photo-excitation of a spin-wave (one-magnon mode, 1M) in the canted antiferromagnetic structure of α-Fe 2 O 3 [66,71]. Spectra collected from different random locations in the reference sample (not shown) did not significantly differ from each other, evidencing the spatially uniform formation of hematite.
Five Raman-allowed phonon modes (A 1g + E g + 3T 2g ) are predicted for the γ-polymorph of iron (III) oxide having a spinel crystalline structure [65,72], similarly to magnetite (Fe 3 O 4 ). Nonetheless, in maghemite, a Fe-deficient form of Fe 3 with denoting a Fe-vacancy and th and oh indicating tetrahedral and octahedral sites, respectively), the peaks are broader [72]. Indeed, three peaks are usually detected (T 1g , E g and A 1g at 360,~510 and~705 cm −1 , respectively) in γ-Fe 2 O 3 , along with the band originating from 1M photo-excitation at~1430 cm −1 [73,74]. In the spectrum of Fe 2 O 3 :Ge NFs (Figure 2), the peaks of both αand γ-phase of the iron oxide were clearly visible, in full agreement with the evidences of XRD analysis. On the contrary, no obvious peak ascribable to germanium dioxide was detected. Spectra recorded at different random locations were different from each other ( Figure S5), indicating that the incorporation of germanium led to the spatially non-uniform formation of αand γ-polymorphs of the oxide. eroatoms) in the hematite lattice [36,42,55,64,66,69,70], while the asymmetrical band has been recently ascribed to its overtone [66,67,69]. The peak at ~830 cm −1 is assigned to the photo-excitation of a spin-wave (one-magnon mode, 1M) in the canted antiferromagnetic structure of α-Fe2O3 [66,71]. Spectra collected from different random locations in the reference sample (not shown) did not significantly differ from each other, evidencing the spatially uniform formation of hematite.  Figure 3a displays the X-ray photoelectron spectra of Fe 2p core level in the electrospun NFs. The spectrum of Fe 2 O 3 NFs consisted of two spin-orbit components at binding energies of 710.5 eV (2p 3/2 ) and 724.0 eV (2p 1/2 ) and of two Fe 3+ shake-up satellites at 718.3 and 732.1 eV, as typical of hematite [63,75]. In agreement with the literature [36,41,56], the incorporation of germanium in the iron oxide lattice did not caused appreciable variations in the spectral profile, but a slight shift (~0.4 eV) of spin-orbit doublets and shake-up satellites towards higher binding energies, for which both co-formation of maghemite [76] and incorporation of germanium [41] could have been responsible. In the Fe 2 O 3 :Ge NFs, the binding energy position of the Ge 2p 3/2 (1220 eV, inset to Figure 3a) and Ge 3d (32 eV, not shown) core levels evidenced the presence of Ge ions with +4 oxidation state [77][78][79]. Five Raman-allowed phonon modes (A1g + Eg + 3T2g) are predicted for the γ-polymorph of iron (III) oxide having a spinel crystalline structure [65,72], similarly to magnetite (Fe3O4). Nonetheless, in maghemite, a Fe-deficient form of Fe3O4 (Fe 3+ th [Fe5/3 3+ €1/3]ohO4 −2 , with denoting a Fe-vacancy and th and oh indicating tetrahedral and octahedral sites, respectively), the peaks are broader [72]. Indeed, three peaks are usually detected (T1g, Eg and A1g at ~360, ~510 and ~705 cm −1 , respectively) in γ-Fe2O3, along with the band originating from 1M photo-excitation at ~1430 cm −1 [73,74]. In the spectrum of Fe2O3:Ge NFs (Figure 2), the peaks of both α-and γ-phase of the iron oxide were clearly visible, in full agreement with the evidences of XRD analysis. On the contrary, no obvious peak ascribable to germanium dioxide was detected. Spectra recorded at different random locations were different from each other ( Figure S5), indicating that the incorporation of germanium led to the spatially non-uniform formation of α-and γ-polymorphs of the oxide. Figure 3a displays the X-ray photoelectron spectra of Fe 2p core level in the electrospun NFs. The spectrum of Fe2O3 NFs consisted of two spin-orbit components at binding energies of 710.5 eV (2p3/2) and 724.0 eV (2p1/2) and of two Fe 3+ shake-up satellites at 718.3 and 732.1 eV, as typical of hematite [63,75]. In agreement with the literature [36,41,56], the incorporation of germanium in the iron oxide lattice did not caused appreciable variations in the spectral profile, but a slight shift (~0.4 eV) of spin-orbit doublets and shake-up satellites towards higher binding energies, for which both co-formation of maghemite [76] and incorporation of germanium [41] could have been responsible. In the Fe2O3:Ge NFs, the binding energy position of the Ge 2p3/2 (1220 eV, inset to Figure 3a) and Ge 3d (32 eV, not shown) core levels evidenced the presence of Ge ions with +4 oxidation state [77][78][79]. O 1s core level of the two samples exhibited evident differences both in terms of binding energy and shape (Figure 3b), hinting at different surrounding cations. Two components at 530.5 (OI) and 532.7 eV (OII)were singled out in the spectrum of Fe2O3 NFs ( Figure S6a); they were ascribed to O2 − ions of the surface lattice oxygen in α-Fe2O3 matrix and to surface hydroxyl (Fe-OH) groups, respectively [80]. The spectrum of Fe2O3:Ge NFs exhibited larger broadening (Figure 3b), mirroring the presence of more kinds of oxygen species [81] and clearly pointing out the incorporation of Ge ions into the γ-Fe2O3 lattice (i.e., solid solutions), as well as the co-presence of segregate phases of γ-Fe2O3 and GeO2 oxides [81,82]. In maghemite, surface lattice oxygen has been reported to contribute to the O1s intensity at lower binding energy compared with hematite [83,84], whereas the O1s peak of germanium dioxide has been detected at 531.8 eV [77]. Thus, in agreement with the results of XRD and MRS analyses, not only Fe-O species in α-Fe2O3 nanocrystals, but also surface lattice oxygen in γ-Fe2O3 and GeO2 phases were identified in the O1s spectrum of O 1s core level of the two samples exhibited evident differences both in terms of binding energy and shape (Figure 3b), hinting at different surrounding cations. Two components at 530.5 (O I ) and 532.7 eV (O II )were singled out in the spectrum of Fe 2 O 3 NFs ( Figure S6a); they were ascribed to O 2 − ions of the surface lattice oxygen in α-Fe 2 O 3 matrix and to surface hydroxyl (Fe-OH) groups, respectively [80]. The spectrum of Fe 2 O 3 :Ge NFs exhibited larger broadening (Figure 3b), mirroring the presence of more kinds of oxygen species [81] and clearly pointing out the incorporation of Ge ions into the γ-Fe 2 O 3 lattice (i.e., solid solutions), as well as the co-presence of segregate phases of γ-Fe 2 O 3 and GeO 2 oxides [81,82]. In maghemite, surface lattice oxygen has been reported to contribute to the O1s intensity at lower binding energy compared with hematite [83,84], whereas the O1s peak of germanium dioxide has been detected at 531.8 eV [77]. Thus, in agreement with the results of XRD and MRS analyses, not only Fe-O species in α-Fe 2 O 3 nanocrystals, but also surface lattice oxygen in γ-Fe 2 O 3 and GeO 2 phases were identified in the O1s spectrum of Fe 2 O 3 :Ge NFs, as well as surface hydroxyl groups ( Figure S6b). The different oxygen species were quantified assuming that the contributions at lower bindings energies (530.1-531.6 eV), O Ia and O Ib , originate from Fe-O and Ge-O lattice oxygen, respectively, whereas the signal at higher binding energy (532.7 eV), O II , is related to either the presence of molecular oxygen weakly bounded or oxygen from adsorbed species (e.g., hydroxyl species) [81]. Accordingly, the concentrations of the three species were estimated to be 57.6% (O Ia ), 31.4% (O Ib ) and 11.0% (O II ). Therefore, considering that the fraction of oxygen relating to the isolated GeO 2 phase (OI b ) is 31.4%, being the atomic surface abundance of oxygen 68.8%, that of germanium 10.9% (Table S1) and the stoichiometric ratio between oxygen and germanium of 2:1, it followed that almost all of the Ge ions on the surface of the Fe 2 O 3 :Ge NFs should be involved in the formation of the oxide phase (Table S2).
Thus, summarizing, the XRD, XPS, and MRS analyses indicated that Fe 2 O 3 NFs can be considered as a single-phase compound; differently, the Fe 2 O 3 :Ge NFs are more complex, both in terms of phase composition and compositional homogeneity, being composed by at least two crystalline phases, and an amorphous component.
To further explore the compositional and structural aspects of Fe 2 O 3 :Ge NFs, X-ray absorption spectroscopy and neutron diffraction analysis were carried out.
XAS measurements were acquired on Fe 2 O 3 :Ge NFs at the Fe and Ge K-edges, aiming at selectively obtaining electronic and structural information on both atomic species. The XANES region gives direct indication of the oxidation state; moreover, its fingerprinting properties allow discerning between different species by rapid comparison with the spectra of reference materials. For this purpose, Figure 4a compares the XANES spectrum of the sample at the Fe K-edge with those of hematite (α-Fe 2 O 3 ) and maghemite (γ-Fe 2 O 3 ) [85]. The energy position of the rising edge is indicative of the oxidation state of the sample: in this case, the edge energy of Fe 2 O 3 :Ge NFs, marked as a black dotted line, was the same as in αand γ-Fe 2 O 3 , indicating that all iron is present as Fe(III), in agreement with results of XPS analysis. All the spectral features in the spectrum were attributable either to maghemite or hematite; for a quantification of the effective composition, a linear combination fit was performed (see Figure S7). The best fit, obtained with 77 (1)% maghemite and 23 (1)% hematite, showed a very good agreement with the experimental signal. However, the presence of a small amount of iron-based oxyhydroxide species, especially γ-FeOOH (lepidocrocite) and ferrihydrite, which present some spectral features in the same energy region as maghemite [86,87], could not be excluded. Fe2O3:Ge NFs, as well as surface hydroxyl groups ( Figure S6b). The different oxygen species were quantified assuming that the contributions at lower bindings energies (530.1-531.6 eV), OIa and OIb, originate from Fe-O and Ge-O lattice oxygen, respectively, whereas the signal at higher binding energy (532.7 eV), OII, is related to either the presence of molecular oxygen weakly bounded or oxygen from adsorbed species (e.g., hydroxyl species) [81]. Accordingly, the concentrations of the three species were estimated to be 57.6% (OIa), 31.4% (OIb) and 11.0% (OII). Therefore, considering that the fraction of oxygen relating to the isolated GeO2 phase (OIb) is 31.4%, being the atomic surface abundance of oxygen 68.8%, that of germanium 10.9% (Table S1) and the stoichiometric ratio between oxygen and germanium of 2:1, it followed that almost all of the Ge ions on the surface of the Fe2O3:Ge NFs should be involved in the formation of the oxide phase (Table S2). Thus, summarizing, the XRD, XPS, and MRS analyses indicated that Fe2O3 NFs can be considered as a single-phase compound; differently, the Fe2O3:Ge NFs are more complex, both in terms of phase composition and compositional homogeneity, being composed by at least two crystalline phases, and an amorphous component.
To further explore the compositional and structural aspects of Fe2O3:Ge NFs, X-ray absorption spectroscopy and neutron diffraction analysis were carried out. XAS measurements were acquired on Fe2O3:Ge NFs at the Fe and Ge K-edges, aiming at selectively obtaining electronic and structural information on both atomic species. The XANES region gives direct indication of the oxidation state; moreover, its fingerprinting properties allow discerning between different species by rapid comparison with the spectra of reference materials. For this purpose, Figure 4a compares the XANES spectrum of the sample at the Fe K-edge with those of hematite (α-Fe2O3) and maghemite (γ-Fe2O3) [85]. The energy position of the rising edge is indicative of the oxidation state of the sample: in this case, the edge energy of Fe2O3:Ge NFs, marked as a black dotted line, was the same as in α-and γ-Fe2O3, indicating that all iron is present as Fe(III), in agreement with results of XPS analysis. All the spectral features in the spectrum were attributable either to maghemite or hematite; for a quantification of the effective composition, a linear combination fit was performed (see Figure S7). The best fit, obtained with 77 (1)% maghemite and 23 (1)% hematite, showed a very good agreement with the experimental signal. However, the presence of a small amount of iron-based oxyhydroxide species, especially γ-FeOOH (lepidocrocite) and ferrihydrite, which present some spectral features in the same energy region as maghemite [86,87], could not be excluded.    for nanosized and/or amorphous samples, and displayed a striking similarity. At the Ge K-edge, the main absorption peak (white line, WL) at 11,112 eV is due to the transition from filled 1s orbitals to empty 4p orbitals. The WL of the spectrum of Fe 2 O 3 :Ge NFs, as well as for quartz-like GeO 2 , was unique and intense, as expected in case of tetrahedral coordination. Indeed, when Ge is in an octahedral environment (e.g., in rutile-like GeO 2 ), the WL is broadened and split into three distinct features [88,89].
In order to gain further evidence on the coordination around Ge, the EXAFS signal and the corresponding Fourier transform (FT) were extracted ( Figure S8a,b). The FT showed an intense peak around 1.7 Å, due to the first Ge-O distance, while all the other contributions above 3 Å became negligible, confirming the amorphousness of the Ge local environment. The x(k) signal resulting from the first shell was isolated through a Fourier-filtering process and fitted, allowing to selectively consider the contribution of the first neighboring atoms around the photoabsorber. The Fourier filtered signal and the fit are shown in Figure S8c,d. The best fit was obtained starting from a GeO 4 structural model; the Ge-O distance was found to be 1.771 (6) Å, in good agreement with the crystallographic distance in quartz-like GeO 2 . The parameters of the EXAFS fit are reported in Table S3. In summary, all the XAS evidences pointed towards the presence of Ge(IV) in fourfold coordination, strongly suggesting that Ge is hosted in the tetrahedral sites of the maghemite spinel structure.
PND data were analyzed on the base of the phases identified by XRD, XAS and MRS; hence, to properly fit the pattern, the hematite and maghemite structures were introduced. The final Rietveld refinement is presented in Figure 5. With respect to the XRD pattern, the neutron data showed strong extra reflections (e.g., 17 • -26 • range, 51 • in Figure 5) associated to the magnetic structure of hematite [90]; this magnetic structure was introduced and considered in the refinement as an independent magnetic phase [91].
Appl. Sci. 2021, 11, x FOR PEER REVIEW 9 of 16 analysis. All the spectral features were visibly broadened in Fe2O3:Ge NFs, as expected for nanosized and/or amorphous samples, and displayed a striking similarity. At the Ge Kedge, the main absorption peak (white line, WL) at 11,112 eV is due to the transition from filled 1s orbitals to empty 4p orbitals. The WL of the spectrum of Fe2O3:Ge NFs, as well as for quartz-like GeO2, was unique and intense, as expected in case of tetrahedral coordination. Indeed, when Ge is in an octahedral environment (e.g., in rutile-like GeO2), the WL is broadened and split into three distinct features [88,89].
In order to gain further evidence on the coordination around Ge, the EXAFS signal and the corresponding Fourier transform (FT) were extracted ( Figure S8a,b). The FT showed an intense peak around 1.7 Å, due to the first Ge-O distance, while all the other contributions above 3 Å became negligible, confirming the amorphousness of the Ge local environment. The x(k) signal resulting from the first shell was isolated through a Fourierfiltering process and fitted, allowing to selectively consider the contribution of the first neighboring atoms around the photoabsorber. The Fourier filtered signal and the fit are shown in Figure S8c,d. The best fit was obtained starting from a GeO4 structural model; the Ge-O distance was found to be 1.771 (6) Å, in good agreement with the crystallographic distance in quartz-like GeO2. The parameters of the EXAFS fit are reported in Table S3. In summary, all the XAS evidences pointed towards the presence of Ge(IV) in fourfold coordination, strongly suggesting that Ge is hosted in the tetrahedral sites of the maghemite spinel structure.
PND data were analyzed on the base of the phases identified by XRD, XAS and MRS; hence, to properly fit the pattern, the hematite and maghemite structures were introduced. The final Rietveld refinement is presented in Figure 5. With respect to the XRD pattern, the neutron data showed strong extra reflections (e.g., 17°-26° range, 51° in Figure 5) associated to the magnetic structure of hematite [90]; this magnetic structure was introduced and considered in the refinement as an independent magnetic phase [91]. As the maghemite phase, cubic (S.G. Fd3m and P4332) and tetragonal (S.G. P41212) polymorphs had been reported in order to account for the ordering of the Fe vacancies [92]. A preliminary Le Bail fit, considering the different maghemite polymorphs, led to identification of the Fd3m phase as the one able to properly account for most of the observed reflections. On this base, Rietveld refinements were performed using the hexagonal α-Fe2O3 and the cubic γ-Fe2O3 structures. The attempts to refine the scale of GeO2 as a possible impurity led to very small amount of the oxide, in good agreement with the results of MRS analysis. This finding indicated that most of germanium was successfully dispersed in the iron oxide phase(s) as a dopant.
Results obtained from XAS clearly indicated the presence of Ge in tetrahedral coordination. Based on these indications, for the refinement of neutron data, germanium was considered only in the maghemite structure (the only one where FeO4 sites are present). As the maghemite phase, cubic (S.G. Fd3m and P4 3 32) and tetragonal (S.G. P4 1 2 1 2) polymorphs had been reported in order to account for the ordering of the Fe vacancies [92]. A preliminary Le Bail fit, considering the different maghemite polymorphs, led to identification of the Fd3m phase as the one able to properly account for most of the observed reflections. On this base, Rietveld refinements were performed using the hexagonal α-Fe 2 O 3 and the cubic γ-Fe 2 O 3 structures. The attempts to refine the scale of GeO 2 as a possible impurity led to very small amount of the oxide, in good agreement with the results of MRS analysis. This finding indicated that most of germanium was successfully dispersed in the iron oxide phase(s) as a dopant.
Results obtained from XAS clearly indicated the presence of Ge in tetrahedral coordination. Based on these indications, for the refinement of neutron data, germanium was considered only in the maghemite structure (the only one where FeO 4 sites are present). However, this scenario was not compatible with the nominal Fe:Ge molar ratio (11.7:1), as all performed refinements led to >85% α-Fe 2 O 3 and <15% γ-Fe 2 O 3 , which should imply a quantitative Ge substitution in the maghemite structure. Nevertheless, the maghemite phase is characterized by a magnetic structure at RT, but the quality of the experimental data did not allow for the determination of such structure and for its subsequent refinement. Due to the complexity of the system, the results obtained from the neutron diffraction gave only a partial vision of the actual phases and composition. It was concluded that the Fe 2 O 3 :Ge NFs were composed by α-Fe 2 O 3 and γ-Fe 2 O 3 , as dominant phases, and of an amorphous component with similar chemical composition. Most of germanium was accommodated in the tetrahedral sites of the maghemite structure, even if also the hematite phase could accept Ge dopant, probably on the defective surface sites as no GeO 6 species (that should be associated to the bulk α-Fe 2 O 3 ) were revealed. These results were consistent with previous report on the possibility to obtain αand γ-phases of iron oxides, respectively oxidized and fully oxidized, upon calcination in air above 500 • C of a hydrated precursor [93]. Results reported in the present work suggested that the incorporation of germanium could drive the phases' formation, with the stabilization of the γ-polymorph, generally not expected at 600 • C.

Electrochemical Properties of the NFs
PCGA measurements were performed to get cathodic and anodic profiles in quasireversible conditions. In this technique, the potential (ranging from 3.0 to 0.01 V vs. Na + /Na) is decreased (cathodic branch) or increased (anodic branch) by 4 mV and, for each step, the charge is collected until the current becomes less than 20 mA g −1 . Considering the theoretical specific capacity of hematite (1005 mAh g −1 ), the current threshold corresponds to C/50, which should allow to collect all the available charge. The corresponding profiles are reported in Figure 6a. The first cycle Coulomb efficiency was 55%, probably due to both the large amount of carbon at the electrode and the conversion reactions already discussed for similar phases [36]. After the first cycle, the profiles showed the typical pseudocapacitive behavior of these composites, with three different regions of different slopes. After the initial cycle, the specific capacities were around 300 mAh g −1 .
Appl. Sci. 2021, 11, x FOR PEER REVIEW 10 of 16 However, this scenario was not compatible with the nominal Fe:Ge molar ratio (11.7:1), as all performed refinements led to >85% α-Fe2O3 and <15% γ-Fe2O3, which should imply a quantitative Ge substitution in the maghemite structure. Nevertheless, the maghemite phase is characterized by a magnetic structure at RT, but the quality of the experimental data did not allow for the determination of such structure and for its subsequent refinement. Due to the complexity of the system, the results obtained from the neutron diffraction gave only a partial vision of the actual phases and composition. It was concluded that the Fe2O3:Ge NFs were composed by α-Fe2O3 and γ-Fe2O3, as dominant phases, and of an amorphous component with similar chemical composition. Most of germanium was accommodated in the tetrahedral sites of the maghemite structure, even if also the hematite phase could accept Ge dopant, probably on the defective surface sites as no GeO6 species (that should be associated to the bulk α-Fe2O3) were revealed. These results were consistent with previous report on the possibility to obtain α-and γ-phases of iron oxides, respectively oxidized and fully oxidized, upon calcination in air above 500 °C of a hydrated precursor [93]. Results reported in the present work suggested that the incorporation of germanium could drive the phases' formation, with the stabilization of the γ-polymorph, generally not expected at 600 °C.

Electrochemical Properties of the NFs
PCGA measurements were performed to get cathodic and anodic profiles in quasireversible conditions. In this technique, the potential (ranging from 3.0 to 0.01 V vs. Na + /Na) is decreased (cathodic branch) or increased (anodic branch) by 4 mV and, for each step, the charge is collected until the current becomes less than 20 mA g −1 . Considering the theoretical specific capacity of hematite (1005 mAh g −1 ), the current threshold corresponds to C/50, which should allow to collect all the available charge. The corresponding profiles are reported in Figure 6a. The first cycle Coulomb efficiency was 55%, probably due to both the large amount of carbon at the electrode and the conversion reactions already discussed for similar phases [36]. After the first cycle, the profiles showed the typical pseudocapacitive behavior of these composites, with three different regions of different slopes. After the initial cycle, the specific capacities were around 300 mAh g −1 . Galvanostatic cycling was performed to evaluate the rate capability and the capacity retention of the electrodes (Figure 6b). The GCPLs at low currents (C/20) confirmed the capacities observed in PCGA, with an average value of 320 mAh g −1 in the anodic process. The large uncertainty of the low current capacities (C/20 and C/10) can be attributed to the intrinsic multi-phase nature of the doped oxide, because the different phases have probably different kinetics affecting the extension of reaction with the sodium ions cycle by cycle. At higher rates, the data precision increases, because the slower phases became less active and do not substantially contribute to the value. The capacity decreased with the C Galvanostatic cycling was performed to evaluate the rate capability and the capacity retention of the electrodes (Figure 6b). The GCPLs at low currents (C/20) confirmed the capacities observed in PCGA, with an average value of 320 mAh g −1 in the anodic process. The large uncertainty of the low current capacities (C/20 and C/10) can be attributed to the intrinsic multi-phase nature of the doped oxide, because the different phases have probably different kinetics affecting the extension of reaction with the sodium ions cycle by cycle. At higher rates, the data precision increases, because the slower phases became less active and do not substantially contribute to the value. The capacity decreased with the C rate, and 270, 240, 205, 175, and 140 mAh g −1 were obtained at C/10, C/5, 1C, and 2C, respectively. On the contrary, the Coulomb efficiencies increased with the current (from 94% at C/20 to 99.2% at 2C), pointing out the presence of parasitic cathodic processes with lower kinetic compared to the charge accumulation reaction. After the 60th cycle (2C), the current was set back to C/20 to evaluate the capacity retention after 70 cycles, which was calculated as 97% with a good and stable recovering of the initial value. Thus, the electrodes showed good specific capacity as anode for sodium ion batteries, with excellent rate capability (140 mAh g −1 at 2C which corresponds to 2 A g −1 ) thanks to the synergistic combination of the electronic transport properties of the complex material, the nanostructured morphologies and the pseudo-capacitive nature of the charge storage mechanism. These performances exceeded those of pure hematite and were in agreement with those obtained from metalloid-doped hematites [36] (Figure S9). Since Fe 2 O 3 :Ge NFs and Fe 2 O 3 :Si NFs were prepared by using the same dopant:iron mass ratio (1:9), the larger relative amount of silicon incorporated in the Fe 2 O 3 lattice may partly account for the higher specific capacity of Fe 2 O 3 :Si NFs at lower rates. In any case, the different percentages of α-Fe 2 O 3 and γ-Fe 2 O 3 (and amorphous) phases present in the Fe 2 O 3 :Si NFs and Fe 2 O 3 :Ge NFs made it hard to make a direct comparison considering only the atomic chemical physical properties.
To test the cyclability of the materials at constant current, 180 cycles at C/10 (70 mA g −1 ) have been performed ( Figure S10). The initial capacity values are in agreement with the rate test of Figure 6b, with an average value of 230 ± 10 mAh g −1 (cycles 1-5). Upon cycling, the material shows a constant decrease of the performances of 0.75 mAh g −1 per cycle, with capacity retentions of 54% and 42% after 100 and 180 cycles, respectively. These unsatisfactory results are mainly due, as already discussed in Ref. [36], to the binder instability, which tends to crystallize upon cycling.

Conclusions
The thorough characterization of the investigated electrospun Fe 2 O 3 and Fe 2 O 3 :Ge NFs via SEM, TEM, MRS, XPS, X-ray and neutron diffraction and XAS allow concluding that: Germanium, mostly dispersed as an ionized impurity (with 4+ oxidation state), occupies the tetrahedral sites of the maghemite lattice and probably the defective hematite surface sites.

•
The synergy between nanostructured morphology and electronic transport properties of Fe 2 O 3 :Ge NFs, together with the pseudo-capacitive nature of the charge storage mechanism, are responsible for their excellent rate capability as anode for SIBs (140 mAh g −1 at 2C).

Supplementary Materials:
The following are available online at https://www.mdpi.com/2076-3 417/11/4/1483/s1, Figure S1: Sketch of the synthesis route followed to produce the electrospun iron oxide NFs, Figure S2: Photo of the as-calcined Fe 2 O 3 and Fe 2 O 3 :Ge NFs., Figure S3: Elemental composition of the NFs, Figure S4: XRD patterns of the investigated NFs, Figure S5: Micro-Raman spectra, as measured at different random locations in the sample, Figure S6: Deconvolution of the O 1s core level profiles of Fe 2 O 3 NFs and Fe 2 O 3 :Ge NFs, Figure S7: Linear combination fit of the spectrum of Fe 2 O 3 :Ge NFs acquired at the Fe K-edge, Figure S8: EXAFS signal and Fourier transform for the spectrum of Fe 2 O 3 :Ge NFs at the Ge K-edge, Figure S9: Results of the rate capability test on Fe 2 O 3 :Ge NFs, compared with previously investigated Fe 2 O 3 and Fe 2 O 3 :Si NFs, Figure S10. Results of the cyclability test on Fe 2 O 3 :Ge NFs. Table S1: Surface composition of the NFs, Table S2: Binding energies and fractional area of the components of O 1s core level in the NFs, Table S3: parameters of the EXAFS fit.
Funding: This research was partially funded by the Italian Ministry of University and Research (MUR) through the project "Towards sustainable, high-performing, all-solid-state sodium-ion" (PRIN 2017MCEEY4).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The raw data and processing aids, required to reproduce the findings of this work, cannot be shared at this time as they form the basis for an ongoing study.