Paraquat is the most important herbicide of the bipyridyl group. The aim of the present study was to compare the removal of paraquat herbicide from aqueous solutions using nanoscale zero-valent iron-pumice/diatomite composites. In this study, nZVI was supported with diatomite and pumice. Scanning electron microscopy (SEM) analysis, X-ray diffraction spectroscopy (XRD), Fourier transform infrared spectrometry (FTIR), and specific surface area tests (BET) were used to evaluate the properties of nanoadsorbents. The residual concentration of paraquat in aqueous solution was detected by high-performance liquid chromatography (HPLC). Then, the effects of different variables including the pollutant concentration, contact time, temperature, adsorbents (D-nZVI and P-nZVI) dose, and pH, were investigated in a lab scale batch system. The results showed that the optimal pH for both processes was 3.74. In optimal conditions, the efficiencies of D-nZVI and P-nZVI were 92.76% and 85.28%, respectively. In addition, isotherm and adsorption kinetics studies indicated that P-nZVI follows the Langmuir and Freundlich isotherm models, and D-nZVI follows the Langmuir isotherm model, and both processes follow pseudo-second-order kinetics. The results indicated that the synthesized nanoparticles were suitable for removing paraquat from aqueous solutions. Both adsorbents were found to be very effective in removing similar compounds at ambient temperature in a short time.

1. Introduction

Currently, in many countries, it is observed that the use of pesticides for agriculture causes problems such as water, soil, and air pollution [1]. Among pesticides, herbicides are widely used in agriculture. Their excessive persistence can state serious risks to human health, aquatic life, and animals [2]. Paraquat is one of the most common herbicides, especially in developing countries. Paraquat is a nonselective herbicide that belongs to the category of bipyridines and is used as a moisture absorber in products such as potatoes, sugarcane, rice, coffee, beans, and to kill weeds [3]. The characteristics of paraquat are listed in Table S1.

The use of this compound may present potential environmental hazards such as respiratory poisoning, hypersensitivity, gastrointestinal obstruction, neurological problems, pulmonary fibrosis, and Parkinson’s disease for humans and animals [46]. The severity of paraquat poisoning depends on many factors and conditions, such as the duration of exposure, the route, and the amount of poison [7]. To protect human health and the environment from the harmful effects of paraquat herbicide, advanced and practical treatment processes are necessary to remove it from the water [8]. Common purification processes for removing paraquat generally include reductive processes such as adsorption on porous solids, membrane filtration, a combination of adsorption and membrane filtration [9], and advanced oxidation processes (AOPs) [10, 11].

Adsorption is the simplest, most economical and most effective method for wastewater treatment among all of the processes [12]. However, more effective materials development is very important to eliminate the effects of paraquat from aqueous solutions. Nanotechnology is one of the fastest growing sectors of the global economy. Based on the theoretical and empirical evidence and studies, zero-valent iron nanoparticles (nZVI) are widely used to remove pesticides, heavy metals, antibiotics, paints, and other contaminants from water due to their specific surface area, environmental safety, catalytic reduction capability, low cost, and easy to use [13, 14].

Zero-valent iron nanoparticles, due to their magnetic interaction and high surface energy, they tend to accumulate in their particles alone, which reduces the specific surface area of nZVI secondary iron contamination, separation and recovery problems after use, and its removal effects are restricted [15, 16]. To prevent the accumulation of nZVI, a variety of materials are produced and used as carriers such as clay, grapheme, carbon, bentonite, and silica [17, 18]. In addition, the porous structure of these compounds leads to high hydraulic conductivity which increases the rate of degradation in aqueous systems [17, 19, 20].

Diatomite is a white, soft, light stone that has a unique natural crystalline structure and is composed of a fossilized diatom skeleton [21]. The highly porous structure of diatomite, low density, high permeability, small particle size, high adsorption capacity, high specific surface area, and its thermal and mechanical stability and hydraulic conductivity have led to its use as a filter, catalyst carrier, and effluent treatment [22]. Diatomite may be used as a carrier for nZVI due to its macroporous structure and excellent thermal and mechanical resistance [23].

Pumice is a porous volcanic rock that has a skeletal structure and a large surface. Pumice has open channels that allow water and ions to travel inside and outside the crystal structure. In addition, pumice can be easily processed. Pumice has been used in water treatment as an inexpensive adsorbent and filter. Pumice can stabilize more nZVI, so it is a better alternative than other supporters [24].

In this context, the present work aimed to prepare, characterize, and apply nZVI-pumice and nZVI-diatomite composites for removing paraquat from aqueous solutions. Furthermore, adsorption kinetics and adsorption isotherms were investigated. To the best of our knowledge, this is the first study that used nZVI-pumice and nZVI-diatomite composites for paraquat adsorption.

2. Materials and Methods

2.1. Materials and Supplies

In this study, iron (II) chloride tetrahydrate (≥99%), sodium borohydride (99%), acetonitrile (98%), and ethanol (99.8%) were purchased from Merck Company (Germany). Paraquat (≥99%), diatomite (99%), and pumice (99%) were purchased from Sigma-Aldrich Company (USA).

2.2. Preparation of nZVI and Diatomite (D-nZVI)

5 g of iron (II) chloride tetrahydrate (FeCl2·4H2O) was added to a mixture of ethanol and deionized water (75 mL of ethanol plus 25 mL of water). Then, 5 g of diatomite was added to this solution. Then, sodium borohydride (NaBH4) was added dropwise under a hood with continuous stirring. After adding all of the borohydride solution, the mixture was left stirring for an additional 10 min. The reduction of iron ions by borohydride ions can be represented according to equation (1).

After the color change of the solution, the solids were separated from the solution by vacuum filtration, washed with ethanol, and dried in an oven at 50°C [25].

2.3. Preparation of nZVI and Pumice (P-nZVI)

1.5 g of pumice was used as a support material. Then, 5 g of FeCl2·4H2O was added to a mixture of ethanol and water (75 mL of ethanol plus 25 mL of water). Thereafter, NaBH4 was added dropwise to the solution under a hood with constant stirring. After the color change of the solution, the solids were separated from the solution by vacuum filtration, washed with ethanol, and dried in an oven at 50°C [26].

2.4. Nanoparticle Characterization

An X-ray diffraction (XRD, PMD Philips X’Pert) analysis was used to identify the crystal structure of the synthesized nanosorbents. The existing functional groups on the prepared nanostructures were evaluated by Fourier transform analysis (FTIR, Bruker, Germany). Adsorbent surface morphology with a field emission scanning electron microscope (FESEM, ZEISS, model Sigma, Germany), elemental analysis of adsorbents by BET analysis (JW-BK132F) was used to measure the surface area of the produced materials.

2.5. Batch Studies

Batch experiments were performed in the Erlenmeyer with 250 mL of aqueous solution. The effects of pH (2–8), adsorbent dose (0.5–2 g·L−1), reaction time (10–100 min), paraquat concentration (0.5–4 mg·L−1), and temperature (15–45°C) on paraquat removal using D-nZVI and P-n-ZVI were investigated. The solutions were shaken in a temperature-controlled shaker (OS-300). Sampling was made at a certain time interval (10–100 min) and then, the samples were filtered with a syringe filter (0.02 μm pore size, Whatman Anotop 10), and stored at 4°C when needed, and analyzed within 48 h of collection [26]. All experiments were performed in triplicates.

2.6. Analytical Methods

Paraquat concentration in the sample was measured by reverse phase high-performance liquid chromatography (HPLC Agilent 1200) using a C-18 column (ODS HYPERSIL, 5 μm, 4.6 × 250 mm, Agilent Technologies) with a mobile phase of water and acetonitrile (ACN: water 75 : 25 for paraquat analysis; HPLC method details: injection volume, V = 10 μL; paraquat RT = 1.63 min). Standard calibration methods were followed. In equation (2), C0 and Ct are the initial and final concentrations of the contaminant in mg L−1; before and after batch experiments, respectively [27, 28].

2.7. Experimental Design

One of the response surface methodology (RSM) approaches, that is widely used‏ ‏in research, is the central composite design (CCD) [29]. In this study, RSM based on the CCD was used to evaluate the interaction effect on the response function (paraquat removal efficiency) and predict the best response rate. In order to design experiments with a CCD, minimal and maximum levels were given to design expert software, and the other levels were determined between the maximum and minimum points by the software. Including three repetitions, 150 experiments were performed for each method (D-nZVI and P-n-ZVI). A total of 300 experiments were performed.

3. Results and Discussion

3.1. Nanoparticle Characterization

According to Figure 1(a) in the D-nZVI sample, there are particles with a spherical morphology throughout the material, the size of which is about 80 to 100 nanometers. It is also known that the structure of this material is relatively dense and has fewer pores than the P-nZVI sample. Figure 1(b) shows that the morphology of the particles is pseudospherical, but this time the particle size is slightly smaller (between 60 and 80 nm), and the particles are chained together to form strands in microscopic images.

To investigate the particle size distribution of these two samples, Image J software was used, and the sizes of 100 different particles in microscopic images were measured at the highest magnification, and the histogram related to this measurement is shown in Figure 2. As seen, about 40% of the visible particles in the D-nZVI sample have a size in the range of 80 to 100 nm and about 65% of the visible particles in the P-nZVI sample have a size in the range of 60 to 90 nm, which indicates a smaller particle size than the first sample. The statistical parameters obtained from this particle size distribution are reported in Table S2.

The values of the standard deviation in Table S2 were obtained from equation (3). In equation (3), Xi is the size of each particle, and is the average value for all of the particle sizes.

According to Table S2, the average particle size in the P-nZVI sample was about 78 nm which was much smaller than the D-nZVI sample which had an average particle size of about 92 nm. Also, the amount of the smallest and largest particles for the P-nZVI sample compared to the D-nZVI sample indicates that the p-nZVI composite is finer than the other sample. Figure 3 also shows the amount of porosity in the structure of these two materials.

According to Figure 3, the percentage of porosity in the D-nZVI sample was 11.69%, while this percentage was 14.37% in the P-nZVI sample, which indicates a higher density of the D-nZVI than the P-nZVI sample. FT-IR and XRD tests were used to investigate the chemical nature and crystal structure of these two composites. The results of the FT-IR test are shown in Figure 4.

Figure 4(a) shows the peaks visible at 473 cm−1, 793 cm−1, and 1088 cm−1 in similar articles as characteristic carbonate ion peaks in the molecular structure of calcium carbonate [30, 31]. The peaks observed at 1088 cm−1 and 793 cm−1 are related to the symmetrical tensile vibration of carbonate ion bonds and the inplane flexural vibration of ions [30, 32], and the tensile vibration of Ca-O bonds in the wave −473 cm3 showed absorption peaks [31]. In addition, the absorption peak visible at 1384 cm−1 is related to the flexural vibration of C-H bonds, and the absorption peaks visible at the wave numbers of 2852 cm−1 and 2923 cm−1, respectively, are symmetric and asymmetric tensile vibrations of the same bonds in the structure of organic compounds [33]. Also, the flexural vibration and tensile vibration of the adsorbed water-related bonds in this composite have shown the absorption peak in the wave numbers of 1630 cm−1 and 3425 cm−1, respectively [34]. Therefore, based on the results, it is clear that the predominant chemical structure in this sample is calcium carbonate, and there are some organic compounds in this sample.

In the spectrum of the P-nZVI sample (Figure 4(b)), the peaks related to the symmetric and asymmetric tensile vibrations of the O-Si-O bonds are shown in the wave numbers of 873 cm−1 and 1020 cm−1, respectively [35, 36]. In addition, the absorption peak appearing at 1450 cm−1 is related to the flexural vibration of N-H bonds, and the absorption peaks visible in the wave numbers of 2853 cm−1 and 2923 cm−1, respectively, are the symmetric and asymmetric tensile vibrations of C-H bonds in the structure of existing organic compounds [33]. Also, the flexural vibration and the tensile vibration of the bonds related to the adsorbed water in this composite have shown the absorption peak in the wave numbers of 1629 cm−1 and 3423 cm−1, respectively [34]. Therefore, it is clear that the predominant chemical structure in this sample is silicon oxide, and some organic compounds are also present in this sample. It should be noted that due to the fact that only chemical bonds can be detected in FT-IR spectra and zero-valent iron has no bond with other elements in the composite, the absence of this element in the FT-IR test results can be justified.

An XRD test was used to evaluate the crystal structure of the samples. In this test, by irradiating X-ray onto the crystal structure of a monocrystalline or polycrystalline material, while entering the X-ray into the crystalline structure of this material, after colliding with different crystal plates and reflecting at a certain angle of reflection, the courier shows. By putting these peaks together and drawing the X-ray diffraction pattern, we can understand the crystalline phases in the material under study [37]. X-ray diffraction patterns are shown in Figure 5. According to the diffraction pattern shown in Figure 5(a) and by adapting this diffraction pattern to the reference patterns by X’Pert software, two calcium carbonate materials (with rhombohedral crystal structure and reference code: JCPDS No: 01-085-0849) and zero-valent iron (with cubic crystal structure and reference code: JCPDS No: 00-001-1267) have been identified. The presence of calcium carbonate in this sample confirms the results of the FT-IR test. The diffraction plate corresponding to each peak is marked on the same peak. In order to obtain the crystal size of each of these phases, equation (4), which is called the Debbie–Scherer relationship, has been used.

In this equation, λ is the X-ray wavelength used (here 1.54 Å), K is the shape factor (approximately equal to 0.9), B is the peak width at half the height, and θ is the peak location. According to the equation, the crystal sizes for the calcium carbonate and zero-valent iron phases are 37.3 and 6.3 nm, respectively. According to the diffraction pattern shown in Figure 5(b) and by adapting this diffraction pattern to the reference patterns, two silicon oxide materials (with tetragonal crystal structure and reference code: JCPDS No: 01-076-0935) and zero-valent iron (with cubic crystal structure and reference code: JCPDS No: 00-001-1267) have been identified. The presence of silicon oxide in this sample confirms the results of the FT-IR test. In the diffraction pattern, the diffraction plate of each peak is specified on the same peak. Also, the values of crystal size obtained by the equation (4) for the phase of silicon oxide and zero-valent iron are 20.7 and 6.5 nm, respectively.

The specific surface area was measured following the Brunauer–Emmett–Teller (BET) N2 method. The BET for D-nZVI and P-nZVI were 59.682 and 40.189 m2g−1, respectively.

3.2. pH Effect

The results of the effect of pH on the removal efficiency of paraquat on adsorbents in Figure 6 indicate that the maximum removal rate was obtained at an acidic pH of 3.74. The highest removal rates for D-nZVI and P-nZVI were 95.14% and 92.07%, respectively. The optimum pH in both processes is 3.74. The pH solution is one of the important factors in controlling the degradation effect on chlorinated pollutants (equation (5)) [38].

The reaction between iron and chlorinated organic compounds occurs according to the equation [39]. As the pH of the solution decreased, the shell of nZVI particles (oxide and hydroxide coatings) separated under acidic conditions, and paraquat access to Fe0 was increased. Under higher pH values, the activity of iron oxide and hydroxide coating on the surface undoubtedly prevents access to the Fe0 level. However, with a further decrease in the pH of the solution, the removal efficiency decreases rapidly, which may be due to greater Fe0 ionization under strong acid conditions [39]. Also, at acidic pH, the soluble part of the iron ion and the hydroxyl radical oxidation power are higher. In a study carried out by Liu et al. to remove heavy metals by pumice with zero-valent iron nanoparticles with pumice, the removal efficiency increased with increasing pH [24]. In a study conducted by Zhang et al. to remove chromium (VI) by diatomite with iron nanoparticles, the best removal efficiency occurred at pH = 2–5 and pH = 3 was selected as the optimum [40].

3.3. Absorbent Effect

The effect of the adsorbent on the removal efficiency of two processes D-nZVI and P-nZVI in 4 concentrations is shown in Figure 7. As shown in D-nZVI, the removal efficiency increases with increasing concentration first and then decreases, in fact, the optimal adsorbent concentration for D-nZVI was 1.42 g·L−1. However, in P-nZVI, the removal rate increased with increasing adsorbent concentration, so that at a concentration of 2 g·L−1, it reached 95.75% efficiency.

The amount of adsorbent is another important factor in the removal process. Based on our results, the removal rate for P-nZVI adsorbent concentration was 95.75% at a concentration of 2 g·L−1 and for D-nZVI, 96.25% at a concentration of 1.42 g·L−1. The results suggest that possible mechanisms for the removal of paraquat from aqueous solutions include the uptake of paraquat by P-nZVI and the reduction of paraquat through the oxidation of Fe0 to Fe3. As well as, the results show that the higher amount of adsorbent and consequently the higher specific surface area lead to the higher removal efficiency. However, for D-nZVI, due to the higher number of available and active sites of the adsorbents, the removal efficiency was higher at first, and then it gradually decreased. It may be related to the occupation of the pores with adsorbate molecules [41]. The presented results are consistent with the results of the Gluer study to remove tetracycline by zero-valent iron nanoparticles with pumice [26]. Also, a similar study carried out by Harman and Genisoglu in 2016 on copper removal by P-nZVI yielded similar results [41]. In a study conducted by Dehestaniathar et al. on the uptake of fluoride by iron oxide nanoparticles stabilized on diatomite, the uptake increased with increasing adsorption [42].

3.4. Effect of Initial Paraquat Concentration

The effect of initial paraquat concentration on the removal efficiency of the two processes is shown in Figure 8. As observed, the highest removal efficiencies (93.27% for P-nZVI and 96.94% for D-nZVI) are related to the concentration of 0.58 mg·L−1 paraquat. In both processes, the removal rate decreased as the initial paraquat concentration increased.

Increasing the paraquat concentration slightly reduces the adsorption efficiency because with increasing the toxin concentration, the amount of paraquat molecules in aqueous media increases and the interference of the intermediate compounds resulting from the initial decomposition increases as a result of competition between molecules over the adsorbent. Find and decrease removal efficiency, in most studies, the initial concentration of the contaminant is constant. In a study conducted by Shokoohi et al. to investigate the removal of phenol by pumice-coated iron from aqueous solutions, the efficiency first increased and then decreased with increasing contaminant concentration due to a decrease in adsorption capacity by the adsorbent [43]. Also, in another study conducted by Dehestaniathar et al. to adsorb fluoride by iron oxide nanoparticles stabilized on diatomite, the removal efficiency decreased with increasing fluoride concentration [42].

3.5. Time Effect

Figure 9 shows the effect of the contact time on the paraquat removal. In the P-nZVI process at the optimal contact time of 100 min, the removal efficiency was 95.27%. But in the D-nZVI process, after 74 min, the removal efficiency is almost constant.

One of the influential factors in the elimination process is reaction time. According to the results, the highest removal percentage (95.25%) was observed for P-nZVI at 100 min. The reason can be attributed to the increase in the amount of OH radical produced as a result of increasing the contact time. Also, the highest removal percentage for D-nZVI was 95.91%. In the D-nZVI process, diatomite prevents the accumulation of iron particles, resulting in increased efficiency, but after a while, due to the saturation of the adsorbent surface, the removal efficiency is almost constant [38, 44].

The study by Dhaouadi and Adhoum investigated the removal of paraquat herbicide by an advanced oxidation process. Their findings showed that the removal efficiency was increased by increasing the contact time [45]. In the study of Babaeifar et al. on the removal of arsenic by pumice-coated iron, except for the first 20 min, the removal efficiency remained constant [46]. In a study conducted by Liu et al. for the catalytic degradation of orange II from aqueous solutions using diatomite/nickel-reinforced iron-reinforced iron, the removal efficiency was increased first over time (by adding diatomite from the aggregation of Fe/Ni nanoparticles) and then remained constant [24]. In the study conducted by Zha et al. to catalytically reduce methyl blue by combining diatomite with ZVI, the removal rate was increased by increasing the contact time [47].

3.6. Temperature Effect

In present study, the effect of paraquat removal efficiency was investigated for both the P-nZVI and D-nZVI processes. According to Figure 10, at 23.69°C, the removal efficiencies were 91.96% and 95.02% for P-nZVI and D-nZVI, respectively.

According to the obtained results, the removal efficiency decreases with increasing temperature. There may be two possible explanations for this phenomenon: (1) paraquat mobility decreases with higher temperatures and (2) as the temperature increases, the activation energy of the decomposition reaction decreases [22, 47]. In a study conducted by Liu et al. to remove heavy metals by nanoparticles with pumice, the removal efficiency was increased with increasing temperature due to the structure of Hg [24]. In a study conducted by Zhang and Wang on the decomposition of methyl orange dye from aqueous solution using carbon nanoparticles with carbon, the efficiency was increased with temperature [48].

3.7. Adsorption Isotherms

In this study, two adsorption models of Langmuir and Freundlich were investigated. The results of the adsorption isotherms are listed in Figure 11 and the relative parameters calculated from the two models are listed in Table 1. For D-nZVI, the higher correlation coefficient of the Langmuir model indicates that the adsorption data are better fitted by the Langmuir model (R2 > 0.99) than the Ferundlich model (R2 > 0.93). This means that the adsorption involved the formation of a monolayer on a homogeneous surface. For the P-nZVI adsorbent, as shown in Figure 11, experimental data excellently fits both the Langmuir and Freundlich isotherm models. Thus, the adsorption process occurred both on homogeneous adsorbent surfaces and reversible heterogeneous surfaces in the adsorption sites. The maximum adsorption capacity (qm) of paraquat for D-nZVI and P-nZVI was 161.3 mg·g−1 and 169.49 mg·g−1, respectively. The results of D-nZVI are consistent with a study conducted by Froutan et al. [49]. The results of P-nZVI are also consistent with the results of a study conducted by Gular to remove tetracycline [26].

3.8. Adsorption Kinetics

To investigate kinetic studies, pseudo-first-order kinetics and pseudo-second-order kinetics were used. The results showed that both processes follow a pseudo-second-order model and show a correlation coefficient above 0.9 (Figure 12 and Table 2). These results showed that the velocity control phase in the adsorption process may be chemical and that paraquat adsorption probably occurs through Van der Waals forces or ion exchange between the adsorbent and paraquat. The results are consistent with the studies of Zhang and Wang [40].

3.9. Comparison of Adsorption Capacity of D-nZVI and P-nZVI Composites

The adsorption capacity of the D-nZVI and P-nZVI composites prepared in the present work was compared with those of various sorbents for removal of paraquat herbicide which are represented in Table 3. Generally, the comparison of the various adsorbents obtained with the others was not simple due to various experimental conditions. One of the important superiority of D-nZVI and P-nZVI composites compared with other sorbents was their high adsorption capacity, while some sorbents had a lower absorption capacity.

4. Conclusion

In the present study, the determination of the optimal values affecting the D-nZVI and P-nZVI processes in order to remove the paraquat herbicide was investigated, and the optimal process conditions were determined. Also, the properties of each adsorbent were determined according to SEM, FTIR, BET, and XDR analyses, which showed that diatomite and pumice were well fixed on iron nanoparticles and prevented their accumulation. The optimal values obtained in the P-nZVI process were 0.5 mg·L−1paraquat herbicide with 2 g·L−1 adsorbent and contact time was 100 min, and the optimal values obtained in the D-nZVI process were 0.5 mg·L−1paraquat herbicide with 1.42 g·L−1 adsorbent at a contact time of 100 min. The optimum pH in both processes was 3.74. Based on our results, it was found that under optimal conditions, P-nZVI and D-nZVI processes, paraquat herbicide can be removed 85.28% and 92.76%, respectively. The contact time was effective in the removal efficiency in both processes, so that as the contact time increased, the removal efficiency also increased. The maximum adsorption capacities for D-nZVI and P-nZVI are 161.3 mg·g−1 and 169.49 mg·g−1, respectively. The mechanism of the adsorption process in D-nZVI follows the Langmuir isotherm (R2 = 0.99) and in P-nZVI follows two models of Langmuir (R2 = 0.97) and the Freundlich isotherm (R2 = 0.98). Adsorption kinetics in both processes are pseudo-second-order models. It can be concluded that the D-nZVI process is more effective for paraquat removal due to its higher removal efficiency. These processes can be used to remove contaminants with similar structures due to their low cost, high adsorption capacity, easy operation, and workability in environmental conditions.

Data Availability

The data used to support the findings of this article are included within the article.

Conflicts of Interest

The authors declare that they have no conflicts of interest.


The authors thank the Shahrekord University of Medical Sciences (grant number: 3243) for supporting this study. The authors of this paper appreciate the authorities of Shahrekord University of Medical Sciences, who have fully involved in this research.

Supplementary Materials

Table S1: characteristics of paraquat. Table S2: statistical parameters obtained from measuring size of 100 particles in the studied samples. (Supplementary Materials)