Adsorption Characteristics of Bixin on Acid- and Alkali-Treated Kaolinite in Aprotic Solvents

The adsorption of bixin in aprotic solvents onto acid- and alkali-treated kaolinite was investigated. Kaolinite was treated three times, for 6 h each, with 8 M HCl or 5 M KOH. The adsorbents were characterized by XRD, FT-IR, EDS, and BET-N2. The effects of contact time and dye concentration on adsorption capacity and kinetics, electronic transition of bixin before and after adsorption, and also mechanism of bixin-kaolinite adsorption were investigated. Dye adsorption followed pseudo-second order kinetics and was faster in acetone than in dimethyl carbonate. The best adsorption results were obtained for KOH-treated kaolinite. In both of the solvents, the adsorption isotherm followed the Langmuir model and adsorption capacity was higher in dimethyl carbonate (q m = 0.43 mg/g) than in acetone (0.29 mg/g). The adsorption capacity and kinetics of KOH-treated kaolinite (q m = 0.43 mg/g, k 2 = 3.27 g/mg·min) were better than those of HCl-treated kaolinite (q m = 0.21 mg/g, k 2 = 0.25 g/mg·min) and natural kaolinite (q m = 0.18 mg/g, k 2 = 0.32 g/mg·min). There are shift in the band position of maximum intensity of bixin after adsorption on this adsorbent. Adsorption in this system seemed to be based essentially on chemisorption due to the electrostatic interaction of bixin with the strong basic and reducing sites of kaolinite.


Introduction
Bixin (methyl hydrogen 9′-cis-6,6′-diapocarotene-6,6′-dioate, C 25 H 30 O 4 ) is a carotenoid dye extracted from the seeds of the tropical shrub annatto (Bixa orellana L.) [1]. It is widely used in industry, cosmetics, and pharmaceutical products and as a food colouring and textile dye [2][3][4][5]. Bixin has photoactive properties and was recently explored as a sensitizing dye in solar cells [6][7][8] and for photodynamic therapy [9]. e potential uses of bixin in these applications are based on the conjugated double bond which can absorb energy in the visible region (400-500 nm), yielding colours in the yellow, orange, and red range. However, as other carotenoids, this double bond renders bixin unstable to light, temperature, and oxygen exposure [10,11]. e poor heat and light stability of carotenoids in vitro is problematic when trying to construct photofunctional materials, such as photosensitized semiconductors and nonlinear optical materials [12]. ere have been reports of e orts to increase the stability of bixin to render it suitable for a broader range of applications, by incorporating this molecule into the surface or interlayer space of clay minerals. Kohno et al. [13] showed that annatto dye/organo-montmorillonites were more photostable than pure annatto dye because the layered structure of the montmorillonite protected the dye molecules from external oxygen. Rahmalia [14] reported the immobilization of bixin on natural kaolinite. e resulting product had slower degradation kinetics in acetone than pure bixin. Furthermore, solar cells sensitized with bixin immobilized on acid-activated kaolinite [7] had a higher energy conversion e ciency than cells sensitized with pure bixin [6]. e incorporation of various organic compounds into clays and clay minerals has been reported, due to the surface properties of these minerals, such as their adsorption capacities, surface charges, their large surface area, charge density, types of exchangeable cations, hydroxyl groups on the edges, silanol groups of crystalline defects or broken surfaces, and Lewis and Brönsted acidity [15][16][17]. Kaolinite (Al 2 Si 2 O 5 (OH) 4 ) is a relatively inexpensive clay mineral that is highly e ective as a carrier material. is behaviour is governed by the extent and nature of the external surface, which can be modi ed by appropriate treatment techniques [18][19][20]. Activation with acid or alkali has been widely studied as a chemical treatment for improving the surface characteristics of natural kaolinite in terms of its interactions with adsorbate [21][22][23]. Kaolinite is also hydrophobic and could therefore easily adsorb hydrophobic organic molecules, such as bixin [24][25][26][27]. e use of an activation process without heating for the ultimate enhancement of energy economics for bixin was investigated.
Clay minerals have been used as solid matrices to enhance the stability of bixin, but no systematic study has described the mechanism of bixin adsorption onto kaolinite. Rapid and e cient dye adsorption is important for industrial applications. e isotherm and adsorption kinetics of bixin were therefore determined on acid-and alkalitreated kaolinite. Untreated kaolinite was also tested for the purpose of comparison. Acetone and dimethyl carbonate were used as solvents because bixin was highly soluble in both these aprotic solvents. We previously reported that the transition energy and molar attenuation coe cient of bixin in dimethyl carbonate were similar to those in acetone [28]. Dimethyl carbonate would therefore be an appropriate alternative to volatile organic solvents like the nonpolar aprotic solvent acetone. Dimethyl carbonate is widely used as dialkyl carbonate with many applications in novel green chemistry [29]. It is a valuable chemical for industrial chemical engineering, due also to its low toxicity [30]. is study identi ed appropriate ketone and carbonate solvents for use in applications of bixin. Such applications, including the synthesis of a new sensitizing kaolinite-bixin dye, have become important in several elds.

Preparation of Acid-and Alkali-Treated Kaolinite.
Activation was achieved by adding 10 g of kaolinite to 100 mL 8 M HCl or 100 mL 5 M KOH separately. e mixtures were incubated at room temperature for 6 h, with constant shaking (300 rpm). e suspension was ltered, and the residue was washed with distilled water until neutral and then dried in an oven at 103°C for 24 h. is process was repeated three times to optimize activation. e nal products obtained are referred to as KA for HCl-treated kaolinite and KB for KOH-treated kaolinite. e untreated sample is referred to as KN.

2.2.2.
Characterization of the Adsorbent. X-ray di raction (XRD) patterns for the samples were obtained with a Bruker X-ray di ractometer (CuKα, λ � 1.54Å, step scan size of 0.02°a nd a count time of 0.5 seconds, 25°C). Samples were prepared by allowing the particles of kaolinite to settle in water to obtain 2 μm particles. ey were then treated with ethylene glycol and prepared as oriented mounts on a glass slide.
Fourier transform infrared spectroscopy (FT-IR) was carried out on a Shimadzu FT-IR spectrophotometer, over a spectral region of 400-4000 cm −1 , with a resolution of 1 cm −1 , and samples were evaluated in powder form mixed with KBr powder.
e BET surface area of the samples was determined by the multipoint N 2 adsorption-desorption method at the temperature of liquid nitrogen (−196°C), with an ASAP 2010 (Micrometrics) instrument.
Changes in elemental composition after treatment were assessed with an Edax Ametek high-resolution energy dispersive X-ray spectroscopy (EDS) detector.

Adsorption Experiment.
Stock solutions of bixin (20 mg/L) were prepared in acetone and dimethyl carbonate separately. Solutions of the required concentration (3-18 mg/L) were prepared by diluting the stock solution. Adsorbent (0.05 g) was then added to 5 mL bixin solution (3-18 mg/L). e mixtures were incubated at room temperature (∼22°C), with shaking at 300 rpm. e samples were withdrawn after 4 h (predetermined equilibrium time), and small aliquots of the supernatant were removed and diluted to an appropriate concentration if required. e absorption spectrum was determined immediately with a Shimadzu UV-1800 UV-Vis spectrophotometer. e bixin concentration of the solutions was determined with a UV spectrophotometer calibrated at 457 nm for acetone and 456 nm for dimethyl carbonate [28]. For the contact time studies, the residual concentration of the 5 mL bixin solution (±10 mg/L) with kaolinite (0.1 g) was determined at various time points, from 5 to 360 min. e experiments were carried out in triplicate.

Characterization of Adsorbents.
e XRD patterns of KN (Supporting Information Figure S1) showed two intense di raction re ections at 2θ values of 12.3 and 24.9°, less intense re ections at 2θ values of 23.2 and 26.6°, and a hump at 2θ values of 19.8-21.5°, associated with kaolinite (PDF 00-058-2001). e di raction re ections of orthoclase were found at 2θ values of 15.4, 21.0, 25.7, 27.5, and 30.1°( PDF 00-022-1212), whereas the di raction re ections of muscovite were found at a 2θ value of 18.0°(PDF 00-058-2036). After treatment with KOH, the re ection width and intensity of kaolinite decreased at 2θ values of 12.3 and 24.9°. is decrease was attributed to a minor structural disorder resulting from alkali treatment, which a ects the crystalline nature of the clay [32].
A di ractogram for KA showed no signi cant di erence with respect to KN, but 2θ values of 15.4 and 30.1°w ere unobservable, and the re ection increased at a 2θ of 12.3°( Figure 1).
is nding may re ect the greater resistance to acid attack of the structure of kaolinite than that of orthoclase. e resistance of clay minerals to acid attack depends strongly on their crystallinity, with more regular crystals associated with greater resistance to acid attack [33]. It may also be due to the elimination of mineral impurities by acid leaching. e higher peak intensity may re ect the presence of larger crystallites or a decrease in mean lattice strain [32,34].  [35,36]. e FT-IR spectra patterns of KN, KA, and KB showed no signi cant di erences between kaolinite before and after treatment, indicating an absence of signi cant change in the kaolinite samples. Infrared absorption spectra, which displayed a fairly sharp absorption band at wave numbers around 911 cm −1 for both samples, showed an absence of change in the composition of the octahedral Al atoms following treatment with acid and alkali in the experimental conditions used.
Changes in elemental composition were investigated by EDS (Table 1). e samples of kaolinite tested were rich in silicon and aluminium. eir composition in terms of impurities, such as Na, Mg, K, and Fe, depended on the type of reagent used. Si and Al contents only slightly decreased after treatment.
is shows the resistance of natural kaolinite minerals to acid and alkali attack.   e results of nitrogen sorption isotherm analysis are summarised in Table 2. e surface area, pore volume, and mean diameter of the untreated kaolinite sample were 7.65 m 2 /g, 3.62 × 10 −2 cm 3 /g, and 18.9 nm, respectively, indicating that the porosity of the original kaolinite was low. Pore volume and mean diameter increased after acid treatment, possibly due to the dissolution of metal ions present in the kaolinite and the rearrangement of its crystal structure, as a result of a reaction between the acid and the clay mineral. Treatment with 5 M alkali at room temperature has been shown to increase speci c surface area but to decrease total pore volume and mean diameter. Belver et al. [21] reported an increase in speci c surface area, presumably because of the disaggregation/separation of kaolinite particles. e decrease in speci c surface area observed for KA may be due to an increase in crystallinity, as indicated by XRD.

Electronic Transition of Bixin.
e absorption spectrum of the supernatant of bixin in acetone and dimethyl carbonate was in the visible region, with peaks at 457 nm and 456 nm, respectively (Figure 3), associated with the 0-1 vibration band position, consistent with the nding of Rahmalia et al. [28]. After adsorption onto KA, the maximum intensity (λmax) of the bixin spectrum was still in the 0-1 band position. However, after adsorption onto KN and KB, the maximum wavelength shifted to shorter wavelengths, associated with a band position of 0-2 for KN and of 0-3 for KB, and intensity decreased. Schoonheydt and Johnston [37] reported that the absorption maxima of dye molecules in nonpolar solvents adsorbed onto the surface of clay minerals could shift to shorter wavelengths because the solvent-molecule interaction was stronger in the ground state than in the excited state. Yariv and Cross [38] suggested that the absorption band of the adsorbed dye displayed a blue shift due to interactions between the π-electrons of the dye and the hybridised orbitals of the surface oxygen atoms, leading to a stabilisation of the π-orbitals and a destabilisation of the π * -orbitals.

E ect of Contact Time.
e e ects of contact time on the amount of bixin adsorbed onto kaolinite were investigated (Figure 4). Kaolinites adsorb bixin with di erent e ciencies, and bixin was rapid and strong during the initial period of contact, between 5 and 60 minutes. During this period, the tendency towards adsorption was high, and the slope of the adsorption curve was steep. is early phase of steep increase was followed by a phase of slow increase between 120 and 360 minutes. During this period, the slope of the adsorption curve gradually attened out, and the bixin adsorption gradually decreased eventually reaching zero.
is corresponded to equilibrium being reached due to the saturation of adsorption sites. e single, smooth, and continuous nature of the curves suggested that the bixin might cover the kaolinite as a monolayer. e percentage dye adsorption was highest on KB, consistent with the BET speci c surface area analysis, which indicated that adsorption was most likely to occur on the external surface of kaolinite. It took 180 minutes to reach equilibrium for bixin in acetone with KN as the adsorbent and 240 minutes for the same mixture but with KA or KB as the adsorbent. It took 240 minutes to reach equilibrium for bixin in dimethyl carbonate, with KN or KB as the adsorbent, and 300 minutes for the same mixture but with KA as the adsorbent. is phenomenon is in uenced by the surface properties of the adsorbent and the chemical and physical constants of the solvents.

E ect of Initial Dye Concentration.
e e ect of initial dye concentration on equilibrium adsorption was investigated at di erent initial bixin concentrations. Initial bixin concentration a ected the amount of bixin adsorbed at equilibrium (Figure 5). At low initial bixin concentrations, the adsorption capacity of KN, KA, and KB increased with initial bixin concentration. It therefore seems likely that an increase in adsorption with initial dye concentration leads to an increase in mass gradient between the solution and adsorbent, thereby driving the transfer of additional dye molecules from the bulk solution to the particle surface [39].

Adsorption Isotherm.
Adsorption properties and equilibrium parameters, commonly known as adsorption isotherms, describe the interaction of the adsorbate with the adsorbents, improving understanding of the nature of the interaction. Isotherms provide information about the optimal use of adsorbents. When optimizing the design of an adsorption system, it is essential to establish the most appropriate correlation for the equilibrium curve. Several isotherm equations are available for analysis of experimental sorption equilibrium parameters. However, Langmuir and Freundlich models are the most widely  used type of isotherm [15, 23-25, 33, 40]. ese models were used to explain the interaction between bixin and kaolinite in this study. ey are the best models for explaining adsorption trends and are based on the rationale that the adsorbents become saturated with adsorbate after su ciently long contact times. e Freundlich isotherm describes the nonspeci c adsorption of a heterogeneous system and reversible adsorption.
e linear form of the Freundlich equation (1) is expressed as follows [41]: where 1/n is a combined measurement of the relative magnitude and diversity of energies associated with a particular sorption process. In the Langmuir model, the mass of solute adsorbed per unit mass of adsorbent increases linearly with solute concentration at low surface coverage, approaching an asymptote as the adsorption sites become saturated. Equation (2) is based on three important assumptions: (1) the energy of adsorption is identical for all sites and is independent of surface coverage, (2) adsorption occurs only at localised sites, with no interaction between adjoining adsorbed molecules, and (3) the sorption maximum represents monolayer coverage. e linear form of the Langmuir equation (2) can be expressed as follows [42]: We calculated the values of the parameters of the Freundlich and Langmuir model (Table 3). e equilibrium data were not consistent with the Freundlich equation for all adsorbents, in either of the two solvents.
e poor t of this model was demonstrated by the very low correlation coe cient (r 2 < 0.95). e values of 1/n < 1 indicates a nonlinear adsorption of the Freundlich model and corresponds to a Langmuir-type isotherm curve, in which marginal sorption energy decreases with increasing surface concentration. e Langmuir equation gave a better t, with r 2 > 0.95, indicating a homogeneous active site and the coverage of the adsorbent surface with a monolayer of bixin. Based on q m values, bixin adsorption to KB was more favourable than its adsorption to KA and KN. e adsorption capacity of the adsorbents appeared to increase with speci c surface area. e capacity of bixin to adsorb to adsorbents may re ect the extent to which the kaolinite was able to swell. e physical swelling of the kaolinite probably depended on the bulk properties of the intervening solvent molecules. In nonpolar solvents, increase in dielectric constant function (R(ε)) is associated with decreases in the volume of the kaolinite, due to lower levels of physical swelling [43]. Adsorption capacity was therefore greater when dimethyl carbonate (R(ε) � 0.412) was used as a solvent, because this solvent has lower dielectric constants than acetone (R(ε) � 0.872). e dimethyl carbonate also has several conformations (at least 3) against acetone which has only one conformation ( Figure 6), causing the possibility of DMC molecules in the highly aggregated solvate. Two structure conformations of DMC (Figure 6b) are favourable between DMC and KB which oxygen of DMC interact with metal atom of Si and Al of KB [44][45][46].

Adsorption Kinetics.
Lagergren's pseudo-rst order and pseudo-second order models were used to investigate the dynamics of bixin adsorption onto kaolinite. e pseudorst order model assumes that the rate of change of solute uptake over time is directly proportional to the di erence in saturation concentration and the amount of solid uptake over time. In most cases, the adsorption reaction involves di usion across a boundary (3) [47]. e adsorption process with chemisorptions controls the rate, according to the pseudo-second order model (4) [48]. k 1 and k 2 were calculated from the intercept of the corresponding plots of log (q e -q t ) against t and t/q t against t. ey are shown in Table 4, along with the values for the correlation coe cients, q e1 and q e2 (calc.) and q e (exp.). e correlation coe cient values for the pseudo-second order rate equation were higher than those for the pseudo-rst order rate equation (Table 4). e r 2 values for the plots were in the range 0.65-0.97 after application of the pseudo-rst order model, but the calculated q e1 values obtained with this model did not give reasonable values because they were lower than the experimental q e values. e q e2 and q e values were very similar for the pseudo-second order model. e adsorption process on all adsorbents in both solvents was found to follow the pseudo-second order kinetic model. ese results suggest that chemisorption predominated in the adsorption occurring in this work [17]. e best results were obtained for KOH-treated kaolinite. Adsorption kinetics was faster with acetone (k 2 � 3.27) than with dimethyl carbonate (k 2 � 1.08) as the solvent. e smaller size of acetone molecules than of dimethyl carbonate molecules and the lower viscosity of acetone (0.295 cP) than of dimethyl carbonate (0.585 cP) may facilitate the di usion of bixin into the interlayer region of kaolinite.

Mechanism of Bixin-Kaolinite Adsorption.
e FT-IR spectra of bixin dye and of bixin-KB following adsorption for di erent times were obtained (Figure 7). e spectrum of bixin dye may be assigned as follows: the -O-H stretching vibration at 3420 cm −1 , the H-C-H bending vibration at 2957, 2917, and 2850 cm −1 , the C�O ester group at 1731 cm −1 , the O-H bending vibration at 1620 cm −1 , the alkene C�C stretching at 1469 cm −1 , C-H bending of the methyl groups at 1378, C�O stretching at 1220 cm −1 , symmetric and asymmetric vibrations of the C-O-C ester group at 1180 cm −1 , and the methylene rocking vibration of ciscarotenoid at 720 cm −1 [49]. Figure 7 shows signi cant frequency modi cations of the absorption bands from bixin in low frequency areas between 3400 and 3700 cm −1 and in high frequency areas between 1600 and 1750 cm −1 . Functional groups COOH and COOR from bixin strongly absorb at strong basic and reducing sites of KN through electrostatic interactions. e frequency of bands at 3420 cm −1 disappears in favour of frequency of bands at 3619 cm −1 . e same phenomenon was observed in high frequency areas; the frequency of bands at 1731 cm −1 disappears in favour of frequency of bands at 1620 cm −1 . is indicated strong interactions between two types of metal  carboxylate groups, which result from a part of the interaction between the carboxylic group of the bixin and the other part of the carboxyester group of bixin, respectively, with Si and Al.

Conclusion
e adsorption characteristics of bixin onto kaolinite, especially for constructing photofunctional materials based on kaolinite-bixin organoclay, have been investigated. is adsorption is more dependent on the speci c surface area of the adsorbent. e adsorption capacity of the kaolinite was considerably improved by an increase in the surface speci c area. Alkali treatment (BET speci c surface area � 8.16 m 2 g −1 ) was therefore more suitable than acid treatment for increasing the capacity of kaolinite to adsorb organic molecules, such as bixin. Selection of the most appropriate aprotic solvent also increased the e ciency of bixin absorption onto kaolinite. Based on UV-visible spectroscopy data, the solvent-molecule interaction was stronger in the ground state than in the excited state. e adsorption isotherm was of the Langmuir-type and was higher in acetone than in dimethyl carbonate. Dye adsorption followed pseudosecond order kinetics and was faster in dimethyl carbonate complex solvate than in acetone. Adsorption in this system appears to be mostly due to chemisorption mediated by the electrostatic interaction of bixin with the strong basic and reducing sites of kaolinite. Finally, dimethyl carbonate has potential as a good solvent with no compound organic volatile for increasing bixin adsorption onto kaolinite.    C o : Initial concentration of the dye solution (mg/L) C e : Concentration of the dye solution at adsorption equilibrium (mg/L) K F : Freundlich adsorption isotherm constant (mg 1-1/n L 1/n g) N:

Nomenclature and Units
Freundlich adsorption isotherm constant K L : Langmuir constant (L/mg) q m : Maximum adsorption resulting in monolayer coverage on the adsorbent surface (mg/g) q t : Amount of dye adsorbed per unit mass adsorbent at any time t (mg/g) k 1 : Pseudo-rst order adsorption rate constant (1/min) k 2 : Pseudo-second order adsorption rate constant (g/mg·min) H: Initial adsorption rate at any time approaching 0 (mg/g·min) R (ε): Dielectric constant function.

Conflicts of Interest
e authors declare that there are no con icts of interest.