Characterization studies
Morphology properties
Morphology properties of different luffa samples with or without hydroxyapatite nanoparticles (HAp) are shown in Fig. 2. The structure of oxidized luffa fibers (S1) and isolated cellulose (S2) was observed using the field emission scanning electron microscope (FE-SEM) Fig. 2 (S1 and S2). The S1 and S2 samples showed long fibrils with rough surfaces, indicating that the oxidation process and chemical treatment have affected the structure of original Luffa fibers, which mainly was due to the removal of wax, non-cellulosic materials, and other extractives (Taimur-Al-Mobarak et al. 2018). The average width of S1 and S2 was 11.4 ± 3.1µm and 13.3 ± 1.0 µm, respectively.
The microstructure and dimensions of oxidized-nanocellulose (S3), oxidized-fibers/hydroxyapatite (S4), cellulose/HAp (S5), and oxidized-nanocellulose/HAp (S6) are determined using TEM imaging, as shown in Fig. 2 (S3-S6). The S3 sample was obtained via oxidation of isolated cellulose by H2O2, resulting in a needle-shaped structure with an average length of 192 ± 37 nm and a width of 25 ± 6 nm. On the other hand, the resulted luffa forms were used as a carrier for loading HAp. It can be seen that HAp nanoparticles were successfully synthesized on the surface of samples, as shown in Fig. 2 (S4-S6). The average length and width of HAp synthesized on the surface of S4 were (56 ± 17 nm and 19 ± 4 nm), S5 (53 ± 14 nm and 22 ± 5 nm), and S6 (40 ± 19 nm and 14 ± 3 nm). It is interesting to note that HAp nanoparticles are presented in abundance but relatively agglomerated on the surface of the S4 sample. However, it is densely and uniformly attached to the surface cellulose (S5). While in the case of sample S6 sample, it showed less content of HAp, which probably was related to the small dimensions of prepared oxidized-nanocellulose. Also, it looks that the presence of HAp helped in an agglomeration of the oxidized-nanocellulose (S6) as compared to without HAp (S3). Similar agglomeration behavior was observed when different ratios of HAp prepared with nanocellulose (Lu et al. 2019). It can be concluded that the size of luffa samples played a significant role in synthesizing and attaching HAp on their surfaces.
FTIR analysis
Change in the chemical structure of luffa samples with or without HAp was performed via FTIR analysis and the results are shown in Fig. 3. The characteristic absorption peaks of oxidized-Luffa fibers (S1), cellulose (S2), and oxidized nanocellulose (S3) are observed at 3334 cm-1 (O–H stretching vibrations), 2898 cm-1 ( CH2 groups of cellulose), 1631 cm-1 (O–H vibration), 1429 cm-1 (O–H vibration), 897 cm-1 (β-glycosidic linkages between glucose units) (Niamsap et al. 2019). The FTIR spectra of Luffa fibers (S1) and oxidized-nanocellulose (S3) samples exhibited a new absorption peak at 1733 cm-1, resulted from the introduction of C=O group after H2O2 treatment (Oun and Rhim 2018).
The characteristic peaks of HAp which loaded on the surface of Luffa samples were detected at 1026 cm-1 due to PO43- group stretching mode (ʋ3, ʋ1) and at 871 cm-1 correspond to ʋ1 CO32- (Yu et al. 2013). It can be seen that the peak intensity of O-H and C–H groups were decreased as shown in Fig. 3 (S4, S5, and S6), probably due to the interaction of these groups with HAp, which helped in attaching HAp on the surface of CNCs (Narwade et al. 2017).
XRD analysis
The XRD analysis was used to determine of crystalline structure and chemical composition of Luffa samples with and without HAp. The XRD diffraction patterns of luffa and luffa/HAp samples are shown in Fig. 4. The characteristic peaks of cellulose were observed at lattice planes (110), (200), and (004), which is related to native cellulose structure (Oun and Rhim 2016). oxidized-luffa fibers (S1), isolated cellulose (S2), and oxidized- nanocellulose (S3) showed different diffraction pattern intensities due to chemical treatments. Treatment of luffa fibers with H2O2 presented less peak intensities, compared to sample S2 and S3, which probably due to the role of H2O2 in removal of lignin only. While in the case of sample S2, the peak intensity was increased which, maybe due to the removal of non-cellulosic parts (hemicellulose and lignin). Compared to sampleS1 and S2, sample S3 presented the highest peak intensity, and this perhaps due to not only removal of non-cellulosic parts but also amorphous regions in cellulose fibers (Oun and Rhim 2018). The crystallinity index (CI) was calculated using Eq. (1) and the results were 73.5%, 82.4%, and 84.4% for samples S1, S2, and S3, respectively.
After loading of HAp on the surface of luffa samples, the intensity of cellulose peaks was significantly diminished, as shown in Fig. 4 (S4-S6). The CI of composite samples was decreased to 70.2%, 66.6%, and 55.9% for S4, S5, and S6, respectively, as compared to samples without HAp. The reduction in the CI of composite samples was probably due to covering of characteristic cellulose peaks by HAp. Similar results were observed when metallic nanoparticles such as AgNPs, CuONPs, and ZnONPs formed on the surface of regenerated cellulose (Shankar et al. 2018).
The inset figure (Fig. 4) shows of XRD analysis of luffa/ HAp composite samples. It worth noting that new peaks have been observed in Luffa/HAp composites at 2θ=26.9°, 32.9°, 40.7°, 47.6°, 49.6°, 53.2°, and 63.9° (Narwade et al. 2017). These new peaks indicate to formation of HAp onto the surface of different forms of luffa samples with different peak intensities (Niamsap et al. 2019).
Adsorption studies
Effect of sorbent type
Converting of cellulosic materials into nanocellulose forms e.g. cellulose nanofibrils (CNFs) and cellulose nanocrystals (CNCs), led to an increase in their surface area, lightweight, and the ability to add different functional groups that improve their adsorption capacities of heavy metal ion and dye (Tshikovhi et al. 2020).
Fig. 5, shows a comparative study of 0.4 g of the prepared sorbents for removing methylene blue (MB) and lead ions (Pb2+ ) from 100 mL aqueous solutions after 120 minutes. The data showed that the removal efficiency of MB by S1, S2, and S4 sorbents is higher than the other sorbents, which reached 85%, 89.6%, and 83.8% respectively (Fig. 5A). The obvious increase in the removal efficiency of MB by these sorbents over the other sorbents probably was due to the presence of abundant functional hydroxyl and carboxyl groups on the surface of nanocellulose that facilitate the interaction of chemical moieties. Also, the web-shape structure of long fibrils may be played an important role as a net and trapped the dye molecules (Li, Ma, Venkateswaran, & Hsiao, 2020). Previously, different cellulose materials have been used for removal of dyes from contaminated water. The removal efficiency of materials has been affected by the source, size, and surface modifications of used cellulose, as well as types of loaded materials (Varghese et al. 2019).
The data also showed that samples S4, S5, and S6 have the highest removal efficiency for lead ions over other sorbent materials, which reached 96.9%, 97.8%, and 96.3%, respectively (Fig. 5B). The higher removal % of these sorbents can be attributed to the presence of hydroxyapatite nanoparticles which have a great affinity to adsorb the heavy metal ions (Bailliez et al. 2004).
Effect of contact time
Fig. 6, display the effect of contact time on the removal efficiency of sorbents for methylene blue (Fig. 6A) and lead ions (Fig. 6B). According to primary experiments, samples S1(oxidized-fibers), S2 (cellulose of luffa), and S4 (oxidized-fibers/HAp) showed the best efficient sorbents in removal of MB as shown in Fig. 5A. While samples S4 (oxidized-luffa fibers/HAp), S5 (cellulose/ HAp), and S6 (oxidized-nanocellulose/HAp) were the best in removal of Pb2+ ions (Fig. 5B). For this, these samples have been chosen to test the effect of contact time on their removal efficiency.
It can be seen that a quick removal within the first 5 min of the adsorption process was obtained with the removal rate of more than 85% for MB and more than 95% for Pb2+ ions. Then, a slower sorption step continued until reaching a state of equilibrium. This behavior possibly is due to the availability of sufficient active sites at the beginning of the reaction, after that the active sites became occupied by MB and Lead ions (Abd El-Aziz et al. 2018).
On the other side, the kinetic studies and rate constants of MB and Pb2+ ions sorption by sorbents were elucidated after applying the pseudo-first-order, pseudo-second-order, intra-particle diffusion model, and Elovich model. The kinetic model's constants and correlation coefficients of MB and Pb2+ ions were calculated and presented in Table (2) and (3), respectively. Interestingly from the data, the kinetics of sorption reaction was perfectly fitted to the pseudo-second-order model for both methylene blue (Fig. 6C) and lead ions (Fig. 6D) which assumes that the rate of solute adsorption is directly proportional to the square of the number of vacant binding sites (Choudhary and Paul 2018). This may be attributed to the higher correlation coefficient value (R2), and the close matching between the experimental and calculated sorption capacities from this model (Kamal et al. 2019).
Effect of the initial MB and Pb2+ ions concentrations
The initial concentration of contaminants is one of the most important factors in adsorption efficiency. Consequently, the removal efficiency was tested using 0.4g of the selected sorbents at different MB and Pb2+ ion concentrations (Fig. 7A and 7). The data presented in Fig. 7A shows that with the increase in the initial concentration of MB from 5 to 225 mg/L, the sorption efficiency decreased gradually from ~100% to 51, 59, and 40 % for S1, S2, and S4, respectively. This behavior can be attributed to the saturation of the most active sites of the sorbents by MB molecules (Aksu and Tezer 2005). While Fig. 7B shows a steady sorption efficiency of around 100 % at a concentration range of Pb2+ ions (200-1000 mg/L) for samples S4, S5, and S6. After increasing the Pb concentration from 1000 mg/L to 4000 mg/L, it was found that the sorption efficiencies gradually decreased to achieve 63, 75, and 70% for S4, S5, and S6, respectively. This higher ability of sorbents to adsorb more Pb2+ compared to MB is attributed to the small size of lead ions (ionic radius of Pb2+ ions (1.33 Å) rather than the large dye molecules (estimated area of MB molecule (130-135 Å)) which allows less adsorption competition on the available sorbent sites (Aljeboree et al. 2017).
On the other hand, to illustrate how the MB and Pb2+ ions interact with the sorbents; Langmuir, Freundlich, Temkin, and Dubinin–Radushkevich (D–R) isotherm models were studied. The constants and correlation coefficients calculated from the isotherm models were listed in Table 2 and Table 3. Remarkably from the data and correlation coefficients, See Fig. 7 (C and D), the sorption of MB and Pb2+ ions were fitted with the Langmuir model which assumes monolayer adsorption of the MB and Pb2+ ions onto active sites of the sorbent's surface (Gupta and Babu 2009). The value of n >1 in Freundlich and E <8 in (D-R) model demonstrating that the adsorption is a physical process (Kumar et al., 2014). Moreover, the separation factor (RL) values were found to be in the range from 0 to 1, which proposing favorable adsorption between sorbents and sorbates.
The maximum MB and Pb2+ ions sorption capacities (qmax) of the selected sorbents which calculated from the Langmuir model were compared with different sorbents in previous studies as presented in Table (4). The data indicated that the prepared sorbents have a good ability to remove MB and Pb2+ ions from the solution.
Table 2 Constants of kinetic models and isotherm models for MB removal.
Constants of kinetics models
|
Constants of isotherms models
|
Pseudo-first-order
|
Langmuir isotherm
|
K1 (min-1)
|
qe(exp.) (mg/g)
|
qe (cal.) (mg/g)
|
R2
|
qmax (mg/g)
|
b (L/mg)
|
R2
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
0.08
|
0.02
|
0.039
|
5.45
|
5.83
|
5.45
|
0.27
|
0.15
|
0.77
|
0.831
|
0.697
|
0.963
|
30.86
|
36.2
|
25.2
|
0.1
|
0.097
|
0.072
|
0.978
|
0.971
|
0.986
|
Pseudo-second-order
|
Freundlich isotherm
|
K2 (g/mg min)
|
qe (exp.) (mg/g)
|
qe (cal.) (mg/g)
|
R2
|
n
|
Kf
|
R2
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
0.46
|
0.83
|
0.18
|
5.45
|
5.83
|
5.45
|
5.53
|
5.83
|
5.49
|
1
|
1
|
0.999
|
1.72
|
1.6
|
2
|
2.45
|
2.56
|
2.46
|
0.894
|
0.895
|
0.974
|
Intra-particle diffusion model
|
Temkin isotherm
|
Kp(mg. g-1min1/2)
|
C
|
R2
|
kt (mol/L)
|
B
|
R2
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
0.037
|
0.019
|
0.076
|
5.17
|
5.6
|
4.7
|
0.6452
|
0.8265
|
0.889
|
1.5
|
1.4
|
2.5
|
5.27
|
6.24
|
3.54
|
0.933
|
0.936
|
0.886
|
Elovich
|
(D–R) isotherm
|
β (mg. g-1min)
|
α (mg.g-1.min-1)
|
R2
|
qmax (mg/g)
|
β
|
E (kJ/mol)
|
R2
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
10.3
|
21.3
|
5.65
|
5.6*10^21
|
2.7*10^50
|
5.6*10^21
|
0.8179
|
0.9602
|
0.8745
|
11.3
|
11.96
|
9.25
|
7*10^-8
|
7*10^-8
|
1*10^-7
|
2.67
|
2.67
|
2.23
|
0.621
|
0.612
|
0.553
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Table 3 Constants of kinetic models and isotherm models for Pb2+ ions removal.
Constants of kinetics models
|
Constants of isotherms models
|
|
Pseudo-first-order
|
Langmuir isotherm
|
|
K1 (min-1)
|
qe(exp.) (mg/g)
|
qe (cal.) (mg/g)
|
R2
|
qmax (mg/g)
|
b (L/mg)
|
R2
|
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
|
0.044
|
0.12
|
0.09
|
125.2
|
126
|
125.9
|
31.8
|
1.0
|
7.9
|
0.996
|
0.703
|
0.957
|
625
|
714.5
|
714
|
0.028
|
0.21
|
0.027
|
0.989
|
0.999
|
0.994
|
|
Pseudo-second-order
|
Freundlich isotherm
|
|
K2 (g/mg min)
|
qe (exp.) (mg/g)
|
qe (cal.) (mg/g)
|
R2
|
n
|
Kf
|
R2
|
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
|
0.004
|
0.3
|
0.04
|
125.2
|
126
|
125.9
|
126.58
|
126.6
|
126.6
|
0.999
|
1
|
1
|
3.78
|
3.78
|
3.74
|
100
|
153.2
|
107.4
|
0.854
|
0.626
|
0.54
|
|
Intra-particle diffusion model
|
Temkin isotherm
|
|
Kp (mg. g-1min1/2)
|
C
|
R2
|
kt (mol/L)
|
B
|
R2
|
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
|
3.3
|
0.16
|
0.86
|
94.04
|
124.7
|
118.5
|
0.876
|
0.365
|
0.659
|
5.41
|
14.15
|
5.42
|
67.1
|
82.3
|
67.1
|
0.963
|
0.814
|
0.821
|
|
Elovich
|
(D–R) isotherm
|
|
β (mg. g-1min)
|
α (mg.g-1.min-1)
|
R2
|
qmax (mg/g)
|
β
|
E (kJ/mol)
|
R2
|
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
S1
|
S2
|
S4
|
|
0.122
|
2.12
|
0.43
|
371591
|
6*10^113
|
8.5*10^21
|
0.988
|
0.594
|
0.888
|
444.8
|
594.5
|
338.9
|
4*10^-7
|
4*10^-7
|
4*10^-7
|
1.11
|
1.11
|
1.11
|
0.898
|
0.928
|
0.10073
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Table 4 Comparison of adsorption capacity of methylene blue and lead ions with several sorbents reported in the literature
Methylene blue
|
Lead ions (Pb2+)
|
Sorbent
|
Adsorption Capacity, (mg/g)
|
References
|
Sorbent
|
Adsorption Capacity, (mg/g)
|
References
|
Neem (Azadirachta indica) leaf powder
|
8.7
|
(Bhattacharya and Sharma 2005)
|
Cellulose-MT-CBM biosorbents
|
39.0
|
(Mwandira et al. 2020)
|
Freeze-dried agarose gel
|
10.4
|
(Seow and Hauser 2016)
|
Cellulose
|
43.9
|
(Aquino et al. 2018)
|
H2SO4 cross-linked magnetic chitosan
|
20.4
|
(Rahmi et al. 2019)
|
Natural clinoptilolite
|
80.9
|
(Günay et al. 2007)
|
Carbon-TiO2 composite
|
25.7
|
(Simonetti et al. 2016)
|
Nanohydroxyapatite
|
192.3
|
(Mohammad et al. 2017)
|
Oxidized-fibers (S1)
|
30.8
|
(Present work)
|
Oxidized fibers/HAp (S4)
|
625.0
|
(Present work)
|
Cellulose (S2)
|
36.2
|
(Present work)
|
Cellulose/HAp (S5)
|
714.5
|
(Present work)
|
Oxidized-fibers/HAp (S4)
|
25.2
|
(Present work)
|
Oxidized nanocellulose/HAp (S6)
|
714.50
|
(Present work)
|