Study of the adsorption of dyes employed in the food industry by activated carbon based on residual forestry

An activated carbon (adsorbent) was prepared from a forestry residual biomass (Capparis scabrida sawdust) by chemical activation with ZnCl2. The adsorbent was tested in kinetic experiments to remove three anionic dyes widely used in the food industry: tartrazine (TR), brilliant scarlet 4R (BS4R) and brilliant blue (BB). The adsorbent was able to remove the dyes in different intensities, and the revealed order of their adsorption ability was BS4R>TR>BB. Most of the kinetic data fit best to the pseudo-second order model; however, high accordance with other models indicates that there is more than one phenomenon to explain the adsorption process. Analyzing the data that fit well to the pseudo-second order model and considering that the equilibrium was reached, the equilibrium adsorption capacity (qe) for TR was 55.3 mg/g (when the AC load was 1 g/l and the TR initial concentration was 50 mg/l); for BS4R, 72.1 mg/g (when the AC load was 1 g/l and the TR initial concentration was 50 mg/l); and for BB, 14.1 mg/g (when the AC load was 1 g/l and the TR initial concentration was 10 mg/l) as the maximum values. AC based on Capparis scabrida residual biomass is a promising material for use in the purification of water polluted by anionic azo dyes.


Introduction
Dyes are widely used in different industries, such as textile [1], food [2] and cosmetic industries. Three of the dyes used extensively in the food industry are tartrazine (TR), brilliant scarlet 4R (BS4R) and brilliant blue (BB) (table 1 shows the chemical structure and some properties of the dyes). A portion of these dyes are disposed of (approximately 10%) in wastewater during production and consumption, even though they are toxic agents for aquatic species [3]. Although these dyes are dangerous for human health [4,5], in most countries they are still permitted [6]. Producing residual forestry biomass is very common in Peru. Currently, some low-cost materials are used to produce byproducts such as construction materials. However, there is another part of this residual biomass that is disposed of in areas without technical and environmental minimum acceptable conditions. In some cases, the residual biomass is burned in open field conditions, causing negative environmental impacts. Capparis scabrida is one of the species from dry forest in northwest Peru that is commonly used to produce handicrafts and cookware. Nevertheless, the residual biomass (sawdust mainly) from Capparis scabrida processing workshops is disposed of elsewhere in areas close to cities and rivers. Despite the fact that there are previous works related to the production of activated carbon (AC) from sawdust for dye removal dissolved in water [1, 11], the removal of different chemical structure dyes with low-cost adsorbents is still a major concern [12]. In this framework, the present work aims to use a sawdust-based activated carbon to remove TR, BS4R and BB from aqueous solutions.

Activated carbon and characterization
ZnCl2-activated carbon was produced from Capparis scabrida sawdust according to the procedure described in previous works [13,14]. Dried, ground and sieved raw material was mixed with ZnCl2 and carbonized at 600 °C in a nitrogen flux for 2 h. The resulting material was washed with an acid solution and abundant distilled water. Finally, the sample was dried and sieved to reach a particle size of less than 0.25 mm.
The produced AC was characterized to determine its textural, morphological and chemical surface properties. Nitrogen physisorption was performed to determine the textural properties of the produced AC. Nitrogen physisorption measurements at 77 K were performed on a 3Flex sorption apparatus (Micromeritics, USA). The specific surface area, SBET, was calculated according to the classical Brunauer-Emmett-Teller (BET) theory for p/p0 = 0.05-0.25. Since the specific surface area, SBET, is not a proper parameter in the case of mesoporous solids containing micropores, the mesopore surface area, Smeso, and the micropore volume, Vmicro, were also evaluated based on the t-plot method using the carbon black STSA , standard isotherm. The net pore volume, Vnet, was determined from the nitrogen adsorption isotherm at maximum p/p0 ~0.99. The mesopore-size distribution was evaluated from the adsorption branch of the nitrogen adsorption-desorption isotherm by the Barrett-Joyner-Halenda (BJH) method via the Roberts algorithm, using the carbon black STSA standard isotherm and the assumption of the slit-pore geometry (characterized by the diameter d of the pores). The micropore-size distribution was evaluated from the low-pressure part of the nitrogen adsorption isotherm (10 -7 < p/p0 <0.05) by application of the Horvath-Kawazoe solution for the slit-pore geometry of carbonaceous materials geometry (characterized by the width w of the pores) [15].
The surface functional groups in the AC were studied by FTIR analyses (Shimadzu IR-Prestige, Japan).
The morphological characteristics of the macropores in the prepared AC were analyzed using SEM micrographs obtained by a Vega3 Tescan (Czech Republic) SEM.

Adsorption kinetic tests
Typical batch kinetic experiments were carried out to test the dye adsorption capacity of the produced AC. TR, BS4R and BB were the dyes used at three initial concentrations (10, 30 and 50 mg/l), and the activated carbon doses were 0.5, 1 and 5 g/l. The volume of solutions in the experiments was 250 ml, and 14 aliquots (10 ml) were taken during the 4 h of experimental time. Dyes concentrations were measured using a Merck Pharo 300 spectrophotometer at wavelengths of 428, 508 and 630 nm for TR, BS4R and BB, respectively.
Kinetic data were evaluated by widely used models: pseudo-first [16] and pseudo-second [17]. The nonlinear form of the models was used, and the parameters were calculated using Origin Pro 2018b software.

Activated carbon characterization
As shown in figure 1a, it is evident that the investigated AC is a microporous-based material, corresponding to the isotherm I type according to the IUPAC classification. Figure 1b indicates that the AC comprises some portion of small mesopores (with pore size < 7 nm) and micropores (inset in figure 1b with pore size ~0.5 nm.   Capparis scabrida-based AC showed a very high specific surface area close to 1676 m 2 /g (table 2) and a high proportion of micropores. The ratio of micropore volume to the net pore volume is ~78%. All determined textural parameters of the investigated AC are summarized in table 2. Table 2 Textural properties and pH point of zero charge (pHPZC) of the investigated activated carbon. As shown in the micrographs of the AC (see figure 2), it is possible to see the macropores. The shape of the macropores is variable, with predominantly ovoid and irregular shapes.    In the FTIR spectrum of the AC (see figure 3), peaks between 2316 and 455 cm -1 are intensive. The peak at 2316 cm -1 related to C≡C stretching vibrations in alkyne groups [18] is present. The peak at 1519 cm -1 could be assigned to the skeletal C=C stretching vibration in aromatic compounds [19,20] or C=O in the quinone structure and carboxylate groups [21]; the peak at 1018 cm -1 was assigned to the C-O stretching vibration in alcohol, phenol, ether or ester [19,20,22] and the peaks at 603 -680 cm -1 were assigned to the O-H bending vibration [20]. Additionally, the peak at approximately 455 cm -1 corresponds to the presence of a semiconductor compound. In a previous work, [14] ZnO and ZnS were detected via XRD in AC derived from Capparis scabrida. ZnO and ZnS are evolved from the chemical activator (ZnCl2) during the activation process.

Kinetic Adsorption tests
Adsorption kinetic data for three different dyes, TR, BS4R and BB, are depicted in figure 4. The adsorbent was able to remove the dyes in different intensities; however, establishing an order of AC adsorption efficiency of the dyes could be BS4R>TR>BB. Different authors have studied the adsorption of anionic dyes [8, [23][24][25], such as the dyes studied in this research, and they coincided in the fact that the best adsorption of TR, BS4R and BB occurs at pH levels less than 3. According to Gautam, Gautam, Banerjee, Lofrano, Sanroman, Chattopadhyaya and Pandey [26], the presence of functional groups such as -OH and C=O over the surface of the adsorbent facilitates the adsorption of anionic dyes at low pH levels. The pH of the initial solution was not measured; however, the usual pH levels of these dye solutions are low, approximately 3.5 [2]. The pHPZC of the adsorbent is 3.5, which means that the pH of the solution is at least similar to pHPZC. This effect supposes that the net charge of the adsorbent is neutral; however, there are positively and negatively charged sites on the surface of the adsorbent. Obviously, protonated dyes can be adsorbed by positively charged sites.
Comparing the kinetic data based on the initial AC load, the equilibrium is quickly reached when the initial AC load is 5 g/l for every initial dye concentration. However, when the initial AC load is 1 and 0.5 g/l, the number of experiments until the equilibrium is reached decreases with the increase in the initial concentration of the dye. This fact is well known, and it is based on the assumption that the higher the amount of adsorbent is, the higher the number of active sites. Another fact that is possible to recognize in the curves is that when the initial concentration of the dye is higher, the equilibrium adsorption value is higher. This effect is based on the affirmation that when the initial adsorbate concentration is higher, the initial driving force is higher as well [2].
In the case of BB adsorption, the process did not reach equilibrium in most cases. The equilibrium is reached when the initial BB concentration is 10 mg/l and the doses of AC are 1 and 5 g/l. This fact could be related to the larger molecular size of the BB molecule compared to that of BS4R and TR, both 1.42 x 0.765 nm [27]. The larger molecular size of the dye could negatively affect the adsorption process since the larger molecular size of the adsorbate does not allow the adsorbate to access small micropores, considering the micropore size ~ 0.5 nm in the investigated AC. Another reason could be related to the pH of the solution during the experiment. The pH of the dye solution during the adsorption process is variable; thus, it could be possible that during BB adsorption, the pH level increases because of the presence of positively charged active site in the BB molecule (N + ), causing a reduction in the adsorption and even fluctuation (increase and decrease) during the time. The presence of positive active sites could cause repulsion with the protonated active sites over the surface of the AC.
The parameters for the dye kinetic adsorptions by AC are shown in table 3. When the initial concentration of dyes is 10 and 30 mg/l, most of the data fit to the pseudo-first and pseudo-second order models (PFO and PSO respectively) with an irregular pattern. In the case of the kinetic data when the initial concentration of dyes is 50 mg/l, the data fit to the both models in the case of TR and BB; however, in the case of BS4R, the data does not fit to the PFO nor PSO when the initial AC load is 0.5 and 1 g/l. When the initial AC load is 5.0 g/l the data fit both models. Despite this tendency, most of the kinetic data present high accordance with more than one kinetic model. This is evidence to postulate that different phenomena (such as mass transfer, chemical bonding, electrostatic forces, etc.) are involved during the dye adsorption process instead of just one. Analyzing the data that fit well to the pseudo-second order model and considering that the equilibrium was reached, the equilibrium adsorption capacity (qe) for TR was 55.3 mg/g (when the AC load was 1 g/l and the TR initial concentration was 50 mg/l); for BS4R, 72.1 mg/g (when the AC load was 1 g/l and the TR initial concentration was 50 mg/l); and for BB, 14.1 mg/g (when the AC load was 1 g/l and the TR initial concentration was 10 mg/l) as the maximum values.       qt = amount of adsorbate adsorbed per mass of adsorbent at time t (mg/g); qe =amount of adsorbate adsorbed per mass of adsorbent at equilibrium (mg/g); k1 = rate constant of the PFO equation (1/min); k2 = rate constant of the PSO equation (g/mg x min); X 2 is chi square and R 2 is the coefficient of determination

Conclusions
The adsorbent (activated carbon) prepared by the chemical activation of Capparis scabrida residual biomass was able to remove tartrazine (TR), brilliant scarlet 4R (BS4R) and brilliant blue (BB) at different intensities; however, the order of adsorption ability was as follows: BS4R>TR>BB. Most of the kinetic data fit best to the pseudo-second order model; however, a high accordance with other models reveals that there is more than one phenomenon to explain the adsorption process. Analyzing the data that fit well to the pseudo-second order model and considering that the equilibrium was reached, the equilibrium adsorption capacity (qe) for TR was 55.3 mg/g (when the AC load was 1 g/l and the TR initial concentration was 50 mg/l); for BS4R, 72.1 mg/g (when the AC load was 1 g/l and the TR initial concentration was 50 mg/l); and for BB, 14.1 mg/g (when the AC load was 1 g/l and the TR initial concentration was 10 mg/l) as the maximum values. AC based on Capparis scabrida residual biomass is a promising material for use in the purification of water polluted by anionic azo dyes.