Removal of Cr(VI) and methyl orange by activated carbon fiber supported nanoscale zero-valent iron in a continuous fixed bed column

Cr(VI) is a common toxic heavy metal ion in wastewater, which is widely used in mining, electroplating, leather tanning, pigment, etc. (Xu et al. 2014; Shang et al. 2014). Cr(VI) may exist in the aqueous solution in three different ionic forms (⁠⁠, or ⁠), the distribution of which depends on solution pH (Kim et al. 2012). It is harmful to most living organisms and difficult to biodegrade. Chromium compounds have been classified as human carcinogens by the International Agency for Research on Cancer (Chen et al. 2018). Among various dyes, methyl orange (MO) is a typical water-soluble anionic azo dye being widely used in various industries such as dyeing, textile, paper, and leather (Subbaiah & Kim 2016; Yan et al. 2016). Azo dyes bring serious harm to the water environment and human health due to their dark color, high toxicity (carcinogenicity, teratogenicity, etc.), and stable complex aromatic molecular structure (Martínez-Huitle & Brillas 2009; Wang et al. 2015; Sakir & Onses 2019). Therefore, effective removal of Cr(VI) and azo dye from wastewater is a hot topic in the field of water purification.

Various physical and chemical methods have been developed to treat heavy metal ions and azo dyes. The optimization of treatment materials has always been the focus of improving the effectiveness of wastewater treatment. Nanoscale zero-valent iron (nZVI) is often used to remove contaminants from wastewater because of its strong reducibility with the redox potential of –0.44 V (Fan et al. 2019) and surface activity (Li et al. 2019). For example, the nZVI can reduce Cr(VI) to low-toxic insoluble Cr(III); meanwhile, the nZVI is oxidized to Fe(II) or Fe(III) (Dong et al. 2016). However, nZVI is usually aggregated because of its high surface energy and intrinsic magnetic interactions (Diao et al. 2016), which will lead to the decrease of specific surface area and the reduction of reactivity. Immobilization of nZVI onto supporting materials is an effective method to solve these problems (Sheng et al. 2016). Catalytic wet hydrogen peroxide oxidation (CWPO) is a simple and efficient oxidation process, which stimulates hydrogen peroxide to produce hydroxyl radicals and then degrade azo dyes using transition metals (such as Fe(II) or Fe(III)) as catalyst (Fenton-like reactions) (Ramirez et al. 2007; Herney-Ramireza et al. 2011). However, it also has some drawbacks, such as difficult recovery of catalyst, easy leaching of metal ions and low reusability (Neamţu et al. 2004; Quintanilla et al. 2007; Yan et al. 2014). To overcome these disadvantages, developing heterogeneous catalysts (namely, transition metal supported on porous materials) has attracted more and more attention in recent years. Activated carbon fiber (ACF) is often used as an adsorbent or catalyst support for the removal of heavy metals and azo dyes because of its developed surface area, large adsorption capacity and high adsorption/desorption rate (Zhang et al. 2015; Carraro et al. 2019; Liu et al. 2019).

Effective combination of nZVI and ACF not only makes use of the high adsorption capacity of ACF and high reduction capacity of nZVI, but also overcomes the disadvantage of aggregation of nZVI. The immobilization of the nZVI on ACF has been used to remove Cr(VI). Huang et al. (2014) investigated the removal performance of activated carbon fiber felt (ACFF), nZVI and ACFF-nZVI by batch experiments, and the results showed the removal ratio of Cr(VI) was 67.0% for ACFF-nZVI, which was higher than that for ACFF (52.6%) and nZVI (59.4%). Qu et al. (2017) studied the removal of Cr(VI) from groundwater using ACF-nZVI composite in a permeable reactive column, and the results showed that ACF-nZVI still kept a Cr(VI) removal rate of more than 90% after five cycles. However, according to the literature, most of the studies were carried out in batch reactors, and there were few reports in a continuous fixed bed reactor. Moreover, according to the mechanism that nZVI can be oxidized to Fe(II) or Fe(III) (Fenton-like catalyst) by Cr(VI), there is no literature on CWPO of MO using ACF-nZVI after reacting with Cr(VI). The objective of this work is to investigate the removal performance of Cr(VI) in a continuous fixed bed with ACF-nZVI and the degradation effect of MO in the same fixed bed. The removal kinetics of Cr(VI) are examined by both Yoon–Nelson and bed depth service time (BDST) models.

Activated carbon fiber felts were purchased from Jiangsu Tongkang Special Activated Carbon Fiber & Fabric Co., Ltd in China and cut into small circles (2 cm diameter). All of the chemicals used were analytical grade.

The morphology of ACF and ACF-nZVI samples was characterized by a scanning electron microscope (SEM, Zeiss EVO10, Germany).

The composition of the ACF-nZVI was determined by X-ray diffraction (XRD, Bruker D8 Advance, Germany) using Cu Ka irradiation at 40 kV and 40 mA with 2θ range from 5° to 90°.

The valence state of iron on the surface of the ACF-nZVI was analyzed by an X-ray photoelectron spectrometer (XPS, Thermo Escalab 250Xi, USA) using an Al Ka radiation source.

Kinetics experiments of Cr(VI) removal were conducted in a continuous fixed bed made of a stainless steel tube (inner diameter = 2 cm, length = 8 cm) at 30 °C and atmospheric pressure.

To obtain a uniform distribution of the inlet Cr(VI) solution inside the fixed bed and ensure the desired temperature, the stainless steel tube was filled with inert glass beads (diameter = 2–3 mm) at both ends and ACF-nZVI (or ACF) in the middle. Cr(VI) solution was carried to the stainless steel reactor by a BT101 L peristaltic pump in up-flow mode, and temperature of the reactor was controlled by a water bath (Figure 1). In the feed tank, the pH of Cr(VI) solution was adjusted with nitric acid. The effects of initial pH value of Cr(VI) solution (2, 3 and 4), bed height (0.8, 1.2 and 1.6 cm) and inlet Cr(VI) concentration (100, 200 and 300 mg/L) on the removal of Cr(VI) were studied. The feed flow rate was fixed at 2 mL/min. The absorbance at the inlet and outlet of the solution was determined by 1,5-diphenylcarbazide spectrophotometry (GB7467-87 1987) at 540 nm using a V5100 visible spectrophotometer. The experimental results are shown in the form of breakthrough curves, i.e., the variations of C/C₀ (outlet concentration/inlet concentration) versus time.

Iron leaching concentration (mg/L) in the outlet solution was measured by using a TAS-990AFG atomic absorption spectrophotometer.

Heterogeneous surface morphology plays an important role in particle adsorption (Cui et al. 2017; Fang et al. 2017). The morphology of ACF and ACF-nZVI samples was observed by SEM, and the images are shown in Figure 2(a). It can be seen that the SEM image of ACF was obviously different from that of ACF-nZVI. The surface of ACF was smooth, but the surface of ACF-nZVI became rough and nZVI particles were well distributed on the surface of ACF.

The component of ACF-nZVI was analyzed by XRD and the result is shown in Figure 2(b). The XRD patterns showed an apparent characteristic peak of carbon at 2θ of 25°, which suggested that the supports were ACFs. The diffraction peak observed at 2θ of 44.5° could be attributed to the existence of zero-valent iron. It indicates that the nZVI particles were successfully supported by the ACFs (Huang et al. 2014; Liu et al. 2014; Zhang et al. 2019).

The valence state of iron on the surface of the ACF-nZVI was analyzed by XPS and the result is shown in Figure 2(c). There was no obvious characteristic peak of Fe⁰, and the characteristic photoelectron peaks at 711.2, 712.59, 714.5, 716.95 and 720.44 eV (Fe1–Fe5) corresponded to the binding energies of Fe 2p_3/2, while 724.07, 725.99, 728.23, 730.54 and 733.78 eV (Fe1′–Fe5′) represented the binding energies of Fe 2p_1/2, which suggests the existence of different iron oxide species such as Fe²⁺ and Fe³⁺ (Fu et al. 2015). Combined with XRD analysis, these results show that the core of the nZVI loaded on the ACF is Fe⁰, and a layer of iron oxides definitely exists on the surface of the Fe⁰. Liu et al. (2014) prepared graphene supported nZVI, and the XPS results showed that oxidation of iron occurred all over the surface of nZVI and little Fe⁰ was left. Similarly, a very weak shoulder peak corresponding to the 2p_3/2 binding energy of Fe⁰ was barely observable due to the high surface sensitivity of XPS (less than 10 nm in depth) for the kaolinite supported nZVI prepared by Zhang et al. (2011).

To study the effect of bed height on the removal performance of Cr(VI), the breakthrough curves of Cr(VI) in the fixed bed with different bed heights (0.8, 1.2, 1.6 cm) were measured at a constant flow rate of 2 mL/min, pH 2 and inlet concentration of 100 mg/L. The results are presented in Figure 4(a). It can be seen that the breakthrough curves at different bed heights showed the same characteristic shapes, but the Cr(VI) breakthrough time was prolonged with the increase of bed height. When the bed height was 0.8, 1.2 and 1.6 cm, the breakthrough concentration reached 2.7%, 2.6% and 5.5% of the inlet concentration at 60, 120, and 250 minutes, respectively. This could be attributed to the fact that the active sites (adsorption and reduction sites) increase with increasing bed height (Chatterjee et al. 2018), i.e., Cr(VI) has more opportunities to be adsorbed by ACF and reacted with nZVI. Therefore, increasing bed height usually results in an increase in breakthrough time.

The World Health Organization recommends the maximum allowable discharge concentration of Cr(VI) into drinking water is 0.05 ppm; however, its concentration in industrial wastewater usually ranges from 5 to 220 mg/L (Ghosh 2009; Mohamed et al. 2016). To investigate the effect of inlet concentration on the removal performance of Cr(VI), the breakthrough curves of Cr(VI) in the fixed bed were measured at different inlet concentrations (100, 200 and 300 mg/L), at a constant flow rate of 2 mL/min, pH 2 and bed height of 1.6 cm. The results are shown in Figure 4(b). As was evident from Figure 4(b), the breakthrough time of Cr(VI) decreased with increase in inlet concentration. When the inlet concentration was 100, 200 and 300 mg/L, the breakthrough concentration reached 11.4%, 11.2% and 11.5% of the inlet concentration at 270, 110, and 60 minutes, respectively. With the increase of inlet concentration, the treated Cr(VI) amount per unit time increases. However, the total number of binding sites available at the same bed height is constant, which may be inadequate at higher inlet concentration (Mishra et al. 2016; Basu et al. 2019). Thus, a faster adsorption saturation of ACF-nZVI in the fixed bed occurred and the breakthrough time was shortened with increasing inlet concentration. Vilvanathan & Shanthakumar (2017) studied continuous adsorption behavior of cobalt ions onto Chrysanthemum indicum in fixed bed columns, and the results showed early breakthrough and quick saturation were achieved at higher ion concentration. Fan et al. (2019) reported that higher initial Cr(VI) concentration led to faster breakthrough of the fixed bed filled with nZVI-biochar adsorbent.

As shown in Figure 5 and Table 1, the experimental data fitted well to the Yoon–Nelson equation, with R² > 0.8. The k_YN values increased with increasing inlet concentration from 100 to 300 mg/L. This may be attributed to an increase in mass transfer coefficients between Cr(VI) solution and ACF-nZVI under higher concentration gradients, and thus the adsorption and reduction rate increases. The adsorbed or reduced Cr(VI) amount per unit time increases with increasing inlet concentration, leading to an earlier breakthrough and saturation. This further explains that the breakthrough time decreased with the increase of inlet concentration. The 50% breakthrough time (τ) increased with decreasing pH or inlet concentration. With decreasing bed height, the τ value decreased. Chen et al. (2012) reported similar fitting results in their literature. The data in Table 1 also show that τ values fitted by the Yoon–Nelson model were close to the experimental data (Figures 3 and 4), suggesting that the Yoon–Nelson model could predict the 50% breakthrough time of Cr(VI) at different pHs, bed heights and initial concentrations without requiring further experiments.

The linear regression analysis of breakthrough curves in the C/C₀ region above 0.05 was carried out by Equation (10). The values of k_AB and N₀ were obtained from the slope and intercept of the time (t) versus ln[(C₀/C) − 1] plot (Figure 6). These parameter values and the R² values are presented in Table 2. For both the ACF and ACF-nZVI at a constant pH 2, the removal capacities (N₀) were 9.9 and 20.1 g/L, respectively. ACF-nZVI showed a Cr(VI) removal capacity two times higher than that of ACF. The values of N₀ increased with decreasing pH and initial Cr(VI) concentration but decreased with decreasing bed height. In addition, the BDST model could predict the breakthrough time (Chatterjee et al. 2018; Du et al. 2018). For example, for ACF-nZVI at pH 2, bed height of 1.6 cm and inlet concentration of 100 mg/L, the predicted and experimental breakthrough time for 11.4% breakthrough was 268 and 270 min, respectively.

To study the stability and reusability of the ACF-nZVI catalysts, three cycle experiments were carried out in the same fixed bed reactor with the EBCT of 1.256 min. After each experiment, the catalysts were recovered, washed and dried in the oven. It can be seen from Figure 8(a) that the conversion ratio of MO remained largely unchanged after three times of reuse. When the catalysts were used for the first time, the second time and the third time, the conversion ratio of MO was 99.2%, 98.3% and 99.2% respectively after 240 min of continuous operation of the fixed bed. From Figure 8(b), it can be found that the iron leaching concentration in the outlet solution fluctuated after the catalysts were reused three times, but the maximum iron leaching concentration was lower than 5.2 mg/L (<5 mg/L after first use). After three cycles, the conversion ratio of COD decreased from 55.7% to 45.3%. Iron leaching is an important factor affecting the deactivation of Fenton-like catalysts (Luca et al. 2018). Hence, low iron leaching concentration leads to high catalyst stability.

In order to further compare the adsorption and catalytic effect of catalyst on MO, ACF and ACF-nZVI after reacting with Cr(VI) solution were used for adsorption experiments. The experimental conditions were the same as those for catalytic wet oxidation of MO, i.e., temperature of 70 °C, EBCT of 1.256 min, pH 2, and inlet concentration of 50 mg/L. The breakthrough curve and adsorption rate are shown in Figure 9. It can be seen from Figure 9 that MO could be adsorbed by unused ACF and the breakthrough concentration of MO was lower than 0.05 in 340 min. The breakthrough concentration of MO reached 0.073, i.e., the adsorption ratio of MO was 92.7% at 360 min. After that, the adsorption rate of MO decreased rapidly and the adsorption ratio was only 4.9% at 480 min. The ACF-nZVI after reacting with Cr(VI) solution had little adsorption effect on MO and the concentration of outlet solution was basically the same as that of inlet solution. However, when ACF-nZVI, after reacting with Cr(VI) solution, was reused for CWPO of MO three times, i.e., 720 min continuously, the conversion ratio of MO remained 99.2%. The conversion ratio of MO remained largely unchanged within the experimental period.

In this study, the application of ACF-nZVI for the continuous removal of Cr(VI) and MO from aqueous solution was fully investigated. The results are as follows:

• (1)

  ACF-nZVI has a better performance for Cr(VI) removal compared with the ACF alone in a fixed bed. The breakthrough concentration reached 6.9% of the inlet concentration at 150 min for ACF, and the breakthrough concentration reached 5.5% at 250 min for ACF-nZVI. The removal of Cr(VI) in the fixed bed was highly dependent on the pH, bed height and inlet concentration. The breakthrough time decreased with the decrease of bed height but increased with the decrease of pH and inlet concentration.

• (2)

  The Yoon–Nelson model was found to fit well with the experimental results. The rate constant (k_YN) values increased with increasing inlet concentration. The 50% breakthrough time (τ) increased with decreasing pH or inlet concentration, but decreased with decreasing bed height. The τ values fitted by the Yoon–Nelson model were close to the experimental data.

• (3)

  The BDST model could predict the breakthrough time and adsorption capacity (N₀). The adsorption capacity of the ACF-nZVI was two times higher than that of ACF. The values of N₀ increased with the decrease of pH and initial Cr(VI) concentration but decreased with the decrease of bed height.

• (4)

  With the decrease of EBCT from 2.512 to 0.628 min, the conversion ratio of MO remained unchanged at first and then decreased sharply. After three successive runs (namely 720 min), the fixed bed with ACF-nZVI still retained high MO conversion (99.2%), but the adsorption ratio was only 4.9% at 480 min for ACF alone.

• (5)

  Compared with the batch reactor, the continuous fixed bed reactor saves the non-reactive time such as feeding and unloading. However, the high flow rate leads to the short contact time between the ACF-nZVI and Cr(VI)/MO, which causes the low utilization rate of ACF-nZVI. How to improve the reaction efficiency of ACF-nZVI and increase the Cr(VI)/MO removal capacity requires future research.

This work was supported by the National Natural Science Foundation of China (No. 41773133), Scientific Research Fund of Hunan Provincial Education Department, China (No. 17A028), Natural Science Foundation of Hunan Province, China (No. 2019JJ50013), and the Science and Technology Development Project of Hengyang City, China (No. 2019yj011171).

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