Extraction of hydrocolloids from Pereskia Aculeata Miller: reuse of process residue as activated carbon for the pigment-removal phase

Hydrocolloids are functional ingredients used in the food industry for the control of microstructure, texture, flavor and shelf-life. (Dickinson, 2003). As a result of the increasing demand for hydrocolloids with specific functionality in recent years, the search for new hydrocolloid sources with appropriate properties is an active area of study (Salehi et al., 2014). Plant-derived hydrocolloids have a friendlier image for consumers compared with hydrocolloids derived from animals (Vardhanabhuti & Okeda, 2006) and number of studies have been carried out to develop hydrocolloids from new vegetable sources (Capitani et al., 2013; Petera et al., 2015).


Introduction
Hydrocolloids are functional ingredients used in the food industry for the control of microstructure, texture, flavor and shelf-life.(Dickinson, 2003).As a result of the increasing demand for hydrocolloids with specific functionality in recent years, the search for new hydrocolloid sources with appropriate properties is an active area of study (Salehi et al., 2014).Plant-derived hydrocolloids have a friendlier image for consumers compared with hydrocolloids derived from animals (Vardhanabhuti & Ikeda, 2006) and number of studies have been carried out to develop hydrocolloids from new vegetable sources (Capitani et al., 2013;Petera et al., 2015).
Hydrocolloids used in the Brazilian food industry are from imported products, despite the fact that the country has native plants with high potential for hydrocolloid production, though their commercial and industrial uses have not been fully investigated (Merce et al., 2001).Aiming to provide a new vegetable source for hydrocolloid production, Lima et al. (2013) explored the hydrocolloid extraction process from Pereskia aculeata Miller.
Pereskia species are considered a general model of the morphology and physiology of ancestral cacti characteristics, preserving some primitive features lost in other genera.In Brazil, Pereskia aculeata is popularly known as ora-pro-nobis (DPN), its leaves are traditionally used in regional cuisine and is also used as an emollient, due to their high mucilage content (Takeiti et al., 2009).Mucilage is a complex polysaccharide with high water absorption capacity that represents a potential source for industrial food hydrocolloids as binder and thickening agent in food.Mucilages from cactus were largely described as arabinogalactan-rich polysaccharides made up of galactose, arabinose, rhamnose and galacturonic acid (Sáenz et al., 2004).DPN leaves are rich in high-quality and highly digestible proteins, with a 25% protein content (Takeiti et al., 2009).
The extraction process of DPN hydrocolloids (DPNH) is shown in Figure 1 (Lima et al., 2013).This process consisted of the hot extraction of raw material previously triturated with water.The solid material resulting was subjected to the Pressing step, and the liquid fraction was mixed with the result of Filtration 1, forming Extract 1.The Filtration 1 step consisted of a vacuum filtration, and Extract 1 was placed in a fixed-bed column with commercial activated carbon to remove pigments (Filtration 2), resulting in Filtrate 2, which was subjected to precipitation in ethyl alcohol.Conceição et al. (2014) adapted this process and performed hydrocolloid extraction considering the optimization of its parameters.Dne of the changes was to use only the DPN leaves, generating solid residues like plant branches in the process.This material could be transformed into activated carbon for the application of reuse in the same process for the removal of pigments, making it sustainable.Adsorption using activated carbon (AC) has proved to be very effective in the removal of dyes.However, complex media, as is the case of hydrocolloid extracts, where components must be preserved, requires the use of adsorbents with specificity and selectivity.Many researchers have studied the production of AC using renewable and cheaper precursors which are mainly industrial and agricultural by-products (Alves et al., 2013;Clark et al., 2012;Reffas et al., 2010).
In view of the aforementioned considerations, the objectives of this work were to evaluate the characteristics of activated carbon based on solid residue from the extraction of DPNH, in a sustainable and cheaper process, and its effects on pigment removal, preserving important components like carbohydrates and proteins.

Materials
DPN was obtained in May 2015 on UFLA Campus (Lavras -Brazil) for use in the present experiments.

Adsorbent preparation
DPN residue (branches) was dried at 105°C, chopped and impregnated with the corresponding treatment for 3 min, followed by 1 h activation in a muffle furnace at 300°C.
The activation solutions tested were 42.5%, 72.2% and 85.0% H 3 PD 4 ; and 10%, 15% and 20% NaDH, considering 100% impregnation rate.The material was rinsed with distilled water until the pH was around 6.5 and dried at 105 °C for 24 h.The dried material was ground and sieved to 1 to 2 mm, the same size of commercial activated carbon (Dinâmica Química -Brazil) (CAC), the standard adsorbent used in the original DPNH extraction process (Conceição et al., 2014).Prior to use, CAC was rinsed with distilled water to remove the fines and ovendried at 105°C.

Preparation of model solutions
Aqueous model solutions were prepared with malachite green (MG) dye to simulate the adsorption of DPNH pigments by the DPN activated carbons with different activating agents and the CAC.
The optimum DPN activated carbon chosen (DPNAC) and CAC were subjected to adsorption tests in different aqueous model solutions based on the material (Extract 1) before the pigment-removal phase (Filtration 2) in the DPNH extraction process shown in Figure 1.These solutions were composed of MG, protein and/or carbohydrate.
Extract 1 is a dark green mucilage rich in carbohydrates and proteins (Conceição et al., 2014).To define the composition of the model solutions, the Extract 1 was produced and the following variables were measuredI: luminosity (L* -Minolta Chroma Meter CR 200 colorimeter, Dsaka -Japan), attributed to the concentration of pigments; carbohydrate content (Anthrone method described by Dische, 1962); and protein content (Bradford, 1976).The single component solutions used were 50 mg L -1 MG (corresponding to L* = 32), 500 mg L -1 protein (bovine serum albumin -BSA -Sigma-Aldrich) and 1500 mg L -1 carbohydrate (anhydrous glucose PA -GLU -CRQ brand).The tertiary aqueous solution considered the mixture of MG, BSA and GLU.For the DPNH extraction process, maximum pigment adsorption and low carbohydrate and protein adsorption are required, since these components are the base of the final product.

Adsorbent characterization
The DPN residue-based carbons activated with the different agents were assessed for density and adsorbent yield.Density was defined as the ratio between the mass and volume of the material studied.Adsorbent yield was calculated using Equation 1.The DPN residue-based activated carbon chosen with the best activating agent was termed as DPNAC.

( ) (
) where W 1 (g) is the weight of dried DPN activated carbon and W 0 (g) is the weight of DPN residue mass (branches dried at 105 °C).
The DPNAC adsorbent surface was evaluated by Scanning Electron Microscopy (Nano Technology Systems, Carl Zeiss, Dberkochen, Germany, Evo 40 VP).Specific surface area and pore volume determinations were based on nitrogen adsorption isotherms at -196.6 °C (Autosorb and Quantachrome NDVA).Specific surface area was calculated by the Brunauere-Emmette-Teller (BET) method; pore size and total volume were calculated by the Barrete-Joynere-Halenda equation; whereas micropore volume was calculated by the t-method.
DPNAC was subjected to thermogravimetric measurements (TG) carried out on a NETZSCH STA409EP instrument (Selb, Germany), considering air dynamic atmosphere and a temperature range of 21-945 °C at 10 °C min -1 .DPNAC was also investigated using Fourier Transform Infrared (FTIR) spectroscopy, before and after chemical activation.The FTIR spectra were obtained and recorded on a FTIR spectrometer (Vertex 70, Bruker, Germany) operating in the range of 400-4000 cm −1 , with a resolution of 4 cm −1 , coupled to a UATR-Universal Attenuated Total Reflectance.

Adsorption studies
Batch adsorption studies were conducted to evaluate the effect of activation on DPN residue.Dne hundred milliliters of each model solution and an adsorbent dose of 10 g L -1 were used in 250 mL Erlenmeyer flasks, as determined by initial experiments on the maximum capacity of adsorption of MG.Flasks were agitated on a shaker at 100 rpm at 25 °C.Aliquots of 0.5 mL were taken at each time interval.Adsorption tests were performed in distilled water (blank) in triplicate to reduce the error of the measurements.The effect of contact time of the different activated carbons was evaluated in a period from 5 min to 6 h, and the initial MG concentration was 10 mg L -1 (Mall et al., 2005).
The sample with the best performance (DPNAC) was subjected to physical-chemical characterization (item 2.4).The effect of contact time on MG, BSA and GLU adsorption by DPNAC and CAC was evaluated in a period from 5 min to 24 h in the single and tertiary aqueous solution (item 2.3).The adsorption isotherms were studied in the concentration range from 10 to 170 mg L -1 MG (24h).
Malachite green (MG) concentration was determined using a UV-Vis spectrophotometer at 620 nm.Protein analyses were performed according to Bradford (1976); BSA was used as standard, and results were expressed in mg equivalent of BSA per gram of adsorbent.Carbohydrates were determined by the Anthrone method (Dische, 1962);GLU was used as standard, and results were expressed in mg equivalent of GLU per gram of adsorbent.All samples were analyzed in a UV-vis spectrophotometer (Varian Cary 50 Probe, California, USA) at the wavelength of 620 nm for carbohydrates or at 595 nm for protein.Calibration curves were prepared for each determination.
The activated-carbon adsorption capacity (q t , mg g -1 ) was calculated using Equation 2.
where C 0 and C t (mg L -1 ) are the adsorbate concentrations at initial and sampling times, respectively, V is the volume of the solution (L), and W is the mass of dry adsorbent used (g).
Evaluation of each model's ability to predict the experimental data was based on both regression correlation coefficient values (R 2 ) and difference between experimental (q e,exp ) and model-estimated (q e,est ) values, evaluated by means of the error measure (Equation 3) e est e exp e exp RMS q q q N (3) where q e,exp and q e,est are the experimental and calculated equilibrium adsorbed amounts, respectively, and N is the number of experimental isotherm points.
Adjustments to equilibrium models were made using Statistica Statsoft 8.0 software for nonlinear equations.

Preliminary results
Preliminary chemical-activation tests were carried out in DPN residue-based carbons in the removal of MG in aqueous solution.The results shown in Table 1 demonstrate that the raw material treated with H 3 PD 4 had statistically higher adsorbent yield compared with those activated with NaDH for the resulting material of 1-2 mm particle size.The carbons activated with NaDH had a lower density than those treated with H 3 PD 4 , indicating that this activation results in materials with a more brittle and fragile structure.The materials activated with H 3 PD 4 had higher MG adsorption capacity (q t ) than those impregnated with NaDH, while carbons activated with 85% H 3 PD 4 showed a more efficient adsorptive behavior even than CAC.
In addition to the positive results of phosphoric acid in the tests, its use in the activation of lignocellulosic materials is well accepted in processes involving foods because of its non-polluting nature (Lim et al., 2010).Considering these results, the DPN-residue activated carbon impregnated with 85% H 3 PD 4 (DPNAC) was chosen for further study.

Adsorbent characterization
Figure 2 represents the microstructure of DPNAC at magnifications of 1000 x (A) and 8000 x (B).This figure shows that the sample has well-defined porous structures on its surface (Figure 2A) with tendency to formation of smaller pores within the larger pores (Figure 2B).Because of machine limitations, only macropores are visible in the images.Thus, BET analyzes were performed to better identify and classify the porous structure.
The DPNAC adsorption/desorption isotherms are shown in Figure 3.The isotherms obtained in the carbon activated with H 3 PD 4 (impregnation rate of 100%) were classified as Type IV because they showed hysteresis, indicating the start of the development of mesopores.Reffas et al. (2010) prepared carbons activated with coffee sediment treated with H 3 PD 4 and observed Type-I isotherm in this material, typical of microporous materials for adsorbents prepared with a low impregnation rate (IR <30%).As impregnation was increased, there was an increase in hysteresis up to the point (IR > 120%) where the response changed and the isotherm assumed Type-IV characteristics, associated with the presence of slit-shaped mesopores, similar to that shown in Figure 3.
The surface parameters and pore structure derived from the nitrogen isotherms are compiled in Table 2. DPNAC presented both micro-and mesoporous structures (41% and 15% of the total surface area, respectively).Specific surface area and total pore volume of DPNAC are comparable to those obtained in the carbons activated with H3PD4 produced from coffee solid residues and corn cobs at high impregnation rates.DPNAC is more microporous than mesoporous even if the impregnation rate is high, but the macroporous area likely corresponds to the microporous area.This difference is attributed to the original porosity of raw materials used in the production of adsorbents (Zhang et al., 2012).The preparation of the adsorbent by activating defective coffee beans press cake (CADC) with 168% impregnation rate showed a much smaller surface area compared with DPNAC, confirming the significant effect that the precursor material has on the physical properties in the preparation of the adsorbent.Furthermore, the impregnation time used in this study, of 3 min, is significantly shorter than the 3 h adopted by Reffas et al. (2010).However, the MG molecule is rather small (1.58 nm) and thus both mesopores (2 to 50 nm average diameter) and micropores (less than 2 nm average diameter) of the adsorbent must be accessible for the dye (International Union of Pure and Applied Chemistry, 1972).
The TG and DTG curves of DPNAC are shown in Figure 4.The thermal degradation of the material showed three major intervals for weight loss, a phenomenon clearly observed in the DTG curve.The first interval between 30 °C and 200 °C showed a mass loss of 16.78% corresponding to the loss of water molecules.Materials composed of lignin activated by acids show stability of the oxygenated groups up to an approximate temperature of 200 °C, suggesting that the weight loss below this temperature is related to water evaporation (Ömeroğlu Ay et al., 2012;García-Rosales & Colín-Cruz, 2010;Polovina et al., 1997).Ligno cellulosic materials have low stability at temperatures above 250 °C, but the acid treatment increases this thermal stability (Álvarez et al., 2005).The main decomposition occurred in the temperature range of 200 °C to 670 °C, associated with the weight loss of 69.83%.The weight loss greater than 200 °C is due to the degradation of oxygen groups of the activated-carbon surface.Carboxylic groups are less stable and decompose at 400 °C, producing CD 2 .The third stage of decomposition of the sample occurs between the temperatures of 670 °C and 920 °C, with a weight loss of 7.16%.Phenolic and carbonyl groups decompose at 800 °C to produce CD and CD 2 (Bandosz, 1999;Hayashi et al., 2000;Polovina et al., 1997;Sun & Webley, 2011;Tancredi et al., 2004).Thus, the highest percentage in mass was observed in the range referring to the oxygen and carboxyl groups, and the lowest in the phenolic and carbonyl groups.The total weight loss of the sample was 93.77% at around 920 °C, resulting in ash residues (Ömeroğlu Ay et al., 2012).V T I: total pore volume; S T I: total surface area; S me I: mesopore surface area; V me I: mesopore volume; S miI: micropore surface area; V miI: micropore volume; CCACI: corn cobs activated carbon; DCACI: defective coffee beans press cake activated carbon; SGAC1I: spent coffee grounds activated carbon with low impregnation rate; SGAC2I: spent coffee grounds activated carbon with high impregnation rate.All adsorbents were impregnated with H 3 PD 4 .Thermogravimetric studies show that during the carbonization of a lignocellulosic material, carboxyl groups are the most thermolabile oxygen groups, as they can be decomposed as CD 2 at temperatures in the range of 100 to 400 °C, while carboxylic anhydrides and lactones decompose in the range of 430 660 °C.The other types of oxygen groups (phenols, ethers, carbonyls and quinones) are decomposed thermally as either CD or CD 2 at temperatures above 600 °C, the pyrone structures being the most thermally stable (decomposing at 900 to 1,200 °C) (Bourke et al., 2007).Thus, to control the acidic or basic nature of the activated carbon produced, except for the intrinsic properties of the precursor material, the temperature and heating rates must be carefully controlled in the carbonization and activation process (Dliveira & França, 2008).
The FTIR spectrum obtained for DPNAC is presented in Figure 5A in comparison with the spectrum obtained for the DPN residue (B).The spectrum of the activated carbon (A) is similar to others shown in the literature for chemical activation of lignocellulosic materials by H 3 PD 4 (Alves et al., 2013;Reffas et al., 2010).The bands between 1300 and 1000 cm −1 , with maximum at 1240 (A) and 1026 (B) cm −1 , are attributed to P=D stretching vibrations (A) and C-D stretching in acids, alcohols, phenols, ethers and esters (B).The significant difference between these bands confirms the effect of the activation procedure.The band at 1240 cm −1 (A) is not present in the spectrum for DPN residue without chemical activation (B) and has been reported to become better defined with an increase in impregnation rate (Reffas et al., 2010).Bands at 1026 cm −1 (C-D) have been reported in association with the presence of lignin and hemicellulose esters (Suárez-García et al., 2002).Notice that these bands do not appear and/or are less evident in the activated carbon (A) in comparison with the DPN residue (B).The non-appearance of this band after treatment with phosphoric acid may be associated with the possibility of the D-H grouping be participating in hydrogen bonding with the activating agent.When the D-H group participates in hydrogen bonding, the C-D bond becomes slightly weaker and, as a result, a reduction or absence of the absorption frequency is  The absorption of MG in the single component solution (Figure 6) increased with the contact time for up to 3 h for DPNAC and 6 h for CAC, with equilibrium achieved after this point.The adsorption equilibrium was similar for both adsorbents, with 4.5 mg g -1 MG.The adsorption occurred rapidly in the initial stage (before 45 min) due to the greater availability of active binding sites (Leng et al., 2015).In the last stage, the adsorption became a controlled binding process due to the lower availability of active sites (Saha et al., 2010).In the tertiary aqueous solution (Figure 6B), DPNAC had a faster adsorption of MG than CAC did for up to 12 h of contact, and after 24 h the adsorption was 4.3 mg g -1 , showing slightly lower MG adsorption than in the single systems.
In the case of BSA and GLU, minimum adsorption is desired for the process under study.For this evaluation, adsorption tests were carried out in single systems.DPNAC had the higher adsorption of BSA, reaching 35.9 mg g -1 , while CAC adsorbed observed (Barbosa, 2008).The effects of heat treatment by FTIR analysis on carbon activated with phosphoric acid was investigated (Guo & Rockstraw, 2007) and concluded that at carbonization temperatures above 300 °C, the incorporation of phosphate grouping was favored because a band around 1250 cm -1 appeared after this temperature.The broad band at 3600-3000 cm −1 is attributed to hydrogen bonds in carboxyl or phenolic groups.A reduction in the intensity of these bands after activation could be associated with interactions between the D-H group and the activating agent (Alves et al., 2013).The band between 3100-2800 cm -1 is related to the C-H bond (Milicevic et al., 2012).

Effect of contact time on adsorption
The effects of contact time on adsorption in single component solution and tertiary aqueous solution by DPNAC and CAC were investigated and results are displayed in Figure 6.   ( ) L e e n L e q K C q K C CAC DPNAC 0.991 0.997 0.164 0.206 q e (mg g −1 )I: equilibrium adsorption capacity; C e (mg L −1 )I: solute concentration in aqueous solution, after equilibrium; q o (mg g −1 )I: maximum adsorption capacity.The remaining constants are empirical parameters associated with each specific model.17.9 mg g -1 after 24 h of contact (Figure 6C).A different response occurred in adsorption of GLU (Figure 6D), with adsorbent CAC having the greatest effect on the adsorption of GLU (43.7 mg g -1 ), whereas DPNAC adsorbed only 11.6 mg g -1 after 24 h.However, as observed in Fig 06 A and  B, the results for the MG-adsorption capacity by DPNAC did not show significant differences between the single and tertiary system, demonstrating the existence of a strong affinity of this adsorbent for the dye.

Adsorption equilibrium
The adsorption isotherms of DPNAC at 25 °C are shown in Figure 7. Dbserving the isotherms, it can be seen that the curve for CAC indicates a favorable process, and for DPNAC a highly favorable process in MG removal.DPNAC has a greater adsorption capacity than the CAC sample does.The activation of rice husks with H 3 PD 4 also obtained good MG-adsorption results (Rahman et al., 2005).
The models used to assess the balance of DPNAC and CAC in MG adsorption were the Langmuir, Freundlich, Redlich-Peterson and Langmuir-Freundlich (Table 3).The study of single component solutions is important for establishing the adsorption mechanisms that occur in the evaluated system, allowing the definition of the adsorption conditions that facilitate the removal of component from the system (Clark et al., 2012).
The data displayed in Table 3 show that the adsorption of MG was best described by the Langmuir-Freundlich model.This model can describe both the Langmuir and Freundlich adsorption behaviors, which are usually better adjusted in adsorbents of heterogeneous surface (Janoš et al., 2009).
For DPNAC, it was observed that the occurrence of micropores prevailed over mesopores, but the surface area of the macropores is probably equivalent to that of micropores (item 3.2).These characteristics of porous structure contribute to the diffusion of solutes by the processes of physical adsorption in mesopores and macropores and chemical adsorption in active sites located within the micropores.The chemical and physical adsorption rates, in the process employing DPNAC as adsorbent, are probably equivalent.
By the Langmuir model, DPNAC had a higher maximum adsorption capacity q 0 (36.008mg/g) than CAC did (19.108mg/g), showing that DPNAC has a high adsorption capacity for MG.The K F (2.379) of DPNAC calculated from the Freundlich model is greater than the K F (2.055) of CAC, confirming that DPNAC has greater affinity for MG.

Conclusions
This study demonstrates the feasibility of using DPN residue in the development of low-cost adsorbent for the removal of MG, as well as promoting sustainability when applied in the process of pigment-removal of DPNH.The chemical activation with 85% H 3 PD 4 (DPNAC) favored the formation of micro-mesoporous structure with predominance of acid functional groups and the increase of selective adsorption capacity of MG compared to CAC.Adsorption isotherms were favorable to the studied process and the Langmuir-Freundlich model which best described the process in equilibrium.

Figure 6 .
Figure 6.Effect of contact time by CAC and DPNAC (25 °C, adsorbent doseI: 10 g L -1 ) onI: (A) MG adsorption in single component solution with initial concentration of 50 mg L -1 MG; (B) MG adsorption in ternary aqueous solution with initial concentrations of 50 mg L -1 MG, 500 mg L -1 BSA and 1500 mg L -1 GLU; (C) BSA adsorption in single component solution with initial concentration of 500 mg L -1 BSA; (D) GLU adsorption in single component solution with initial concentration of 1500 mg L -1 BSA.

Figure 7 .
Figure 7. Adsorption isotherms for the adsorption of MG by DPNAC and CACI: Langmuir-Freundlich fit.

Table 1 .
Adsorbent yield, density and adsorption capacity (q t ) of the activated carbons of DPN residue.
* 10 mg L -1 MG in aqueous solution, 10 g L -1 of adsorbent, 6 h, 100 rpm.Data subjected to analysis of variance and Tukey's mean test at 5% significance level.Means in the column followed by the same letter are statistically equal to each other.

Table 3 .
Adsorption isotherm models and fitting parameters.