Formation mechanisms of Fe3−xSnxO4 by a chemical vapor transport (CVT) process

Our former study reported that Fe-Sn spinel (Fe3−xSnxO4) was easily formed when SnO2 and Fe3O4 were roasted under CO-CO2 atmosphere at 900–1100 °C. However, the formation procedure is still unclear and there is a lack of theoretical research on the formation mechanism of the Fe-Sn spinel. In this work, the reaction mechanisms between SnO2 and Fe3O4 under CO-CO2 atmosphere were determined using XRD, VSM, SEM-EDS, XPS, etc. The results indicated that the formation of Fe3−xSnxO4 could be divided into four steps: reduction of SnO2 to solid phase SnO, volatilization of gaseous SnO, adsorption of gaseous SnO on the surface of Fe3O4, and redox reaction between SnO and Fe3O4. During the roasting process, part of Fe3+ in Fe3O4 was reduced to Fe2+ by gaseous SnO, and meanwhile Sn2+ was oxidized to Sn4+ and entered into Fe3−xSnxO4. The reaction between SnO2 and Fe3O4 could be summarized as Fe3O4 + xSnO(g) → Fe3−xSnxO4 (x = 0–1.0).


Results
Determination of the phase composition of the roasted samples with Fe 3 O 4 and SnO 2 . Mixed samples (natural magnetite and cassiterite powders) were roasted at 950 °C under an atmosphere of 15 vol.% CO/ (CO + CO 2 ) for different time, and the roasted samples were then prepared for XRD, VSM, XPS and SEM-EDS analyses. Figure 1 demonstrates the XRD patterns of the samples roasted at 950 °C for the time varying from 15 min to 600 min, and the sample roasted for 120 min in 100 vol.% N 2 atmosphere was also measured. It can be seen from Fig. 1 that the main phase constitutions of the samples were magnetite, cassiterite and Fe-Sn spinel (Fe 2.6 Sn 0.4 O 4 ) under 15 vol.% CO atmosphere. However, the diffraction peaks of Fe-Sn spinel remarkably enhanced as the roasting time prolonged, which indicated the gradual conversion of magnetite into Fe-Sn spinel. Interestingly, no diffraction peak of Fe-Sn spinel was found in the XRD pattern of the samples roasted under 100 vol.% N 2 atmosphere, revealing that there was no reaction happening between Fe 3 O 4 and SnO 2 at 950 °C. Based on the above results, it is inferred that the CO-CO 2 atmosphere plays an important role in the formation of Fe-Sn spinel.
In order to investigate the transformation process of Fe-Sn spinel, the magnetization hysteresis loops of the above-mentioned samples (No. 1#~4#) were studied by VSM at room temperature, and the results are displayed in Fig. 2. The results in Fig. 2 showed that the saturation magnetization (M S ) of Sample 1# was about 47.6 emu/g, because the magnetite was stable when roasted under N 2 atmosphere and there was no Fe-Sn spinel formed during the roasting. The Ms values of the samples (2#~4#) were much lower than that of Sample 1#. As the roasting time increased from 15 min to 120 min, the Ms value decreased obviously from 36.2 emu/g to 7.3 emu/g. In addition, it was observed from Fig. 2 that the coercivity field (the value of Hc when Ms is equal to zero) also decreased markedly with the increase of roasting time. Previous studies showed that Sn 4+ could replace the Fe 3+ in the magnetite to form Fe-Sn spinel, which resulted in a smaller hysteresis as well as the coercivity field 10,11 . As reported, Sn 4+ could enter into the octahedral sublattice of magnetite, and then Fe-Sn spinel was easily formed under CO-CO 2 atmosphere, which led to the decrease of saturation magnetization with the roasting time increasing.  To analyze the element distribution and composition of the Fe-Sn spinel formed in the roasted sample (Sample 3#), the backscattered micrographs of Fe-Sn spinel and the corresponding elements' area distribution images by SEM-EDS are shown in Fig. 3. As seen from Fig. 3a, the major phases in the sample were magnetite (Spot B), cassiterite (Spot D) and Fe-Sn spinel (Spot A and C). Moreover, the Fe/Sn atomic ratio of Spot A and Spot C was similar to the value of 2.6: 0.4, which was coincident with the result presented in Fig. 1. In addition, it was amazing to find that the Fe-Sn spinel displayed as a thin layer and enwrapped the magnetite compactly. Based on the results in Fig. 3b and e, the corresponding Sn element's distribution indicated that Fe-Sn spinel was formed on the outside surface of magnetite particle. Obvious elemental gradient of Sn from the outside surface to the inner was observed in Fig. 3b, and there was almost no Sn element existing in the center part of the magnetite particle. However, Fe element was not found on the surface of cassiterite as shown in Fig. 3i. Therefore, there existed the mass transfer of Sn from SnO 2 to Fe 3 O 4 , and the formation mechanism would be further researched.
X-ray photoelectron spectroscopy (XPS) was then applied to check the chemical state of the samples' surfaces. The Fe 3p, Fe 2p and Sn 3d photoelectron spectra of the Raw material (magnetite and cassiterite powders were blended as mass ratio of 4:1) and Sample 3# are shown in Fig. 4. Based on the reported XPS studies of Fe 3p and Fe 2p, the photoelectron peaks of Fe are always associated with satellite peaks and background noise, which are complicated to distinguish definitely 30,31 . As shown in Fig. 4a and b, the binding energy of Fe 3p and Fe 2p 3/2 in Sample 3# shifted obviously from 55.79 eV to 55.29 eV and 711.09 eV to 710.69 eV, respectively. The decrease of the Fe binding energy was attributed to the replacement of Fe 3+ by Sn 4+ in Fe 3 O 4 , and then the Fe 3+ in Fe 3 O 4 was partially converted into Fe 2+ for the charge balance [8][9][10][11] . As observed from Fig. 4c, the Sn 3d photoelectron peak of the Raw Material was well matched with the peaks of pure SnO 2 in the previous literatures 32,33 . However, the XPS photoelectron peak of Sn 3d in Fig. 4c can be resolved into Sn 2+ and Sn 4+ . And both of Sn 3d 5/2 and Sn 3d 3/2 clearly showed two groups of Sn chemical bonding energies of 486.6 eV and 495.0 eV for Sn 4+ , and 494.3 eV and 485.9 eV for Sn 2+ 32,33 . Our previous studies on the reduction roasting of SnO 2 have proved that there is no SnO existing in the roasted samples 14,20,21 . Therefore, the resolved peaks of Sn 2+ (Fig. 4c)  2) was carried out rapidly and no SnO (s) was found in the roasted samples.
In this section, the reaction between Fe 3 O 4 and gaseous SnO was investigated and the schematic diagram of the experiment was shown in Fig. 5. A platinum wire screen was used to separate the cassiterite and magnetite particles, and then the samples were placed into an electrically-heated vertical-tube furnace and roasted at 950 °C for 60 min under 15 vol.% CO atmosphere. In this system, the solid-solid reactions between Fe 3 O 4 and SnO 2 were impossible to proceed, so that the effect of gaseous SnO on the formation of Fe 3−x Sn x O 4 could be investigated. The SEM-EDS analyses of the roasted magnetite particles are shown in Fig. 6. It was observed from Fig. 6 that Fe-Sn spinel layer with a thickness of about 5 μ m was formed at the outer sphere of the magnetite particles. The microstructure of the roasted sample in Fig. 6 was similar to that in Fig. 3. The corresponding elemental distributions of Sn and Fe indicated that the reactions between Fe 3 O 4 and gaseous SnO took place as a typically unreacted core model, so an obvious product layer was formed outside the magnetite particles. Occasionally, a crack throughout the magnetite particle was found in Fig. 6, and the enrichment of Sn element propagated along with the crack. The results further confirmed our inference that gaseous SnO was the vital medium for the mass transfer of Sn during the formation of Fe-Sn spinel. Gaseous SnO was a volatile substance, which played an important role in the CVT process.  Table 1, and the ∆ G θ -T equations are also listed in Table 1 and Fig. 7.

Formation mechanisms of Fe 3−x Sn x O 4 . The above-mentioned results indicated that the Fe-Sn spinel was formed via the reactions between gaseous SnO and Fe 3 O 4 , and the reaction could be summarized as
The standard Gibbs free energy (∆ G θ ) change of the related reactions was calculated as follow: where R is the ideal gas constant (8.3144 J/mol·K), T is the temperature in kelvin (K), and K θ is the standard equilibrium constant. In the reactions between gaseous SnO and FeOx, K θ is equal to the reciprocal of the standard vapor pressure of gaseous SnO. Then, gas-phase equilibrium diagram of FeOx under different SnO partial pressure was calculated and plotted in Fig. 8 The schematic diagram of the formation process of Fe 3−x Sn x O 4 by a CVT process is summarized in Fig. 9. Under the conditions of 15 vol.% CO atmosphere and roasting temperature of 950 °C, the reaction procedure between SnO 2 and Fe 3 O 4 could be described as follows: (a) SnO 2 is reduced to solid phase SnO while Fe 3 O 4 is stable under this condition; (b) SnO is volatilized as gaseous phase, and this process is much fast because no SnO (s) is observed in the roasted samples 14,20,21  During this CVT process, it was found that formation of gaseous SnO was the critical step, which had obvious effect on the mass transfer of Sn and the redox reaction between SnO and Fe 3 O 4 .

Eq.
Reactions ∆G θ -T (KJ/mol)    Method Natural magnetite and cassiterite powders used in this study were the same as those given in our previous study 14 . The theoretical Fe 3 O 4 and SnO 2 contents of the samples were 98.7 wt.% and 98.5 wt.%, respectively. The purity of gases (CO, CO 2 and N 2 ) used in the tests was higher than 99.99 vol.%. All the roasting tests were conducted in a vertical-tube furnace. The natural magnetite and cassiterite powders were first blended at mass ratio of 4:1. Then, the mixed sample was put into a corundum crucible and roasted in the furnace. The CO/(CO + CO 2 ) content was fixed at 15 vol.% and the roasting temperature was kept at 950 °C. The CO content refers to the CO volume concentration in the CO-CO 2 mixed gas (i.e., CO/(CO + CO 2 )). After roasted for different time, the samples were taken out and quenched into liquid nitrogen rapidly. Finally, the cooled samples were used for analysis.