Introduction

The industrialisation and urbanisation of nations sometimes sacrifice the environment for short-term economic gains. Water pollution is a serious concern, and without proper management will lead to irreversible environmental damages. Textile industry is a very water-intensive industry and is ranked as top ten most polluting industry in respect to water pollution (Kant 2012; Rasul et al. 2006). Improper practice of discharging textile wastewater into rivers is still happening, especially in many lower income countries (de Lima et al. 2007; Rasul et al. 2006).

Many dye wastewater methods have been innovated and improved. To name a few remediation methods, there are adsorption, phytoremediation (Kooh et al. 2016a), photocatalyst (Saravanan et al. 2013) and nanoparticles (Agarwal et al. 2016). Adsorption is currently the most popular and most researched method in the last decade (Wang et al. 2011) due to its simplicity in design and can be easily integrated into factory. The cost of wastewater remediation by adsorption depends on the type of adsorbents. To ensure economic feasibility of the wastewater remediation process, one may choose materials that are renewable, sustainable, abundantly available or materials with little values such as agricultural wastes (Gupta and Nayak 2012; Kooh et al. 2016b) and weeds (Lim et al. 2014). Activated carbon may have been ideal materials for certain dyes and heavy metals; however, to produce them may be energy intensive.

This study investigates the pitcher and the leaves of Nepenthes rafflesiana (NR), a species of pitcher plant, as potential adsorbents for removal of methyl violet 2B (MV) dye. Nepenthes is a carnivorous plant, usually with a jug-like leaf which attracts, traps and digests animal to meet its nutritional needs (Clarke 1997). Nepenthes is distributed in Northern Australia, South East Asia and Southern China, while vast majority occurred in Islands of Borneo and Sumatra. NR is one of the most common varieties of Nepenthes, found in open, sandy, wet areas, cleared areas, peat swamp forest and kerangas forest (heath forest) (Clarke 1997; Osunkoya et al. 2007). NR is characterised by its huge size and can be measured up to 35 cm long and 15 cm wide (Clarke 1997). NR also contains lignocellulosic materials (Osunkoya et al. 2008) which are known to adsorb dyes (Parab et al. 2010). Currently, there is no report of using Nepenthes as an adsorbent for remediation of dye wastewater. MV is a purple organic dye, belonging to the triphenylmethane class. MV is chosen as a dye of focus because of its industrial importance in textile, rubber, inkjet, paper and petroleum products (Sabnis 2010). MV is also a known irritant to eyes, skin, gastrointestinal and respiratory tract and is also a known carcinogenic and mutagenic agent (Michaels and Lewis 1985; Vachalkova et al. 1995). The novelty of this study is the use of pitcher plant as an adsorbent as no studies on the use of this plant as an adsorbent have previously been reported.

Materials and methods

Preparation of adsorbent and dye stock solution

Nepenthes pitcher (NP) and Nepenthes leaves (NL) were obtained from the hills outside the compound of Universiti Brunei Darussalam, Brunei Darussalam, Northern Borneo. Samples were washed with distilled water, sonicated and then dried in an oven at 70 °C. Dried samples were blended into fine powder using a kitchen blender and then sieved size of <355 μm were collected and used for the adsorption experiment.

Methyl violet 2B dye (MV) (M r 393.95 g mol−1, 80% dye content, Sigma-Aldrich) was used without further purification. A 500 mg L−1 stock solution was prepared and serial dilutions were carried out to obtain lower dye concentrations.

Characterisations of adsorbents

Fourier transform infrared (FTIR) spectra (KBr disc approach) were carried out using a Shimadzu Model IRPrestige-21 spectrophotometer. The surface morphology analyses of the adsorbents were carried out using a scanning electron microscope (SEM) (Tescan Vega XMU). The determination of the adsorbents’ point of zero charge (pHpzc) was done by salt addition approach using 0.1 mol L−1 KNO3 solutions (Zehra et al. 2015). All measurements of pH were done using a Thermo-Scientific digital pH meter.

Batch adsorption procedures

In general, batch adsorption experiments were carried out by simply agitating 0.04 g of adsorbents in 20 mL dye solutions using an orbital shaker set at 250 rpm, unless stated otherwise.

Preliminary experiments were carried out to investigate the effect of adsorbent dosage, using 100 mg L−1 MV (unadjusted pH) with the studied adsorbent dosage ranged from 0.01 to 0.06 g. The optimum adsorbent dosage was found to be 0.04 g for both NP and NL, and this was used throughout the experiment. The data is not shown for brevity.

The effects of pH (2–9), ionic strength (0–0.8 mol L−1), dye concentrations (20–500 mg L−1), and duration of agitation (0–180 min) were investigated. One parameter was changed at a time, while the rest were kept constant.

The adsorption capacity, q e (mg g−1), and removal efficiency (%) were used for measuring the amount of dye adsorbed by the adsorbents. The equations are as followed:

$$q_{\text{e}} = \frac{{\left( {C_{\text{i}} - C_{\text{e}} } \right) V}}{m}$$
(1)
$${\text{Removal efficiency }} = \frac{{\left( {C_{\text{i}} - C_{\text{e}} } \right) \times 100 \% }}{{C_{\text{i}} }}$$
(2)

where C i is the initial dye concentration (mg L−1), C e is the dye concentration of the filtrate after adsorption (mg L−1), V is dye volume (L) and m is the adsorbent mass (g).

Isotherm and kinetics modelling

The adsorption data were fitted into six isotherm models, namely the Langmuir (1916), Freundlich (1906), Dubinin–Radushkevich (D–R) (Dubinin and Radushkevich 1947), Tempkin (Tempkin and Pyzhev 1940), Redlich–Peterson (R–P) (Redlich and Peterson 1959) and Sips (1948) models, and also to various kinetics models such as the pseudo-first-order (PFO) (Lagergren 1898), pseudo-second-order (PSO) (Ho and McKay 1999) and Weber–Morris intraparticle diffusion (WMID) (Weber and Morris 1963) models. The equations and linear plots of these models are summarised in the Table 1.

Table 1 Equations of the isotherm and kinetics models
Full size table

Error analyses

The kinetics and isotherm models that best characterised the adsorption data are commonly chosen based on the values of the coefficient of determination (R 2). However, the linearisation of kinetics and isotherm models from their non-linear forms may violate the normality assumptions of standard least squares, leading to error (Ho 2004; Ratkowsky 1990). Chi-square test (χ 2) and average relative error (ARE) are included in this study, to determine the best-fitted models. The equations of χ 2 and ARE are as follow:

$$\chi^{2 } : \mathop \sum \limits_{i = 1}^{N} \frac{{(q_{\text{e,exp}} - q_{\text{e,cal}} )^{2} }}{{q_{\text{e,exp}} }}$$
(12)
$${\text{ARE}}: \frac{100}{n}\mathop \sum \limits_{i = 1}^{N} \left| {\frac{{q_{\text{e,exp}} - q_{\text{e,cal}} }}{{q_{\text{e,exp}} }}} \right|_{i}$$
(13)

where q e,exp is the q e value obtained experimentally, q e,cal is the calculated value dived from the isotherm or kinetics models and n is the number of data points included in the calculation.

Thermodynamics experiments

The thermodynamics nature of the adsorption process was investigated using the Van’t Hoff equation (Eq. 14), and the studied temperature ranged from 25 to 65 °C. The equations are as follow:

$$\Delta G^{\text{O}} = H^{\text{O}} - T\Delta S^{\text{O}} = - RT\; {\text{ln }}k$$
(14)
$$k = \frac{{C_{\text{i}} - C_{\text{e}} }}{{C_{\text{e}} }} .$$
(15)
$$\ln k = \frac{{\Delta S^{\text{o}} }}{R} - \frac{{\Delta H^{\text{o}} }}{RT}$$
(16)

where T is the temperature (K), ∆G o (kJ mol−1) is the Gibbs’ free energy, ∆S o (J mol−1 K−1) is the entropy change, ∆H o (kJ mol−1) is the enthalpy change, k is the distribution coefficient for adsorption, and R is the gas constant (8.314 J mol−1 K−1). The ∆S o and ∆H o were calculated from the linear plot of ln k vs 1/T.

Regeneration experiment

In this study, distilled water and 0.1 mol L−1 NaOH were used for regenerating the spent adsorbents. The detailed procedure is available in our previous work (Dahri et al. 2014). Briefly, the spent adsorbents were washed in distilled water until no desorption occurs, and the regenerated adsorbents were dried in an oven at 70 °C overnight. To wash using NaOH, the spent adsorbents were agitated for 30 min, followed by repeated distilled water washing until the pH of the washed solution was near neutral and the regenerated adsorbents were then dried in an oven at 70 °C.

Results and discussion

Characterisation of adsorbents

As summarised in Table 2, functional group analyses using FTIR revealed that both NP and NL contain O–H, –NH3, aliphatic C–H, N–H, phenyl, C–O–C, and –COO groups. For NP-MV and NL-MV, shifts of O–H, –NH3 and –COO bands were observed which indicate that these groups may have interacted with the MV dye molecule.

Table 2 Summary of kinetics data fitted into PFO, PSO and WMID models
Full size table

Functional group

Wavenumber (cm−1)

NP

NP-MV

NL

NL-MV

O–H and –NH3

3384

3409

3378

3382

Aliphatic C–H stretch

2917

2919

2918

2918

N–H bending

1620

1620

1619

1620

Phenyl

1451

1451

1450

1450

C–O–C

1035

1036

1036

1036

–COO anti-symmetric stretching

1519

1519

1524

1514

–COO symmetric stretching

1377

1368

1370

1368

The surface morphology of untreated and MV-treated adsorbents were as shown in Fig. 1. The surface of NP and NL appeared to be irregular in shape with some noticeable fibres. After dye treatment, there are no significant changes in the structure of both the adsorbents, therefore, did not result in any major distinguishable differences.

Fig. 1
figure 1

SEM images of a NP, b NP-MV, c NL, and d NL-MV

Full size image

Effects of pH and ionic strength

The three main mechanisms for dye interaction with adsorbent are electrostatic interaction, hydrophobic–hydrophobic interaction and hydrogen bond. Electrostatic interaction is the attraction between the charged functional group of the adsorbent with the dye molecule of opposite charge. Hydrophobic–hydrophobic interaction occurs between the non-polar groups of dyes and the non-polar groups on an adsorbent’s surface (Hu et al. 2013).

In adsorption study, pH and ionic strength are very important parameters to investigate. The pH can directly affect the functional groups on the adsorbents, as well as affecting the charge of the dye molecules, while the ionic strength can influence the electrostatic interaction between the dye and the adsorbent and the hydrophobic–hydrophobic interaction (Al-Degs et al. 2008; Hu et al. 2013).

The point of zero charge, pHpzc, for NP and NL were determined to be 4.5 and 3.9, respectively. Theoretically, the functional groups on the surface of NP will be predominate negatively charged at pH higher than 4.5, and predominate positively charged at pH lower than 4.5. As seen in Fig. 2a, there was little influence of medium pH on adsorption of MV onto NP as the adsorbent. At pH 2.6, the predominant positive charged surface of NP should lead to electrostatic repulsion with the cationic MV molecules, however, removal efficiency of 78% was obtained, which indicates that electrostatic interaction may not be the dominant mechanism for the interaction of NP-MV system. Similar observation was seen for the NL-MV system. Similar behaviour was also observed in our previous work on the removal of MV and malachite green dyes using Azolla pinnata (Kooh et al. 2015; Kooh et al. 2016c).

Fig. 2
figure 2

Plots showing the effects of a pH (0.04 g adsorbents, 20 mL 100 mg L−1 dye), and b ionic strength (0.04 g adsorbents, 20 mL 100 mg L−1 dye, unadjusted pH)

Full size image

The effect of ionic strength is as shown in Fig. 2b, where the increase of ionic strength led to reduction of the removal efficiency. The removal efficiency decreased from 72 to 44% for NP, while NL decreased from 86 to 58%, as NaCl concentration increased from 0 to 0.8 mol L−1. The reduction in removal efficiencies is due to the suppression of the electrostatic interaction as the surplus Na+ ions compete with MV molecules for the adsorption sites (Hu et al. 2013). The increase of ionic strength also compressed the electrical double layer which moves the adsorbents and the dye molecules closer together which may enhance the hydrophobic–hydrophobic interactions (Binnie et al. 2002; Hu et al. 2013). Despite the suppression of electrostatic interaction, for the adsorption to maintain high removal efficiency at high ionic strength is an indication that hydrophobic–hydrophobic interaction may play larger role in NP-MV and NL-MV systems. Furthermore, the hydrophobic interaction can be enhanced by the presence of salt, whereby the electric double layer decreases in thickness allowing the dye molecules and adsorbent particles to move closer to each other (Binnie et al. 2002).

Effect of contact time and kinetics modelling

The effect of contact time is to ensure that the duration of agitation time is sufficient to attain the equilibrium of the adsorption process. As seen in Fig. 3a, b, the equilibrium for both NP-MV and NL-MV systems were achieved within 120 min. The rapid increase in dye uptake during the first 60 min is due to the vacant adsorption sites, which become occupied and less available over time, resulting in the much lower increase in dye uptake and eventually reaching equilibrium.

Fig. 3
figure 3

Plots showing the effect of contact time for a NP-MV, b NL-MV, using three different concentrations of MV, and the WMID plot of c NP-MV and d NL-MV

Full size image

The kinetics data were assessed using PFO and PSO kinetics models and the data are summarised in Table 2, where it can be observed that the R 2 for PSO is close to 1.0 when compared to PFO values. When comparing the q e,cal and q e,exp, the values were also much closer for PSO, while for PFO the values are much further apart. Lastly, the χ 2 and ARE values for PSO are very low compared to PFO. Therefore, based on R 2, comparison of q e,cal and q e,exp, and the error functions values, the kinetics data for the removal of MV by both the adsorbents are best described by the PSO model. This indicates the rate of adsorption of NP-MV, and NL-MV systems follows the second-order rate law, however, referring to the amount of adsorption sites on adsorbent’s surface and not the amount of adsorbate in bulk solution (Liu 2008).

To describe the diffusion process, WMID is used as PFO, and PSO models were not applicable. The diffusion process can be generally divided into three phases: the rapid external diffusion phase, the intraparticle diffusion phase and the slow equilibrium phase. As seen in Fig. 3c, d, multi-linearity of the WMID plots observed indicates the presence of different phases of adsorption (Wang and Li 2007). The rapid external diffusion phase was not observed in the WMID plot and it is most likely to be completed within the first 5 min of agitation. The first linear region as seen in WMID plot is the intraparticle diffusion phase, while the second linear region is the slow equilibrium phase. The WMID model states that if the linear WMID plot crosses the origin, then the diffusion mechanism is controlled by intraparticle diffusion. As seen in Table 2, the intercepts, C, for both NP-MV and NL-MV adsorption systems are non-zero, therefore, indicating that the intraparticle diffusion is not the rate-limiting step. The C is also proportional to the thickness of the boundary layer.

Effect of initial dye concentrations and isotherm modeling

The effect of initial dye concentrations is as shown in Fig. 4, where it can be observed that increase of dye concentration resulted in higher dye adsorption. This behaviour is attributed to the Fick’s diffusion law which stated that the concentration gradient provides driving force for the mass transfer rate, hence higher C i resulted in higher q e (Frijlink et al. 2015).

Fig. 4
figure 4

Plots showing the effect of initial dye concentration for NP-MV and NL-MV systems

Full size image

The experimental data on the adsorption of MV onto NP and NL were fitted to five different isotherm models. These isotherm models are useful to describe the adsorption process and each of the isotherm models’ applications and theories are discussed in literature (Foo and Hameed 2010). Table 3 summarises the parameters for the isotherm models, as well as the error analyses. Comparing the five isotherm models’ R 2 values for NP, it can be seen that Langmuir > Freundlich > Sips > Tempkin > R–P > D–R while for NL, Langmuir > Sips > Tempkin > Freundlich > D–R > R–P. This indicates that the Langmuir is the best fit isotherm model for the experimental data of both the adsorbents which is further supported by the low χ 2 and ARE values. Thus, according to the Langmuir model, the monolayer adsorption is involved.

Table 3 Summary of parameters of six isotherm models
Full size table

The Freundlich model has poor fitting into the experimental data as it is more suitable to be used on low concentration of dye (Foo and Hameed 2010; Rushton et al. 2005). D–R model was originally used to describe adsorption of gas onto activated carbon with van der Waals forces as the main mechanism. In this study, the adsorption process involved many mechanisms, which is why the D–R model displayed poor fitting into the data. Tempkin model excels in gas phase equilibrium prediction but is not suitable to be used in complex adsorption system (Kim et al. 2004).

The R L values for NP and NL are both determined as 0.1 which suggested that the adsorption process is favourable. The q m value for NP is 288 mg g−1 while for NL is 194 mg g−1. Both adsorbents’ MV removing capacities are better than peanut straw char (101 mg g−1) (Xu et al. 2011), Posidonia oceanic leaf (119 mg g−1) (Cengiz and Cavas 2010), jackfruit seed (126.7 mg g−1) (Dahri et al. 2016) and Casuarina equisetifolia needle (165.0 mg g−1) (Dahri et al. 2013). NP ability to remove MV is on par with pu-erh tea (286 mg g−1) (Li et al. 2010) while NL is on par with soyabean waste (180 mg g−1) (Kooh et al. 2016b) and Azolla pinnata (194 mg g−1) (Kooh et al. 2015), however, lower than water lettuce (267.6 mg g−1) (Lim et al. 2016) and duckweed (332.5 mg g−1) (Lim et al. 2014).

Thermodynamics experiment

Thermodynamics experiments provide useful insights on the thermodynamics nature of the adsorption process. The ∆G o obtained for the NP-MV system at the temperature 25, 35, 45, 55 and 65 °C are −2.27, −2.55, −2.42, −2.86 and −3.24 kJ mol−1, respectively, while for the NL-MV system are −3.58, −4.55, −5.19, −4.77 and −5.33 kJ mol−1. The negative ∆G o values indicates that the adsorption process for both NP-MV and NL-MV systems are spontaneous. The ∆H o for NP-MV and NL-MV systems are 3.91 and 7.40 kJ mol−1, respectively, which indicate that the adsorption processes for both the adsorbents is endothermic in nature. The ∆S o for NP-MV and NL-MV systems are 20.5 and 37.7 J mol−1 K−1, respectively, where the positive values indicate favourable and increase of randomness in the adsorption.

Regeneration experiment

Disposal of spent adsorbent by conventional methods may not be ideal, i.e., sending to the landfill will cause the hazardous wastes to leach from the spent adsorbent, while incineration may lead to release of poisonous gases. Alternative method such as the use of regenerated spent adsorbent is not only esthetic, but also provided added value to the adsorbent in respect to its reusability. The data of regeneration experiment is as shown in Fig. 5. Washing using distilled water and the base were able to regenerate the spent adsorbents and achieved removal efficiencies higher than unused adsorbent. Regeneration using base was able to achieve removal efficiency higher than distilled water washing. This behaviour is not uncommon and could be due to removal of lignin and low molecular weight wax caused by repeated washing, leading to the exposure of adsorption sites favourable for adsorption of MV dyes and has also been observed in our previous work (Kooh et al. 2016c).

Fig. 5
figure 5

Regeneration experiments showing several wash cycles using distilled water and NaOH for a NP and b NL

Full size image

Conclusions

In this study, we report the use of two new adsorbents derived from Nepenthes rafflesia for the removal of toxic methyl violet 2B dye. Both the adsorbents gave high maximum adsorption capacities towards MV dye as compared to many reported natural adsorbents. Kinetics of the adsorption process followed the pseudo-second-order and intraparticle diffusion was found not to be the rate determining step. Both the adsorbents were also able to maintain its removal efficiency under different medium pH. High q m values coupled with successful regeneration of the spent adsorbents gave added values for them to be utilized as potential adsorbents for the removal of MV dye in wastewater treatment.