Hydrodechlorination of carbon tetrachloride with nanoscale nickeled zero-valent iron @ reduced graphene oxide: kinetics, pathway, and mechanisms

Carbon tetrachloride (CT) was widely used in the past as a cleaning and degreasing solvent in industry due to its ability to dissolve a wide range of organic substances. The widespread use of CT led to its dispersion in the environment persistently (Doherty 2000). Due to the adverse health effects of CT on humans and animals, such as central nervous system depression, liver and kidney damage, CT has been established as a class of priority pollutants prohibited by the US EPA (Hua & Hoffmann 1996). However, chloroform (CF) as a reduction product of CT is more toxic (Semprini & McCarty 1992). Therefore, it is of great significance to develop an efficient method for transforming CT into non-hazardous products.

In general, degradation of CT via biological methods always requires quite a long time (Hashsham et al. 1995), and may have the risk of poisoning the organism in some situations (Santharam et al. 2014), which strongly restricts its application field. Boronina & Klabunde (1995) found that three zero-valent metals (magnesium, tin and zinc) can be used to reduce CT in aqueous solutions. Feng & Lim (2005) found that although Zn could reduce CT with a high reaction rate compared with Fe powder, it was hard to reduce CT to complete dechlorination products (CDCPs) during the reaction. Due to the low-cost and high efficiency, zero-valent iron (ZVI) has been widely investigated in the wastewater treatment of chlorinated organic compounds (COCs) (Gillham & Ohannesin 1994; Matheson & Tratnyek 1994; Balko & Tratnyek 1998; Cwiertny et al. 2006). Among these, nanoscale zero-valent iron (nZVI) has particularly attracted considerable attention due to its large specific surface area (Zhang 2003). However, this advantage could not be fully exploited while in practice (Liu & Lal 2012), such as nZVI easily formed a magnetic passive layer which blocked the reactive sites and resulted in aggregation of the nanoparticles (NPs) (Zhang et al. 2002). Therefore, it is highly required to develop a new composite catalyst to improve the catalytic activity of nZVI.

Recently, various support materials have been developed to prevent particle aggregation (e.g. MWCNTs, activated carbon, SiO₂, carboxymethylcellulose, Fe₃O₄) (Kustov et al. 2014; Jin et al. 2018; Lv et al. 2018; Wu et al. 2018; Chen et al. 2019). Beyond these, graphene as a supporting material holds excellent mechanical, electrical and thermal characteristics, which is superior to other traditional materials (Dreyer et al. 2009). In order to obtain a similar structure to pristine graphene, reduced graphene oxide (rGO) was synthesized from graphite flake via a chemical method. It has some excellent characteristics: (1) the high specific surface area of the two-dimensional structure could allow nanoparticle catalyst to be efficiently loaded on it (Bai & Shen 2012), (2) the conjugated structure of graphene could enhance the adsorption capacity of the reactant molecules (Zhang et al. 2012), (3) the oxygen-containing groups could make the layers hydrophilic (Pei & Cheng 2012), (4) the characteristics of electrical conductivity were favorable during the catalytic reaction (Stankovich et al. 2007).

Deposition of a second metal (e.g. Ni, Pd, Cu, Co, Au and Pt) as a co-catalyst onto the surface of iron is a general process to enhance the dechlorination efficiency and reaction rate (Cwiertny et al. 2006). Therein, Pd/Fe bimetal greatly improved the dechlorination efficiency of COCs; however, nickel as a non-precious metal could replace palladium metal without loss of performance (Zhang et al. 1998). As a result of lower hydrogen overpotential, Ni as a co-catalyst accelerated the hydrogen production process compared with single Fe, and the produced H₂ could reduce COCs indirectly with the existence of an effective catalyst (e.g. Pd, Ni, Cu and Co) (Matheson & Tratnyek 1994; Schreier & Reinhard 1995; Lowry & Reinhard 1999; Schrick et al. 2002). For this reason, we chose Ni as a cost-effective metal dopant on nZVI particles to improve the reactivity of the catalyst.

In this study, nanoscale nickeled zero-valent iron @ rGO (NF@rGO) composite was synthesized, characterized and used to degrade CT. Batch experiments evaluated the removal capacity and the catalytic activity of laboratory-synthesized catalysts; besides that, the reduction products of CT and its reaction kinetics were studied. Catalyst characterization was carried out by X-ray diffraction (XRD), X-ray photo-electron spectroscopy (XPS) and scanning electron microscope (SEM) with X-ray energy dispersive spectroscopy (EDS). The purpose of our work was to develop a highly active catalyst, and provide a theoretical reference for the catalytic hydrodechlorination of COCs.

All the chemicals used were of analytical reagent (AR) grade from commercial sources. Graphite flake (purity ≥ 99.8%, −325 mesh), NaNO₃ (≥99.5%), H₂SO₄ (5 and 98%), KMnO₄ (≥98.0%), H₂O₂ (30%) and HCl (5%) were used to synthesize graphene oxide. FeSO₄·7H₂O (≥99.5%). NiSO₄·6H₂O (≥98.5%) and NaBH₄ (≥ 99.5%) were used to synthesize NF@rGO. All degradation experiments used a certain volume of freshly made stock solution of CT (15 mg/L). All deionized (DI) water was prepared with an Aquelix 5 (Millipore) pure water system and bubbled with nitrogen (99.2%) before use.

The graphene oxide (GO) was prepared by Hummers method, and the whole process was carried out in an Erlenmeyer flask with water bath and magnetic stirring. Firstly, graphite flake (1 g) and NaNO₃ (0.5 g) powder were mixed by using H₂SO₄ aqueous solution (23 mL, 98%). After 1 h stirring at 0 °C, 3 g of KMnO₄ was divided into 6 parts and add to the dispersion within 2.5 h at 10–15 °C. Subsequently, the temperature of the water bath was adjusted to 35 °C and it was stirred for 24 h, and then the temperature was increased to 90 °C. Meanwhile, H₂SO₄ aqueous solution (100 mL, 5%) and deionized water (150 mL) was added to the mixture respectively for further dilution of the dispersion. After the H₂O₂ aqueous solution (5 mL, 30%) was added to the suspension, the resultant GO was immediately filtrated (using 20 μm qualitative filter paper) until the suspension stopped bubbling, and it was washed with HCl aqueous solution (5%) and DI water (until no SO₄²⁻ was detected in the supernate fluid). Finally, the dried GO powder was obtained with a vacuum oven (60 °C for 48 h).

At last, the NF@rGO NPs were washed and freeze-dried for further use. Similarly, nZVI and Ni/Fe bimetallic NPs were also prepared with this procedure. The preparation method of 2N₄F@MWCNTs (Fe/MWCNTs = 2 wt./wt., Ni/Fe = 4 wt.%) was according to our previous work (Chen et al. 2019).

The surface morphology of nZVI, GO and NF@rGO was characterized by scanning electron microscope (SEM) (Hitachi, SU-8010, Japan), the Ni content of Ni/Fe bimetal was quantified by X-ray energy dispersive spectroscopy (EDS) (Horiba, Japan). The mineralogical properties of nZVI, fresh and used NF@rGO, were investigated by X-ray diffraction (XRD) (Bruker, AXS, D8 Advance). The chemical states of NF@rGO were examined by X-ray photo-electron spectroscopy (XPS) (Thermo Scientific, ESCALAB 250 XI).

A series of batch experiments were conducted to study the catalytic activity of laboratory-synthesized catalysts and investigate the intermediate products of CT degradation. All of the experiments were carried out in 100 mL serum bottles with a rotary shaker (220 rpm), each bottle was double sealed, Teflon-lined and had an aluminum cap, and each experiment was run in triplicate. Before the experiment, 0.01 g of laboratory-synthesized catalysts were added in the serum bottle respectively with 15 mg/L of CT stock solution until no headspace was left. At specific time intervals, 1 mL of samples were filtered with 0.22 μm filters and withdrawn to bottle headspace for analysis the concentration of CT and its degradation products by gas chromatography spectrometry (GC-2014, Shimadzu, Japan). In addition, the generated CH₄ and CO in the bottle were gathered with an aluminum foil gas-collecting bag.

The sample from bottle headspace was determined by GC with electron capture detector (ECD). Samples were separated and evaluated with a Rtx-1 column (30.0 m × 0.25 mm ID × 0.25 μm thickness, Restek, USA), with nitrogen (99.99%) as the carrier gas. The oven temperature was programmed from 40 to 100 and finally to 200 °C at the rate of 8 and 6 °C/min, with a 5 min initial hold and 10 min final hold. The temperature of injector and detector were 220 and 320 °C, respectively.

Besides, the sample from the gas-collecting bag was determined by GC with a thermal conductivity detector (TCD). Samples were separated and evaluated with a TDX-01 column (2.0 m × 3.00 mm, China), with helium (99.99%) as the carrier gas. The oven temperature was 120 °C, held for 7 min, and the temperatures of the injector and detector were 100 and 150 °C, respectively. Blank experiments and repeated experiments were carried out and the results are described in the Supplementary Information.

The XRD patterns in Figure 1 were determined to identify the crystal structures of nZVI, fresh and used 4N₄F@rGO (Fe/GO = 4:1 wt./wt., Ni/Fe = 4 wt.%). The sharp and strong diffraction peaks at 2θ = 44.62° and 65.03° corresponded to the standard card of α-Fe (JCPDS, No. 06-0696). However, the intensity of the reflection peak at 2θ = 44.62° decreased significantly in fresh 4N₄F@rGO, this may be caused by the small crystallites of Fe⁰ particles.

Besides, it was noteworthy that the reflection peaks at 2θ = 27.55°, 36.29° and 46.91°, which belonged to aged 4N₄F@rGO, corresponded to the standard card of lepidocrocite (γ-FeOOH) (JCPDS, No. 44-1415) and were assigned to the (210), (301) and (020) planes, respectively. The reflection peaks at 2θ = 30.07°, 35.87°, 56.91°, 62.50° corresponding to magnetite (Fe₃O₄) (JCPDS, No. 19-0629) were assigned to the (220), (311), (511) and (440) planes, and/or maghemite (γ-Fe₂O₃) (JCPDS, No. 25-1402), assigned to the (206) and (119) planes. The results exposited that the Fe⁰ was oxidized and deposited on the surface of 4N₄F@rGO in the process of reaction.

However, no obvious characteristic diffraction peaks of Ni or graphite were detected either in fresh or recycled 4N₄F@rGO, which was due to the diffraction peaks of nickel metal being very close to iron, and graphene sheets were prevented from forming a regular crystal structure as a result of the inserted NPs.

In order to get more information on the chemical composition of NF@rGO, XPS analyses were performed to evaluate the chemical states including Ni, Fe, O and C (Figure 2(a)).

According to Ni 2p spectrum (Figure 3(b)), the photoelectron peaks located at 852.4 and 869.0 eV demonstrated the existence of Ni (Ni2p_3/2, 852.3 eV; Ni2p_1/2, 869.7 eV) (Moulder et al. 1979), and the peaks of NiO (Ni2p_3/2, 853.3 eV; Ni2p_1/2, 871.7 eV) (Moulder et al. 1979) were also detected at 854.2 and 870.6 eV.

Meanwhile, Figure 2(c) shows the high-resolution spectrum of Fe2p, two photoelectron peaks at 710.5 and 723.9 eV demonstrate the existence of oxidized iron (Fe(III) and Fe(II)), which are similar to magnetite (Fe₃O₄) (Fe2p_3/2, 710.6 eV; Fe2p_1/2, 724.1 eV) (Fiedor et al. 1998), hydrated ferric oxide (FeOOH) (Fe2p_3/2, 711.2 eV; Fe2p_1/2, 724.9 eV) (Echigo et al. 2012), hematite (α-Fe₂O₃) (Fe2p_3/2, 711.0 eV; Fe2p_1/2, 724.6 eV) (Yamashita & Hayes 2008) and maghemite (γ-Fe₂O₃) (Fe2p_1/2, 710.9 eV) (Zhang et al. 2011). Moreover, Fe⁰ showed an obvious peak at 706.7 eV followed by its satellite peak at 719.5 eV (Fe2p_3/2, 706.75; Fe2p_1/2, 719.95 eV) (Moulder et al. 1979).

The O (1 s) spectrum (Figure 2(d)) showed three types of oxygen species, which corresponded to the anionic oxygen in Fe_XO_Y (530.7 eV), the oxygen-containing functional groups in rGO (531.9 eV) (Fan et al. 2011), and the adsorbed H₂O (536.0 eV) (Ertl et al. 1997).

The C (1 s) spectra (Figure 2(e)) presented four types of carbon bonds (Moulder et al. 1979; Ertl et al. 1997; Hayes 2001): (1) non-oxygenated carbon (including C-C, C = C, and C-H) (284.4 eV), (2) the carbon in C-O (C-O-C and/or C-OH groups) (285.4 eV), (3) the carbonyl carbon in C = O (287.8 eV), (4) carboxylate carbon in O-C = O (288.9 eV). The peak area of C-C was greater than other carbon-based oxygen-containing functional groups, which proved that GO was successfully reduced to rGO by NaBH₄ during the preparation of the catalyst.

The SEM image of nZVI (a), rGO (b), 4N₄F@rGO (c, d) is presented in Figure 3; the partition ratio of the alloy element (corresponding to the box in part d) was measured and is presented in Figure 3(e). It can be seen that nZVI particles interconnected to form a necklace-like structure due to surface tension and magnetic interactions between each particle. Figure 3(b) exhibits the sheet-like structures of GO. Moreover, it can be clearly seen 4N₄F@rGO from Figure 3(c) and 3(d) that the Ni/Fe bimetal particles were evenly distributed on the surface of rGO. The EDS analysis indicated that the NPs were Ni/Fe bimetal and reached a desirable ratio.

Various mass of Ni/Fe bimetallic NPs (Ni/Fe = 4 wt.%) were loaded onto the surface of rGO. As shown in Figure 4, the reduction of CT followed pseudo-first-order kinetics, 4N₄F@rGO (Fe/GO = 4 wt./wt., Ni/Fe = 4 wt.%) reached a higher efficiency in CT degradation.

Furthermore, the properties of four laboratory-synthesized catalysts were investigated by calculating the removal capability (mg _CT/g _catalyst). The results of the calculation are listed in Table 1; 4N₄F@rGO could degrade CT with a high removal rate in 35 min, the observed rate constant (k_obs) indicated that the reaction rate of 4N₄F@rGO was 10 times that of nZVI. By comparison with the removal capability and reaction rate, 4N₄F@rGO achieved better results than 2N₄F@MWCNTs (Fe/MWCNTs = 2 wt./wt., Ni/Fe = 4 wt.%) in CT degrading, which means rGO as a supporter could be loaded with more metal to improve its catalytic performance.

Figure 5(d) clearly demonstrates that nickel as a co-catalyst decreased the activation energy of 6.59 kJ/mol in N₄F compared with nZVI, and rGO as a conductive material could further reduce the activation energy of 4.73 kJ/mol by accelerating electron transfer during the redox reaction.

In order to explore the reduction pathway of CT, various products were monitored by GC in the process of reaction, including chloroform (CF), dichloromethane (DCM), carbon monoxide (CO) and methane (CH₄), trace amounts (around the detection limit of GC) of perchloroethylene (PCE) and trichloroethylene (TCE) were also detected after a certain time of reaction. Among these, CO and CH₄ may be considered as complete dechlorination products (CDCPs).

As shown in Figure 6, for CF as a daughter product from CT, the amount decreased significantly after 10 min in 2N₄F@MWCNTs and 4N₄F@rGO catalytic system; however, DCM was hardly further reduced by these two catalysts within 2 h through independent experiments, so the generated CDCPs were attributed to the reduction products of CF. It is noteworthy that CDCPs were not detected in the nZVI catalytic system during the experimental period, which means Ni as a co-catalyst could promote the generation of CDCPs. Besides, benefiting from the high reactivity of 2N₄F@MWCNTs and 4N₄F@rGO, considerable amounts of CDCPs were produced in 2 h.

Table 3 reveals that all CT were reduced through the hydrogenolysis pathway to CF and DCM by these four catalysts. However, the presence of nickel not only accelerated the reduction of CT to CF, but also promoted the complete reduction pathway of CF. Moreover, N₄F obtained more reactive sites due to immobilization on rGO, greatly increasing the reaction rate of CT reduction, including reduction of CT to CF (k₁) and reduction of CF (k₂+k₃). As a result, rGO could support more N₄F NPs, 4N₄F@rGO has the capability of reducing more CF to CDCPs (k₃).

In this research, 4N₄F@rGO has been successfully synthesized by a two-step method and its removal capacity and CT reduction rate was assessed and compared with three other laboratory-synthesized catalysts. The results suggested that 4N₄F@rGO can degrade CT with a high reaction rate and selectivity towards non-chlorinated products. Furthermore, nickel metal reduced the active energy and accelerated the complete dechlorination reaction by the H-radical produced, and the Ni/Fe bimetal gained more reactive sites by loading on rGO to further accelerate the reduction of CT to CDCPs.

Based on the kinds of degradation products of CT and its concentration variation in reduction experiments, it was inferred that the reduction process of CT was initiated by dissociative electron transfer based on carbon-centered radicals. Meanwhile, the mechanism of CT reduction by 4N₄F@rGO was also discussed, it was concluded that rGO as an excellent electroconductive supporter could not only disperse the bimetal but also participated in the galvanic interaction with Fe⁰ to reduce CT indirectly.

The catalytic activities of two catalysts (4N₄F@rGO and 2N₄F@MWCNTs) with different supporters were compared. It was found that rGO could load more Ni/Fe and provide a better fluidity for substrate, which was conducive to reduce more intermediate products to CDCPs.

This work was supported by the National Natural Science Foundation of China (50578151), the National Science and Technology Major Project of China (2009ZX07207-008, 2009ZX07419-002, 2009ZX07207-001), the Beijing Municipal Education Commission School-Enterprise Cooperation Projects (51900265005), portable, in car, on-line monitoring instrument development and demonstration for focusing on prevention and control of heavy metals like mercury, chromium, lead, cadmium, arsenic (2012YQ060115), the Fundamental Research Funds for the Central Universities (2652013101, 2652013086, 2652013087), and the Key Project of Air Pollution Causes and Control Technology Research (2016YFC0209205).

All relevant data are included in the paper or its Supplementary Information.