Batch equilibrium and kinetics of mercury removal from aqueous solutions using polythiophene/graphene oxide nanocomposite

Water pollution through mercury emissions is of great environmental concern due to its high toxicity and detrimental effect on human and aquatic life. Pollution of water by mercury occurs mainly through discharge of effluent streams from industries such as paint making, electrical, pulp and paper, fertiliser, rubber processing, chloro-alkali, coal-fired power plants and ore mining (Moghaddam & Pakizeh 2015). In aqueous solution, mercury exists mainly in the form of elemental mercury (Hg) and ionic mercury (Hg⁺ and Hg²⁺) species (Lottermoser 2010). Mercury is considered as one of the most toxic heavy metals on earth, and is listed as a priority pollutant by the US Environmental Protection Agency (USEPA 2014). Mercury has the ability to bind to the amino acid cysteine proteins, resulting in serious neurological damage to both human and aquatic life. Due to its high toxicity, the World Health Organization (WHO) has set the tolerance limit for mercury in drinking and surface water at 1 μg/L and 10 μg/L, respectively (WHO 1990). Thus, environments having a mercury concentration above the recommended limits require application of a proper treatment regime.

The conventional treatment methods used in the removal of mercury from effluent streams include sulphide precipitation (USEPA 2007), the Blue PRO^® reactive filtration process (Shafeeq et al. 2012), membrane processes (Sharma et al. 2015), coagulation (Henneberry et al. 2011), biofilms (Bhattacharya et al. 2016) and the adsorption process (Yu et al. 2016). Among these treatment technologies, adsorption has been reported as the most robust due to its simplicity, ease of operation and environmental benignity (Cui et al. 2015). A number of adsorbents such as activated carbon (Ekinci et al. 2002; Yardim et al. 2003), bicarbonate treated peanut hull carbon (Namasivayam & Periasamy 1993), thiol-functionalised mesoporous silica microspheres (Bibby & Mercier 2002), iron hydrous oxide gel (Kinniburgh & Jackson 1978), and thiol derivatised single wall carbon nanotubes (Bandaru et al. 2013), have been employed in the removal of Hg²⁺ from aqueous solutions. However, the aforementioned materials suffer from limitation such as inherently low capacity, slow removal kinetics and poor selectivity. Consequently, there is an urgent need to design and develop novel materials with desirable attributes such as enhanced capacity, fast removal kinetics, and high reusability.

Recently, a lot of effort has been geared towards use of nanocomposite (NC) materials for environmental remediation due to the synergistic effects of the individual components and the improved affinities towards metal ions. The development of NC focuses mainly on tailoring the physicochemical properties of materials to enhance contaminant selectivity. Conducting polymers have been used to decorate functional groups such as amine, thiol and carboxyl onto nanoadsorbent surfaces. These functional groups are beneficial in the decontamination of effluent streams. Among the conducting polymer family of polypyrrole (PPy), polyaniline (PANI) and polythiophene (PTh), few studies have reported the use of PTh for heavy metals removal from aqueous environments. PTh, and their composites, have been extensively used for optical, electronic and mechanical applications (Uygun et al. 2009; Wayne et al. 2010; Bobade 2011; Kumar et al. 2015). Thus it is important to investigate the possibility of using PTh-based materials in the removal of contaminants. The structure of PTh contains abundant sulphur atoms bonded to two carbon atoms on either side, leaving two un-bonded lone pairs of electrons exposed. Thus it is capable of donating a pair of electrons to a Lewis acid such as Hg²⁺, hence forming a chelate. In this case, mercury acts as an electron acceptor while the sulphur atom acts as an electron donor, according to the principle of hard soft acids and bases (HSAB) (Pearson 1963; Nabais et al. 2006; Wajima & Sugawara 2011). However, PTh has poor dispersion characteristics and a low surface area, resulting from particle agglomerations. Hence, a support material is required to improve its properties.

Carbon-based nanomaterials, for instance GO, are currently receiving a great deal of attention as adsorbents and also as polymer host materials for abatement of inorganic and organic pollutants from aqueous solutions. Graphene oxide (GO) offers utility in various applications because of its unique 2D features and bond structure (Sheet et al. 2014). Remarkable properties such as the presence of surface functional groups (carboxyl, hydroxyl, epoxide and carbonyl) and large specific surface area make GO and their composites highly potential adsorbents for water purifications (Nuengmatcha et al. 2014). Therefore, a NC prepared by combining GO and PTh could result in an adsorbent material with dramatic improvements in physicochemical properties such as high surface area, improved dispersion characteristics and enhanced mercury ion pollution clean-up. Although a few sulfurised adsorbents for Hg²⁺ removal from aqueous solutions have previously been reported (Nabais et al. 2006; Wajima & Sugawara 2011), their preparation procedures were tedious and time-consuming. Moreover, toxic H₂S gas and very high temperatures were utilised, necessitating careful control. Similarly, the reported GO was mainly prepared through hazardous methods that are susceptible to explosions and release of toxic gases such as NO₂, N₂O₄, and/or ClO₂ (Marcano et al. 2010). Therefore, there is an urgent need to develop green methods for preparation of high yield dispersible PTh/GO NC with improved properties suitable for Hg²⁺ adsorption.

Consequently, this study seeks to synthesise, characterise and evaluate the performance of PTh/GO NC in the removal of mercury ions from aqueous solutions. GO sheets were synthesised from graphite flakes using a modified method reported by Marcano et al. (2010), and thereafter, the as-prepared GO was functionalised by PTh via an in situ interactive polymerisation method. The influence of pH, adsorbent dosage, initial concentration and temperature on Hg²⁺ adsorption were explored by performing batch equilibrium isotherms and kinetics experiments. Lastly, desorption and competitive adsorption were studied and a plausible Hg²⁺ removal mechanism was proposed.

Graphite flakes (325 mesh, 99.95% purity), anhydrous iron (III) chloride (FeCl₃), potassium permanganate (KMnO₄), chloroform (CHCl₃), thiophene (Th) monomer (C₄H₄S, density = 1.051 g/cm³, ≥99%), hydrogen peroxide (H₂O₂, 30 wt.% ACS reagent) and mercury dichloride (HgCl₂) salt were all purchased from Sigma-Aldrich (Germany). A stock solution (1 L) of 1,000 mg/L of Hg²⁺ was prepared by dissolving 1.354 g of HgCl₂ in deionised water (Purite water system, Model Select Analyst HP40, UK) acidified with 5 mL concentrated HNO₃ to prevent hydrolysis. All pH adjustments were done with either 0.1 M HNO₃ or 0.1 M NaOH solution. Other chemicals were of analytical reagent (AR) grade.

GO nanosheets were prepared using an improved method described elsewhere (Marcano et al. 2010). In brief, the following stoichiometric ratios were used in the synthesis of GO sheets; H₂SO₄: graphite flakes (100:1, v/w); KMnO₄: H₂SO₄ (1:20, w/v); H₃PO₄: H₂SO₄ (1:9, v/v). First, 3.6 g was added to 400 mL of H₃PO₄/H₂SO₄ mixture in a 1,000 mL glass beaker. KMnO₄ (18 g) was then added gradually into the graphite-acid mixture to prevent overheating. The mixture was kept under stirring and the temperature rose to about 35–40° C. The mixture was heated further to 50 °C and maintained at that under constant stirring for 12 h. The brown mixture was then poured into 460 mL of ice and 5 mL of 30 wt.% H₂O₂, resulting in a bright-yellow colour and temperature increase to about 98°C. Finally, GO nanosheets were obtained through ultrasonic agitation after the mixture had been filtered and thoroughly washed using HCl and deionised water, until pH of the filtrate was 7.

PTh/GO NC was prepared by first dispersing 0.4 g of GO nanosheets in 30 mL of chloroform in a conical flask under sonication for 30 minutes. Thereafter, 0.95 mL of thiophene monomer was injected into the reaction mixture while stirring magnetically for further 30 minutes. To this mixture, FeCl₃ (8 g) was poured in one go and left for 6 h under stirring at room temperature. Thereafter, the obtained suspension was vacuum filtered and repeatedly rinsed with deionised water and methanol, until the pH of the colourless filtrate was 7. Finally, the solid materials (PTh/GO NC) were vacuum dried at 60 °C for 24 h. The total mass of NC was 1.4 g, and the compositions of GO and PTh were 28.5 wt.% and 71.5 wt.%, respectively. Similarly, PTh/GO NC was prepared using different masses (0.1, 0.2, 0.6, 0.8 g) of GO and were designated as 9.1, 16.7, 37.6 and 44.4 wt.%, respectively, based on GO composition in the NC.

Brunauer–Emmett–Teller (BET) measurements were recorded on PTh and PTh/GO NC with a Micromeritics TRISTAR 3,000 surface area analyser using the N₂ adsorption-desorption method (−196°C). The morphological characteristics and elemental compositions were analysed using a field emission scanning electron microscope (FE-SEM, JEOL JSM-7600F) equipped with energy dispersive X-ray (EDX). Information on functional groups was obtained by Fourier transform infra-red (FT-IR) spectrometer, while zeta potentials were measured using a ZETASIZER nano-series (Nano ZS-90, Malvern Instruments, UK) at different pHs.

Desorption and regeneration tests were conducted by first adding 0.08 g of the PTh/GO NC in different 50 mL solutions of Hg²⁺, contained in 100 mL plastic sample bottles. Each solution had an Hg²⁺ concentration of 100 mg/L and pH 6. These samples were shaken for 24 h at constant temperature (25 °C) to attain equilibrium. Thereafter, the samples' contents were mixed together and the mercury-bound PTh/GO NC was separated by centrifugation, rinsed with distilled water to remove any unadsorbed Hg²⁺ ions, and dried in a vacuo at 60 °C. This mercury-bound PTh/GO NC was dispersed in 0.5 M solutions of either HCl, H₂SO₄ or HNO₃ (serving as eluting agents) and placed in the thermostatic shaker for 24 h. After that, the adsorbent was separated and the supernatant was measured for Hg²⁺ ions. Then desorption efficiency was calculated by performing appropriate mass balances. In another set of experiments, the used-adsorbent was first activated by suspending it in 1 M NaOH solution for 3 h, before conducting four subsequent adsorption-desorption experiments.

Specific surface area is one of the key parameters that determine the performance of any adsorbent. The BET surface area, total pore volume and pore size of PTh and PTh/GO NC are listed in Table S1 (supplementary file, available with the online version of this paper). The specific surface area of PTh/GO NC (42.81 m²/g) was found to be higher than that of PTh (18.23 m²/g). The increased surface area can be attributed to the synergistic effect of GO in the NC, and it correlated well with the enhanced Hg²⁺ adsorption performance onto PTh/GO NC. Also, the total pore volume and the average pore size decreased from 0.2487 to 0.0950 m³/g and from 43 to 36 nm, respectively. These results are consistent with previous works (Setshedi et al. 2015).

To verify the surface charges on the as-prepared adsorbents, zeta potentials were measured at different pHs. As shown in Figure S2 (supplementary file), zeta potentials of PTh/GO NC were less negative than those of GO over the whole measured pH range. For both materials, the zeta potential declines from positive values at acidic pHs to negative values. This decline can be attributed to protonation and deprotonation effects. The isoelectric point of PTh/GO NC was found at pH 5, and the composite exhibited significantly enhanced the carrying capacity of hydrogen ions because of the ability of sulphur atoms to adsorb hydrogen ions. Consequently, many hydroxyl ions were required to neutralise hydrogen ions on PTh/GO NC, resulting in a remarkable increase in isoelectric point. On the other hand, GO did not exhibit any isoelectric point within the studied pH range (Yan et al. 2014). Therefore, at pH > 5, PTh/GO NC was negatively charged and at pH < 5, it was positively charged. These observations are consistent with the increased Hg²⁺ ion removal as pH increases.

Mercury uptake increased with increase in temperature. This can be attributed to the fact that an increase in kinetic energy of mercury ions in aqueous solution enhances the solid-liquid interaction. Moreover, at high temperatures, there is less resistance offered to the fast moving Hg²⁺ ions by viscous forces in the aqueous phase.

Adsorption isotherms provide useful information regarding the capacity of an adsorbent as well as the description of the functional dependence of capacity on the concentration of the pollutant. The equilibrium data were analysed using Langmuir, Freundlich, Tempkin and Dubinin–Radushkevich (D-R) (Foo & Hameed 2010), listed as Equations (S1)–(S4), respectively (supplementary file, available with the online version of this paper). Non-linear fits of Langmuir and Freundlich models (inset is linear Langmuir model) are shown in Figure S3(a), while the linear forms of D-R, Freundlich and Tempkin models are plotted in Figure S3(b), Figures S4(a) and S4(b) (supplementary file), respectively. Parameters extracted from linearised and non-linearised isotherm models are presented in Table 1 and Table S3 (supplementary file), respectively. Based on the linear regression analysis, R² values obtained from the Langmuir isotherm model were higher (R² > 0.99) than those from Freundlich and Tempkin models, suggesting that the experimental data are best described by the Langmuir model. Further, values of separation factor (R_L) (not shown) were within 0 < R_L < 1, range, confirming that adsorption is favourable. The Langmuir adsorption capacity for the PTh/GO NC was 113.64 mg/g which is comparably high to other adsorbents reported in the literature (Table S4 in the supplementary file). On the other hand, non-linear regression analysis gave low values of R² for both the Langmuir and Freundlich models, although the Freundlich model gave slightly higher values of R². This discrepancy of R² values obtained might be due to the inherent limitations of linear regression method. Realistically, the Freundlich isotherm, which is based on the assumption of a heterogeneous surface, described the adsorption of mercury onto PTh/GO NC satisfactorily and the adsorption was also favourable (0 < 1/n < 1). In addition, the mean free energy values obtained from the D-R model (7.89, 8.39 and 12.61 kJ/mol) suggested that the adsorption of mercury onto PTh/GO NC may have occurred through a physicochemical mechanism (Raji & Pakizeh 2013).

Table 1

Linearised isotherms' and thermodynamic parameters

Temp .	Langmuir .			Freundlich .			Tempkin .			D-R .				Thermodynamics .
°C .	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
25	113.64	0.109	0.993	0.185	39.22	0.965	18.22	1.617	0.960	7.89	8.03E − 09	124.18	0.956	−0.166	35.25	118
35	135.68	0.117	0.992	0.212	41.87	0.933	23.86	1.103	0.951	8.39	7.11E − 09	153.85	0.973	−1.473
45	166.67	0.131	0.992	0.214	52.42	0.989	28.57	1.453	0.992	12.61	6.29E − 09	185.69	0.992	−2.541

Temp .	Langmuir .			Freundlich .			Tempkin .			D-R .				Thermodynamics .
°C .	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.	.
25	113.64	0.109	0.993	0.185	39.22	0.965	18.22	1.617	0.960	7.89	8.03E − 09	124.18	0.956	−0.166	35.25	118
35	135.68	0.117	0.992	0.212	41.87	0.933	23.86	1.103	0.951	8.39	7.11E − 09	153.85	0.973	−1.473
45	166.67	0.131	0.992	0.214	52.42	0.989	28.57	1.453	0.992	12.61	6.29E − 09	185.69	0.992	−2.541

Note: All isotherms' parameters and their respective units are defined in the supplement (available with the online version of this paper).

View Large

Thermodynamic parameters given in Table 1 were obtained from Van 't Hoff's plot (Figure S5) and Equations S5 and S6 (supplementary file). The values of Gibb's free energy decreased from −0.166 to −2.5407 kJ/mol, when the temperature was increased from 25 to 45 °C, implying that the adsorption process was feasible and spontaneous in nature. The positive values of enthalpy and entropy changes stated that the adsorption process was endothermic, and the randomness at the solid-liquid interface increased during adsorption, respectively.

The adsorption kinetic data were fitted to the following kinetic models; pseudo-first-order and pseudo-second-order models (Qiu et al. 2009) listed as Equations (S7) and (S8) in the supplementary file (available online). Figure 6 shows non-linear kinetics plots while Figure S6 (supplementary file) shows the linear kinetic plots for the removal of Hg²⁺ from aqueous solution. The kinetic parameters obtained from both non-linear and linear models are summarised in Table 2 and Table S5 (supplementary file), respectively. Comparatively, the pseudo-second-order model had the highest coefficient value (R² > 0.99) among the tested kinetic models, suggesting that the adsorption proceeds via the pseudo-second-order kinetics. Moreover, the theoretical equilibrium capacities are also close to the experimental ones ⁠, suggesting better fitting to the pseudo-second-order kinetic model. However, the relatively high values of regression coefficient (0.98) for the pseudo-first-order model could mean that physical adsorption was present, thus Hg²⁺ adsorption onto PTh/GO NC may have proceeded through a physicochemical process.

Table 2

Non-linear kinetic models parameters for the adsorption of Hg²⁺ on PTh/GO NC

.	Concentration (mg/L) .
Kinetic model/parameters .	50 .	100 .	200 .
(mg/g)	31.0	60.2	81.6
Pseudo-first-order
⁠, (mg/g)	33.46	67.49	89.13
k1, (min−1)	0.085	0.035	0.030
R2	0.988	0.984	0.997
Pseudo-second-order
(mg/g)	31.7	60.4	79.3
k2 (g mg−1 min−1)	3.5 × 10−3	0.7 × 10−3	0.5 × 10−3
h (mg g−1 min−1)	3.52	2.55	3.14
R2	0.997	0.990	0.989

.	Concentration (mg/L) .
Kinetic model/parameters .	50 .	100 .	200 .
(mg/g)	31.0	60.2	81.6
Pseudo-first-order
⁠, (mg/g)	33.46	67.49	89.13
k1, (min−1)	0.085	0.035	0.030
R2	0.988	0.984	0.997
Pseudo-second-order
(mg/g)	31.7	60.4	79.3
k2 (g mg−1 min−1)	3.5 × 10−3	0.7 × 10−3	0.5 × 10−3
h (mg g−1 min−1)	3.52	2.55	3.14
R2	0.997	0.990	0.989

View Large

The adsorption of mercury ions onto PTh/GO NC can be considered to have proceeded via a complexation or chelation mechanism between S atoms of PTh/GO NC and the mercury ions, according to HSAB and ligand field theories (Pearson 1963). PTh is a chelating agent with S groups (soft base), and hence has very high affinity for mercury ions (soft acid). Thus, the 6 s unoccupied orbitals contained in mercury ions may be adsorbed by chelating with a lone pair of electrons available in the S atom of the PTh/GO NC through electron pair sharing, resulting in the formation of coordination compounds with a tetrahedral pattern. Also, it is possible that there could be some ion exchange and electrostatic attraction between Hg²⁺ and H⁺ and Hg²⁺ with OH⁻ considering the results on the effect of pH (protonation and deprotonation effects). The adsorption mechanism is further supported by the adsorption isotherms and kinetics models, which suggested that more than one mechanism was involved. Moreover, the FT-IR results showed a shift in C-S-C wavenumber and a change in the intensity and shape of the peaks after mercury adsorption, suggesting there was some interaction between the S atom and Hg²⁺ Similar results were reported by Tahmasebi et al. (2013) and Chen et al. (2015).

This study evaluated the performance of the prepared PTh/GO NC in the removal of Hg²⁺ ions from aqueous solutions. The uptake and equilibrium studies indicated that the adsorption of Hg²⁺ ions onto PTh/GO NC follows the Langmuir and Freundlich isotherms, while the kinetics fitted the pseudo-second-order model. The thermodynamic studies shows that the adsorption of Hg²⁺ onto PTh/NC is endothermic in nature and proceeded by physicochemical process. Further, the results from the adsorption-desorption studies indicated that HCl and NaOH could be used as eluent and activator, respectively, and the regenerated adsorbent could be reused for three cycles without significant loss in its adsorption capacity. The main possible removal mechanisms involved were complexation, ion-exchange and neutralisation. These findings demonstrated that PTh/GO NC is a promising candidate material for mercury ion remediation.

Financial support was received from Rand Water, South Africa, through the Rand Water Chair in Water Utilisation (Tshwane University of Technology). We wish to gratefully acknowledge Jacob Kitinya for assisting in proofreading the manuscript.