Surface modification of date seeds (Phoenix dactylifera) using potassium hydroxide for wastewater treatment to remove azo dye

Water shortage and contamination rank equivalent to environmental change as the most ecologically disruptive issues in the 21st century. Textile industry effluents contain many synthetic dyes, which are of major concern, and importantly many are harmful pollutants (Wong et al. 2018). Therefore, it is important to manage dye concentrations in wastewater before release to the environment. Synthetic dye removal is difficult since most are non-biodegradable and highly resistant to oxidation. The removal methods available for dealing with dyes, organic matter, heavy metals, etc, include advanced oxidation processes, membrane treatment, photocatalysis and adsorption. Adsorption is one of the most popular methods among these, especially using waste materials as the adsorbent(s) because of their low cost.

Agricultural waste residues, especially fruit shells, seeds, peel and husks are important for use as adsorbents because using them in wastewater treatment enhances their value as well as reducing disposal problems (Adegoke & Bello 2015; Ahsaine et al. 2018; de Souza et al. 2018). Activated carbon is widely used in wastewater treatment to remove contaminants. It is obtained primarily from materials with a high carbon content, like wood, lignite, coal, waste sludge, etc. Various groups are working on the preparation of appropriate adsorbents from such wastes (Ganesapillai et al. 2017; Danish & Ahmad 2018; Danish et al. 2018) and, because of its higher specific surface area and more active sites, normally such activated carbon is among the best adsorbents available for removing pollutants from wastewater. The disadvantage of activation is that it requires a significant energy input and is expensive. Therefore, there is a real need to find new, low-cost materials for use as adsorbents (Abbas & Ahmed 2014).

Date palm trees occur widely in many regions, especially the Middle East. The seeds constitute about 10 to 15% of the entire fruit. Because of the large-scale production of dates, seed disposal is a growing concern in the region. Instead of using the seeds as a direct fuel or as animal feed – e.g., for cattle or sheep (Al-Farsi & Lee 2011; Demirbas 2017) – they can be used in wastewater treatment, with surface modification (Daniel et al. 2012), under simple conditions. Date seeds consist mainly of cellulose, hemicelluloses and lignin (Nabili et al. 2016). Both lignin and cellulose are organic polymers present in plant species, but they differ in chemical structure. Cellulose comprises complex organic carbohydrate monomers, while lignin consists of organic non-carbohydrate monomers that are difficult to degrade.

Pyrolysis is an irreversible process involving the thermal decomposition of matter at high temperatures under anoxic conditions. When date seeds are pyrolyzed, their cellulose and hemicellulose components decompose, giving rise to pores that directly increase the surface area, improving their application as adsorbents. Surface modification of the carbon materials is achieved with thermal energy on its own or in combination with chemical energy. Chemicals commonly used for surface modification are inorganic salts (Yadavlli et al. 2017), phosphoric acid (Wu et al. 2018) or potassium hydroxide (Gottipati & Mishra 2013; Araga et al. 2017). Modification of carbon surfaces gives rise to positive charges, which enhance the adsorption of anionic adsorbates.

Methyl orange (MO) is difficult to degrade since it is an azo dye containing an group, the presence of which causes serious problems in water reuse and creates environmental threats on discharge (Hameed et al. 2017).

In this study, low-cost and readily available date seeds were modified to become an adsorbent by applying minimal heat and chemical (potassium hydroxide – KOH). The efficiency of this material as an adsorbent for removing MO was then tested. The aim was to achieve the optimum condition for removing MO dye using the modified carbon (MC) in a 2³ factorial design.

Medjool date (Phoenix dactylifera) seeds were purchased from a local market in UAE. KOH was procured from Sisco Research Laboratories (SRL) Pvt. Ltd., India, and used directly as the surface modifying agent. Deionized water was used in all experiments. All experiments were performed at least three times and average values are reported.

Pyrolysis was carried out using an electric muffle furnace (Labtech, Korea) at 150 °C for one hour at a heating rate of 2 °C/min to prepare the MC. This was done by soaking the date seed carbon (DSC) in 1 M KOH for 24 hours, washing with distilled water and filtering using Whatman No. 1 filter paper until the pH was neutral (pH = 7). The MC was then dried in a hot-air oven at 100 °C for 30 minutes. The Fourier transform infrared (FT-IR) spectra of the DSC and MC were recorded between 500 and 4,000 cm⁻¹ using a Perkin Elmer Spectrum BX FT-IR spectrophotometer (Waltham, MA, USA). Scanning electron microscopy (SEM) images were also obtained using a Quanta FEG-250 scanning electron microscope for both samples. X-ray diffraction (XRD) analyses were performed of DSC and MC using a Rigaku Ultima IV X-day diffractometer (Tokyo, Japan).

A 2³ factorial designs are used to find the optimal value of the affecting factors in a specified range without wasting time, energy and chemicals (Rathinam et al. 2011). It was employed in MINITAB 17 to evaluate the main effects of all factors and their interactions. Optimal values were determined by conducting experiments (two trials) of all possible factorial combinations – i.e., initial MO concentration, adsorbent dosage and temperature. The factorial maxima and minima, as evaluated, are shown in Table 1. The process was optimized by analyzing the main and interaction effects, Pareto and normal probability charts, and residual analysis.

Before studying the use of MC in MO removal by adsorption, the materials were characterized. Fourier Transform infrared (FTIR) analysis was used to determine the surface-functional groups and the results are shown in Figure 1. The peaks observed match well with those reported in the literature for in aliphatic hydrocarbons (Cuhadaroglu & Uygun 2008). Peaks in the range 2,800 to 2,980 cm⁻¹ indicate the presence of primary aliphatic (⁠⁠) symmetric and asymmetric stretching, and the peak intensity indicates that the DSC has more aliphatic bonds than the MC. Peaks between 2,300 and 2,400 cm⁻¹ represent carbon dioxide in the samples. Figure 1 shows that the MC has more carbonyl groups (>C = O) than the DSC. The peaks at 1,452 and 1,370 cm⁻¹ indicate the presence of aliphatic bending bonds. Both samples exhibit medium peaks between 1,330 and 1,420 cm⁻¹. The peaks between 700 and 720 cm⁻¹ are very weak, and indicate 1,3-disubstituted alkanes.

Figure 2 shows the X-ray diffraction patterns for the DSC and MC, and both showed broad peaks indicating that they are largely amorphous. For the DSC, the peak was observed at around 20.9°, corresponding to 0.42556 nm d-spacing with hexagonal structure. After KOH treatment, however, the material's crystallinity increased, giving a comparatively sharp peak at 26.412° with 0.33746 nm hexagonal d-spacing. This indicates the formation of a [002] plane of a disordered graphite structure, possibly arising from the intercalation of into the carbon matrix, leading to lattice expansion and a more porous structure.

Figure 3 shows the SEM images of DSC and MC material. The low magnification images are shown in Figures 3(a) and 3(b). After modification, a well-developed porous structure was observed in the high-magnification images, as shown in Figures 3(c) and 3(d). Irregular pores of 6.25 ± 0.35 μm diameter are present in the modified material and are thought to be channels in the microporous network. The SEM images also show that both the DSC and MC have rough textures with heterogeneous surfaces, but there are many more randomly distributed pores in the modified material, which increase the surface area.

Table 2 shows the fits to the Langmuir and Freundlich isotherm equations for MO adsorption on MC. Comparison of the R² values shows that the adsorption follows the Langmuir adsorption isotherm model. Using this model, the equilibrium adsorption of MO was calculated at 66.26 mg/g of adsorbent, which is quite high compared to values given in the literature (Krika & Benlahbib 2015).

The effects of the three factors (C: adsorbate concentration, X: adsorbent dosing and T: temperature) were studied in all possible combinations with respect to MO removal through adsorption by MC – see Table 3. The factorial design was used to evaluate the effects of C, X and T independently, in two-way (C-T, T-X, X-C) and three-way combinations (C-X-T). The results are shown in Table 4. Any change in a factor level causes a change in the response on MO dye removal. All factors, including their interactions, were significant with probability, P < 0.05 at a confidence interval of 95%. The R² value is 99.67%, while the adjusted and predicted R² values are 99.38% and 98.68% respectively (Table 5). A high value of predicted R² indicates that responses for new observations are likely to be accurate.

An increase in the factors whose regression coefficients are positive results in an increase in removal efficiency, while an increase in those factors whose regression coefficients are negative causes a decrease (Hegazy et al. 2014). Within the range specified, an increase in initial MO concentration from low to high level increases the removal efficiency by 3.94%. The initial dye concentration is an important driving force in overcoming the mass transfer resistance between adsorbent and adsorbate (Safa & Bhatti 2011). Hence, as the initial MO concentration increases, the number of MO molecules competing for the vacant adsorption sites increases, as does removal efficiency (Rehman et al. 2012). Similarly, increasing the adsorbent dosage from a low to high level increases removal efficiency by 9.38%, and raising the temperature increases such efficiency by 11.3%. Normally, the energy of the adsorbate increases at higher temperature so desorption also increases. The temperature range used in the study was comparatively low however, at 25 to 45 °C, so adsorption increased at the higher temperature.

The response can be studied in terms of variations in the main effects – see Figure 4. The longer the sloping line, the greater the effect of the corresponding factor (Palanikumar & Davim 2009). It is clear that the most important factor affecting removal efficiency is temperature, followed by adsorbent dosage and then initial MO concentration.

The effects of interaction between factor pairs on MO removal affect the line orientation – see Figure 5. When the two lines are parallel, the specific interaction concerned reduces the effect. When the lines intersect, the interaction is more significant (Mathialagan & Veeraraghavan 2005). Hence, as can be seen in Figure 5, the interaction between initial MO concentration and adsorbent dosage is the most significant. When the initial MO concentration is increased from 100 to 400 mg/L, removal efficiency increases from 8.76 to 15.53% at an adsorbent dosage of 1 g/L, and from 21.41 to 36.2% at 5 g/L adsorbent dosage.

The effect of the individual parameters, two-way and three-way interaction on the MO removal efficiency was also verified using Pareto and normal probability charts. Both show similar results to the main and interaction effects. The Pareto chart (Figure 6) shows the different values for each effect and the minimum statistically effective value (Alcântara et al. 2016). At a 95% confidence interval, the minimum value is determined as 2.31 using Student's t-test, which is comparable to values reported in the literature (Carmona et al. 2005; Abdel-Ghani et al. 2009; Raja 2012). The Y-axis (Figure 6) represents the Student's t-test values for the main effects, and the two-way and three-way interactions. All the horizontal bars crossing the minimum statistically effective value correspond to 2.31 (vertical line), indicating that the effects of these three parameters are significant (Hegazy et al. 2014). The length of each bar shows the relative effect of the individual parameter. On the normal probability plot (Figure 7) of standardized effects, points very close to the line produce insignificant effects while those much further away produce highly significant effects (Geyikci & Buyukgungor 2013). All points to the right of the line have positive effects while those to the left have negative effects (Palanikumar & Davim 2009). Both charts confirm that temperature is the most important of the three individual effects, whereas, among the interactions between two parameters, that between initial MO concentration and adsorbent dosage is much more significant than the other interaction pairs.

Residuals are the differences between the experimental and fitted values. Figure 8 shows the normal probability plot of residual values for MO removal efficiency. The distribution is normal, i.e., the points are in reasonable alignment. Figure 9 is a residual plot of MO removal efficiency with different data fitted. There are no outliers and all points lie within the range +1.5 to −1.5. There is minimal deviation between the fitted and observed values (Rathinam et al. 2011).

A cost-effective adsorbent was prepared from date seeds, which are available locally, without excessive energy use. MC was produced from date seeds without activation with an MO adsorption capacity of 66.26 mg/g at 318 K. MO adsorption follows the Langmuir adsorption isotherm.

Adsorption was optimized using a 2³ factorial design to study the effect of three factors – initial MO concentration, adsorbent dosage and temperature. ANOVA and Student's t-test showed that the main effects and their interactions up to the third order were all significant.

The most significant main effect was temperature, while the most significant interaction was between initial MO concentration and adsorbent dosage. This is in accordance with the Pareto and normal probability charts. It is concluded from the study that an effective carbon adsorbent can be prepared, without activation, from cheap materials available locally.

The authors acknowledge Dr. R. Krishan, Central Analytical Laboratory, Birla Institute of Technology and Science, Hyderabad Campus, India, and The Petroleum Institute, Abu Dhabi for their kind support in characterizing the samples by XRD and SEM respectively.