Surface modification of date palm activated carbonaceous materials for heavy metal removal and CO₂ adsorption

3.1.1 Electron microscopy analysis

The SEM micrographs at different magnifications for K₂CO₃ and Cs₂CO₃ AC samples doped with nitrogen before (Fig. S1 a,b and e,f, respectively) and after (Fig. S1 c,d and g,h, respectively) adsorption of Pb(II) ions are given in Fig. S1. As expected, the samples exhibit irregular shapes and a presence of macropores/holes (as per SEM resolution) of different sizes and shapes which is generated during the inert pyrolysis process as a result of evolving of variety of gases. It is worth mentioning here that, not many differences are observed by SEM micrographs as a result of N-enriching or the use of different activation agents. Such microstructural changes can only be revealed by N₂ sorptiometry and transmission electron microscopy (if the materials are still stable when bombarded by high energy electrons).

3.1.2 N₂ Sorptiometry

Nitrogen sorptiometry isotherms for the studied AC samples are shown in Fig. 1 which depicts the effect of different activators on the surface characteristics of the ACs. The S_BET, V_tot, d_p are determined and found, depending on the activating agent used, to be in the range of 319–2704 m²/g, 0.23–1.59 cm³/g, and 2.0–4.9 nm, respectively (Table 1). Each activator proved to have a different effect on the isotherms. Ps-K₂CO₃-AC sample (Fig. 1a) exhibit Type Ib isotherm according to F. Rouquerol et al. new classification of isotherms (Rouquerol et al., 2013). Significant increase in the low-relative pressure range (P/P° less than 0.2) indicates a majority of micropores with complete overlap between adsorption and desorption curve (Xie et al., 2016; Xu et al., 2018). However, at higher relative pressure (P/P° > 0.2; i.e. above the knee), the nitrogen adsorption curves slowly increased up to a relative pressure of 0.9, then a larger increase at relative pressure above 0.9. This suggests that some extent of wider porosity (mesopores) and multilayers N₂ adsorption also present with minimum external surface area (S_ext). This sample (Ps-K₂CO₃-AC) shows no hysteresis which confirms the presence of majority of microporosity as clearly seen in the data reported in Table 1. Isotherms of AC with different activators are compared in Fig. 1a. The isotherm of Ps-Na₂CO₃-AC is of Type IIb which displays a H4 hysteresis loop at high relative pressure. It is worth mentioning here that Type IIb is a new classification proposed by F. Rouquerol et al. for isotherms similar to Type IV but with no plateau at high P/P° (Rouquerol et al., 2013). The emerging hysteresis loop within P/P° range from 0.4 to 0.9, commonly related to capillary condensation in the mesopores (Shoaib & Al-Swaidan, 2016; Meng et al., 2017). The presence of Types H4 hysteresis verifies the presence of slit-shaped pores, mostly in micropores range, which are commonly found in activated carbon. Isotherm for Ps-CaCO₃-AC is also of Type IIb but the hysteresis is of Type H3 (Rouquerol et al., 2013) with an open-end. H3 hysteresis is usually associated with also a slit-shape pore but mostly in the meso-range. Such open end hysteresis could be due to incomplete combustion during precursor carbonization resulting in decrease of micropores generation (Tobi et al., 2019). The specific surface area was highest for the Ps-Cs₂CO₃-AC sample (Fig. 1b & Table 1). More mesopores area (S_meso = 1647 m²/g) were evident than micropores area (S_micro = 812 m²/g) for this AC sample, as compared with the remaining tested activators (see Table 1). This different feature is clearly due to the presence of quite profound mesopores being reflected into a clear hysteresis as a result of capillary condensation. In Fig. 1b, the isotherm (in black color) starts with a moderate rise at low relative pressures, indicating the existence of micropores followed by a clear large hysteresis loop extended from P/P° = 0.4 to 0.9 which also indicates the presence of a large amount of mesopores (Xie et al., 2016; Cazetta et al., 2018). To the best of our knowledge, this sample has the highest S_BET of 2704 m²/g for studied palm-originated AC. Upon doping with urea, the surface area dropped dramatically, and the hysteresis size decreased due to a major reduction in the mesopores (see the data in red in Fig. 1b and Table 1). From Fig. 1c one can see that there is an occurrence of pore shrinkage in the case of Ps-HA and Ps-NNDDA ACs though the Ps-U, Ps-DDAB, and Ps-EA ACs showed similar well-developed pore structures.

Close

Fig. 1

N₂ sorptiometry isotherms showing (a) effect of different activators on Ps-ACs (b) before and after urea treatment for Ps-Cs₂CO₃-AC and (c) effect of different doping reagents on Ps-K₂CO₃-AC.

Export to PPT

3.1.3 Surface parameters and pore size distribution

Table.1 allows us to compare the specific surface area and pore size characteristics for AC originated from palm stem/leaf and for the synthesized N-enriched ones. It was found that the porosities of all studied carbons possess bimodal pore structure (micro and meso) but with Cs₂CO₃ activator, the mesoporosity is more profound. Ps-Cs₂CO₃-AC also showed the highest S_BET value of 2704 m²/g followed by Ps-K₂CO₃-AC with 1056 m²/g, then Ps-Na₂CO₃-AC with 663 m²/g and finally Ps-CaCO₃-AC with 319 m²/g. The dramatic increase of surface parameters in Ps-Cs₂CO₃-AC is due to the presence of extermly high amount of mesopores (almost 70 % mesopores and 30 % micropores; see its S_micro/S_BET% value in Table 1) whereas Ps-K₂CO₃-AC is characterized by a majority of micro- over mesopores (highest S_micro/S_BET% of 79). Under the same activation conditions, the adsorption average pore diameter, d_p, is ranged between 2.0 and 2.5 nm for all AC except for Ps-CaCO₃-AC which was around 5 nm. These results reveal that incomplete dissolution of CaCO₃ nanopowder during the activation step has altered the pore development. The reduction in S_BET was observed in all N-doped ACs which is due to a great reduction of the mesoporosity (see Table 1) as the use of N-enrichment reagents and the second 600 °C thermal treatment caused blocking/collapsing of the already developed mesopores. The N₂ sorptiometry parameters of ACs produced out of palm leaf (Pl samples) showed the same trend as that of stem-originated ones (Ps samples).

The trend could be further demonstrated by the pore size distribution obtained by the Horvath-Kawazoe method (Macías-García et al., 2006) (Fig. 2). On contrary to the other prepared ACs, Fig. 2 shows the presence of a distinctive high amount of mesopores in Ps-Cs₂CO₃-AC and Ps-Cs₂CO₃-U (data in black color). So, the use of Cs₂CO₃ activating agent resulted in ACs of bimodal extensive porosity that was reflected on its S_BET values even after N-enriching using urea (see also Table 1). It is noted that, apart from the Cs₂CO₃-activated samples, the average pore size distribution is around 2 nm for all the N doped ACs, whereas it extends to slightly larger pore size ranges in the non-enriched ones (except Ps-K₂CO₃-AC sample which remains at 2 nm due to large decrease in the microporosity). From the surface area and pore size distribution of the studied samples one can conclude that some of the prepared ACs (e.g. Ps-Cs₂CO₃-AC and Ps-Cs₂CO₃-U) would possess high adsorption capacity.

Close

Fig. 2

Pore size distribution using HK (Horvath-Kawazoe) method for ACs with (a) different activating agents and (b) different nitrogen doping reagents.

Export to PPT

3.1.4 X-ray diffraction

The samples produced from palm stem and leaf parts showed very similar X-ray diffraction with characteristics comparable to those found in AC. The XRD patterns of normal and N-doped ACs are compared in Fig. S2. One strong peak at 2θ = 23° and one weak peak at 43° were emerged corresponding to d₀₀₂ reflection of graphitic (sp² hybridized) carbon and d₀₀₁ corresponding to sp³ hybridized lattice reflection, respectively. The broad pattern clearly depicts the amorphous nature of activated carbon (Hashemian et al., 2014; Cazetta et al., 2018). There is a close relationship between the degree of graphitization and the interplanar spacing d₀₀₂ of carbon materials, i.e. the larger the diffraction angle, the smaller will be the degree of graphitization (Xie et al., 2014). In this work, similar degree of graphitization in the N-enriched samples was found if compared to the non-enriched AC.

3.4.1 X-ray photoelectron spectroscopy

To clarify the surface chemical states of elemental content, XPS spectra of the ACs were examined. The characteristics of the AC surface after the introduction of N into the carbon structure are illustrated by C1s, N1s, and O1s deconvoluted XPS peaks of urea-treated AC which is the sample with the highest N content (Fig. 4). C1s spectrum can be assigned to C—C/C⚌C, C⚌N/C—O, C—N/C⚌O, and O⚌C–O at 284.6, 286.0, 287.2, and 288.7 eV, respectively. The main C1s peak was at 284.6 eV which corresponds to graphite like sp² and sp³ hybridized carbon atoms. Also, peaks at N1s 398.3 and 400.3 eV can be attributed to N⚌C and N—C bonds, respectively (Marton et al., 1994; Rossi et al., 1994). Hence, the two observed nitrogen functional groups are sp² pyridinic (lower amount) and sp³ pyrrolic/pyridonic (higher amount) (Pietrzak, 2009; Ilnicka & Lukaszewicz, 2015; Rehman et al., 2019). A peak at 401.5 eV, which is corresponding to graphitic N, is absent in the prepared N-doped samples. In other words, out of the typical N-bonding configurations (pyridinic N, pyrrolic N/pyridine N, graphitic N, quaternary N, and oxidative N) which may be formed in the carbon networks, only the former two groups were able to be identified by XPS in our N-modified Pl/Ps-ACs. It can be presumed that the N-modification by urea, mainly incorporates pyridinic and pyrrolic N into the carbon skeleton and these two groups can act as the adsorption sites for heavy metals (Wang et al., 2019) and CO₂ (Lee et al., 2016). The O1s core level spectrum has a slightly wide peak at 532.9 eV which is assigned to phenolic groups (C-OH) and/or ether groups (C—O—C) (Xu, et al., 2018).

Close

Fig. 4

XPS envelope and deconvoluted spectra of N-enriched sample (activated with K₂CO₃).

Export to PPT

3.4.2 FTIR and Raman

The FTIR spectrum of raw samples and basic AC are given in Fig. 5a and that of nitrogen modified ACs is given in Fig. 5b. The raw samples have the most crowded spectrum. A typical strong and broad absorption band between 3600 and 3400 cm⁻¹ can be seen in the spectra shown, which is the characteristic of antisymmetric stretching vibration of free and intermolecular hydroxyl groups (O—H) in the cellulose, hemicellulose and lignin, and that of adsorbed water in the material. The broadening of this band is due to the high degree of hydrogen bonding (Ibrahim et al., 2006; Shoaib & Al-Swaidan 2016; Basheer et al., 2019). The spectrum of the raw sample is very much comparable to the one reported by Shoaib & Al-Swaidan (2016) for raw date palm fronds. We can assign the peaks at 3368 cm⁻¹ to the stretching mode of –OH in the raw sample, 2919 cm⁻¹ to aliphatic (–CH₂) symmetrical stretching, 1735 cm⁻¹ to C⚌O stretching (Cuhadaroglu & Uygun, 2008) and 1626 cm⁻¹ to C⚌C and aromatic ring stretching (Sevilla et al., 2017) and scissoring bend of –OH (Xiao et al., 2011). The peak at 1426 cm⁻¹ and 1373 cm⁻¹ corresponds to –CH bend and rock (Shoaib & Al-Swaidan, 2016), respectively, present in alkanes. Peaks at 1330, 1245 and 1052 cm⁻¹ are characteristic of –C—O (Shoaib & Al-Swaidan, 2016) and C—C stretching. The peak for the hydroxyl group at 3600–3400 cm⁻¹ in the inactivated sample has disappeared or significantly diminished in the spectra of all ACs. The main three peaks in these spectra are in the range 1570 cm⁻¹, 1400 cm⁻¹ (very small), and 1100 cm⁻¹ for all the basic ACs which can be attributed to C⚌C aromatic ring stretching, phenolic vibrational elongation, and C—O stretching, respectively (de Andrade et al., 2015; Shoaib & Al-Swaidan, 2016). It is evident from the FTIR results that there is no fundamental variation in the surface functional groups related to the difference in the activators used. FTIR spectra of N-enriched ACs produced are shown in Fig. 5b. There are two main peaks found in all the functionalized samples. One in the range of 1100–1200 cm⁻¹ which can be ascribed to C—N (aliphatic) or C—O (carboxyl, hydroxy, ether, and ester) stretching vibration, and the other 1550–1570 cm⁻¹ which could be due to stretching vibration of C⚌O, C⚌C and also to N—H (secondary amine) bend. The DDAB sample has a peak at 3413 cm⁻¹, which could be assigned to N—H symmetric stretching vibration or O—H stretching vibration (Wang et al., 2014; Tseng et al., 2015).

Close

Fig. 5

FTIR spectra of the ACs (a) Before N modifications (b) After N modifications.

Export to PPT

Fig. 4S presents Raman spectra for K₂CO₃ and Cs₂CO₃ ACs before and after nitrogen treatment. The well-known D and G bands of graphite are in 1330 and 1577 cm⁻¹ regions respectively. The G peak corresponds to the only allowed Raman transition in this region of the spectrum of graphite (Escribano et al., 2001). The line entered at 1577 cm⁻¹ (G line) corresponds to the Raman allowed E_2g mode in the ideal graphite and the line centered at 1330 cm⁻¹ is the disorder-induced line which is due to the disordered portion in the carbon (Zhao et al., 2009). The I_D/I_G ratios for all studied samples are relatively high indicating a large extent of defects. Also, the samples treated with N-agent and processed with two-step thermal treatment (see the urea-doped ACs samples in Fig. S4) exhibit a clear higher structure defect (higher I_D/I_G ratio) due to introducing nitrogen into the ACs and/or the second thermal treatment (K. Li et al., 2019; Wang et al., 2019).

3.7.1 Effect of pH on removal

The pH value of the system is generally a key parameter during the adsorption process and has a significant effect on the removal of ions from an aqueous solution as it controls the surface charges on the adsorbent as well as the chemical nature of metallic cation. The pH of a solution can affect the surface of the adsorbent as well as the species of metal ions in the solution. Here, the removal of Cr(VI) at different pH (at 2, 4, 6, and 8) for Cs₂CO₃ and K₂CO₃ ACs was studied (Fig. 7a). A maximum adsorption was observed at pH = 2 for both ACs which was expected owing to previous literature (Ammar et al., 2021). The decrease in adsorption with an increase in pH can be inferred to be due to competition of HCrO₄⁻ and CrO₄²⁻ ions with OH⁻ ions on adsorption active sites (Chand et al., 2009; Zheng et al., 2011; Ren et al., 2019). The effect of the initial concentration of Cr(VI) ions was also studied and observed that the adsorption percentage increases with an increase in initial Cr(VI) ions concentration.

Close

Fig. 7

pH dependence of Cr(vi) and Pb (ii) for Ps.Cs₂CO_3.Urea AC (a and b respectively). Weight of adsorbent = 10 g/L, Cr(VI) and Pb(II) solutions concentration = 10 ppm, contact time = 60 min.

Export to PPT

A pH range from 2 to 7 was also investigated for Pb(II) removal (Fig. 7b). Higher pH solutions were excluded from the study as the precipitation of Pb(OH)₂ observed in the solution for pH > 7 which made the removal of Pb²⁺ not possible (Sekar et al., 2004; Momčilović et al., 2011; Soliman et al., 2016). It was observed that the maximum adsorption (Pb²⁺ removal percentage of 99.99 %) occurred at pH 7 when Pb(II) concentration is 10 ppm and AC amount is 10 g/L. Almost same percentage of removal was obtained at pH 6. Hence, pH 6 and 7 can be considered optimal for Pb(II) ions adsorption. Because of that, all further Pb(II) adsorption studies were carried out at a neutral pH of 7. In the highest sorption range, the dominant species of adsorption are Pb²⁺ and PbOH⁺. In an acidic solution, the AC’s surface being positively charged, the adsorption of these species is not favored. This explains the lower adsorption at a low pH range.

3.7.2 Adsorption isotherms

Mathematical isotherm models such as Langmuir, Freundlich, and Temkin were applied for describing the equilibrium established between the metal ions on the ACs and the ions remaining in the solution, linear forms of which can be expressed as Eqs. (3), (4), and (5) respectively (Sun et al., 2013).

where C_e (mg/L) is the equilibrium concentration of metal ion, q_e (mg/g) is the amount adsorbed under equilibrium, q_max (mg/g) is the maximum adsorption amount for the corresponding heavy metal, K_L (L/mg), K_F (mg/g(L/mg)^1/n), K_T (L/g) the Langmuir, Freundlich and Temkin isotherm constants, respectively. The constant n in Freundlich model is known as the intensity parameter. The parameter B = RT/b where b is a Temkin’s model constant related to the heat of sorption in J/mol. The adsorption isotherms of Cr(VI) adsorption on urea-treated Ps-Cs₂CO₃ ACs are represented in Fig. 8(a,b,c). The calculated Langmuir, Freundlich and Temkin constants are summarized in Table 3. Adsorption of Cr(VI) was also investigated before doping with N. It was observed that the enrichment caused a decrease in Cr(VI) uptake. Also, the adsorption capacities were found to increase with the equilibrium concentrations of metal ions. Based on R² value, the adsorption results for all the adsorbents were well-fitted with Freundlich isotherm which, unlike Langmuir isotherm and Temkin, indicates a multilayer adsorption mechanism for heterogeneous surface (Lin et al., 2019). The ln K_F (the intercept) is a measure of adsorption capacity and the slope 1/n gives the adsorption intensity (Eq.4). The n value ranging from 2 $-$ 10 implies good adsorption. n value ranging from 1 $-$ 2 gives moderately good adsorption and less than 1 will be for poor adsorption (El Nemr et al., 2020).

Close

Fig. 8

Langmuir, Freundlich, and Temkin isotherms for Cr(VI) adsorption in a, b and c respectively and that for Pb(II) adsorption in d, e and f respectively by Urea functionalized Ps-Cs₂CO₃ AC.

Export to PPT

The Ps-Cs₂CO₃ AC was found to have much more uptake ability compared to that of K₂CO₃ AC which can be attributed to the excellent surface area and structural properties of this AC compared to other ACs produced by different activators. The Pb(II) ions adsorption was found to increase greatly after enriching the AC with urea because of the presence of surface functional groups. The existence of different acidic groups, such as carboxyl, lactone, carbonyl and phenolic functional groups, also may contribute to the large adsorption capacity of these carbons. Thus, although chemical modification resulted in the reduction of S_BET value, the added functional groups arising from the modification compensated for the decrease in sorption brought about by the reduction in the S_BET. This encouraged us to investigate further the AC prepared using Cs₂CO₃ activation.

It is to be also noted here that adsorption capacities increased with the equilibrium concentrations of metal ions as observed in the case of Cr(VI) adsorption. The adsorption isotherms (Langmuir, Freundlich and Temkin) are given in Fig. 8 (d,e,f) and their parameters are listed in Table 3 in order to understand better the adsorption behavior based on varying concentrations of adsorbate ions solutions. Based on R² value, the adsorption results for Pb(II) ions on all the adsorbents were best fitted with Langmuir isotherm (0.9974) than Freundlich and Temkin isotherm mechanisms indicating saturated monolayer formation of Pb(II) ions on the adsorbent. The value of K_F, which is related to the adsorption capacity, was found to be 32.07 for Pb(II) adsorption which was much higher compared to Cr(VI) adsorption (2.77). The value of K_L also was much higher for lead (23.28) compared to chromium (6.52). The sorption equilibrium constants listed in Table 3 can be used to predict whether the sorption of Pb(II) and Cr(VI) ions onto Ps-Cs₂CO₃-AC was favorable or unfavorable using the dimensionless separation parameter defined as,

As depicted, the separation factor values were always between 0 and 1 confirming the favorability of the metal ion sorption.

3.7.3 Adsorption kinetics

Adsorption kinetics of Cs₂CO₃ AC (N-doped) was monitored (Fig. 9) using pseudo-first order and pseudo-second order rate equations which can be represented by Eqs. (7) and (8) respectively (Sivaprakasam & Venugopal, 2019),

Close

Fig. 9

(a,d) Effect of contact time on adsorption (b,e) Pseudo-first-order model and (c,f) Pseudo-second-order model for Cr(VI) and Pb(II) respectively onto Ps-Cs₂CO₃ enriched using Urea.

Export to PPT

The amount of the metal ion adsorbed at time t was given by q_t (mg/g) and first-order rate constant was given by k₁ (min⁻¹). q_e and k₁ is calculated from the intercept and slope of the plot, respectively,

where k₂ is the rate constant of second-order kinetics. The plot t/q_t against t should be a straight line if pseudo-second-order kinetics was applicable whereas, the slope and intercept of the plot would give q_e and k₂, respectively. As obtained (Fig. 9a,b,c and Table 3), the Cr(VI) adsorption mechanism was found to be controlled by pseudo second-order kinetics with a large coefficient (R² = 1). Moreover, the calculated q_e values were also close to q_e values obtained from the experimental data, suggesting high accuracy for pseudo-second order model (Karnjanakom & Maneechakr, 2019). The validity of the kinetic models was tested in the case of Pb(II) adsorption as well (Fig. 9d,e,f and Table 3) from which it is clear that pseudo-second order kinetics is also obeyed by the ACs under investigation as the application it leads to much better regression coefficient (R² = 1) which is supported by the majority of literature (Aroua et al., 2008; Momčilović et al., 2011; Karnjanakom & Maneechakr, 2019). The values of the q_e, estimated from the first order kinetic model significantly differ from experimental q_e values. However, in pseudo-second order kinetic model, the q_e values were very close to the experimental q_e value. Similar to the finding in this study, the pseudo-second order model is extensively applied for studying physiochemical interaction in numerous environmental adsorption studies. The analysis of kinetic data by other researchers also showed that the pseudo-second order rate equation was a good fit for Pb²⁺ adsorption (Banat et al., 2003; Aldawsari et al., 2017). Such findings indicate that the adsorption kinetics depend on both the adsorbate and the adsorbent.

3.7.4 Thermodynamics of lead adsorption

Since the N-functionalized ACs were very efficient in removing Pb(II) ions, the thermodynamic parameters of its adsorption were studied. The standard Gibb’s energy,

Where, R is ideal gas constant (=8.314 J mol⁻¹ K ⁻¹) and T is the temperature (K).

The apparent equilibrium constant (K) is given,

The unit of the thermodynamic equilibrium constant (K) (Eq.10) should be corrected to become dimensionless before using the Van 't Hoff equation. This is carried out using the calculation suggested by Lima et al. (2019). After such correction one can calculate the enthalpy and the entropy of the system using Eq. (11). Thus, the calculated ΔG (at 298, 308, 318, 328 and 338 K), ΔH and ΔS at adsorbate concentrations 10–80 ppm are displayed in Table 4.

The plot of lnK^corr as a function of 1/T provides a straight line at all 8 adsorbate concentrations investigated (Fig. S6). The ΔH and ΔS can be calculated from the slope and intercept, respectively. The thermodynamic parameters are also depicted in Table 4.

Since diffusion is an endothermic process, the increase in uptake of Pb(II) observed with temperature may be due to the desolvation of the adsorbing species, the changes in the size of pores, and/or the enhanced rate of intraparticle diffusion of adsorbate (Zarei et al., 2019). The increase in K^corr with an increase in temperature indicates the endothermic nature of the process. As we know, physical adsorption is generally an exothermic reaction. Hence, the logical cause of the observation is that the adsorption should include endothermic chemical process. ΔG at all concentrations were negative at all the temperature ranges studied. Thus, it can be inferred that the adsorption of Pb(II) by the AC under investigation is feasible and spontaneous. Also noted a steady, but small, increase in the ΔG values with increasing temperature, i.e., a better adsorption was obtained at higher temperatures. From the standard free energy change values, a multilayer adsorption can be assumed as the ΔG for multilayer adsorption falls between − 20 kJ mol⁻¹ and zero. In addition, the endothermic nature of Pb(II) adsorption with irreversible process was concluded from the positive values of ΔH and ΔS (Karnjanakom & Maneechakr, 2019; Sivaprakasam & Venugopal, 2019). The heat evolved during chemisorption and physisorption falls into the range of 80–200 kJ mol⁻¹ and 2.1–20.9 kJ mol⁻¹, respectively. The ΔH for 10 ppm and 20 ppm batches were found to be 30.67 and 26.27 kJ mol⁻¹ respectively showing the occurrence of mainly physisorption at this concentration level. The ΔH for adsorption at higher adsorbate concentrations, however, were found to be within the range of physisorption (Zarei et al., 2019). In addition, the small positive ΔS values signify the increased randomness at the solid-solution interface upon adsorption of Pb(II) on Ps-Cs₂CO₃ ACs.