Graphene oxide and thiol functionalized natural zeolite for the removal of lead from water

Lead is a relatively corrosion-resistant, dense, ductile and malleable metal that has been used by humans for a variety of applications for at least 5,000 years (Brown & Margolis 2012). During 1999–2004, the estimated average blood lead level was 1.9 μg/dL for the population younger than five years old in the United States, that is approximately 100 times higher than ancient background levels, indicating that substantial sources of lead exposure exist in the environment (Brown & Margolis 2012; Centers for Disease Control and Prevention 2013). Lead in drinking water is a major public concern (Deshommes et al. 2010), as it is known to damage the kidney, liver and reproductive system, basic cellular processes and is a potent neurotoxin (Inglezakis et al. 2007; Mah & Jalilehvand 2012; Hanna-Attisha et al. 2016; Taylor et al. 2017). Childhood lead poisioning has an impact on many developmental and biological processes, most notably intelligence, behavior, and overall life-achievements (Hanna-Attisha et al. 2016). Switching the drinking water source from Detroit's Lake Huron to Flint River water in April 2014 created perfect conditions for lead leaching into drinking water (Guyette 2015). The blood lead levels for children younger than five years before 2013 and after the 2015 water source change were reviewed. The incidence of elevated blood lead levels increased from 2.4% to 4.9% after the water source change (Hanna-Attisha et al. 2016). Similarly, the incidence of elevated blood lead for children aged ≤1.3 years in Washington, DC, increased more than four times within 2001–2003 when lead in water was higher compared with 2000 when lead in water was lower (Edwards et al. 2009). Considering the irreversible, life-altering, costly and disparate impact of lead exposure, primary prevention is necessary to eliminate exposure (Centers for Disease Control and Prevention 2012). Although lead is now removed from paint and gasoline, lead contamination of drinking water may still be increasing because of lead-containing water in infrastructures, changes in water sources and change in water treatments (Miranda et al. 2007; Edwards et al. 2009).

Several treatment processes are available in water treatment for removal of lead, such as chemical precipitation, ion exchange, adsorption, membrane filtration, coagulation, flotation and electrochemical treatment (Fu & Wang 2011). Chemical precipitation and ion exchange are the most common methods for removal of heavy metals. The latter is more advantageous than the former due to high treatment capacity, high removal efficiency and fast kinetics, high selectivity, less sludge volume produced and recovery of metal value (Ali & El-Bishtawi 1997).

Among these processes, adsorption processes by natural zeolites have received attention as they are environmentally and economically acceptable hydrated aluminosilicate materials with exceptional ion-exchange and sorption properties (Margeta et al. 2013). The microporous structures are made of three-dimensional frameworks of [SiO₄]⁴⁻ and [AlO₄]⁵⁻ tetrahedra, linked by sharing oxygen atoms (Chiang & Chao 2001; Treacy & Higgins 2007). Because of the excess negative charge on the surface of zeolite, which results from isomorphic replacement of silicon by aluminum in the primary structural units, natural zeolites belong to the group of cationic exchangers (Colella 1999; Motsi et al. 2009). Zeolite's selectivity is an important property in the water treatment process. The selectivity of zeolite toward heavy metal cations exists in the series: Pb²⁺ > Cd²⁺ > Cu²⁺ > Co²⁺ > Cr²⁺ > Zn²⁺ > Mn²⁺ > Hg²⁺, and selectivity by anions exists in the series: SO₄²⁻ > I⁻ > NO₃⁻ > Br⁻ > Cl⁻ > OH⁻ (Langella et al. 2000; Armbruster 2001). To increase the ion exchange capacities, natural zeolites are commonly functionalized by physical or chemical methods before use (Taffarel & Rubio 2009).

In this paper, a graphene oxide coated zeolite and thiol functionalized zeolite were prepared by using graphene oxide and cystamine dihydrochloride under basic conditions using a fast, environmentally friendly and low-cost method in order to improve the lead removal efficiency in natural zeolite. The adsorption experiments were carried out for lead to investigate the effect of solution pH and contact time on the adsorption efficiency. Kinetics and isothermal studies were completed to understand the interactions between heavy metal ions and adsorbents. Diffusion was also studied to evaluate the rate-determining step during the process.

Natural clinoptilolite zeolite (0.7–1.0 mm) was obtained from Zeolite Australia PTY. Single-layer graphene oxide powder (ACS Materials) and 96% cystamine dihydrochloride (Sigma-Aldrich) were used without further modifications. Lead standard of 1,000 ppm was purchased from Ricca Chemical and was used to prepare initial solutions. Deionized water (DI water) and extra-pure water were prepared in the laboratory. HNO₃ and NaOH solution were used to adjust the solution pH.

Two methods of zeolite functionalization were developed. Their fabrication methods are described below.

Graphene Oxide (GO) Functionalized Zeolite. A 2.5 mg/mL graphene oxide suspension was prepared by adding 25 mg of GO to 10 mL of deionized water. The suspension was sonicated for two hours. Then 10 g of acid-treated zeolite (ATZ) (method described in SM) was added to the suspension on a shaking table at 150 rpm for three hours. The zeolite was separated from the mixture using vacuum filtration and dried in the oven at 100 °C for three hours.

Thiol Group Functionalized Zeolite. A 0.05 M solution of cystamine dihydrochloride was prepared by adding 2.25 g (0.01 mol) of cystamine dihydrochloride crystals to 200 mL DI water. Then 10 g of sodium-treated zeolite (NaCl-Z) (method described in SM) was added to the solution and refluxed at 85 °C for 48 hours. Treated zeolite was washed with DI water and dried at 100 °C for 12 hours.

The Fourier transform infrared (FTIR) spectra of each sample were obtained at room temperature by means of the Shimadzu FTIR Tracer-100 using LabSolution IR software. FTIR spectra were recorded in a range of 4,000–400 cm⁻¹. Raman measurements of the natural and GO coated zeolite were carried out using a Horiba Raman microscopy/spectroscopy instrument (Xplora Plus BX41TF). Thermo-gravimetric analysis (TGA) was performed using a TA Instrument (Shimadzu DTG-60AH). All the adsorbent materials were heated from 25 °C to 850 °C in a nitrogen atmosphere. Scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS) was performed on a Hitachi S-4800 equipped with a Burker EDS detector to generate images and provide chemical compositions of the samples. Imaging was performed under 15.0 kV accelerating voltage with an extraction current of 15 μA. Inductively coupled plasma mass spectroscopy (ICP-MS) (Thermo Scientific™ Element 2™ high resolution) was performed in water samples for determination of lead concentration, preceded by filtration through a 0.2 μm filter and preservation in 0.2% nitric acid.

Adsorption experiments were performed at two different initial concentrations: 560 ppb and 12 ppm. Firstly, the pH values of the 560 ppb solutions were adjusted to 4, 7 and 10 HNO₃ and NaOH to study the effect of initial solution pH on the adsorbents’ removal efficiency. To do this, 0.50 g of adsorbent was added to 100 mL of the solution while on the shaking table at 250 rpm for four hours. Samples were collected over time and the residual concentration of lead was determined by ICP-MS. Secondly, the adsorption experiments were performed using a 12 ppm solution at pH level of 6 to evaluate the maximum adsorption capacity of the adsorbents. For this purpose, 0.20 g of adsorbent was added to 1,000 mL of the solution while on the shaking table at 250 rpm for 72 hours. Samples were collected over time and the residual concentration of lead was determined by ICP-MS.

Exploring the adsorption kinetics is an essential topic since it provides information about the mechanism of adsorption, which is critical for optimizing the efficiency of the process. Adsorption isotherms in general show how the adsorbent materials behave when the adsorption process reaches an equilibrium condition. By employing a suitable kinetic model, quantifying the changes in adsorption over time can be investigated (Beyki & Shemirani 2015). In this study, the Langmuir isotherm, and pseudo-first-order, pseudo-second-order along with diffusional models have been employed to investigate lead adsorption kinetics. An interruption test was employed for distinguishing between particle diffusion and film diffusion control in the batch experiment (Helfferich 1962). In this experiment, 0.20 g of adsorbent was added to 1,000 mL of Pb solution with a concentration of 12 ppm at pH 6. The batch was stirred for 72 hours. The adsorbent materials were removed from the solution for a brief period of time and were then immersed in a fresh lead solution and stirred for another 72 hours. The pause gave time for the concentration gradients in the adsorbent material particles to level out.

To confirm the presence of the functional groups for each of the materials, material characterization was completed using FTIR, Raman, TGA and SEM-EDS. Figure 1 shows the FTIR spectrum of CDHZ and CZ. The presence of cystamine ions could be confirmed by the observation of bands in CDHZ assigned to C-H vibrations in 2,850–2,930 cm⁻¹ and of stretching bands attributed to the S-H and C-S vibrations at 2,555 and 686 cm⁻¹, respectively (Angell & Schaffer 1966; Lagadic et al. 2001). Si-C and Si-O-Si bands were also observed at 1,100 and 1,024 cm⁻¹, respectively.

The Raman spectrum of GOZ is shown in Figure 2. Two typical peaks were presented: the G band at 1,600 cm⁻¹, which corresponds to in-phase vibration of the graphite lattice, and the D band at 1,350 cm⁻¹, which is attributed to structural defects that are presented in the carbon phase. Hydroxyl groups yield a series of bands around 1,300 cm⁻¹, which is exactly in the same range that the D band is located. Due to the closeness of the hydroxyl groups band and the D band, both types of bands will tend to merge into a broad spectral feature (Kudin et al. 2008).

The thermogravimetric analysis showed a water loss of (CZ = 3.31%, GOZ = 3.61%, CDHZ = 2.83%) at 60–150 °C (Figure 3). This water loss is due to the weakly bound water. The TGA thermograms showed another weight loss of (CZ = 2.18%, GOZ = 0.80%, CDHZ = 2.05%) at 150–250 °C, due to the water located in zeolite cavities and bound to the nonframework cations. There was a third water loss of (CZ = 0.27%, GOZ = 0.30%, CDHZ = 0.52%) at 450–500 °C corresponding to structural water (Castaldi et al. 2005). GOZ showed a rapid weight loss at 100–200 °C that other than water loss might mainly be related to oxidation of carbon and removal of oxygen-containing groups (Cui et al. 2015). An important weight loss of cystamine dihydrochloride at 200–300 °C has been reported (Gebremedhin-Haile et al. 2003). CZ and CDHZ showed 3.1% and 5.76% at 200–800 °C, so the difference might be attributed to the decomposition of thiol functional groups on CDHZ.

As suggested by SEM images of CZ, GOZ and CDHZ (Supplementary Material, Figure S1), it appears that the functionalization does not lead to any observable surface changes as the functionalization is only a surface treatment technology. In order to confirm the presence of carbon and sulfur, EDS analysis was performed for GOZ and CDHZ respectively, and carbon and sulfur distribution is observed in the functionalized zeolite structure (Figure 4). Through EDS analysis, it is possible to verify that the C/Si ratio in GOZ, and S/Si ratio in CDHZ, have been increased compared with CZ (Supplementary Material, Table S1).

According to the Pourbaix diagram, a shift in pH will result in a different species of Pb. The pH 4 shows Pb²⁺, pH 7 shows PbOH⁺, and pH 10 shows Pb(OH)₂ (Takeno 2005). A lead solution with an initial concentration of 560 ppb was used and the solution pH adjusted to 4, 7 and 10. Solutions at different pHs with CZ, GOZ, and CDHZ were sampled at minutes 3, 10, 30, 90, 180, and 240 (Figure 5). Residual concentration of Pb at solution pH 4, 7 and 10 for CZ, GOZ, and CDHZ was measured. Results indicate that 90 minutes is a sufficient contact time to reach the equilibrium state for CZ, GOZ, and CDHZ.

By increasing the pH level from 4 to 7, Pb²⁺ uptake was slightly affected by the H⁺ concentration (Figure 5). Sorption of lead on natural clinoptilolite has been shown to be high in particular in an acidic environment (Perić et al. 2004; Oter & Akcay 2007). Because of higher zeolite selectivity toward Pb²⁺ than H⁺, Figure 5 illustrates that the fabricated materials in this research can effectively remove Pb²⁺ from contaminated waters at pH 4 and 7. These results are not in agreement with findings published by others related to other types of zeolites. In Kabwadza-Corner et al. (2015), maximum adsorption capacities were increased from 666 to 2,500 mmol.kg⁻¹ for A4, 1,000 to 2,000 mmol.kg⁻¹ for Faujasite X, 333 to 588 mmol.kg⁻¹ for Faujasite Y, and 123 to 179 mmol.kg⁻¹ for Mordentie by increasing the pH level from 3 to 5. Berber-Mendoza et al. (2006) reported a decrease of 1.5 times in maximum exchange capacity of zeolite toward Pb²⁺ when solution pH was decreased form 4 to 2. Data in Joshi (2016) showed lead removal is more favorable at pH 7 among the three tested pHs of 5, 6, and 7. Increasing the pH from 4 to 7 showed an increase in the exchanged Pb²⁺ on GOZ. This can be explained by considering the role of proton concentration on lead adsorption; increasing H⁺ concentration causes a decrease in the adsorption efficiency, which confirms the competition between H⁺ and Pb²⁺ ions.

Increasing solution pH from 7 to 10 resulted in a decrease in the performance of CDHZ. This can be attributed to the Pb(OH)₂ precipitates formed in this pH range, shown in Figure 6. These precipitates limited the accessibility of the Pb²⁺ ions to the thiol functional groups in the CDHZ structure. Although the concentration of thiol functional groups is a factor affecting the immobilization of Pb⁺² ions, the accessibility of metal ions to these binding sites is also a controlling factor in the process (Lagadic et al. 2001).

Lead solution with a concentration of 12 ppm at pH level of 6 was employed for a batch test. Lead removal tests were conducted using 0.20 g of CZ, GOZ, and CDHZ and solutions were sampled at hours 0.5, 1.5, 3, 6, 12, 24, 48 and 72. After 72 hours the solution was replaced with a fresh lead solution with 12 ppm concentration at pH 7 and the batch continued for another 72 hours. Figure 6 shows the residual concentration according to time for CZ, GOZ, and CDHZ. In the low-concentration-level experiments (560 ppb) (Figure 5), tests were performed at different pH levels (acidic, neutral, and basic). In the high-concentration-level experiments (12 ppm) (Figure 6), the tests were only performed at neutral pH level. At this pH level both GOZ and CDHZ showed higher adsorption capacity compared with CZ.

In a sorption experiment there are two potential rate-determining steps: interdiffusion of ions in the adsorbent (particle diffusion); and interdiffusion of ions in the liquid film (film diffusion) (Helfferich 1962). There are two other rate-determining steps that have been investigated: counter ion exchange across the interface between ion exchanger and solution, and actual chemical exchange reaction. The first one is very unlikely for theoretical reasons and also is not supported by experiments. The second one has been ruled out for ordinary ion exchange process (Helfferich 1962).

Isotherm adsorption models for CZ, GOZ, and CDHZ are shown in Figure 7. The ion exchange rate immediately after reimmersion was found to be greater than prior to the interruption, confirming that the ion exchange process is a particle diffusion controlled process.

Kinetic models showed that the pseudo-second-order has the best linearity for Pb adsorption, and according to the theoretical calculation (Supplementary Material, Table S2), the q_e obtained from the model showed the least deviation from the experimental value, indicating that this model can be accepted as the kinetic mechanism for lead adsorption. In addition, the summations of adsorption capacity of each adsorbent, before and after interruption, CZ (72.95 mg/g), GOZ (85.21 mg/g) and CDHZ (98.14 mg/g), indicate that the addition of graphene oxide and thiol functional groups increased the lead removal capacity of natural clean zeolite by 16.81% and 34.53%, respectively. However, at this point we cannot compare the roles of thiol groups with oxygen groups on GOZ in terms of maximum adsorption capacity. Multiple factors such as GO thickness and the amount of the GO loaded on the parent zeolite affect the adsorption of lead on the GOZ. So far, this research has confirmed that GO was successfully introduced to the zeolite structure and by means of the oxygen functional groups in GO the adsorption capacity was improved. However the adsorption capacities of both GOZ and CDHZ are not claimed as their maximum adsorption capacities as the aforementioned factors need to be further explored and optimization of the materials can be performed.

• Zeolite can be successfully functionalized by cystamine dihydrochloride and graphene oxide as functional groups to remove lead from aqueous solutions.

• The addition of graphene oxide and thiol functional groups increased the lead removal capacity of natural clean zeolite by 16.81% and 34.53%, respectively.

• Treated zeolites show their best performance in lead adsorption under the experimental conditions when the solution pH is 4 and 7. The decrease in adsorption capacity at solution pH = 10 is attributed to the Pb(OH)₂precipitates at this pH.

• The functionalized zeolite is capable of removing lead by adsorption and ion exchange mechanism.

• Batch tests show that it takes 90 minutes to establish equilibrium in lead adsorption for the initial solution concentration of 560 ppb.

• Kinetics analysis of the lead adsorption show that the adsorption process follows the pseudo-second-order as the kinetic model. Diffusion studies confirm particle diffusion as the dominant rate controlling step in the adsorption process.

Authors thank Dr Heather. A. Owen for technical support with SEM analyses at the UWM Electron Microscope Laboratory, and Patrick Anderson from the School of Freshwater Sciences for technical assistance with ICP-MS analyses. This research study was supported by the National Science Foundation Industry/University Cooperative Research Center on Water Equipment & Policy located at University of Wisconsin-Milwaukee (IIP-1540032) and Marquette University (IIP-1540010).

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