Biopolymer-Grafted, Magnetically Tuned Halloysite Nanotubes as Efficient and Recyclable Spongelike Adsorbents for Anionic Azo Dye Removal.

The quest for sustainable development and green chemistry had led to the design and synthesis of advanced adsorbent materials for efficient removal of pollutants in industrial effluents. Magnetic halloysite nanotubes with chitosan nanocomposite sponges were prepared by combining solution-mixing and freeze-drying. Magnetic@chitosan/halloysite (Fe3O4-HNT/CS) and spongelike chitosan/halloysite (HNT/CS) were used as adsorbents for the removal of Congo red dye in aqueous solution in a batch process. The as-prepared composites were characterized using scanning electron microscopy, energy-dispersive X-ray analysis, X-ray diffraction, vibrating-sample magnetometry, thermal gravimetry-differential scanning calorimetry, and Fourier transform infrared spectroscopy. Data from kinetic study were analyzed with pseudo-first-order and pseudo-second-order models, whereas the mechanism was analyzed using Bangham's, Elovich's, intraparticle, and double-exponential diffusion models. The equilibrium data were evaluated using Langmuir, Freundlich, Temkin, and Dubinin-Radushkevich isotherm models. The adsorption kinetics of dye removal followed the pseudo-first-order model with average rate constants of 0.260 and 0.196 min-1 for Fe3O4-HNT/CS and HNT/CS, respectively. The Langmuir adsorption isotherm best fitted the equilibrium data with R 2 > 0.9 with maximum adsorption capacities of 41.54 and 54.49 mg g-1 obtained for HNT/CS and Fe3O4-HNT/CS, respectively. Negative values of ΔG° obtained from thermodynamic studies revealed that the adsorption process was spontaneous. The values of ΔH° and ΔS° obtained for Congo red dye removal were 69.46 and 39.54 kJ mol-1 and 240.5 and 145.1 J mol-1 K-1 for HNT/CS and Fe3O4-HNT/CS, respectively. The results indicated that CS-HNT is an excellent adsorbent; however, its magnetic modification further improved its recyclability and enhanced the performance for the removal of Congo red dye from aqueous solution.

Introduction

Industrial and domestic processes require water that is free from colors and related compounds. On the contrary, however, effluents from industries such as textile, paper and pulp, dye and dye intermediates, pharmaceutical, tannery, craft bleaching, food technology, hair coloring, rubber, paper, plastic, cosmetics, etc. contain dyes, organic colorants, and phenolic compounds.[1] These industries consume majority of tens of thousands of chemically different types of dyes that are currently manufactured.[2] When untreated effluents from these industries are released in aquatic environment, development of aquatic animals and plants is inhibited because of outright blocking of sunlight, resulting in reduced photosynthesis, increased biological oxygen demand, and reduced dissolved oxygen to sustain the aquatic life. Furthermore, these pollutants exhibit toxic effects on microbial and aquatic populations and become toxic and carcinogenic to mammals when they enter the food chain.[3]

The first contaminants to be recognized in wastewater effluents are colorants, and the presence of a synthetic dye in water even in a very small amount (<1.0 mg L–1) is typically highly visible, affecting the aesthetic merit and transparency of the water bodies. The complex structure and synthetic origin of dyes made their removal through conventional wastewater treatment a colossal task. Moreover, they are designed to resist breakdown with time and defy exposure to harsh conditions such as sunlight, water, soap, and oxidizing agents, thus making their removal challenging during wastewater treatment. The Congo red (CR) dye is a sodium salt of 3,3′-([1,1′-biphenyl]-4,4′-diyl)bis(4-aminonaphthalene-1-sulfonic acid), and it is the first synthetic dye produced with a complex chemical structure and capability of dying cotton directly;[4] hence, its application in a wide field of industrial processes involve dyeing. Unfortunately, it metabolized to benzidine, a compound implicated as a carcinogenic and a mutagenic agent,[5,6] therefore, exposure to this dye posed a potential danger of bioaccumulation and allergy.

Various methods including physical, chemical, physicochemical, and biological methods have been developed for the removal of dye and other contaminants from wastewater. Adsorption has been the most common physicochemical treatment method with potential applications because of its low cost, effectiveness, and efficiency. Mostly, activated carbon has been widely accepted in commercial systems as an adsorbent for the removal of the contaminants in wastewater including dyes because of its significant adsorption capacity; nevertheless, its high capital cost, complicated synthesis process, low capacity, slow kinetics property, and the harmful byproducts of the treatment process are major setback for using activated carbon.[7−9]

The quest for sustainable development and green chemistry had led to the design and synthesis of advanced adsorbent materials for efficient adsorption, separation, and purification. These include natural or synthetic zeolite,[10] organically modified porous silica,[11] natural or modified clays,[12] macroporous polymeric adsorbents,[13] and so on. However, high costs in terms of operation and consumption time, low selectivity, limited adsorption capacities, and difficulties in their regeneration and reuse are major challenges for such materials. Therefore, it is of great theoretical and practical significance to develop high-performance, low-cost, and recyclable adsorbents.

Significant attention had also been paid to polymer–clay nanocomposites, which are low-cost, environment-friendly, and abundant materials with good chemical, mechanical, and structural stability.[14,15] Most clays have cationic exchange capacity, thereby limiting their applications to cationic pollutants; therefore, functionalization is envisaged to confer enhanced performance. Halloysite (HNT), an aluminosilicate clay with a hollow nanotube structure, has been used as a nanocomposite, as a catalyst, in molecular hydrogen storage, and as an adsorbent because of the presence of hydroxyl groups and large pore volume and surface area.[14] Chitosan is a biopolymer derived from chitin, an abundant precursor, by deacetylation. Its richness in hydroxyl and amino functional groups makes it one of the substances that had been most exploited in the field of adsorption of substances.[16] Its use has been limited by its pH sensitivity and low mechanical properties; hence, for practical operations, these properties require improvement to increase the adsorption capability of chitosan.[17] The introduction of clay into the polymer structure increases not only the adsorption capacity of the adsorbent but also the mechanical and thermal stability, thereby resulting in an adsorbent with superior properties compared to those of either the clay or the polymer; nevertheless, the difficulty encountered from separation of the suspended polymer–clay composite after use is a big challenge. This challenge can be circumvented by incorporation of iron oxide nanoparticles into the clay–polymer moiety, thereby creating adsorbents with magnetic properties that significantly impact the adsorption process on the surface with accelerated separation and improved efficiency of water treatment.[18] Although various reports on the use of the HNT/CS composite are available in the literature, the inclusion of magnetic particles is less reported.

This study reports the synthesis and characterization of a biopolymer-grafted, magnetically tuned halloysite composite (Fe3O4–HNT/CS) and its unmagnetized precursor. Their performances as adsorbents for the removal of Congo red dye were also investigated. The effects of the initial concentration of dye, contact time, pH, and adsorbent dosage on Congo red dye removal as well as recyclability were reported. Kinetic and equilibrium data as well as thermodynamics data from the batch adsorption studies were analyzed with appropriate models and reported.

Results and Discussion

Characterization of the Adsorbent

The surface morphologies of CS, HNT/CS, and Fe3O4–HNT/CS were observed by scanning electron microscopy (SEM) presented in Figure . It is clear from the figure that the presence of HNT modified the surface morphology of the CS (Figure a). The rough surface of the chitosan becomes more collapsed, and the uniformly smooth surface is gradually transformed to three-dimensional morphologies as HNT and Fe3O4 were incorporated into the composite.

Figure 1

SEM images of (a) CS, (b) HNT/CS, (c) Fe3O4–HNT/CS, and (d) Fe3O4–HNT/CS after adsorption.

Energy-dispersive X-ray (EDX) analysis of CS, HNT/CS, and Fe3O4–HNT/CS is shown in Figure . The presence of O and C in chitosan was revealed by Figure a and the incorporation of HNT was confirmed by the presence of Al Na and Si in Figure b, whereas the presence of Fe in Figure c affirmed the incorporation of magnetic particles into the structure of Fe3O4–HNT/CS.

Figure 2

EDX analysis of (a) CS, (b) HNT/CS, and (c) Fe3O4–HNT/CS.

The decomposition process of CS/HNT and Fe3O4–CS/HNT could be described in three main steps as shown by the thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) curves in Figure a,b. The weight loss in TGA curves of CS/HNT is shown at around 91, 285, and 527 °C. The weight loss at these temperatures could be attributed to the dehydration due to loss of surface and interlayer water, decomposition of the remaining low-molecular-weight species, and dehydration and dehydroxylation of structural water. Similarly, Fe3O4–CS/HNT displayed weight loss at 108, 257, and 484 °C; whereas the first could be attributed to the evaporation of adsorbed water, the last two stages could be attributed to decomposition of the chitosan in the hybrid material.

Figure 3

DSC and TG analyses of (a) HNT/CS and (b) Fe3O4–HNT/CS.

The Fourier transform infrared (FTIR) spectra of HNT/CS and Fe3O4–HNT/CS before and after adsorption are presented in Figure . The double peaks at 3600 cm–1 and above are due to the stretching vibration of the surface hydroxyl group on the two composites. The peaks at 471, 529, and 928 cm–1 in HNT/CS and those at 464, 558, and 913 cm–1 in Fe3O4–HNT/CS were assigned to deformation of Si–O–Si, Si–O–Al, and OH of inner hydroxyl groups. The peaks at 1087 and 1014 cm–1 in HNT/CS and Fe3O4–HNT/CS, respectively, are assigned to Si–O. The NH2 vibration frequency could be assigned to the peaks at 1588 and 1515 cm–1 of HNT/CS and Fe3O4–HNT/CS, respectively, whereas the peak at 594 cm–1 found in Fe3O4–HNT/CS is due to Fe–O. A shift in band position and broadening of peaks were prominent after the adsorption, suggesting further interaction between the dye and the adsorbents.

Figure 4

FTIR spectra of HNT/CS and Fe3O4–HNT/CS before adsorption and after adsorption.

X-ray diffraction (XRD) spectra of CS, HNT, HNT/CS, and Fe3O4–HNT/CS are presented in Figure . The characteristic peaks of CS are shown at 14.4 and 20.5°. The XRD spectra of HNT/CS showed that CS is well incorporated into the HNT as all of the peaks in the individual samples were retained in their composite. The XRD patterns of Fe3O4–HNT/CS displayed distinct peaks at 2θ values of about 30.6, 35.9, 43.8, 57.2, and 63.1, as marked in the figure. These peaks are characteristics of Fe3O4, thereby indicating that the composite had been magnetically incorporated in agreement with previous studies.[19]

Figure 5

XRD spectra of the Fe3O4–HNT/CTS composite.

Magnetic properties of the Fe3O4–HNT/CTS composite are shown in Figure . The curves showed that magnetization increased with an increase in the magnetic field, and also they displayed symmetrical nature and passed through the origin. The hysteresis loop shows zero coercivity with zero remanence, implying that the composite has a superparamagnetic property.[20,21] When compared with pure Fe3O4 with a magnetization value of 0.78 emu mg–1 (inset), Fe3O4–HNT/CS shows a lower saturation magnetization of 0.12 emu mg–1, and this can be attributed to the presence of larger fractions of CTS and HNT in the composite.

Figure 6

Magnetic properties of the Fe3O4–HNT/CTS composite.

Effects of Contact Time and Initial Concentrations on Adsorption Capacities

The effects of contact time and initial concentrations on the uptake of CR by HNT/CS and Fe3O4–HNT/CS are presented in Figure a,b. A rapid adsorption of dye was noticed in about 10–15 min of the experiment, then a quasi-equilibrium was established, and the process displayed no significant increase in dye uptake after 30 min in the solutions with the two adsorbents, inferring a dynamic equilibrium (30–180 min). The adsorption capacity at equilibrium increases from 9.2 to 35.0 and 8.7–30.9 mg g–1 for Fe3O4–HNT/CS and HNT/CS, respectively, as the initial concentrations of CR increase from 50 to 200 mg L–1.

Figure 7

Effects of contact time and initial concentrations on adsorption of CR by (a) Fe3O4–HNT/CS and (b) HNT/CS.

Effect of pH on the Removal Efficiency of the Adsorbent

The chemistry of the adsorption medium, to certain extent, is a determinant in the adsorbent and adsorbate interaction, thereby making adsorption highly pH-dependent. The efficiency of CR removal from its aqueous solution as a function of pH between pH values of 2 and 10 is presented in Figure . The pKa of CR is 4.1;[22] therefore, at lower pH, the negatively charged molecules of the dye predominate in the solution (Figure a), making part of the adsorbent soluble and leading to inefficiency of the composite. As the pH increases, the composite becomes more stable and the surface charge leads to an increase in electrostatic interaction between the dye molecule and the adsorbent. Fe3O4–HNT/CS adsorbed slightly higher than the HNT/CS composite, which may be due to additional surface charges provided by the magnetic nanoparticles.

Figure 8

(a) pH pzc for Fe3O4–HNT/CS and (b) effect of pH on CR adsorption.

Effect of Adsorbent Dosage on Removal Efficiency

The effect of adsorbent dosage on the removal efficiency of CR by Fe3O4–HNT/CS and HNT/CS is shown in Figure . The increase in adsorbent dosages from 10 to 40 mg of Fe3O4–HNT/CS led to a corresponding increase in adsorption efficiency from 97.9 to 99.6%, whereas the efficiency of HNT/CS increased from 96.07 to 97.85% as the dosage increased from 10 to 30 mg. These increase in efficiencies may be attributed to increment in the number of adsorption sites available for adsorption.[23] However, a further increase in adsorbent dosages led to reduced efficiency; these observations may be a result of particle aggregation, leading to a decrease in the total surface area of the adsorbent and an increase in the diffusional path length.

Figure 9

Effects of adsorbent dosage on adsorption of CR by Fe3O4–HNT/CS and HNT/CS.

Adsorption Isotherms

Adsorption models are mathematical and fundamental equations used to quantify and explain the mechanism of the adsorbent–adsorbate equilibrium. In this study, the equilibrium data were analyzed with nonlinear models of Langmuir, Freundlich, Dubinin–Radushkevich, and Tempkin isotherms presented in Table . The details of these models had been given elsewhere.[24,25]

Table 1

Isotherm Models for the Adsorption Study

name	model	parameters
Langmuir		Qmax, b, RL
	RL = 1/(1 + bC0)
Freundlich	Qeq = KFCe1/n	KF, n
Tempkin		aT, bT
Dubinin–Radushkevich	Qe = Qs e–βε2	Qs, b, E
	E = (2β)−0.5

The isotherm models were fitted into the experimentally determined Qe vs Ce by nonlinear regression using the Scientist 2.0 software package for Windows (Micromath Scientific Software, Salt Lake City, UT). The adequacy and fitness of the isotherm equation to the adsorption data were based on the values of the correlation coefficients, R2.[26]Figure shows isotherm fits for CR adsorption by HNT/CS and Fe3O4–HNT/CS, and they were obtained through the parameters presented in Table . From the figure, it is obvious that the experimental and the theoretical data displayed a higher level of agreement, which is evident in the values of R2 obtained for isotherm parameters. The Langmuir isotherm, which proposed monolayer adsorption capacities of 41.54 and 54.49 mg g–1 for HNT/CS and Fe3O4–HNT/CS respectively, slightly increased the adsorption capacity displayed by the magnetized composite, which could be attributed to the presence of the magnetic particles. The values of RL obtained for the adsorbents implied that the duo displayed favorable adsorption potentials toward CR. The empirical parameter, 1/n, of the Freundlich isotherm showed that the HNT/CS sample displayed a lower surface heterogeneity than that of Fe3O4–HNT/CS and hence its higher affinity; however, both adsorbents have 1/n values less than 1, confirming their favorable adsorption potentials toward CR.[27] The Tempkin isotherm further confirms the earlier observation of surface heterogeneity; the difference in energies of these heterogeneous surfaces justifies the higher adsorption capacity of Fe3O4–HNT/CS, which has lower energy of adsorption, as shown in Table . The Dubinin–Radushkevich isotherm allows the estimation of the energy of adsorption on energetically heterogeneous sites. This isotherm gave the lowest R2 when compared with others, and the value of energy obtained showed that the adsorption of CR on HNT/CS and Fe3O4–HNT/CS is by chemisorption.

Figure 10

Isotherm fits for the adsorption of CR on (a) HNT/CS and (b) Fe3O4–HNT/CS.

Table 2

Isotherm Parameters for the Adsorption of CR by HNT/CS and Fe3O4–HNT/CS

isotherms	parameters	HNT/CS	Fe3O4–HNT/CS
Langmuir	Qmax (g mg–1)	41.535	54.491
	b (L mg–1)	0.133	0.024
	RL	0.075	0.299
	R2	0.999	0.996
Freundlich	KF (g mg–1 min–1/n)	9.499	2.267
	1/n	0.373	0.558
	R2	0.995	0.994
Tempkin	bT (J mol–1)	278.645	223.282
	aT (L mg–1)	1.337	0.262
	R2	0.998	0.997
Dubinin–Radushkevich	QS (g mg–1)	30.293	28.737
	β × 103 (mol kJ–1)2	1.97	2.19
	E (kJ mol–1)	15.914	15.126
	R2	0.982	0.993

Adsorption Kinetics Studies

The description of the adsorption rate at the solid/solution interface has been accomplished using several kinetic models. The mechanism of adsorption and the potential removal rate were analyzed by subjecting the kinetic data to pseudo-first-order and pseudo-second-order kinetic models.[28] The intraparticle diffusion model was used to investigate the diffusion-rate-controlling steps affecting the surface reaction, to properly recognize the adsorption kinetics.[29,30] The Elovich kinetic model was used to characterize the effect of activated chemisorption on the adsorption kinetics,[31] and the Bangham model was used to study the influence of pore diffusion on the adsorption kinetics, whereas double-exponential kinetic models properly analyzed the two-step adsorption mechanism, i.e., the rapid and slow adsorbed fractions of the entire process. The mathematical expressions of these models are presented in Table , whereas their extensive discussion abound in the literature.[24,25,30]

Table 3

Kinetic Models for the Adsorption Study

name	model	parameters
pseudo-first-order	Qt = Qe(1 – e–k 1t)	Qeq, k1
pseudo-second-order		Qeq, k2
Elovich	Qt = 1/β ln(αβ × t)	α, β
intraparticle diffusion	Qt = Kidt0.5 + Ci	Kid, Ci
Bangham	Qt = kbts	kb, s
double-exponential		Qe, D1, D2, kD1, kD2

The kinetic models were fitted into the experimentally determined Qt vs t by nonlinear regression, as described for the isotherm fits. The adequacy as well as the fitness of the kinetic equations to the experimental data was based on the values of the correlation coefficients, R2, and sum square error function (% SSE).

The pseudo-first-order fits (Figure a,b) and second-order kinetic fits (Figure c,d) for the adsorption of CR on Fe3O4–HNT/CS and HNT/CS are presented, and the corresponding parameters are also shown in Tables and 5. Pseudo-first-order adsorption kinetics by the Fe3O4–HNT/C composite displayed correlation coefficient, R2, ranging between 0.982 and 0.995 with the experimental values of Qe showing a wide deviation from the calculated values (Table ). The pseudo-second-order parameter R2 ranged between 0.989 and 0.999, the experimental values of Qe are a bit close to the calculated values, and the %SSE showed that the adsorption process for this adsorbent is best explained by the pseudo-second-order kinetic model. The rate constant k2 is in the magnitude of 10–3; it decreases with an increase in concentration, and the lower values obtained suggest a longer equilibrium time.[32] The adsorption by HNT/CS displayed a contrary mechanism (Table ); the %SSE and calculated Qe supported the pseudo-first-order kinetic mechanism.

Figure 11

Pseudo-first-order (a, b) and pseudo-second-order (c, d) fits for the adsorption of CR on Fe3O4–HNT/CS and HNT/CS, respectively.

Figure 12

Elovich (a, b) and Bangham kinetic model (c, d) fits for the adsorption of CR on Fe3O4–HNT/CS and HNT/CS, respectively.

Table 4

Kinetic Parameters for the Adsorption of CR by Fe3O4–HNT/CS

pseudo-first-order kinetics model parameters						pseudo-second-order kinetics model parameters
Co (mg L–1)	50	75	100	150	200	Co (mg L–1)	50	75	100	150	200
Qe exp (mg g–1)	10.13	15.886	18.95	26.807	41.51	Qe	10.13	15.886	18.95	26.807	41.51
Qe cal (mg g–1)	9.798	15.01	17.853	23.395	35.777	Qe cal (mg g–1)	10.042	14.167	19.429	27.083	39.578
k1 (min–1)	0.525	0.149	0.199	0.184	0.244	k2 × 103 (g mg–1 min–1)	17.656	2.971	1.674	0.827	0.858
R2	0.992	0.995	0.982	0.988	0.989	R2	0.999	0.989	0.999	0.995	0.996
% SSE	0.107	0.304	0.335	1.62	1.907	% SSE	0.008	1.171	0.064	0.011	0.217
Elovich Model Parameter						Bangham Model Parameters
Qe cal (mg g–1)	10.885	16.725	19.819	26.83	41.088	Qe cal (mg g–1)	9.543	12.29	15.518	19.763	31.626
α (mg (g min)−1)	129 681	500.707	2742.43	906.472	3854.9	kb (mg g–1 min–s)	8.684	8.168	11.247	13.034	22.142
β (g mg–1)	1.337	0.476	0.489	0.302	0.225	s	0.035	0.151	0.119	0.154	0.132
R2	0.998	0.997	0.998	0.999	0.999	R2	0.998	0.999	0.989	0.993	0.989
% SSE	0.00555	0.00279	0.00211	1 × 10–6	0.0001	% SSE	0.003	0.051	0.033	0.069	0.057
Intraparticle Diffusion Model Parameters						Double-Exponential Model Parameters
K1d (mg g–1 min–0.5)	2.574	3.104	4.133	5.199	8.43	Qe cal (mg g–1)	10.151	15.973	18.964	28.034	42.888
C1 (mg g–1)	1.027	0.915	1.065	1.632	2.8	D1 (mol L–1)	1.272	7.528	10.61	24.402	16.212
R2	0.995	0.986	0.979	0.958	0.973	D2 (mol L–1)	8.878	8.444	8.35	7.21	26.676
%SSE	0.733	3.449	0.983	3.533	2.132	kD1 (min–1)	0.05	1.186	0.826	0.044	0.022
K2d (mg g–1 min–0.5)	0.077	0.47	0.382	1.051	1.48	kD2 (min–1)	1.057	0.057	0.063	0.044	0.988
C2 (mg g–1)	9.371	11.323	15.153	15.97	26.007	R2	0.999	0.998	0.999	0.999	0.998
R2	0.999	0.996	0.999	0.999	0.999	% SSE	0.0004	0.0029	0.0001	0.1917	0.1033
% SSE	0.206	2.977	1.495	6.37	5.539

Table 5

Kinetic Parameters for the Adsorption of CR by HNT/CS

pseudo-first-order						pseudo-second-order
Co (mg L–1)	50	75	100	150	200	Co (mg L–1)	50	75	100	150	200
Qe exp (mg g–1)	10.94	13.03	24.392	32.316	40.708	Qe	10.94	13.03	24.392	32.316	40.708
Qe cal (mg g–1)	11.078	12.904	24.97	32.661	42.518	Qe cal (mg g–1)	11.17	13.442	26.827	34.963	45.788
k1 (min–1)	0.191	0.277	0.142	0.189	0.179	k2 x 103 (g mg–1 min–1)	3.874	4.756	0.907	0.978	0.667
R2	0.994	0.998	0.998	0.999	0.998	R2	0.996	0.999	0.997	0.998	0.995
%SSE	0.016	0.009	0.056	0.011	0.198	%SSE	0.044	0.1	0.997	0.671	1.557
Elovich Model Parameters						Bangham Model Parameters
Qe cal (mg g–1)	11.439	13.316	25.1	33.301	43.365	Qe cal (mg g–1)	9.547	11.841	20.206	28.216	36.165
α (mg (g min)−1)	115.005	5400.46	73.209	548.771	364.25	kb (mg g–1 min–s)	6.78	9.475	13.59	20.76	25.923
β (g mg–1)	0.747	0.949	0.284	0.273	0.192	s	0.126	0.082	0.146	0.113	0.123
R2	0.998	0.999	0.991	0.994	0.989	R2	0.998	0.999	0.989	0.993	0.989
%SSE	0.0021	0.0005	0.0008	0.0009	0.0043	%SSE	0.016	0.008	0.029	0.016	0.012
Intraparticle Diffusion Model						Double-Exponential Model
K1d (mg g–1 min–0.5)	2.488	3.249	5.718	7.84	10.866	Qe cal (mg g–1)	13.977	13.53	25.7	34.101	44.119
C1 (mg g–1)	0.73	0.889	0.178	1.166	0.092	D1 (mol L–1)	3.362	4.597	18.178	17.573	27.605
R2	0.98	0.982	0.997	0.991	0.991	D2 (mol L–1)	10.134	8.934	7.523	16.54	16.542
%SSE	0.275	0.114	0.719	0.059	0.13	kD1 (min–1)	0.004	0.073	0.09	0.08	0.095
K2d (mg g–1 min–0.5)	0.293	0.189	0.274	0.455	0.584	kD2 (min–1)	0.183	0.751	0.779	0.498	0.388
C2 (mg g–1)	8.975	11.659	22.424	28.976	37.552	R2	0.999	0.998	0.999	0.999	0.998
R2	0.998	0.999	0.999	0.999	0.999	%SSE	4.722	0.136	0.259	0.274	0.598
%SSE	0.575	0.24	0.138	0.238	0.048

The fitting results by the Elovich model (Figure a,b) and Bangham model (Figure c,d) were as displayed, whereas the corresponding parameters are given in Tables and 5. The Elovich model displayed a better fit for adsorption by the two adsorbents; variation in β was noted as the CR concentration increased, which could be attributed to the relationship between the surface activation energy and dye concentration; and the R2 and %SSE values showed that the model perfectly fitted the data from the two adsorbents. The Bangham model displayed good agreement with the experimental data from the adsorption processes, and the Bangham constant kb depicts the adsorption rate of the adsorbent. From the tables, it is obvious that the rate increases as the concentration of the dye increases. The correlation coefficient R2 and %SSE obtained for this model indicate that the model perfectly explains the kinetics of adsorption as pore-diffusion-controlled adsorption.[32,33]

The intraparticle diffusion and double-exponential models’ fitting results are displayed in Figure a–d for further understanding of adsorption of CR on HNT/CS and Fe3O4–HNT/CS, respectively. The corresponding parameters in the tables indicate high correlation coefficient, R2, and reasonably lower values of %SSE, suggesting better fittings. The intraparticle model fitting and parameters for the two adsorbents show that the curves do not pass through the origin, i.e., C1 ≠ 0, which is a confirmation that although the process involved intraparticle diffusion, it is not the sole rate-controlling step. The adsorption process is best described by the two-step mechanism, as confirmed by the intraparticle diffusion model. Using the double-exponential model, the fits from this model (Figure c,d) displayed good agreement with the data and the process could be described as having the first step of rapid dye uptake involving external and internal diffusion followed by a prevailing slow step controlled by intraparticle diffusion adsorption until equilibrium is reached.

Figure 13

Intraparticle diffusion (a, b) and double-exponential model (c, d) fits for the adsorption of CR on Fe3O4–HNT/CS and HNT/CS, respectively.

Thermodynamics of the Adsorption Process

The thermodynamics of the adsorption processes was studied by subjecting equilibrium data at various temperatures to respective thermodynamic eqs –3. These equations enable the determination of the free energies (ΔG°), enthalpies (ΔH°), and changes in entropy (ΔS°) associated with the adsorption processes.[33]where m is the adsorbent dosage, Qe is adsorption capacity at equilibrium, and Ce is the equilibrium concentration of the dye.The van’t Hoff plot for the adsorption of CR by Fe3O4–HNT/CS and HNT/CS is shown in Figure , whereas the thermodynamics parameters are presented in Table .

Figure 14

van’t Hoff fit for adsorption of CR by Fe3O4–HNT/CS and HNT/CS.

Table 6

Thermodynamic Parameters for Adsorption of CR by Fe3O4–HNT/CS and HNT/CS

	Fe3O4–HNT/CS					HNT/CS
temp (K)	Kd	ΔG (kJ mol–1)	ΔH (kJ mol–1)	ΔS (J mol–1 K–1)	R2	Kd	ΔG (kJ mol–1)	ΔH (kJ mol–1)	ΔS (J mol–1 K–1)	R2
303	1.132	–2.852	69.469	240.458	0.949	1.635	–4.118	39.547	145.129	0.973
313	3.246	–8.447				2.449	–6.372
323	3.505	–9.411				2.733	–7.339
333	3.838	–10.625				3.091	–8.557
343	4.316	–12.308				3.595	–10.251

The thermodynamic parameters obtained for the two adsorbents revealed that the enthalpies of adsorption, ΔH°, were positive, connoting an increase in the adsorption efficiency with an increase in the process temperature, whereas the positive values of entropies, ΔS°, indicated that there was an affinity between the CR molecules and adsorbent surfaces and that the degree of dispersion of the process increased with an increase in temperature. The spontaneity of the adsorption of CR by the adsorbents was revealed by the negative Gibbs free energy (ΔG) values obtained at various temperatures, suggesting favorable adsorption of CR by the adsorbents at a higher temperature.

Regeneration Study

One of the important factors usually considered for the selection of adsorbent is reusability; the higher the reusability, the lower the cost.[20] To investigate the reusability of Fe3O4–HNT/CS and HNT/CS, the adsorption regeneration test were carried out according to the method described by Wang et al.[34] The solution of Congo red dye was poured into a syringe previously loaded with the adsorbents, and the syringe was then inverted into a sample collection bottle (Figures a and 16). The solution from outlet becomes clearer and the concentration of the dye in the outlet by comparing its absorbance with that of dye solution. The strong electrostatic interaction between the adsorbent and the dye molecules makes it difficult for the dye to be eluted by deionized water (Figure b); however, squeezing with 0.1 M NaOH solution for about 30 s and washing several times with distilled water until neutral pH restores the adsorbent for reuse (Figure c). Similar processes were repeated for the magnetic composite (Figure d) with similar observations. A series of compression and recycling procedures were performed, and it is amazing to note that the adsorbents were recovered with little or no deformation, with intact efficiency of the adsorbents until after about 6 cycles when it was reduced to ≈95 and 87% for HNT/CS and Fe3O4–HNT/CS, respectively. This implied that the adsorbent exhibited excellent resilience, recyclability, and durability with better reusability; therefore, the composites are promising adsorbents and highly economical for CR removal.

Figure 15

(a) Image of CR removal by HNT/CS sponge. (b) HNT/CS sponge after adsorption, (c) HNT/CS sponge after elution. (d) CR removal by Fe3O4–HNT/CS sponge, and (e) regeneration study of Fe3O4–HNT/CS and HNT/CS.

Figure 16

Structure of Congo red dye.

Conclusions

The synthesis of Fe3O4–HNT/CS and HNT/CS was successful according to the results of characterizations, as well as their use as adsorbents for the removal of CR from aqueous solution. However, the adsorption processes for the removal of CR dye depend on the contact time, initial pollutant concentration, solution pH, adsorbent dosage, and temperature. The isotherm fits well with the adsorption data with the Langmuir isotherm model showing the maximum monolayer adsorption capacities of 41.54 and 54.49 mg g–1 for adsorption of CR by Fe3O4–HNT/CS and HNT/CS, respectively. The kinetic models showed that intraparticle diffusion was not the sole rate-determination step for the adsorption process; rather, rapid external and internal diffusion followed by a prevailing slow step controlled the rate. The models are suitable for the prediction of the adsorption processes. The thermodynamic parameters of free energy, enthalpy, and entropy changes proved that the present adsorption process is feasible, spontaneous, endothermic, and random in nature. The overall analysis by kinetics, isotherms, and thermodynamics suggested that the adsorption of CR by Fe3O4–HNT/CS and HNT/CS is through chemisorption involving interaction between dye molecules and the functional groups present on the adsorbent surfaces. Consequently, it is obvious that Fe3O4–HNT/CS and HNT/CS are effective adsorbents for the removal of Congo red dye from aqueous solution.

Materials and Method

Materials

Halloysite clay nanotube (HNT), chitosan (90% deacetylation degree), iron(III) chloride hexahydrate (FeCl3·6H2O), and iron(II) chloride tetrahydrate (FeCl2·4H2O) (AR grade) were products of Sigma-Aldrich. Hydrochloric acid and sodium hydroxide were procured from Merck, India. Congo red dye was a product of Loba Chemicals, India. Other reagents were of analytical grade, and MilliQ water was used for all of the experiments.

Synthesis of the Spongelike HNT/CS Composite

HNT was activated by suspending 4 g of the powder in 40 mL of NaOH (24% w/v) and sonicated at 50 °C for 1 h. It was washed with distilled water and dried at 105 °C overnight, and the resulting solid was ground to <200 μm mesh size. CS was also activated by dissolving 1 g of the powder in 25 mL of acetic acid (2% v/v) solution under continuous stirring. Then, 1 g of the activated HNT was dispersed into the solution of the chitosan sonicated for 25 min and stirred further for 2 h. At that point, the arrangement was poured into a round and hollow plastic shape. Subsequently, it was frozen into ice at −20 °C, kept overnight in a freezer, and then lyophilized at −80 °C using a freeze dryer. The product was washed with distilled water until pH is neutral and dried in a vacuum oven overnight at 60 °C.

Preparation of the Fe3O4–HNT/CS Composite

The mixed solution of 2 mmol of FeCl3·6H2O and 1 mmol of FeCl2·4H2O (50 mL) was prepared and stirred for 1 h, and 1 g of prepared HNT/CS was immersed in the solution mixture under mechanical shaking. It was then transferred into 50 mL of ammonia solution (35% v/v) to allow coprecipitation of the magnetic particles. The mixture was transferred into a water bath at 60 °C where 2 mL of glutaraldehyde (25% v/v) was slowly dropped into the reaction system and stirred for another 1 h. The product was washed thrice with ethanol and distilled water and then dried overnight in a vacuum oven.

Characterizations

The morphology and elemental composition of the prepared spongelike HNT/CS and Fe3O4–HNT/CS were investigated using SEM (Hitachi, Japan, S-3000H) equipped with energy-dispersive X-ray (EDX). X-ray diffraction (XRD) patterns of these samples were obtained by an X-ray diffractometer (PANalytical, X’Pert PRO, Netherlands) using Cu Kα (γ = 1.54178 Å) radiation. Fourier transform infrared (FTIR) spectra were recorded from 400 to 4000 cm–1 in a TENSOR 27 spectrometer (Bruker, Germany) using the KBr pellet technique. The magnetic properties of the Fe3O4 and Fe3O4–HNT/CS composites were characterized with a vibrating-sample magnetometer (VSM) (Lake Shore, 735 VSM, model 7304), whereas thermal stabilities were evaluated using thermogravimetric analysis (TGA) recorded by an SDT Q600 V8.3 Build 101 simultaneous DSC–TGA instrument.

Preparations of Aqueous Solution of the Dye

A stock solution containing Congo red dye was prepared by dissolving accurately weighed solute such that the solution contained 1 g equivalent of the dye in 1 L of MilliQ water, and working standard solutions were then prepared from the stock by dilution. The pH of the working solution was maintained with aliquot of HCl or NaOH prior to the adsorption study.

Adsorption Studies

The batch equilibrium and kinetic adsorption studies were conducted using Erlenmeyer flasks containing 25 mL of dye solutions with a concentration range of 50–200 mL–1 and with 0.4 g L–1 adsorbents. The contents were placed in a regulated water bath (30 ± 1 °C) with a shaker at 150 rpm, samples were collected at pre-set time intervals, and the dye concentrations in aqueous media were recorded by reading the absorbance at 497 nm on a UV–vis spectrophotometer (UV–vis–NIR, Varian 500 Scan Cary). The amounts of dye adsorbed (mg g–1) by the adsorbents as a function of time (Qt) and at equilibrium (Qe) were estimated according to eqs and 5 belowwhere Co, Ct, and Ce are the initial, time t, and equilibrium concentrations (mg L–1) of the dye, respectively; V is the volume (L) of the solution; and m is the mass (g) of the adsorbent.