Adsorption of phenol and chromium (VI) from aqueous solution using exfoliated graphite: Equilibrium, kinetics and thermodynamic studies

2.1 Materials

In order to obtain the exfoliated graphite, Grade ES250 B5 expandable graphite (Qingdao Kropfmeuhl Graphite) was exposed to a temperature shock at 600 °C for at least 5 min in a Thermo-power electric furnace (Focke et al., 2014). Potassium chromate and sodium hydroxide were obtained from Merck (Germany) while phenol and 1.5-diphenylcarbazide were obtained from Saarchem and Sigma-Aldrich (South Africa), respectively. Hydrochloric acid was supplied by Glassworld, South Africa.

2.2 Batch adsorption studies

Aqueous Cr (VI) and phenol solutions with varying concentrations were made by diluting appropriate volumes of 1000 mg/L stock solutions with deionized water. The optimum adsorbent loading was determined by using 5 to 30 g/L of EG and contacting it with 30 mg/L of Cr(VI) or phenol solutions for 48 h. The effect of initial pH on the removal of the pollutants was studied in the 2 to 12 pH range. The pH was altered with hydrochloric acid and sodium hydroxide. Adsorption isotherms were determined at the optimum conditions for each pollutant, and the concentration was varied from 10 to 100 mg/L. The kinetics were determined by drawing an appropriate sample volume at pre-determined time intervals and filtering before analysis.

2.3 Phenol and Cr (VI) analysis

Changes in Cr (VI) concentration were tracked using the standard colorimetric method detailed in our previous work (Tichapondwa et al. 2018). The phenol concentration was determined using high performance liquid chromatography. The phenol separation was performed on a Waters PAH C18 (4.6 × 250 mm, 5 µL) column and the resultant data was analysed using proprietary Empower software. The mobile phase was composed of two solvents namely, a 1% acetic acid and water mixture and a 1% acetic acid in acetonitrile mixture.

2.4 Characterization of materials

The surface morphology of EG before and after contact with the respective pollutants were viewed on a Zeiss Ultra Plus field emission scanning electron microscope (FEG SEM). The nature of the phases in the adsorbent material was determined through X-ray diffraction (XRD) measurements which were recorded using a Bruker D8 Advanced solid powder diffraction fitted with Lynx eye detector. The surface area was determined using a Micrometrics TriStar BET analyser. A ca. 7-fold increase from 2.4 to 22.4 m²g⁻¹ was observed upon exfoliation (Tichapondwa et al., 2018).

2.5 Equilibrium study

The equilibrium adsorption capacity $q_e\left(\frac{\mathit{mg}}{g}\right)$ was calculated using equation (1) (Belhachemi and Addoun, 2011) with V (L) and W (g) the volume of the solution and mass of the EG respectively.

The Redlich-Peterson, Freundlich and Langmuir isotherms were adopted to evaluate the removal process. Equation (2) gives the Langmuir isotherm with $Q_0$ (mg/g) the maximum amount of a pollutant that EG can adsorb and $K_L$ (L/mg) is the Langmuir constant.

The Langmuir isotherm assumes a homogeneous, monolayer adsorption mechanism (Riahi Samani et al., 2016). It was then assumed that once a Cr (VI) or a phenolate ion attaches on a site, no further attachment can take place on the occupied area.

The Freudlich isotherm is an empirical equation formulated to accurately predict the behavior of heterogeneous systems. The isotherm implies that an infinite attachment of pollutant molecules can take place (Foo and Hameed, 2010). The isotherm also deals with non-ideal and reversible adsorption (Singh et al., 2018). A higher n value means a lower affinity. The closer the value of n to zero, means that the system is heterogeneous. A value of n typically shows the degree of non-linearity with values of n<1 indicating that the adsorption is a chemical process (Desta, 2013). The Freudlich model is given in equation (3) where $K_F$ , is the Freundlich constant and 1/n indicates the heterogeneity of the EG’s surface.

The Redlich-Peterson isotherm is an empirical equation with 3 parameters that include features from Langmuir and Freundlich isotherms (Singh et al., 2018) and can also be employed to predict adsorption equilibria. The Redlich-Peterson model is given in equation (4) where $K_{\mathit{RP}}$ , ${\propto }_{\mathit{RP}}$ and $\beta$ are the Redlich-Peterson constants. When $\beta =1$ this equation becomes Langmuir and when $\beta =0$ it becomes Henry’s law equation.

Due to inherent errors brought about by using linearized models for the isotherms, a non-linear regression method has been applied using the built-in function (solver) in Microsoft excel 2013 to determine the isotherm constants. In this study, the two error functions, sum of the squares of errors (ERRSQ) and Marquardt’s percent standard deviation (MPSD) were employed to accurately predict the constants.

The ERRSQ in equation (5) is common and preferred by many researchers for its simplicity (Gimbert et al., 2008).

Equation (6) gives the MPSD error function. The smaller the MPSD value the more accurate the estimation of $q_e$ (Kumar et al., 2008).

The kinetic behaviour of this adsorption process was evaluated using kinetic models, namely pseudo second order, Elovich’s rate equation and pseudo first order. The pseudo first order rate equation describes a dynamic process of the liquid–solid phase adsorption (Dehghani et al., 2016). Equation (7) gives the linear form of Lagergren’s rate equation where $q_e$ and $q_t$ $\left(mg/g\right)$ are the equilibrium adsorption capacities with time, $t\left(min\right)$ and $k_{p1}\left({\mathit{min}}^{-1}\right)$ the pseudo first order rate constant.

The adsorption studies of various pollutants have been successfully projected using the widely used pseudo second order rate equation also known as the Ho’s equation (Qiu et al., 2009). The linear form of the equation is detailed in equation (8) with $V_0=k_{p2}{q}_e^2\left(\frac{\mathit{mg}}{g.min}\right)$ which gives the initial adsorption rate.

Elovich’s equation describes the kinetics where chemisorption is the dominant force between the adsorbates and solids with heterogeneous sites. This equation over the years has seen its application extend to the liquid–solid adsorption processes. The linearized form of the Elovich equation is presented in equation (9) where $t_0=\frac{1}{\alpha a}$ , $\alpha$ is the initial adsorption rate and $a$ is the desorption constant (Qiu et al., 2009).

Liquid-solid adsorption processes are characterised by film diffusion, intra-particle diffusion and mass transfer. Equation (10) gives the intra-particle diffusion model used in the present study where $k_{\mathit{int}}$ is the intra-particle diffusion rate constant. $${\mathrm{q}}_{\mathrm{t}}=k_{\mathit{int}}{t}^{\frac{1}{2}}+\mathrm{I}$$

For the thermodynamic study, isotherm data were obtained at 25, 45 and 65 °C. A magnetic stirrer fitted with a temperature controller was used to maintain the target temperature within a ±2 °C range.

3.1 Characterization of exfoliated graphite

The scanning electron microscope (SEM) micrographs of exfoliated graphite before adsorption, after contact with Cr (VI) and after contact with phenol are shown in Fig. 1. The surface of the EG prior to contact with either Cr (VI) or phenol is clear and smooth as shown in Fig. 1 (a) with the typical worm-like accordion structure. After contact with the respective pollutants, the initial worm-like structure collapses with smaller platelets breaking off from the main structure suggesting that delamination of the EG into graphite nano-platelets occurred. The surface of EG became packed with visible spherical particles attached after Cr (VI) adsorption, Fig. 1(b) and even more packed after phenol adsorption Figure (c).

Close

Fig. 1

The SEM images of (a) exfoliated graphite before adsorption studies (b) after Cr (VI) adsorption and (c) after phenol adsorption.

Export to PPT

Qualitative XRD analysis was performed on the EG surface. Fig. 2 shows the XRD spectra of the EG prior to adsorption studies and compares it to the spectra after contacting with Cr (VI) and phenol, respectively. The main graphite peak was observed at a 2θ position of 30° for the samples before contact with phenol and Cr (VI) and after contact. This indicated the existence of graphitic structures in all samples (Ravulapalli and Kunta, 2018). Furthermore, the intensity of the peak increased after contacting the adsorbent with the respective pollutants. This confirms that EG delamination occurred after contacting it with the pollutants in aqueous solution. The resultant graphite nano-platelets had less imperfections and higher crystallinity compared to the EG, hence the sharper peaks.

Close

Fig. 2

XRD patterns for EG before and after contacting respective pollutants both at 10 mg/L.

Export to PPT

3.2 Effect of adsorbent loading

The optimum EG loading resulting in maximum phenol and Cr (VI) removal was 30 g/L as shown in Fig. 3. The amount removed for phenol and Cr (VI) increased proportionally with an increase in the amount of adsorbent used before levelling off. Thereafter no noteworthy change in the amount removed was observed. It should also be noted that loadings beyond 30 g/L were not investigated due to mixing difficulties similar to those reported in our previous study (Tichapondwa et al., 2018). The increase in percentage pollutant removal with an increase in EG loading can be ascribed to the high number of adsorption sites active and available (Garg et al., 2007).

Close

Fig. 3

Effect of adsorbent loading on pollutant removal C_o = 30 mg/L, 48 h.

Export to PPT

3.3 Effect of pH

The initial pH of the aqueous medium affects the adsorption of both Cr (VI) and phenol since pH controls the surface charge of the adsorbent and degree of ionization of the adsorbate (Senturk et al., 2009). Fig. 4 shows the effect of solution pH on Cr (VI) and phenol removal by EG in the pH range 2 to 12. The maximum Cr (VI) removal (89.8%) was found under acidic conditions at pH 2. The amount of Cr (VI) removed decreased sharply after pH 7. This can be explained by considering the variation in the dominant Cr (VI) species with pH (Rakhunde et al., 2012) and (Szabó et al., 2018. According to Gładysz-Płaska et al. (2012), the prevalent form of Cr (VI) ions between pH 2 and 7 were $C{r}_2{O}_7^{2-}$ and $\mathit{HCr}{O}_4^{-}$ ions whereas at pH values higher than 7, $\mathit{Cr}{O}_4^{2-}$ ions dominate. In the acidic region, the EG surface becomes positively charged due to strong protonation and attracts the negatively charged ions thus significantly enhancing adsorption (Gupta and Balomajumder, 2015). Above pH 7, it is assumed that there was a competition between $\mathit{Cr}{O}_4^{2-}$ and ${\mathit{OH}}^{-}$ ions which resulted in the significant drop in the percentage pollutants removed. Similar observations have been reported by (Gupta and Balomajumder, 2015).

Close

Fig. 4

Effects of pH on Cr (VI) and phenol removal at C_o = 30 mg/L of solution and 30 g/L loading respectively.

Export to PPT

Phenol showed less affinity to EG than Cr (VI) and an increase in pH slightly improved the percentage removal of phenol from about 50% to just above 62% at pH 12. The amount of phenol removed was a weak function of the pH. This suggests that phenol removal was mainly due to the weak Van der Waals forces and weak pi bonds that exist between the phenol structure and the EG surface (Gupta and Balomajumder, 2015). Fig. 4 shows that increasing the value of pH increases the percentage of phenol removed. At pH < 7, the phenol molecule is mainly in its natural form, $C_6{H}_{5}OH$ which is neutral and very stable and this negatively affected the adsorption process (Gładysz-Płaska et al., 2012, Ge et al., 2019). At pH greater than 7, the phenol molecule is largely in the form of the phenolate ion, $C_6{H}_5{O}^{-}$ and the strong base, $O{H}^{-}$ ions in solution attract the $H^{+}$ ions forming water and leaving the adsorption sites for the phenolate ions (Ge et al., 2019). Ge et al. (2019) highlighted similar observations whilst using modified magadiite to absorb phenol.

It should also be noted that (Focke et al., 2014) who used the same precursor expandable graphite as the one used in this study confirmed the presence of oxygen functional groups on the exfoliated graphite. These functional groups were postulated to result from the partial oxidation of graphite during intercalation and exfoliation. During the intercalation process, the delocalized electrons in the conjugate pi orbitals of graphite interact with reacting species forming ionic compounds graphite salts. These, species can either donate electrons to graphite or accept electrons from graphite (Noel and Santhanam, 1998). For example, strong oxidizing agents will accept electrons from graphite to form acceptor compounds leading positively charged carbon layers on the graphite lattice. It is thus plausible that Cr (VI) in the form of the oxyanions interact with graphite this manner, during adsorption. Also, it has been reported that interactions exist between conjugated pi-electron electron systems (Wang et al., 2014; Yu et al., 2017). It is therefore plausible that the aromatic ring of phenol and graphite interact this manner (Yu et al., 2017; Tang et al., 2018). These mechanisms together with the contribution of the pH effects posited to be responsible for the removal of the target pollutants in the present study.

3.4 Adsorption isotherms

The mechanism of interaction between the adsorbates and the exfoliated graphite can be further elucidated by determining the equilibrium adsorption isotherms and kinetics of the systems. The model parameters were calculated by minimizing equation (5) and validated using equation (6). The results obtained through the fitting of the various adsorption isotherms to experimental data are presented in Fig. 5 while their corresponding constants are shown in Table 1. The Redlich-Peterson isotherm best described the Cr (VI) adsorption as confirmed by the regression coefficient °(R^2°= 0.9965) and a low MPSD value (2.8532) obtained. A Redlich-Peterson exponential constant, β that lies between 0 and 1 indicates good adsorption (Gupta and Balomajumder, 2015). Fig. 5 (a) shows a close relationship between the fit of Langmuir and Redlich-Peterson models which can be attributed to the high exponential constant (β = 0.8982). As this constant approaches’ unity, the Redlich-Peterson model reduces to the Langmuir model. For phenol adsorption, Fig. 5 (b), the Redlich-Peterson model reduced to the Langmuir model (β = 1). Several researchers (Aksu and Akpinar, 2000, Wong et al., 2004, Gimbert et al., 2008, Shahbeig et al., 2013) reiterated that the exponential constant cannot be greater than 1, hence this was used as a constraint in the solver function. Consequently, the Langmuir isotherms best describes the phenol adsorption on the EG with (R² = 0.9975) with a smaller MPSD value (MPSD = 9.3608). Parameters obtained from the Langmuir isotherm such as K_L were used to establish the separation factor (R_L) which indicates the likelihood of the adsorption process taking place. This separation factor is calculated using equation (11)

Close

Fig. 5

Adsorption isotherms for (a) Cr (VI) at pH 2, adsorbent loading of 30 g/L and 48 h contact time and (b) Phenol at pH 12, adsorbent loading of 30 g/L and 48 h contact time.

Export to PPT

For ${0<R}_L<1$ , the adsorption is favorable, $R_L>1$ , the adsorption is unfavorable, $R_L=1$ , adsorption is linear and $R_L=0$ , adsorption is irreversible (Belhachemi and Addoun, 2011, Gupta and Balomajumder, 2015). All the $R_L$ values lie between 0 and 1 for both phenol and Cr (VI) and decrease as the concentration increases. This is indicative of a favorable adsorption process especially at lower concentration. The $R_L$ values of phenol are much higher than those of Cr (VI) for the same initial concentration suggesting that indeed the Langmuir model best describes the phenol adsorption when compared to the Cr (VI) adsorption. The Freundlich model resulted in the lowest regression coefficient values for phenol (R² = 0.9774) and Cr (VI) (R² = 0.9666) and which yielded the highest values of MPSD for phenol (MPSD = 30.98) and Cr (VI) (MPSD = 22.50) and suggesting a poor fit and no significant representation of the experimental data. The model assumes a heterogeneous system, which as proven by the heterogeneity factor is not true for this adsorption process since the heterogeneity factor for phenol and Cr (VI) is above 0.1. The current adsorption results have significantly improved from our initial results from an adsorption capacity of 0.55 mg/g and 0.73 mg/g to 2.14 mg/g and 2.23 mg/g for phenol and Cr (VI) respectively. This is mainly attributed to the optimized conditions of pH. Even though the results shows a good fit and high percentage removal, the adsorption capacity of the EG remains significantly lower as compared with other adsorbents such as activated carbon (7.8 mg/g for Cr (VI) (Rai et al., 2016) and 98.6 mg/g for phenol (Lütke et al., 2019)), chitosan (12.4 mg/g for Cr (VI) (Ferrero et al., 2014)), polymer materials (500 mg/g for Cr (VI) (Yu et al., 2013), 51.92 mg/g for phenol (Makrigianni et al., 2015) as well as 38.2 mg/g for Cr (VI) (Tshemese et al., 2020)), clay materials (320.0 mg/g for Cr (VI) (Akram et al., 2017)), honeycomb carbon materials (332.5 mg/g for Cr (VI) (Liang et al., 2019)) and graphene (28.3 mg/g for phenol (Li et al., 2012)). The EG material however has adsorption capacity comparable to other adsorbents such as saw dust (3.7 mg/g for phenol (Meniai, 2012)), clay materials (4.3 mg/g for Cr (VI) (Zhao et al., 2005), soil (9.5 mg/g for phenol (Subramanyam and Das, 2009, Păcurariu et al., 2013)). and carbon nanotubes (12.5 mg/g for Cr (VI) (Huang et al., 2015)). Li et al. (2015) also reported adsorption capacities within our range for the removal of Cr (VI) using Zeolitic imidazolate framework-67 micro-crystals. They reported 5.9, 9.3 and 13.3 mg/g for initial solutions of 6, 10 and 15 mg/L respectively. The adsorption capacities from the present study were higher than 3 adsorbents reported by (Mekonnen et al., 2015), namely avocado kernel seeds (1.0 mg/g), Juniperus procera saw dust (1.3 mg/g) and papaya peels (0.8 mg/g).

3.5 Adsorption kinetics

The kinetic experimental data was fitted to the models represented in equations (7) to (10). The results are shown in Fig. 6. The parameters of pseudo first order, pseudo second order and the intra-particle diffusion model were obtained using linear plots and determining the slope and intercept whereas the parameters of Elovich’s equation were obtained through the Solver function in excel 2013 by minimizing the error. The associated constants are shown in Table 2.

Close

Fig. 6

Kinetic data fits using (a) pseudo first order (b) pseudo second order (c) Elovich's equation (d) Intra-particle diffusion model. The triangles (Δ) represents raw data for Cr (VI) and the diamond (◊) represents the raw data for phenol. The dashed lines represent the model fits for the respective models.

Export to PPT

Both the pseudo first order and Elovich’s equation did not adequately represent the experimental data. This is in agreement with the relatively low regression coefficient value, R² < 0.99 and the inconsistencies between the fitted $q_e$ and experimental values. To determine the nature of diffusion, the intra-particle diffusion model was adopted for Cr (VI) and phenol respectively. Based on equation (10), a plot of $q_{t}vs{t}^{0.5}$ is a straight line when the intra-particle diffusion is a rate-limiting step. In case the intra-particle diffusion is the sole rate-limiting step, the plot will pass through the origin. In this study however, this is not the case, suggesting that the kinetics could be dominated by film diffusion and intra-particle diffusion at the same time (Qiu et al., 2009, Gupta and Balomajumder, 2015). The pseudo second order model adequately depicts the kinetic behavior of the removal of these pollutants on EG. This implies that the removal process is predominately driven by chemisorption and also depends on the physiochemical interactions between the EG and respective pollutants (da Gama et al., 2018).

3.6 Thermodynamic modelling

An understanding of the thermodynamics of the system is necessary in order to deduce if a process is spontaneous, endothermic or not. To determine this, non-linear isotherms of adsorption at three predetermined temperatures were conducted. Using the best fitting model, an equilibrium constant was obtained for the respective temperatures. The corresponding thermodynamic parameters of adsorption were obtained using the Van’t Hoff’s equation (12) which measures the changes in equilibrium constants with variations of the temperature. A graph of $\ln Ke$ vs 1/T was constructed to determine the values of the standard enthalpy change ( $\mathrm{\Delta }{H}^o)$ and standard entropy change ( $\mathrm{\Delta }{S}^o$ ) from the slope and intercept of the linear plot while the standard Gibb’s free energy change ( $\mathrm{\Delta }{G}^o)$ was determined using equation (13) where

T is the absolute temperature (K), R is the universal gas constant (8.314 J/ K.mol) and K_e is the thermodynamic equilibrium constant.

The equilibrium constant obtained in the respective isotherms was expressed in the units (L/mg) and was converted to a dimensionless constant (K_e) using the method detailed by (Lima et al., 2019).

The thermodynamic study of both phenol and Cr (VI) adsorption resulted in positive values of $\mathrm{\Delta }{H}^o$ indicating an endothermic adsorption process (Table 3). The negative values of $\mathrm{\Delta }{G}^o$ at all temperatures for both pollutants indicated a spontaneous and a feasible process. High positive values of $\mathrm{\Delta }{S}^o$ particularly for Cr (VI) were obtained, this reflects a high affinity of EG towards the respective pollutants. The positive $\mathrm{\Delta }{S}^o$ values also indicate increased randomness in the solid–liquid interface and increased degree of freedom of the sorbate and a favorable adsorption process (Ghosal and Gupta, 2017).