A. EG/Fe₂O₃ samples composition

According to Ref. Reference Lutfullin, Shornikova, Pokholok, Sorokina and Avdeev9, iron oxide particles are randomly distributed on the EG surface and mostly concentrated on defects of worm-like particles. The size of iron oxide particles varies within a wide range from tens of nanometers to units of micrometers. According to XRD patterns, the phase composition of all samples is represented by graphite phase padding, and in the case of EL-900 and PRO-900 samples, by peaks corresponding to the α-Fe₂O₃ phase. The iron content in the samples depends on the preparation routs: ω(Fe) = 12.0% for EL-350, ω(Fe) = 14.0% for EL-900, and ω(Fe) = 9.7% for PRO-900.

To determine the valent state and localization of iron in EG/Fe₂O₃ samples, Mossbauer spectroscopy was applied (Fig. 1). The corresponding hyperfine parameters are listed in Table I. The Mossbauer spectrum of EL-350 sample is composed of the major doublet with spectral contribution S = 80% and sets of minor sextets. According to its IS of 0.36 mm/s and quadrupole splitting of 0.71 mm/s, the former is to be attributed to ferrihydrite.^{Reference Long14} Ferrihydrite is an amorphous hydrated iron oxide Fe₂O₃·nH₂O, the chemical formula of which can be written as Fe₅HO₈·4H₂O or FeOOH·2H₂O.^{Reference Long14} The minor sextets (with the spectral contributions of about 20%), according to their hyperfine parameters, correspond to α-Fe₂O₃ and Fe₃O₄.^{Reference Long14}

FIG. 1.

Mossbauer spectra of EG/Fe₂O₃ samples: (a) before reduction and (b) after reduction by hydrogen.

TABLE I.

The parameters of Mossbauer spectra of EG/Fe₂O₃ samples.

IS—Isomer shift related to α-Fe, QS—quadrupole splitting, H—magnetic field, S—spectral contribution.

By contrast, the major components in the Mossbauer spectra of the other samples (EL-900 and PRO-900) were α-Fe₂O₃ sextets. There were also minor sextets and doublets corresponding to Fe₃O₄ and Fe³⁺, respectively. Unlike the above samples, the Fe³⁺ doublets are presumably related to superparamagnetic Fe₂O₃ particles with a particle size <10 nm,^{Reference Zboril15,Reference Kundu, Mukherjee and Chakravorty16} because these samples were obtained at high temperatures.

Besides iron-containing particles, the EG/Fe₂O₃ samples contain oxygen functional groups. According to FTIR analysis, carbonyl, hydroxyl, and nitro groups were observed in the EL-350 sample, whereas the EL-900 sample bears carbonyl and hydroxyl groups and the PRO-900 sample, only hydroxyl groups (Fig. 2). As long as expanded graphite is a black solid, the quantitative determination of amounts of each group is impossible, although the total content of oxygen functional groups could be described as C/O atomic ratio. Herein, the C/O atomic ratio was determined for pure EG samples on the assumption that iron compounds do not influence the determined parameter. According to elemental analysis, the total amount of oxygen-containing functional groups decreases in the sequence of EL-350-pure > EL-900-pure > PRO-900-pure (Table II). Finally, the combination of gravimetry, elemental and Mossbauer analyses allows to estimate the compositions of EG/Fe₂O₃ samples shown in Table III.

FIG. 2.

FTIR spectra of EG/Fe₂O₃ samples.

TABLE II.

The elemental compositions of EG samples.

TABLE III.

The compositions of EG/Fe₂O₃ samples.

B. Reduction of EG/Fe₂O₃ samples in the dynamic regime

DSC investigations of the reduction of EG/Fe₂O₃ samples in hydrogen atmosphere were performed at temperatures up to 600 °C. Reduction of EL-350 sample was found to be a two-step process, which included an exothermic reaction (Fig. 3, peak 1) in the temperature range of 200–278 °C, with a thermal effect of ∆H ₁ = −27 J/g, and an endothermic reaction (Fig. 3, peak 2) in the temperature range of 310–378 °C with a thermal effect of ∆H ₂ = 66 J/g. Reduction of the samples EL-900 and PRO-900 was a one-step process, occurring in the temperature range of 350–420 °C (Fig. 3, peaks 3 and 4), with thermal effects of ∆H ₃ = 76 and ∆H ₄ = 63 J/g, respectively.

FIG. 3.

The DSC curves of the reduction of EG/Fe₂O₃ samples.

As long as the investigated EG/Fe₂O₃ materials consist of EG matrix and iron-containing particles, the reduction of EG and Ceram/Fe₂O₃ samples were carried out in identical conditions. The reduction of EG samples did not exhibit any thermal effects, except a small exothermic effect for the EL-350-pure sample in the range of 200–220 °C (∆H ₅ = −12 J/g) [Fig. 4(a)]. This effect is attributed to the decomposition of functional groups of its carbon. Presumably, the decomposition of functional groups in EG samples is similar to that in graphite oxides. The partial decomposition of functional groups, mostly hydroxyl and carboxyl groups, takes place spontaneously at temperatures >200 °C, accompanied by evolving of heat and gases (CO, CO₂, and H₂O). At the same time, the functional groups can be directly reduced by hydrogen. The decomposition of the majority of functional groups is observed in the 200–250 °C temperature range, accompanied by intense exothermic effects. At higher temperatures, further decomposition of residual functional groups occurred, but it was significantly less intensive without the visible exothermal effect.^{Reference Pei and Cheng17,Reference Jeong, Lee, Jin, Kim, Bae and Lee18} Thus, during the reduction of EG and EG/Fe₂O₃ samples in hydrogen, the decomposition of oxygen-containing groups took place. The thermal effect of this process was observed only for EL-350-pure [Fig. 4(a), peak 5] and EL-350 (Fig. 3, peak 1) samples with low C/O atomic ratios.

FIG. 4.

The DSC curves of (a) EG samples and (b) Ceram/Fe₂O₃ samples.

In parallel, the reduction of Ceram/Fe₂O₃ was investigated. According to XRD and Mossbauer spectroscopy, the Ceram-350 sample, similar to the EL-350 one, contains ferrihydrite and α-Fe₂O₃, whereas Ceram-900, like the EL-900 and PRO-900 samples, contains almost only α-Fe₂O₃. The following endothermic peaks are observed on DSC curves of Ceram/Fe₂O₃ samples: in the range of 300–372 °C, for Ceram-350 [Fig. 4(b), peak 7]; and in the range of 330–400 °C, for Cearm-900 [Fig. 4(b), peak 8]. According to literature,^{Reference Pei and Cheng17–Reference Baldokhina, Suzdaleva, Prusakova, Burnazyana, Korneevb, Kovalenkob and Folmanisb19} these effects are attributed to the reduction of FeOOH and Fe₂O₃, respectively, that supposes endothermic effects in EG/Fe₂O₃ samples are entirely the result of the reduction of iron-containing phases (Fig. 3). Additionally, two-humped exothermic effect is observed in the case of Ceram-350 [Fig. 4(b), peak 6]. To understand this process, the emitted gas analysis was performed during the heating of Ceram-350 in inert atmosphere. In the 130–300 °C range, the emission of NO₂ was prevailing [Fig. 5(a)]. The peak 6 on the DSC curve [Fig. 4(b)] corresponds to further reduction of evolved gas by hydrogen. Nitrogen dioxide can be formed from the decomposition of nitric acid adsorbed in the ceramic matrix, which, in turn, was formed during flash heating at 350 °C of the ceramic matrix impregnated with iron (III) nitrate.

FIG. 6.

SEM images and energy dispersive x-ray spectra of EL-350 sample after reduction.

To summarize, the exothermic peak on DSC curves of the EL-350 sample represents the two processes: (i) the decomposition of oxygen containing functional groups and (ii) the reduction of gases evolved during the decomposition of nitric acid adsorbed in EG/Fe₂O₃ [Fig. 5(b)]:

(1)$${\rm{4HN}}{{\rm{O}}_{\rm{3}}} = {\rm{4N}}{{\rm{O}}_{\rm{2}}} + {{\rm{O}}_{\rm{2}}} + {\rm{2}}{{\rm{H}}_{\rm{2}}}{\rm{O, }}\,\,\Delta H\,{\rm{ > 0}}\quad {\rm{,}}$$

(2)$${\rm{N}}{{\rm{O}}_{\rm{2}}} + {{\rm{C}}_{{\rm{graphite}}}} = \;{\rm{NO}} + x{\rm{CO}} + {\rm{1}} - x{\rm{ }}{{\rm{C}}^{\rm{ + }}}{\rm{graphite, }}\,\,\Delta H\,\,{\rm{ \lt \,\,0}}\quad {\rm{.}}$$

At the same time, further interaction of evolved gases with hydrogen results in the following exothermal effects:

(3)$${\rm{2NO}} + {\rm{2}}{{\rm{H}}_{\rm{2}}} \to {{\rm{N}}_{\rm{2}}} + {\rm{2}}{{\rm{H}}_{\rm{2}}}{\rm{O, }}\,\,\Delta H\,\,{\rm{ \lt\,\, 0}}\quad {\rm{,}}$$

(4)$${\rm{CO}} + {\rm{3}}{{\rm{H}}_{\rm{2}}} = {\rm{C}}{{\rm{H}}_{\rm{4}}} + {{\rm{H}}_{\rm{2}}}{\rm{O, }}\,\,\Delta H\,\,{\rm{ \lt\,\, 0}}\quad {\rm{,}}$$

(5)$${{\rm{O}}_{\rm{2}}} + {\rm{2}}{{\rm{H}}_{\rm{2}}} \to {\rm{2}}{{\rm{H}}_{\rm{2}}}{\rm{O, }}\,\,\Delta H\,\,{\rm{ \lt\,\, 0}}\quad {\rm{.}}$$

Reactions (3)–(5) occur at significant rates even at a temperature of 200 °C because active gases, such as atomic oxygen, radicals of nitrogen, and carbon oxide, formed during the decomposition process, interact with hydrogen very intensively.

The endothermic peaks in EG/Fe₂O₃ samples correspond to the reduction of iron-containing phases only. According to SEM, by energy dispersive x-ray analysis, both small and large particles of iron oxides are reduced to Fe (Fig. 6). From Mossbauer spectroscopy, 76–90% of iron oxides converts to α-Fe during the reduction of EG/Fe₂O₃. The residual iron is in form of superparamagnetic particles of iron oxide (III) with particles size <10 nm [Fig. 1(b) and Table IV]. The estimated compositions of the EG/Fe₂O₃ samples after reduction are listed in Table V.

FIG. 5.

Profile of the amount of gases evolving during the heating of samples in an inert atmosphere: (a) Ceram-350, (b) EL-350.

TABLE IV.

Parameters of Mossbauer spectra of EG/Fe samples.

TABLE VI.

Calculated parameters for the best kinetic models of the reduction.

As one can see, the origin of expanded graphite matrix affects the completeness of iron reduction. The degree of iron reduction in EL-350 is about 76%, in EL-900 about 85%, and in PRO-900 about 90%. The increase of C/O atomic ratio in the EG/Fe₂O₃ samples from 13 to 35 (Table II) leads to increase in the degree of iron reduction from 76% to 90% (Table V). This can be explained by the interaction of iron-containing phases with EG matrix in EG/Fe₂O₃. Iron compounds are bound to a matrix by chemical forces through oxygen atoms of functional groups.^{Reference Phillips20} The high amount of functional groups can limit the process of reduction, especially the growth of nuclei. During the growth of the metallic iron phase, small particles of iron oxides could remain to be strongly connected to the matrix that lowers the degree of iron reduction.

TABLE V.

The compositions of EG/Fe samples.

But, at the same time, EG matrix does not influence the temperature range and thermal effects of iron oxide reduction. Endothermic peak 2 on the DSC curve of the EL-350 sample (Fig. 3) corresponds to reduction of FeOOH with impurities of Fe₂O₃ and Fe₃O₄:

(6)$${\rm{FeOOH}} + {\rm{1}}{\rm{.5}}{{\rm{H}}_{\rm{2}}} = {\rm{Fe}} + {\rm{2}}{{\rm{H}}_{\rm{2}}}{\rm{O, }}\,\,\Delta H = {\rm{60}}\,{\rm{J/g}}\quad ,$$

(7)$${\rm{F}}{{\rm{e}}_{\rm{2}}}{{\rm{O}}_{\rm{3}}} + {\rm{3}}{{\rm{H}}_{\rm{2}}} = {\rm{2Fe}} + {\rm{3H2O, }}\,\,\Delta H{\rm{ = 11}}\,{\rm{J/g}}\quad ,$$

(8)$${\rm{F}}{{\rm{e}}_{\rm{3}}}{{\rm{O}}_{\rm{4}}} + {\rm{4}}{{\rm{H}}_{\rm{2}}} = {\rm{3Fe}} + {\rm{4}}{{\rm{H}}_{\rm{2}}}{\rm{O, }}\,\,\Delta H = {\rm{6}}\,{\rm{J/g}}\quad {\rm{.}}$$

Considering the content of iron compounds in EG samples and the degree of iron reduction, the theoretical enthalpy is to be equal to $\Delta H_{\rm{2}}^{{\rm{theoretical}}}$ = 77 J/g (±10%), coinciding, within error, with the experimental enthalpy ∆H ₂ = 66 J/g (±5%).

The endothermic effects observed on the DSC curves of the samples EL-900 and PRO-900 (Fig. 3, peaks 3 and 4) correspond to the reduction of iron oxide (III) with impurities of Fe₃O₄:

(9)$${\rm{F}}{{\rm{e}}_{\rm{2}}}{{\rm{O}}_{\rm{3}}} + {\rm{3}}{{\rm{H}}_{\rm{2}}} = {\rm{2Fe}} + {\rm{3}}{{\rm{H}}_{\rm{2}}}{\rm{O}}\quad {\rm{,}}$$

(10)$${\rm{F}}{{\rm{e}}_{\rm{3}}}{{\rm{O}}_{\rm{4}}} + {\rm{4}}{{\rm{H}}_{\rm{2}}} = {\rm{3Fe}} + {\rm{4}}{{\rm{H}}_{\rm{2}}}{\rm{O}}\quad {\rm{.}}$$

The experimental values enthalpy are agreed well with the calculated values: $\Delta H_{\rm{3}}^{{\rm{theoretical}}}$ = 80 J/g (±10%) ≈ ∆H ₃ = 76 J/g (±5%) and $\Delta H_{\rm{4}}^{{\rm{theoretical}}}$ = 60 J/g (±10%) ≈ ∆H ₄ = 63 J/g (±5%).

C. Kinetics of the EG/Fe₂O₃ reduction

The kinetics of reduction was investigated on the example of the PRO-900 sample. DSC curves of reduction at heating rates of 3, 5, and 10°/min are given in Fig. 7. The NETZSCH Thermokinetics 3.1 software was used to evaluate kinetic models and Arrhenius parameters. Several one-stage and two-stage kinetic models were tested. There are several models, e.g., phase-boundary-controlled reaction models, n-dimensional diffusion models, n-dimensional growth of nuclei models, etc., which are widely utilized to fit the DSC data of iron compound reduction. It was found that the two-stage model, viz., the expanded Prout–Tompkins autocatalysis reaction model followed by n-dimensional Avrami–Erofeyev growth of nuclei one, fits best the experimental data (Table VI). It is to note that the suggested mathematical models for the reduction of iron oxides (III) on the EG matrix appeared to be the same as the models for their reduction without graphite matrix.^{Reference Lin, Chen and Li21,Reference Pineau, Kanari and Gaballah22} This is consistent with the theoretical consideration. In the beginning, the reduction process is limited by chemical reaction factors only. On formation, a new α-Fe phase adsorbs hydrogen molecules and catalyzes the reduction—this corresponds to the autocatalytic Prout–Tompkins model. During the formation of large α-Fe nuclei, the reduction becomes limited by topological factors of growth of the nuclei.

FIG. 7.

The approximation of the mathematical model to the DSC curves of the sample PRO-900.

D. The reduction of EG/Fe₂O₃ samples in static regime

In this part, we investigated the reduction of materials at constant temperatures. Figure 8 shows the XRD patterns of PRO-900 heated in hydrogen atmosphere at temperatures of 300, 400, 500, and 600 °C for 2 h. As it follows from the figure, the reduction onset of iron oxides was at 400 °C. The increase in the reduction temperature results in the increase of XRD peak intensities of α-Fe as well as their narrowing, reflecting the enlargement of iron particle. At 600 °C, as determined by Mossbauer spectroscopy, the maximum degree of iron reduction of 98% is reached. The two other EG/Fe₂O₃ samples (EL-350 and EL-900), with the degree of iron reduction at 600 °C higher than 96%, demonstrate similar behavior. Thus, during the prolonged reduction of EG/Fe₂O₃ samples at 600 °C, the influence of expanded graphite nature had almost vanished as iron-containing phases were nearly completely transformed into α-Fe.

FIG. 8.

XRD patterns of the sample PRO-900 after the reduction at constant temperatures.

The EG/Fe samples obtained are characterized by the high values of saturation magnetization, which linearly depends on the α-Fe content [Fig. 9(a), Table VII]. The experimental magnetic saturation values are consistent with the theoretical ones calculated as $M_{\rm{s}}^{{\rm{theoretical}}}$ = M _s (bulk α-Fe)·ω(α-Fe). It implies, that α-Fe particles in EG/Fe are more than tens of nanometers in size for all samples.

FIG. 9.

(a) The magnetization of EG/Fe samples in dependence on magnetic field strength, (b) the pore structure distribution of EG/Fe samples.

TABLE VII.

Magnetic and sorption properties of EG/Fe samples.

The EG/Fe materials are considered to be very promising magnetic sorbents for liquid hydrocarbons sorption applications, e.g., for oil spill response. These materials possess very low bulk density, viz., 2.0 kg/m³ for EL-350 and EL-900 and 2.5 kg/m³ for PRO-900. It is well known that the pore structure of EG/Fe is mainly comprised of macropores. As it was impossible to investigate pore structure of EG/Fe because of their low bulk densities, we used compressed samples with densities of 40 kg/m³ to perform it by standard contact porosimetry [Fig. 9(b), Table VII].^{Reference Avdeev, Sorokina and Tverezovskaya12} The total pore volume V _pore (40 kg/m³) did not depend on the nature of EG/Fe and was about 22–23 cm³/g. Macropores with a radius of 100 nm–10 μm contribute to approximately 70% of the total volume and corresponds to the elliptical and crevice inside pores of EG/Fe worm-like particles. Macropores with radius of 10–100 μm contribute to approximately 30% of the total volume and are spaces among worm-like particles. Average pore sizes are in the range of 110–180 nm, and the values of specific surface are in the range of 30–150 cm²/g (Table VII). In our previous work, it was shown that samples EL-350 and EL-900 possessed very high sorption capacities to liquid hydrocarbons: up to 85 g/g of octane and up to 110 g/g of motor oil^{Reference Lutfullin, Shornikova, Vasiliev, Pokholok, Osadchaya, Saidaminov, Sorokina and Avdeev5} that did not depend on iron content. Iron phases were found not to affect the porous structure of materials. In this work, we determined the sorption capacity for crude oil with a viscosity of 1.8 mPa s. The values of sorption capacities for EG/Fe with bulk densities of 2–2.5 kg/m³ are in the range of 89–93 g/g, and they do not depend on the nature of expanded graphite or specific surface but only on total pore volume (Table VII). The EG/Fe materials with sorbed oil can be attracted by magnets with a magnetic field strength of up to 4 kOe and a gradient of field strength of up to 1 kOe/mm.