The influence of an external magnetic field on the dynamics of magnetite reduction with hydrogen

The kinetics of hydrogen reduction of magnetite was investigated in different magnetic fields. The magnetic moment measurements in situ were used for the control of the reaction. A strong difference in the magnetic properties of the reaction results was obtained for applied strong and weak magnetic fields. The X-ray diffraction and Mössbauer spectra of the reduced samples confirmed their different composition. The mechanism of the magnetic field effect is discussed.


Introduction
Any chemical process in which carbon can be replaced by molecular hydrogen H 2 as a reducing agent is potentially important for solving the problem of global CO 2 emissions. For example, in the metallurgical industry, 1.8 tons of CO 2 are accounted for each ton of produced steel, and this gas comprises 8% of global CO 2 emissions. 1 The use of hydrogen for the direct reduction of iron ore makes it possible to obtain steel with a low carbon content, that signicantly improves its quality. 2 At atmospheric pressure and T > 450 C, the reduction process includes three stages (hematite-magnetite-wustiteiron), while at lower temperatures wustite (FeO) does not form. 3 The experimentally it is conrmed the presence of the thermodynamically unstable FeO oxide as an intermediate recovery product is relatively rarely mentioned in the literature. The wustite formation was observed during the reduction of magnetite at temperatures above 500 C 4 Metastable FeO is formed in iron-supported catalysts due to the stabilizing action of carriers (MgO, SiO 2 , Al 2 O 3 ). The carrier surface stabilizes the wustite formation due to the strong oxide-oxide interaction. 5 The formation of FeO at T > 570 C as an intermediate in the reduction of magnetite with hydrogen, regardless of kinetic or diffusion limitations, was conrmed by the authors of. 6 In the temperature range 350-570 C, instead of direct reduction of magnetite to iron, a two-stage process on the surface of magnetite is possible, at least on an atomic scale. It is assumed that in the reversible disproportionation reaction, 4FeO ¼ Fe 3 O 4 + Fe, FeO is formed at the interface between Fe 3 O 4 and Fe. 6 The effects of the magnetic eld on the reduction reactions of metal oxides were investigated in a number of works 7,8 mainly devoted to the reduction reaction of hematite. However, the reliability of the obtained experimental data and the lack of explanation of the nature of the effect should be noted.
The hypothesis explaining the inuence of the magnetic eld on topochemical reactions was put forward by Bulgarian researchers -G. P. Visokov and D. G Ivanov 9 and is based on the assumption of a change in the magnetic moments of the starting materials and products during the chemical reaction. The positive effect of applying magnetic elds on the heat conduction, reaction kinetics, and hydrogenation time of a lanthanum nickel bed was presented in. 10 A large amount of experimental data was obtained by a group of researchers led by Rowe. 11,12 They investigated the effect of the external magnetic eld ( 4200 Oe) on the reduction reactions of iron, cobalt and nickel oxides. The authors used the  thermogravimetric method. In a number of studies, saturation magnetization was additionally measured, from which the composition of the reaction mixture was calculated. The difference in the values obtained by those two methods, on average, did not exceed several percent. The authors did not nd an explanation of the observed phenomenon nature. Subsequently, the results of these studies were called into question. 13 It should be noted that in none of the cited works the method of direct control of the magnetization of the reaction mixture during the reaction was not used. The inuence of an external magnetic eld on the kinetics of magnetite reduction has not been studied. In the present work, an attempt was made for the rst time to study the effect of an external magnetic eld on the kinetics of hydrogen reduction of magnetite.

Experimental section
Magnetite (Fe 3 O 4 ) nanopowder, 50-100 nm particle size, 97% trace metals basis Sigma-Aldrich was used as the object of study. The experiment was carried out on a vibration sample magnetometer in which a owing quartz microreactor served as a magnetic sample. 14 The magnetic moment calibration was made with a pure cobalt bulk sample of 10 mg. The mass of the magnetite sample was 10 mg in all experiments. Before kinetics studies, magnetite samples were calcined with Ar ow 10 cm 3 s À1 at linear heating of 10 C min À1 to the temperature of 600 C in the external magnetic eld of 60 Oe. For further research, hydrogen (99.993%) was used without preliminary purication. Isothermal experiments were carried out in the temperature range 340-450 C. The external magnetic eld was varied in the range from 60 Oe to 5000 Oe with the external electromagnet. The saturation magnetization was measured at the magnetic eld from of 7000 to 8000 Oe by extrapolation to the zero eld.
Diffraction patterns were recorded on a PANalytical Empyrean diffractometer in Bragg-Brentano geometry (mode -2 scans, 40 mA, 40 kV) in increments of 0.026 degrees, in the range of angles from 5 to 100 degrees using a Pixel3D detector. The anode material is Cu. Diffraction patterns were analyzed using HighScore Plus PANalytical soware with full-prole analysis. Magnetite samples for research were prepared according to the method described above. Aer cooling in hydrogen, the samples were passivated in a stream of technical Ar, and then transferred to a diffractometer.
Mössbauer spectroscopy measurements were carried out on a standard spectrometer with constant acceleration in the transmission geometry with a 57 Co source in Rh. The Mössbauer spectra were decomposed into components using the least squares method using the program "UnivemMS". In the eld of 5000 Oe at the temperature of 310 C, an increase in magnetization is observed which at T ¼ 400 C is replaced by the magnetization decrease (Fig. 1a). In contrast, in the external eld of 60 Oe (Fig. 1b), a nonmonotonic decrease in the magnetization is observed with temperature increasing. The Fig. 2 The dependence of magnetization on temperature in the heating-cooling mode on the value of the external field 5000 Oe (a) and 60 Oe (b) in the hydrogen flow. Fig. 3 The dependence of the relative magnetization on the recovery time at 350 C in the different magnetic field. The curves were normalized to the initial magnetic moment. saturation magnetization aer cooling the sample reduced in the 5000 Oe eld showed that magnetite was completely reduced to iron. At the same time, a similar procedure performed with the sample of magnetite reduced in the 60 Oe eld indicates a partial reduction of magnetite. When the sample reducing in H 2 was heated to the temperature of 350 C and then cooled, the magnetization aer cooling was greater (Fig. 2a) or less (Fig. 2b) than the initial magnetization before. The reduction process depends on the applied external magnetic eld.

Results and discussion
The relative increase in the magnetization aer cooling of the sample restored in the eld of 5000 Oe (Fig. 2a) indicates the appearance of a phase in the system with a higher specic magnetization than the initial magnetite. The decrease in the magnetization with respect to the initial one (Fig. 2b) indicates the transition of part of magnetite to the weak magnetic phase. In Fig. 3 the dependences of the relative magnetization (J/J 0 where J 0 is the initial value of magnetization) on time of isothermal reduction at T ¼ 350 C are shown in the different magnetic elds. From the data presented it follows that in the magnetic eld being stronger than 500 Oe the magnetization increase is observed with time.
To analyze the composition of the reduced sample the X-ray diffraction was used. In Fig. 4 the X-ray diffraction patterns of magnetite samples reduced in H 2 for 60 minutes are presented for magnetic elds of 60 Oe and 5000 Oe at the temperature of 450 C. Aer reduction, the samples were passivated in the technical Ar stream 10 cm 3 s À1 at the room temperature during 15 min. This procedure allowed us to avoid oxidation of iron nanoparticles in the process of obtaining diffraction patterns.
From the data presented in Fig. 4, it follows that the iron concentration in the sample reduced in the magnetic eld of 5000 Oe is completely composed of iron. Quantitative phase analysis of the diffraction pattern of the sample reduced in the magnetic eld of 60 Oe consists of 70% magnetite and only 30% of iron.
Similarly prepared reduced samples were studied with Mössbauer spectroscopy: the Mössbauer spectra of the samples reduced in the magnetic eld of 60 Oe and 5000 Oe are presented in Fig. 5.
The common component of two spectra is a sextet with hyperne magnetic eld about 330 kOe. It corresponds to the metallic a-iron. 15 The spectrum of the sample reduced in the eld of 60 Oe is shown in Fig. 5a where two more sextets are visible, the parameters of which (the hyperne magnetic elds 4884 and 4569 kOe) characterize the magnetite 16 remaining due to incomplete reduction. Besides there is a small doublet component (2%) corresponding to wustite. 17 It follows from the Mössbauer data that 58% of magnetite is reduced in low eld of 60 Oe to a-iron (see Table 1).
The spectrum of the sample obtained in the eld of 5000 Oe is shown in Fig. 5b. It contains the main component of metallic a-iron with 84% intensity, and the relaxation component with The sample was prepared in the magnetic field 5000 Oe. a-Fe H hf ¼ 330 kOe P ¼ 84%.
very large line width, the appearance of which is associated with ultra-small superparamagnetic iron particles (less than 10 nm) and with traces of magnetite. Thus, in a magnetic eld of 5000 Oe, more than 84% of magnetite is reduced to a-Fe.
From the above presented X-ray diffraction and Mössbauer spectroscopy data, it follows that the reduction of magnetite in the magnetic eld of 60 Oe and in the eld of 5000 Oe leads to signicantly different results. In our opinion, the decrease in magnetization upon reduction in a 60 Oe eld is due to the formation of the antiferromagnetic phase of wustite FeO as a result of the topochemical process The non-monotonic character of the decrease in magnetization (Fig. 1b) in the temperature range from 300 C to 400 C indicates the multi-stage recovery process and the existence of several non-stoichiometric FeO 1Àx phases (where x < 1) differing in total magnetic moments. One of the reasons for the nonmonotonic change in magnetization can be the disproportionation reaction of wustite: The detailed mechanism of the process, apparently, includes the interaction of surface oxygen (O) s with hydrogen H 2 followed by the formation of an anionic vacancy (V) s A fundamentally different situation is observed in the eld of 5000 Oe (Fig. 1a). In this case, at T $ 310 C, a monotonic increase in magnetization occurs up to 400 C. At 400 C, the increase in magnetization is replaced by a decrease, which indicates the completion of the recovery process, and a further change in the magnetization is due to the temperature dependence of the iron magnetization near the Curie temperature. Measurement of the saturation magnetization of the sample aer cooling indicates a complete reduction of magnetite to iron. The increase in magnetization at T $ 310 C indicates the predominant role of a one-stage process Thus, an external magnetic eld changes the mechanism of the recovery process and already in a eld exceeding 1000 Oe a noticeable contribution of reaction (3) and a decrease in the contribution of reaction (1) are observed. In all cases, the process includes a nucleation stage. Nucleation, in turn, suggests the presence of active centers on the surface. The comprehensive theoretical model and also experimental study describing the action of an external magnetic eld on a chemical process in a heterogeneous reaction on solid has not been clearly reported. We suggest that an external magnetic eld may affect the concentration of nucleation centers. The work 18 shows that spin-structure can strongly affects orbital occupation of the surface and especially charge-transfer between the solid surface and the adsorbate.

Conclusion
The magnetic moment sensitive technique was used to investigate the iron oxides reduction by hydrogen in the magnetic eld. It is shown that H 2 adsorption and dissociation are modied by changes in spin-structure. Thus, the external magnetic eld can affect the rate of a topochemical reaction.

Author contributions
PCpurpose of investigation and chemical reactions, data analysis, writing the manuscript, discussion the text; NK -X-ray diffraction, data analysis and corresponding description, discussion the text; VA -Mössbauer measurements and analysis, corresponding description, discussion the text; YP -Mössbauer data analysis; AN -X-ray diffraction and Mössbauer data analysis, discussion the text; NPmagnetic properties analysis, manuscript preparation, discussion the text.

Conflicts of interest
There are no conicts to declare.   ). Full line width at half maximum (G). Relative areas of spectral components (A) represent relative contents of the corresponding Fe forms assuming a common recoilless fraction for all forms in a sample.