Amorphous Mo5O14-Type/Carbon Nanocomposite with Enhanced Electrochemical Capability for Lithium-Ion Batteries

An amorphous MomO3m−1/carbon nanocomposite (m ≈ 5) is fabricated from a citrate–gel precursor heated at moderate temperature (500 °C) in inert (argon) atmosphere. The as-prepared Mo5O14-type/C material is compared to α-MoO3 synthesized from the same precursor in air. The morphology and microstructure of the as-prepared samples are characterized by scanning electron microscopy (SEM), X-ray diffraction (XRD), and Raman scattering (RS) spectroscopy. Thermal gravimetry and elemental analysis indicate the presence of 25.8 ± 0.2% of carbon in the composite. The SEM images show that Mo5O14 is immersed inside a honeycomb-like carbon matrix providing high surface area. The RS spectrum of Mo5O14/C demonstrates an oxygen deficiency in the molybdenum oxide and the presence of a partially graphitized carbon. Outstanding improvement in electrochemical performance is obtained for the Mo5O14 encapsulated by carbon in comparison with the carbon-free MoO3.


Introduction
Due to their numerous applications including their vital role as light weight, long life and high-energy density power sources, special attention has been directed towards lithium-ion batteries (LIBs). These power sources find major applications from portable electronic appliances to electric vehicles. As the positive electrode plays a major role in the electrochemical performance of the LIBs, many efforts of research are currently made on active electrode materials with high capability and long cycling life [1]. Among the molybdenum-oxide series evaluated as electrode materials, α-MoO 3 and MoO 2 are thermodynamically stable phases. They crystallize in a unique layered structure (orthorhombic, Pbnma space group) and a rutile-like structure (monoclinic P2 1 /c space group), respectively [2]. In the same Mo-O binary system, the oxygen deficient Mo m O 3m−1 phases include Mo 4 O 47 , which crystallize in the so-called Magnéli phases described in details by Kihlborg [3]. These oxides have been identified as derived from MoO 3 by shear mechanism with networks of MoO 6 octahedra. These units are connected in three dimensions with empty parallel tunnels involving the presence of potential sites for hosting Li + ions. They are performant cathode materials in non-aqueous lithium cells [4][5][6][7][8]. Few works reporting the lithiation of Mo 5 O 14 are available in the literature [5,[9][10][11]. Preliminary patterns of the fully performant cathode materials in non-aqueous lithium cells [4][5][6][7][8]. Few works reporting the lithiation of Mo5O14 are available in the literature [5,[9][10][11]. Preliminary patterns of the fully lithiated Mo5O14 crystal suggest that the unit cell expands in the b direction and contracts in both the a and c directions upon insertion of lithium [9]. Cignini et al. reported a specific capacity of 310 mA g −1 for Mo5O14 as the positive electrode of a primary Li cell discharged at 0.5 mA cm −2 and 1.0 V cut-off. [5]. Nazri and Julien [10] showed that MoO2.8, an oxygen-deficient Mo5O14 prepared by dehydration and annealing treatment of molybdic acid powder (MoO31H2O) at 750 °C, exhibited a high electrical conductivity of 10 −2 S cm −1 . Such a material displayed an excellent reversible capacity (1. 45 Li/Mo) in cells with 1 mol L −1 LiClO4 in propylene carbonate (PC) electrolyte in the potential range 3.3-1.3 V vs. Li + /Li.
However, MoO3 and MomO3m−1 lithiated oxides suffer from a large volume expansion (about 104% for MoO3). During Li + insertion, pulverization of the cathode occurs at a potential of less than 1.5 V and disconnection from the current collector. As a result, the electrochemical performance is poor [12]. Different strategies can be used to improve the cyclability as follows. Nanosized MomO3m−1 particles were prepared with low crystallinity [13,14]. Blended hybrid was fabricated [15]. Composites with addition of an electronically-conductive carbonaceous materials were synthesized [12,[16][17][18][19]. The carbon has two beneficial effects: it opposes the aggregation of the active particles and it accommodates the volume change of electrode material during cycling [17]. Different synthesis techniques were used to fabricate MoO3/C composites: electrospinning method [18], hydrothermal route [12,16,19,20], template method [21], spray pyrolysis [22] and ball milling [23].
In this paper, a nanocomposite of oxygen-deficient MomO3m−1/carbon is synthesized by a sol-gel method using citric acid as chelating agent. The calcination of dried precursor in argon atmosphere results in the formation of the amorphous-like Mo5O14 (MoO2.8) oxide immersed in a carbon matrix. This new material is characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), surface area analysis (BET) and Raman scattering (RS) spectroscopy. Finally, electrochemical properties of MoO3/C composite for lithium ion batteries were also investigated for comparison.

Materials and Methods
Pure MoO3 (P-MoO3) sample and MomO3m−1/C nanocomposite were prepared by a citrate-gel method using ammonium molybdate tetrahydrate (AMT, (NH4)6Mo7O24•4H2O). First, a stoichiometric amount of AMT was dissolved in de-ionized water (DI) and stepwise added to the stirring aqueous solution of citric acid (CA) with metal/CA ratio of 1:1 under neutral pH adjusted with ammonium hydroxide; Second, the overall solution was stirred for 3 h to form homogenous mixture through reaction between the metal ion and the chelating agent. Transparent gel was formed after a slow evaporation of the solution heated at 70 °C. Furthermore, continuous heating with stirring led to the transformation of the gel to xerogel, which converted to powder by further drying at ca. 120 °C. Pure MoO3 crystals were obtained by calcining the milled precursor at 500 °C for 5 h in air, while the MomO3m−1/C composite was formed under argon atmosphere. The overall synthesis process is presented in Scheme 1.  The composition was determined by thermal gravimetric analysis (TGA) and differential scanning calorimetry (DSC) using an analyzer model TGA-7 series (Perkin Elmer, Waltham, MA, USA,) in the temperature range of 30-1000 • C in air at a heating rate of 10 • C min −1 . Elemental analysis was performed using a CHNOS analyzer (Elementar Analysen Systeme GmbH, Langenselbold, Germany). The structure of the samples was analyzed by X-ray diffraction (XRD) using a X'Pert diffractometer (Philips, Hamburg, Germany) equipped with a CuK α X-ray source (λ = 1.54056 Å). Data were collected in the 2θ range of 10-80 • . The particle morphology was examined by scanning electron microscopy (SEM) (JEOL microscope, Akishima, Japan). Raman spectra were collected at a spectral resolution of 2 cm −1 with a double monochromator (Jobin-Yvon model U1000) using the 514.5 nm line of an Ar-ion laser (Spectra-Physics 2020) in a backscattering geometry. The laser power was kept below 25 mW to prevent the sample degradation by the laser spot. The specific surface area was measured by nitrogen adsorption/desorption at 77 K using the Brunauer-Emmett-Teller (BET) method (Quantachrome NOVA Automated Gas Sorption).
Electrochemical properties were investigated by cyclic voltammetry (CV) and galvanostatic charge-discharge (GCD) techniques. The cathodes were prepared by mixing 80% (w/w) of the active material, 10% (w/w) super C65 carbon (TIMCAL, Lac-des-Iles, Canada) and 10% (w/w) polyvinylidenefluoride (Solef PVdF 6020 binder, Solvay, Brussels, Belgium), in N-methyl-2-pyrrolidone (NMP, Sigma-Aldrich, St. Louis, MO, USA) to get a slurry, which was coated on copper foil at a loading of 2.1 mg cm −2 . After drying over night at 80 • C, discs were punched out of this film with a diameter of 1.4 cm. The 2032-type coin cells were assembled in an argon-filled glove box with lithium foil (Alfa Aesar, Haverhill, MA, USA) as the anode, 1 mol L −1 LiPF 6 in ethylene carbonate:dimethyl carbonate (EC:DMC) = 50:50 (v/v) (LP30, Sigma-Aldrich, battery grade) as the electrolyte, and glass microfiber filters (Whatmann ® -GF/D 70 mm Ø) as the separator. A multi-channel potentiostat (VMP3, Bio-Logic, Seyssinet-Pariset, France) was used to test the electrochemical performance of the electrodes at 25 • C in the voltage range of 0.6-3.5 V vs. Li + /Li. Figure 1a shows the XRD pattern of the molybdenum oxide/carbon composite obtained by calcination of ammonium molybdate tetrahydrate in argon atmosphere. This diagram exhibits four broad peaks centered at 2θ angles of 25.7, 35.1, 43.5 and 62.2 • , which suggest a quasi-amorphous texture of the Mo oxide in a carbon matrix obtained by the decomposition of the chelating agent in absence of oxygen. The pattern shown in Figure 1a compares well to that of the tetragonal Mo 5 O 14 phase (P4/mbm space group, a = 2.2989 nm, and c = 0.3936 nm) reported by Kihlborg [24]. Note this composition deduced form this XRD pattern implies the molybdenum valence distribution Mo 6+ 3 Mo 5+ 2 . This valence mixing in turn implies a high concentration of free carriers and thus a large electrical conductivity, which will be beneficial to the electrochemical properties. The crystal structure of Mo 5 O 14 can be depicted as complexes of linked MoO 6 octahedra and MoO 7 pentagonal bipyramids [25]. The peak at 2θ = 25.7 • corresponds to the overlapping (001), (060) and (504) lines of the Mo 5 O 14 lattice, while the less intense peaks at 2θ = 35.1 • and 43.5 • coincide with the (541) and (002) Bragg lines of the tetragonal structure, respectively [26]. The crystallite size calculated by the Scherrer's formula from the full-width of the broad line at 2θ = 25.7 • is found to be 2.9 nm. The presence of the disordered carbon (low degree of graphitization) is evidenced by the main signal at 2θ ≈ 26 • ((002) Bragg line) probably overlapping with the main signals of the Mo 5 O 14 phase. However, the amorphous nature of carbon deposited on the Mo oxide at a low temperature of 500 • C is confirmed by the absence of the (100) and (101) carbon-related peaks in the 40-45 • 2θ range [16]. The formation of an amorphous Mo 5 O 14 /C composite under argon gas is due to the presence of CO and CO 2 gases generated by the combustion reaction of the carbon source using citric acid, which reduces the Mo 6+ ions of the ammonium molybdate. The TGA experiments performed during the synthesis (Figure 2a)  the organic compound C6H8O7 accompanied by the decomposition of (NH4)6Mo7O24 (partial desorption of H2O and NH3) into oxides. At this stage, a weight loss of almost 50% was recorded.  The XRD pattern of P-MoO3 ( Figure 1b) displays the typical pattern of a well-crystallized α-MoO3 phase that can be indexed in the orthorhombic Pbnm space group (JPCDS card 76-1003). The coarse-grained powder presents a preferred orientation of (0k0) planes, as the (020), (040) and (060) Bragg lines have a large intensity. Rietveld refinement provided lattice constants of a = 0.3965 Å, b = 1.3861 Å, and c = 0.3697 Å.

Structure and Morphology
The carbon content of the Mo5O14/C composite synthesized by sol-gel method was determined using thermal gravimetry (TG) and the differential scanning colorimetric (DSC) experiments. Figure  2b,c show the TG curves of the as-prepared P-MoO3 and the Mo5O14/C composite, respectively, recorded in the range 25-1000 °C in air. The thermal behavior of P-MoO3 does not show obvious weight loss until the decomposition of MoO3 (melting point at ca. 795 °C) [27]. The situation for the Mo5O14/C composite is different. The TG curve shows a strong endothermic peak at ~500 °C with a weigh loss of 26% corresponding to the evolution of CO2 through reaction between carbon in the composite and the atmospheric oxygen. Thus, the composite contains 26% carbon coming from the pyrolysis of citric acid. Upon further heating, the small exothermic peak appearing at ca. 800 °C is attributed to the melting of Mo5O14 in the composite similarly to the feature observed for pure MoO3 crystal. Further elemental determination of carbon was performed using a CHNOS analyzer that gave the same percentage of carbon (25.8 ± 0.2%) in the composite.
The morphology of the Mo5O14/C composite was analyzed by scanning electron microscopy and compared with that of pure MoO3 synthesized from the same precursor ( Figure 3). While the SEM images (a,b) of P-MoO3 show uniformly distributed and well-developed packed platelets of the layered α-MoO3 structure, the SEM images (c,d) of the Mo5O14/C composite display a honeycomb carbon matrix with small pores, and irregular particles of molybdenum oxide are dispersed inside the matrix.  the organic compound C6H8O7 accompanied by the decomposition of (NH4)6Mo7O24 (partial desorption of H2O and NH3) into oxides. At this stage, a weight loss of almost 50% was recorded.  The XRD pattern of P-MoO3 ( Figure 1b) displays the typical pattern of a well-crystallized α-MoO3 phase that can be indexed in the orthorhombic Pbnm space group (JPCDS card 76-1003). The coarse-grained powder presents a preferred orientation of (0k0) planes, as the (020), (040) and (060) Bragg lines have a large intensity. Rietveld refinement provided lattice constants of a = 0.3965 Å, b = 1.3861 Å, and c = 0.3697 Å.
The carbon content of the Mo5O14/C composite synthesized by sol-gel method was determined using thermal gravimetry (TG) and the differential scanning colorimetric (DSC) experiments. Figure  2b,c show the TG curves of the as-prepared P-MoO3 and the Mo5O14/C composite, respectively, recorded in the range 25-1000 °C in air. The thermal behavior of P-MoO3 does not show obvious weight loss until the decomposition of MoO3 (melting point at ca. 795 °C) [27]. The situation for the Mo5O14/C composite is different. The TG curve shows a strong endothermic peak at ~500 °C with a weigh loss of 26% corresponding to the evolution of CO2 through reaction between carbon in the composite and the atmospheric oxygen. Thus, the composite contains 26% carbon coming from the pyrolysis of citric acid. Upon further heating, the small exothermic peak appearing at ca. 800 °C is attributed to the melting of Mo5O14 in the composite similarly to the feature observed for pure MoO3 crystal. Further elemental determination of carbon was performed using a CHNOS analyzer that gave the same percentage of carbon (25.8 ± 0.2%) in the composite.
The morphology of the Mo5O14/C composite was analyzed by scanning electron microscopy and compared with that of pure MoO3 synthesized from the same precursor ( Figure 3). While the SEM images (a,b) of P-MoO3 show uniformly distributed and well-developed packed platelets of the layered α-MoO3 structure, the SEM images (c,d) of the Mo5O14/C composite display a honeycomb carbon matrix with small pores, and irregular particles of molybdenum oxide are dispersed inside the matrix. The coarse-grained powder presents a preferred orientation of (0k0) planes, as the (020), (040) and (060) Bragg lines have a large intensity. Rietveld refinement provided lattice constants of a = 0.3965 Å, b = 1.3861 Å, and c = 0.3697 Å.
The carbon content of the Mo 5 O 14 /C composite synthesized by sol-gel method was determined using thermal gravimetry (TG) and the differential scanning colorimetric (DSC) experiments. Figure 2b,c show the TG curves of the as-prepared P-MoO 3 and the Mo 5 O 14 /C composite, respectively, recorded in the range 25-1000 • C in air. The thermal behavior of P-MoO 3 does not show obvious weight loss until the decomposition of MoO 3 (melting point at ca. 795 • C) [27]. The situation for the Mo5O14/C composite is different. The TG curve shows a strong endothermic peak at~500 • C with a weigh loss of 26% corresponding to the evolution of CO 2 through reaction between carbon in the composite and the atmospheric oxygen. Thus, the composite contains 26% carbon coming from the pyrolysis of citric acid. Upon further heating, the small exothermic peak appearing at ca. 800 • C is attributed to the melting of Mo 5 O 14 in the composite similarly to the feature observed for pure MoO 3 crystal. Further elemental determination of carbon was performed using a CHNOS analyzer that gave the same percentage of carbon (25.8 ± 0.2%) in the composite.
The morphology of the Mo 5 O 14 /C composite was analyzed by scanning electron microscopy and compared with that of pure MoO 3 synthesized from the same precursor ( Figure 3). While the SEM images (a,b) of P-MoO 3 show uniformly distributed and well-developed packed platelets of the layered Since XRD experiments are not efficient tools to determine the structure of ill-crystallized materials, we investigated the short-range local structure of Mo5O14/C composite using Raman spectroscopy. Figure 4 compares the Raman patterns of the pristine MoO3 (a) and the Mo5O14/C composite (b). The Raman spectrum of Mo5O14/C composite prepared at 500 °C in argon atmosphere can be considered as the sum of the Mo5O14 and carbon contribution. In the low-frequency region, the spectrum shows features of the molybdenum oxide phase, while the high-frequency range (greater than 1200 cm −1 ) displays the fingerprint of the carbon matrix. Note that the weak intensity of Mo5O14 Raman pattern is due to the presence of carbon environment absorbing the scattering light. The five low-frequency Raman bands located at 972, 780, 700, 430 and 225 cm −1 can be identified as the vibrational patterns of tetragonal Mo5O14. The Mo-O stretching mode appears at 972 cm −1 , the Mo-O-Mo bridging modes are recorded at 780, 700 and 430 cm −1 , while the deformation modes of the Mo-O bonds appear at ca. 225 cm −1 . This deformation mode was observed at 229 cm −1 in Mo4O11. It should be noted that this Raman spectrum definitely differs from those of either α-MoO3 (Figure 4a) or Mo4O11 phase [28,29]. For example, in the spectrum of Mo5O14, the broad peak at ~780 cm −1 replaces the sharp peak of α-MoO3 at 820 cm −1 , which reveals the presence of sub-stoichiometric MomO3m−1 crystals. Such a blue shift of the stretching mode is attributed to the modification of Mo-O bonds induced by the reduction of Mo 6+ to Mo 5+ next to the oxygen vacancies [30]. Nazri and Julien [31] showed that the intensities and positions of vibrational bands of MoO3nH2O are extremely sensitive to small changes of the Mo-O polyhedra. A loss of the translational symmetry is expected because of the oxygen vacancies present in the MomO3m−1 lattice. Thus, the highest stretching mode appears at 996 cm −1 in MoO3, while it is shifted to 985 cm −1 in Mo4O11. The Hardcastle-Wachs approach allows to correlate the Raman shifts with the Mo-O bond distances [32]. The stretching at 972 cm −1 observed in the Raman spectrum of Mo5O14 appears at the same frequency as that of MoO3nH2O with a Mo-O distance of 1.687 Å against 1.671 Å for anhydrous α-MoO3. The broad bands observed in the range 1200-1600 cm −1 correspond to the D-and G-bands of carbon. The D-band at 1354 cm −1 is associated with the disorder-allowed A1g zone-edge mode of graphite, whereas the G-band centered at 1573 cm −1 corresponds to the optically allowed E2g zone center mode of crystalline graphite. The broadening of the G-and D-bands are characteristics of localized in-plane sp 2 domains, and disordered graphitic-like carbon, respectively. The peak intensity ratio ID/IG of approximately 1.19 confirms the low degree of graphitization of the carbon matrix and the small size of the graphitic domains La of about 4.8 nm according to the empirical equation La (nm) = 4.4 ID/IG [33]. The limited graphitization is due to the low temperature used for the preparation of the composite (i.e., 500 °C), but seems enough to be beneficial for electron transport from/to the semiconducting Mo5O14 active material. Since XRD experiments are not efficient tools to determine the structure of ill-crystallized materials, we investigated the short-range local structure of Mo 5 O 14 /C composite using Raman spectroscopy.  [28,29]. For example, in the spectrum of Mo 5 O 14 , the broad peak at~780 cm −1 replaces the sharp peak of α-MoO 3 at 820 cm −1 , which reveals the presence of sub-stoichiometric Mo m O 3m−1 crystals. Such a blue shift of the stretching mode is attributed to the modification of Mo-O bonds induced by the reduction of Mo 6+ to Mo 5+ next to the oxygen vacancies [30]. Nazri and Julien [31] showed  Figure 5 shows the nitrogen adsorption-desorption isotherms and pore-size distributions (PSD) for both P-MoO3 and Mo5O14/C composite. The uptake of nitrogen adsorption of both samples increases gradually with the increase of relative pressure (Figure 5a). Note that the quantity of nitrogen adsorbed for the Mo5O14/C composite is much larger than that of P-MoO3 sample. The adsorption rises noticeably at low relative pressures of less than 0.5, indicating that micropores are present. Instead of reaching a plateau, the uptake of nitrogen further increases with relative pressure until P/P0 approaches unity, which suggests that a substantial amount of external surface area was created in Mo5O14/C sample. As shown in PSD curves (Figure 5b), the distribution of the pore diameters has peaks at 4.8 nm for Mo5O14/C composite. The Mo5O14/C composite exhibits a broader pore-size distribution with the pore diameter in the range 2−7 nm. It has been widely reported that physical and chemical activation of carbon materials help to open and widen the microporosity even with a conversion to mesopores [34,35]. The material porosity, i.e., pore size distribution and pore volume, has been quantitatively evaluated using the Barrett-Joyner-Halenda (BJH) method [36]. Results are listed in Table 1 and show that the BET surface area, the BJH total pore volume and the porosity are higher in Mo5O14/C than in P-MoO3. The increase of the specific surface area is attributed to the presence of 26 wt.% carbon. The higher pore size in the Mo5O14-type/C composite also favors better electrochemical performance of the electrode.    Figure 5 shows the nitrogen adsorption-desorption isotherms and pore-size distributions (PSD) for both P-MoO 3 and Mo 5 O 14 /C composite. The uptake of nitrogen adsorption of both samples increases gradually with the increase of relative pressure (Figure 5a). Note that the quantity of nitrogen adsorbed for the Mo 5 O 14 /C composite is much larger than that of P-MoO 3 sample. The adsorption rises noticeably at low relative pressures of less than 0.5, indicating that micropores are present. Instead of reaching a plateau, the uptake of nitrogen further increases with relative pressure until P/P 0 approaches unity, which suggests that a substantial amount of external surface area was created in Mo 5 O 14 /C sample. As shown in PSD curves (Figure 5b), the distribution of the pore diameters has peaks at 4.8 nm for Mo 5 O 14 /C composite. The Mo 5 O 14 /C composite exhibits a broader pore-size distribution with the pore diameter in the range 2−7 nm. It has been widely reported that physical and chemical activation of carbon materials help to open and widen the microporosity even with a conversion to mesopores [34,35]. The material porosity, i.e., pore size distribution and pore volume, has been quantitatively evaluated using the Barrett-Joyner-Halenda (BJH) method [36]. Results are listed in Table 1 and show that the BET surface area, the BJH total pore volume and the porosity are higher in Mo 5 O 14 /C than in P-MoO 3 . The increase of the specific surface area is attributed to the presence of 26 wt.% carbon. The higher pore size in the Mo 5 O 14 -type/C composite also favors better electrochemical performance of the electrode.  Figure 5 shows the nitrogen adsorption-desorption isotherms and pore-size distributions (PSD) for both P-MoO3 and Mo5O14/C composite. The uptake of nitrogen adsorption of both samples increases gradually with the increase of relative pressure (Figure 5a). Note that the quantity of nitrogen adsorbed for the Mo5O14/C composite is much larger than that of P-MoO3 sample. The adsorption rises noticeably at low relative pressures of less than 0.5, indicating that micropores are present. Instead of reaching a plateau, the uptake of nitrogen further increases with relative pressure until P/P0 approaches unity, which suggests that a substantial amount of external surface area was created in Mo5O14/C sample. As shown in PSD curves (Figure 5b), the distribution of the pore diameters has peaks at 4.8 nm for Mo5O14/C composite. The Mo5O14/C composite exhibits a broader pore-size distribution with the pore diameter in the range 2−7 nm. It has been widely reported that physical and chemical activation of carbon materials help to open and widen the microporosity even with a conversion to mesopores [34,35]. The material porosity, i.e., pore size distribution and pore volume, has been quantitatively evaluated using the Barrett-Joyner-Halenda (BJH) method [36]. Results are listed in Table 1 and show that the BET surface area, the BJH total pore volume and the porosity are higher in Mo5O14/C than in P-MoO3. The increase of the specific surface area is attributed to the presence of 26 wt.% carbon. The higher pore size in the Mo5O14-type/C composite also favors better electrochemical performance of the electrode.

Electrochemical Properties
The cyclic voltammetry (CV) behavior for pristine MoO 3 and Mo 5 O 14 /C composite electrodes performed in the potential range 0.6-3.5 V vs. Li + /Li at the scanning rate of 0.05 mV s −1 are displayed in Figure 6a,b, respectively. The choice of the low voltage cut-off of 0.6 V was intentional because of the formation of metallic Mo and 3Li 2 O, which takes place as a plateau at around 0.4-0.5 V upon further Li + insertion [37,38]. The CV curves of pristine MoO 3 exhibit well-defined redox peaks, due to the fact that this material is well-crystallized. In the first cycle, there are two strong cathodic peaks at ca. 2.7 and 2.3 V, with one reversible anodic peak around 2.6 V. The cathodic peak at 2.7 V is irreversible and disappeared after the first lithiation during subsequent cycles (Figure 6a). Such a behavior has been evidenced by preparing pre-lithiated Li x MoO 3 samples using the chemical insertion by LiI in the range 0.1 ≤ x ≤ 1.0. For x ≈ 0.49, the CV curve displays only one reversible redox couple that is an indication of the disappearance of the pure α-MoO 3 phase with similar patterns for subsequent cycles [39]. The mesoporous Mo 5 O 14 /C composite reveals a featureless voltammetry response over a broad potential range; this pattern arises from its amorphous-like structure, which leads to numerous insertion sites for Li + ions. After the first cycle, a high reversibility is observed from overlapping CV curves (Figure 6b.) Nanomaterials 2020, 10, x FOR PEER REVIEW 7 of 13

Electrochemical Properties
The cyclic voltammetry (CV) behavior for pristine MoO3 and Mo5O14/C composite electrodes performed in the potential range 0.6-3.5 V vs. Li + /Li at the scanning rate of 0.05 mV s −1 are displayed in Figure 6a,b, respectively. The choice of the low voltage cut-off of 0.6 V was intentional because of the formation of metallic Mo and 3Li2O, which takes place as a plateau at around 0.4-0.5 V upon further Li + insertion [37,38]. The CV curves of pristine MoO3 exhibit well-defined redox peaks, due to the fact that this material is well-crystallized. In the first cycle, there are two strong cathodic peaks at ca. 2.7 and 2.3 V, with one reversible anodic peak around 2.6 V. The cathodic peak at 2.7 V is irreversible and disappeared after the first lithiation during subsequent cycles (Figure 6a). Such a behavior has been evidenced by preparing pre-lithiated LixMoO3 samples using the chemical insertion by LiI in the range 0.1 ≤ x ≤ 1.0. For x ≈ 0.49, the CV curve displays only one reversible redox couple that is an indication of the disappearance of the pure α-MoO3 phase with similar patterns for subsequent cycles [39]. The mesoporous Mo5O14/C composite reveals a featureless voltammetry response over a broad potential range; this pattern arises from its amorphous-like structure, which leads to numerous insertion sites for Li + ions. After the first cycle, a high reversibility is observed from overlapping CV curves (Figure 6b.)  Figure 7a,b present the galvanostatic charge-discharge (GCD) profiles of the pristine MoO3 and Mo5O14/C composite electrodes, respectively, recorded at C/10 rate (70 mA g −1 ). In a fresh cell, pristine MoO3 and Mo5O14/C composite are electrodes in a "charged state" with all Mo ions with the 6+ oxidation state. Thus, the first process of lithiation occurs during discharge (reduction of Mo to 5+ and 4+ states) providing a capacity Q. During the subsequent charge Q', Mo is oxidized, but not all Li ions can be extracted from the lithiated phase, thus less electrons are transferred and Q'<Q. Definitely, the electrochemical behaviors of these electrodes are different. The difference between the shape of charge/discharge curves of pristine MoO3 and Mo5O14/C is mainly due to the difference in the structure and morphology of the active material. In contrast with the 2D structure of α-MoO3, Mo5O14 has a 3D structure with empty site in the tunnels inducing a different Gibbs energy that results in unlike voltage profile. In addition, the voltage profile of an amorphous material is a S-shape curve due to the random distribution of the available site in the framework.
In the first discharge process, the crystallized pristine MoO3 shows two typical plateaus ( Figure  7a), in accordance with the cathodic peaks in the CV curve (Figure 6a). The insertion of lithium proceeds in two steps at 2.75 and 2.30 V vs. Li + /Li within the capacity range of 0-50 and 100-250 mAh g −1 , respectively. The disappearance of the first plateau at the second cycle is related to an irreversible structural change, suggesting that part of Li + ions cannot be extracted during the charge process. This structural modification has been described by the phase transition from MoO3 to LixMoO3 (where x ≤  . In a fresh cell, pristine MoO 3 and Mo 5 O 14 /C composite are electrodes in a "charged state" with all Mo ions with the 6+ oxidation state. Thus, the first process of lithiation occurs during discharge (reduction of Mo to 5+ and 4+ states) providing a capacity Q. During the subsequent charge Q', Mo is oxidized, but not all Li ions can be extracted from the lithiated phase, thus less electrons are transferred and Q'<Q. Definitely, the electrochemical behaviors of these electrodes are different. The difference between the shape of charge/discharge curves of pristine MoO 3 and Mo 5 O 14 /C is mainly due to the difference in the structure and morphology of the active material. In contrast with the 2D structure of α-MoO 3 , Mo 5 O 14 has a 3D structure with empty site in the tunnels inducing a different Gibbs energy that results in unlike voltage profile. In addition, the voltage profile of an amorphous material is a S-shape curve due to the random distribution of the available site in the framework. During next cycles, the discharge capacities slightly decrease to reach 325 mAh g −1 at the 50th cycle. The cyclability of pristine MoO3 and Mo5O14/C composite electrodes was tested at C/10 rate in the cell voltage range of 0.6-3.5 V as shown in Figure 8. The Mo5O14/C composite shows an almost stable electrochemical behavior up to 85 cycles with a rate of capacity fade of 0.8% per cycle. Results of the rate capability are presented in Figure 9. The high performance of the Mo5O14/C composite is clearly revealed with a specific discharge capacity of 155 mAh g −1 at 10C rate (7 A g −1 ).   In the first discharge process, the crystallized pristine MoO 3 shows two typical plateaus (Figure 7a), in accordance with the cathodic peaks in the CV curve (Figure 6a). The insertion of lithium proceeds in two steps at 2.75 and 2.30 V vs. Li + /Li within the capacity range of 0-50 and 100-250 mAh g −1 , respectively. The disappearance of the first plateau at the second cycle is related to an irreversible structural change, suggesting that part of Li + ions cannot be extracted during the charge process. This structural modification has been described by the phase transition from MoO 3 to Li x MoO 3 (where x ≤ 0.25), which induces a pronounced expansion of the interlayer spacing (from 0.69 to 1.175 nm) [39,40]. Similar behavior was previously observed for lithium intercalation in the layered V 2 O 5 host [41]. The specific capacity of the P-MoO 3 electrode decreases continuously over subsequent discharge-charge cycles from 330 mAh g −1 (1st cycle) to 136 mAh g −1 (25th cycle). This large capacity loss (2% per cycle) reveals the poor electrochemical stability of the α-MoO 3 phase prepared by citrate-gel method. In contrast, the mesoporous Mo 5 O 14 /carbon composite exhibits better electrochemical performance. GCD curves are featureless (without plateau) showing a steady decay of the cell voltage upon Li insertion. Such discharge-charge profiles are characteristic of an amorphous phase, which implies a wide variation in energy for the Li sites available for insertion [42,43]. For the first discharge, this electrode delivers a specific capacity of 703 mAh g −1 , decreasing to 388 mAh g −1 at the 2nd cycle. During next cycles, the discharge capacities slightly decrease to reach 325 mAh g −1 at the 50th cycle. The cyclability of pristine MoO 3 and Mo 5 O 14 /C composite electrodes was tested at C/10 rate in the cell voltage range of 0.6-3.5 V as shown in Figure 8. The Mo 5 O 14 /C composite shows an almost stable electrochemical behavior up to 85 cycles with a rate of capacity fade of 0.8% per cycle. Results of the rate capability are presented in Figure 9. The high performance of the Mo 5 O 14 /C composite is clearly revealed with a specific discharge capacity of 155 mAh g −1 at 10C rate (7 A g −1 ). Nanomaterials 2020, 10, x FOR PEER REVIEW 8 of 13 cycles from 330 mAh g −1 (1st cycle) to 136 mAh g −1 (25th cycle). This large capacity loss (2% per cycle) reveals the poor electrochemical stability of the α-MoO3 phase prepared by citrate-gel method. In contrast, the mesoporous Mo5O14/carbon composite exhibits better electrochemical performance. GCD curves are featureless (without plateau) showing a steady decay of the cell voltage upon Li insertion. Such discharge-charge profiles are characteristic of an amorphous phase, which implies a wide variation in energy for the Li sites available for insertion [42,43]. For the first discharge, this electrode delivers a specific capacity of 703 mAh g −1 , decreasing to 388 mAh g −1 at the 2nd cycle. During next cycles, the discharge capacities slightly decrease to reach 325 mAh g −1 at the 50th cycle. The cyclability of pristine MoO3 and Mo5O14/C composite electrodes was tested at C/10 rate in the cell voltage range of 0.6-3.5 V as shown in Figure 8. The Mo5O14/C composite shows an almost stable electrochemical behavior up to 85 cycles with a rate of capacity fade of 0.8% per cycle. Results of the rate capability are presented in Figure 9. The high performance of the Mo5O14/C composite is clearly revealed with a specific discharge capacity of 155 mAh g −1 at 10C rate (7 A g −1 ).

Discussion
The structural and electrochemical properties of a composite formed of oxygen-deficient Mo m O 3m−1 and carbon have been examined for the first time. Mo m O 3m−1 compounds are particularly interesting, mainly due to the structure with large tunnels accessible for Li + ions. In addition, they have a high electrical resistivity, i.e., ρ = 1 Ω cm for Mo 8 O 23 and ρ = 78 Ω cm −1 for Mo 18 O 52 that are semiconductors, while η-Mo 4 O 11 is metallic with ρ = 1.7 10 −4 Ω cm [44]. In the work presented here, the reduction of the layered structure MoO 3 obtained by the decomposition of citric acid as chelating agent and ammonium molybdate tetrahydrate as raw material has been performed in closed atmosphere. The as-prepared mesoporous Mo 5 O 14 /C composite was successfully obtained by a moderate annealing temperature of 500 • C. This new material is composed of a honeycomb carbon matrix, in which amorphous Mo 5 O 14 nanoparticles are introduced. It is assumed that the Mo 6+ ions were reduced by the presence of CO, CO 2 and NH 3 gases coming from the chelate and raw material forming the non-stoichiometric (oxygen-deficient) Mo 5 O 14 /C compound. This tunnel-like structure accommodates oxygen deficiency via oxygen vacancies and crystallographic shear planes. The formation of an intermediate compound Mo m O 3m−1 between Mo VI O 3 and Mo IV O 2 during the reduction of MoO 3 has been reported many times [15,45,46], but here, despite the previous claims, we obtained a stable phase with the O/Mo ratio of 2.8.
Also, this work illustrates the synergy between a molybdenum oxide and carbon that exhibits excellent electrochemical performance of the composite. Also, it reveals the importance of electronic conductivity and the mesoporous nature for substantial lithium insertion in tunnel-like molybdenum-oxide compounds. In contrast to the large capacity fade due to the interlayer spacing expansion in crystalline α-MoO 3 , the excellent electrochemical behavior of the Mo 5 O 14 /C composite electrode is due to the formation of large mesopores. The unique cycling performance of the composite is attributed to the presence of carbon, which plays a dual role in enwrapping the tunnel-like Mo 5 O 14 nanoparticles and accommodating the volume change. Graphitized carbon provides a large contact surface and acts as an excellent conductive agent facilitating the charge-discharge process. The porosity formed by carbon facilitates the transport of charge carriers and the delocalized insertion site in the amorphous matrix enhances the kinetics of Li + ions. According to its high reversible capacity and good cycling stability, the Mo 5 O 14 /C composite is a promising alternative electrode material for application in lithium-ion batteries.
Several studies have demonstrated the improved cycling performance of oxygen-deficient Mo m O 3m−1 compounds compared with stoichiometric MoO 3 . In an early work, Besenhard and Schöllhorn [4] have shown that discharge-charge reactions of Mo 18 14 can deliver a specific capacity of 310 mA g −1 when discharged at 0.5 mA cm −2 in the voltage range 2.5-1.0 V [5]. Christian and co-workers [9] reported that for the oxides Mo 4 14 ) displays a stepped behavior with a voltage plateau at ca. 2.2 V followed by a potential decline for x > 0.7Li/Mo and a total Li uptake of 1.45Li/Mo for a cut-off voltage of 1.0 V. For all Mo ions reduced to Mo 4+ , the expected lithium uptake would be 8/5(1.6) Li/Mo; a value close to 258 mAh g −1 measured by electrochemical titration [11]. Jung et al. [14] showed that during the first discharge, MoO 2.895 prepared by ball milling delivered a specific of ca. 600 mAh g −1 that decreased to 420 mAh g −1 (2nd cycle) in the voltage range 3.5-0.5 V. For comparison, electrochemical properties of composites formed by a Mo oxide and a carbonaceous material are listed in Table 2. According to the GCD profile presented in Figure 7b, the initial discharge capacity and the initial Coulombic efficiency of the Mo 5 O 14 /C composite are 703 mAh g −1 and 55%, respectively. In the same voltage range, a composite electrode made of MoO 3 nanocrystals distributed in an amorphous carbon matrix delivered a specific capacity of approximately 630 mAh g −1 and an initial Coulombic efficiency of 75% [47]. An initial discharge capacity of approximately 550 mAh g −1 with a Coulombic efficiency of 69% was reported for crumpled graphene-MoO 3 composite cycled at a constant current density of 2 A g −1 [22]. Furthermore, the higher performance of the Mo 5 O 14 -type/C composite electrode is attributed to the accommodated volume expansion. Christian et al. [9] have shown that the unit cell of the fully lithiated Mo 5 O 14 crystal expands in the b direction and contracts in both the a and c directions upon insertion of lithium. Also, we believe that the presence of 26 wt.% carbon homogeneously distributed and associated with the intrinsic properties of the active particles minimizes the volume expansion.

Conclusions
Pure α-MoO 3 and mesoporous Mo 5 O 14 -type/C composite were successfully prepared by a citrate-gel method from ammonium molybdate precursor. Crystalline orthorhombic α-MoO 3 is observed for xerogel calcined in air, while the Mo 5 O 14 /C composite is prepared under argon atmosphere. The latter one is a poorly crystalline material that is composed of Mo 5 O 14 nanograins incorporated into a carbon matrix as noticed from X-ray diffraction and Raman spectroscopy. The carbon content of approximately 26 wt.% was estimated from TGA and elemental analysis. Analysis of the vibrational features of D-and G-Raman bands of the carbon reveals that it is moderately graphitized. BET measurements show that the Mo 5 O 14 /C composite exhibits a surface area of 7.1 m 2 g −1 , average pore size of approximately 4 nm and total pore volume of 0.07 cm 3 g −1 , which are favorable for enhanced electrochemical insertion of Li + ions into the mesoporous lattice.
The Mo 5 O 14 /C composite appears to be electrochemically stable showing a specific capacity of 325 mAh g −1 and a capacity fade of 0.8% after 50 cycles at 0.1C. Such a good performance is attributed to the mesoporous nature of the electrode material and the presence of carbon, which suppresses the aggregation of molybdenum-oxide particles and thus increases their structural stability during cycling, while keeping the particles electrically connected.