Carbon microspheres decorated with iron sulfide nanoparticles for mercury(II) removal from water

All starting materials and chemicals used in this work were reagent grade from commercial sources and used without further purification. A VARION BIM-7 styrene-based cation exchange resin containing 7% divinylbenzene cross-linker, 2% acrylonitrile modifier, and iminodiacetate (–N(CH₂COOH)₂) functional groups was used as the starting material. This resin had a nominal binding capacity of 1 mol divalent metal cation per 1 dm³ resin. The resin was saturated with aqueous Fe₂(SO₄)₃ solution according to the method described in our previous work [14]. Briefly, the resin was conditioned with 1 M aqueous NaOH solution, washed with distilled water, saturated with threefold excess of 0.5 M aqueous Fe₂(SO₄)₃ solution, and finally washed with water. The exchanged resin was dried in air for several days and then in a drying box at 120 °C for 1 day.

The loaded resin, 2–3 grams placed into a porcelain combustion boat, was carbonized in a very slow, high-purity nitrogen stream in a horizontally aligned quartz tube using a tube furnace (see also Refs. [14, 15] for more details). The furnace was heated up to the desired temperature in 30 min, kept at this temperature for 4 h, and then left to cool down naturally.

Mössbauer spectroscopic (MoS) measurements were taken at room temperature using a KFKI Mössbauer spectrometer in constant acceleration mode with a ⁵⁷Co(Rh) source of 1.5 GBq activity. Spectra were analyzed assuming Lorentzian line shapes using the MossWinn 3.0i XP software, and isomer shifts were determined relative to α-Fe reference.

Thermogravimetric (TG)/Mass (MS) measurements were taken on a modified PerkinElmer TGS-2 thermo balance and a HIDEN HAL 2/301 PIC quadrupole mass spectrometer. The dried resin, typically 2.5 mg placed into a platinum sample pan, was heated at a heating rate of 20 °C min⁻¹ up to 900 °C in an argon atmosphere. Portions of the volatile products were introduced into the mass spectrometer operated at a 70 eV electron energy through a glass-lined metal capillary transfer line heated to 300 °C.

Scanning electron microscopy (SEM) and energy-dispersive X-ray analysis (EDX) were performed using an FEI Quanta 3D high-resolution microscope. Resolution of the instrument at 30 keV accelerating voltage and in high vacuum was ≤ 1.2 nm and ≤ 2.5 nm using the secondary electron detector (SED) and backscattered electron detector (BSED), respectively. The energy resolution of the X-ray detector (EDX) was 130 eV at Mn Kα.

Raman spectra were recorded on a confocal Horiba Jobin Yvon LabRAM HR Raman microscope using He–Ne excitation (632 nm) and a laser power of 0.1 mW.

BET specific surface area (SSA) of carbon microspheres was determined using an ASDI RXM-100 Catalyst Characterization instrument, nitrogen gas at liquid nitrogen temperature, and the volumetric method. Nitrogen sorption experiments were done using a Quantachrome Autosorb 1C static volumetric apparatus at 77 K by measuring the increase in volume at equilibrium as a function of the relative pressure. Adsorption data were obtained applying successive doses of the selected gas until p/p₀ = 1 relative pressure was reached. Microspheres were outgassed prior to measurements at 300 °C for 24 h.

Iron ion concentrations in aqueous samples were determined using a Spectro Genesis ICP-OES simultaneous spectrometer with axial plasma observation. The mercury ion determination in aqueous samples, however, was not possible using this ICP-OES due to unrealistic results. Therefore, mercury ion concentrations in aqueous samples were determined using the spectrophotometric method based on the absorbance of the Hg–thiocyanate–rhodamine B (RB) complex Hg²⁺(SCN⁻)₄(RB⁺)₂ as described in the literature [16]. Measurements were taken using a PerkinElmer UV/Vis Lambda 35 spectrometer applying a slit width of 1 nm.

The adsorption performance of microspheres was determined using stock Hg(NO₃)₂ solutions with known Hg²⁺ concentration of 10, 20 and 40 mg dm⁻³. Typically, 30 ml of stock solution was treated with 10, 25, or 50 mg adsorbent at 25 ± 1 °C at pH 6.5 for one day. The Hg²⁺ concentration before and after the treatment was determined using UV–Vis spectroscopy by measuring the absorbance of Hg–thiocyanate–RB complex at 610 nm. The amount of Hg²⁺ adsorbed on microspheres (q_t, mg g⁻¹) was calculated using Eq. (1)where A₀ and A_t are the initial absorbance and absorbance at time t, c₀ is the initial Hg²⁺ concentration (mg dm⁻³), V is the volume of the solution (dm³), and m is the mass of adsorbent (g).

Kinetic measurements were taken at 25 ± 1 °C and at pH 6.5. The pseudo-second-order rate equation [17] was expressed as follows (2)where t = time, q_eq = adsorbed amount of Hg²⁺ at equilibrium (mg g⁻¹), q_t = adsorbed amount of Hg²⁺ at time t (mg g⁻¹), and k_ad = adsorption rate constant.

For determining the adsorbed amount of mercury, three replicate syntheses and analyses were carried out and the average value is reported with uncertainty values of ± 4 mg g⁻¹.

Our aim was to synthesize iron sulfide NP-modified carbon microspheres; therefore, critical questions were whether the sulfate ion stays in the resin as a counterion after cation exchange and whether iron sulfide NPs form during carbonization. The Fe-to-S ratio of the dried resin was determined on several points of its surface using EDX, and an average ratio of about 1:1 was obtained. (Note that EDX is a semiquantitative method; Fe-to-S ratios varied between 1.06:1 and 0.93:1; an example is shown in Figure S1 in Supporting Material.) The functional groups of the loaded resin can thus be described as [–NH⁺(CH₂COO⁻)₂Fe³⁺(H₂O)_x]SO₄²⁻. The iron content on the surface of loaded resins was measured to be 7.2% by EDX. To find the right carbonization temperature, the thermal stability of iron sulfate-loaded resin was studied by TG–MS. The weight loss of the resin started steadily about 187 °C, and the main decomposition occurred at about 432 °C (see Fig. 1). The weight loss up to 900 °C was 69.4%. MS monitoring of the decomposition identified CO₂ and H₂O as major decomposition products at lower temperatures what indicated decarboxylation of the resin and partial loss of coordinated water (Fig. 1). The polymer backbone decomposition and hydrocarbon (e.g., benzene and styrene) formation occurred in the main decomposition step (Fig. 1). Hydrogen formation was detected at high temperatures above 500 °C, possibly due to water–carbon reaction. It has to be noted that the sensitivity of the mass analyzer is gradually decreasing with increasing mass; therefore, the relative ion currents do not represent their relative amounts. The weight loss of carbonized resin was small above 500 °C; therefore, the carbonization of the resin on a larger scale using a furnace was investigated in the 500–1000 °C temperature range in 100 °C increments.

The phase compositions of carbon microspheres prepared by carbonization at various temperatures were determined using MoS (Figs. 2 and S2–S7), PXRD (Figures S8–S13), and Raman microscopy (see below). Non-stoichiometric iron monosulfide (FeS) was identified as the major iron compound. Magnetite (Fe₃O₄) appeared as a minor component at lower temperatures between 500 and 700 °C, and it was replaced by elemental iron (both α and γ) and cementite (Fe₃C) at higher temperatures (see Table 1). At least two or more phases of non-stoichiometric and stoichiometric iron monosulfide (FeS) formed during thermal decomposition of the resin. It was not possible, however, to distinguish between these FeS phases due to their similarity. PXRD patterns of various pyrrhotite Fe_1−xS (x = 0.875–0.92) and troilite FeS phases are very similar, particularly for nonzero x [18]. We note that neither troilite nor any pyrrhotite single phase could perfectly describe the experimental PXRD patterns, although any of them provided a reasonable fit. Two hexagonal pyrrhotite phases were therefore used to describe PXRD patterns for microspheres obtained at 500, 600, and 700 °C, and only one for samples obtained at 800, 900, and 1000 °C. The average pyrrhotite crystallite size is observed to increase slightly by increasing the carbonization temperature and estimated to be around 20 nm at 500 °C and 25 nm at 1000 °C (Table S2). Graphitization was not observed by PXRD. MoS spectra were very complicated due to the large number of overlapping sextets making the fit for FeS phases ambiguous. MoS spectra could be described by fitting an additional superparamagnetic pyrrhotite component (Table 1, Fig. 2). The latter together with the normal component indicated a wide crystallite size distribution.

Raman microscopy confirmed the carbonization of the resin. Raman spectra recorded on several spots of microspheres exhibited two strong bands, a broader and stronger at 1331–1351 and a narrower with smaller intensity at 1591–1597 cm⁻¹, assigned to carbonic D and G bands, respectively (Fig. 3). The D band is associated with the vibrational mode of sp³-bonded carbon atoms, while the G band is related to graphitic sp² carbon structures. The D band exhibited a shoulder at around 1155–1173 cm⁻¹ which may arise due to sp²–sp³ bonds or polyene-like structures [19]. Weak second-order Raman bands are also detected in the 2500–3000 cm⁻¹ region. The appearance of these first- and second-order Raman bands, as well as the D/G band intensity ratio, is characteristic of amorphous carbon [14]. Raman measurements indicate a low degree of graphitization for synthesized microspheres, in agreement with PXRD. The detected Raman bands appear on top of a broad luminescence background at 500 °C carbonization, which indicates imperfect carbonization at this temperature, but this background disappears with increasing carbonization temperature. The measured Raman intensity, in general, decreases with increasing carbonization temperature, which indicates gradual replacement of sp³ carbon atoms by sp² carbon atoms. Since the laser spot of our Raman microscope was about 500 nm, larger crystallites on the surface of microspheres could be detected. Mapping microspheres’ surfaces, Raman bands characteristic of pyrrhotite [20, 21] were identified at 316, 336, 369, and 379 cm⁻¹ (see Fig. 3, inset). Immersing microspheres into 18% aqueous hydrochloric acid, the characteristic smell of H₂S could be instantly observed, what provided further support for the formation of pyrrhotite crystals on the microspheres’ surfaces.

According to SEM investigation, the synthesized microspheres inherited the spherical morphology of ion exchange resins (original diameter between 600 and 900 μm), but shrunk during the heat treatment to 300–500 μm (Figs. 4 and S14–S19). Macropores and small crystallites were well observable on the surface of microspheres. The size of crystallites gradually increased and crystallites became more exposed on the surface by increasing carbonization temperature. Relatively large crystallites even with a diameter up to 250 nm were also obtained (Fig. 4), and possibly these were detected with Raman spectroscopy (see above). These crystallites could be removed from the surface by 18% aqueous hydrochloric acid treatment (see Figure S20). Iron ions were leaching out from the surface into the solution, and their concentration was determined by ICP-OES analysis to be between 5.1 and 6.3%. Although surface iron leachable by acid gradually increased by increasing the carbonization temperature, differences were surprisingly small between microspheres prepared at different temperatures. It is worth noting that acid treatment is very effective to remove all iron-containing phases where the crystallite is not perfectly surrounded by the inert carbon matrix.

The BET specific surface areas of microspheres were determined using nitrogen adsorption measurements, and results are shown in Fig. 5 and listed in Table S3. The SSA of microspheres obtained by carbonizing at 500 °C was 278 m² g⁻¹, and the SSA gradually decreased to 72 m² g⁻¹ at 1000 °C with increasing temperature, possibly due to sintering the surface. SSA decreased by about 30% by removing NPs from surfaces using hydrochloric acid (see Table S3). Nitrogen adsorption–desorption isotherm was measured for microspheres prepared at 900 °C because they exhibited the best performance for mercury removal (see below). The isotherm (Fig. 5) and pore volume distribution (Figure S21) indicated poorly developed micro- and mesopores. (The calculated total pore volume for micro- and mesopores was only 0.022 and 0.17 cm³ g⁻¹ at p/p₀ = 0.97939; no special pore width was obtained.) According to IUPAC classification, the isotherm is a type IV with an H3 hysteresis loop, characteristic of non-rigid aggregates of particles on the surface or not completely filled macropores of the pore network [22].

The mercury removal capacity and efficiency of synthesized microspheres were studied by treating 30 ml aqueous Hg(NO₃)₂ solution with a Hg²⁺ concentration of 10, 20, or 40 mg dm⁻³ at pH 6.5 for one day using 10, 25, or 50 mg adsorbent load. The Hg²⁺ concentration before and after the treatment was determined by UV–Vis spectroscopy. Hundred percentage mercury removal efficiency was obtained for solutions with 10 or 20 mg dm⁻³ Hg²⁺ even at 10 mg adsorbent load for the best performers; therefore, the solution with Hg²⁺ concentration of 40 mg dm⁻³ was used as the case study. Adsorption capacities were obtained to be relatively small for microspheres prepared at lower temperatures, but they strongly increased by increasing carbonization temperatures. They were 20, 23, 22, 42, 104, and 100 ± 4 mg g⁻¹ for microspheres carbonized at 500, 600, 700, 800, 900, and 1000 °C, respectively. This trend is just the opposite of that of SSA (Fig. 5), which indicates that not the SSA is the determining factor in mercury removal. Investigating microspheres after mercury adsorption by SEM, deposits of new crystallites were detected (Figs. 6 and S22). EDX indicated a 1:1 Hg-to-S ratio in the deposited material (Figure S23) and Raman microscopy unambiguously identified HgS [23]. (Strong and weak characteristic bands of HgS are observed at 253 and 344 cm⁻¹, respectively; see Fig. 6.) During the adsorption of mercury ions, iron was leaching out of the surface into the solution, and it was measured by ICP-OES to be 4, 5, 5, 10, 27, and 27 mg g⁻¹ for microspheres obtained at 500, 600, 700, 800, 900, and 1000 °C, respectively. The mechanism of Hg²⁺ removal by iron sulfide nanoparticles is generally described by three processes, namely adsorption, ion exchange, and precipitation [24‐27]. The latter two result in iron leaching into the solution. In addition, it cannot be ruled out that small amounts of iron NPs exist on the microsphere surfaces (see Table 1 for phase composition). A clear correlation is observed between the mercury ion removal from solution and iron dissolution in our experiments, namely larger adsorption capacity associated with larger iron dissolution. The amount of Hg²⁺ ions removed from the solution by chemical reactions can be estimated by assuming a 1:1 replacement of iron by mercury (FeS + Hg²⁺ → HgS + Fe²⁺, FeS + xHg²⁺ → Fe_1–xHg_xS + xFe²⁺, and Fe + Hg²⁺ → Hg + Fe²⁺ reactions), and calculating the deposited mercury and leached iron from adsorption capacities and ICP-OES data, respectively. According to this, 70–90% of mercury ions are estimated to be removed from the solution due to chemical reactions and 10–30% by physisorption, indicating that precipitation and ion exchange are important processes in the mercury removal mechanism (also note deposited HgS detected by Raman). Microspheres obtained by carbonizing the resin at 900 °C exhibited the highest sorption capacity of 104 mg g⁻¹. The mercury ion removal efficiency of this material was obtained to be 86.4, 99.6, and 100.0% using a sorbent load of 0.33, 0.83, and 1.67 mg sorbent per 1 dm³ Hg²⁺ solution, corresponding to 5.44, 0.16, and 0.0 mg dm⁻³ remaining Hg²⁺ ion concentration in treated solutions, respectively. The sorption kinetics of this material was studied using a sorbent concentration of 0.33 g dm⁻³. Experimental data could be well fitted using a pseudo-second-order kinetic model [17] (see Fig. 7), from which the sorption rate constant was calculated to be 3.69 × 10⁻⁴ g mg⁻¹ min⁻¹. The initial adsorption rate was obtained to be 4.01 mg g⁻¹ min⁻¹. 50% of the adsorbed mercury ions were removed in 13 min. Several carbon-based materials have been synthesized, activated, and tested for mercury removal to date [6, 12]. A comparison of the sorption performance of synthesized microspheres to those of relevant thiolated carbonaceous sorbents published in the literature is shown in Table 2. Microspheres have relatively high adsorption capacity, even without using any activation method. The Hg²⁺ removal efficiency, namely 100%, as well as the production and application simplicity of the novel microspheres is very advantageous for wide-scale applications compared to previously synthesized materials.