Study of adsorption isotherms and physicochemical characterization of natural food colourants

Synthetic dyes are widely used by the food industry, as they enable intense colours and are stable. However, some studies evidence adverse reactions associated with the consumption of these food additives, such as the development of some types of cancer, allergic reactions, induction of hyperactivity and behavioural changes in children (McCann et al., 2007; Stevens et al., 2011; Kus & Eroglu, 2015). Consequently, their safety has been questioned with increasing intensity, leading to the prohibition of the use of some of these dyes in food.


Introduction
Colour is one of the main sensory attributes that influence the choice and acceptance of food by consumers and is also considered an important parameter in quality control.During processing, the addition of either or both natural and synthetic dyes aims to restore the original colour lost or to give colour to colourless foods (Aberoumand, 2011).
Synthetic dyes are widely used by the food industry, as they enable intense colours and are stable.However, some studies evidence adverse reactions associated with the consumption of these food additives, such as the development of some types of cancer, allergic reactions, induction of hyperactivity and behavioural changes in children (McCann et al., 2007;Stevens et al., 2011;Kus & Eroglu, 2015).Consequently, their safety has been questioned with increasing intensity, leading to the prohibition of the use of some of these dyes in food.
Consumer concerns regarding the safety of consuming synthetic dyes have forced the food industry to find natural sources as replacements.The use of natural dyes is a current market trend, not only for its colour function but also for the potential health benefits attributed to natural pigments, which include antioxidant properties and prevention of cardiovascular diseases and neurological conditions (Pascual-Teresa, 2014;Esatbeyoglu et al., 2015;Rodriguez-Amaya, 2016).
Curcumin, anthocyanins and betalains are among the most used natural dyes for commercial application in foods.Anthocyanins are water-soluble pigments, widely distributed in the plant kingdom.The hue of anthocyanins varies as a function of pH, and their stability is dependent on factors, such as temperature, and the presence of light and oxygen.Betalains are red, yellow, pink or orange pigments, derived primarily from beetroot.Curcumin is a yellow pigment obtained from the roots of turmeric.Ot shows limited water-solubility and its stability depends mainly on exposure to light and pH changes (Akram et al., 2010;Buchweitz et al., 2013;Güneşer, 2016;Mondal et al., 2016;Celli & Brooks, 2017).
The interest in the possible beneficial effects of natural pigments for human health has motivated the evaluation of their various aspects, such as structure, stability and antioxidant capacity (Khalil et al., 2012;Piyarat et al., 2014;Sui et al., 2014;Xu et al., 2015).However, moisture adsorption isotherms data on natural dyes commercially available for use in foods is scarce.
On the present work, it was obtained for the first time, to the best of our knowledge, the moisture adsorption isotherms from natural food dyes that are available for commercial use (anthocyanins, betalains and curcumin).The moisture, water activity (a w ), hygroscopicity, colour and antioxidant capacity of the natural dyes were also determined.

Materials and methods
The natural dyes investigated included anthocyanins from grapes, curcumin from turmeric roots and betalains from beetroot, all kindly provided by Christian Hansen Ondustria e Comércio Ltd (Valinhos, São Paulo, Brazil), in the form of soluble powders and without additional purification.Potassium sorbate and reagent grade citric acid and disodium phosphate anhydrous were purchased from Sigma-Aldrich Company (St Louis, MD, USA).
The humidity was determined gravimetrically, vacuum oven drying at 70 °C until constant weight, according to the Association of Dfficial Analytical Chemists (2006).The a w was measured automatically with a Testo 650 apparatus (Hannover, Germany), at 25.2 ± 0.03 °C.Hygroscopicity was determined by the static gravimetric method with saturated NaCl solution (75% relative humidity (RH), a w = 0.75) (Taylor, 1961), and was defined as the equilibrium moisture reached by the dyes when exposed to 75% RH.All determinations were performed in quadruplicates.
The samples were initially submitted to drying at 70 °C in a vacuum oven (TE 395-1, Tecnal) and for heat removal, placed in desiccator containing silica gel and cooled to room temperature.After weighing, the dried samples were conditioned in hermetically sealed glass desiccators containing the saline solutions, simulating different levels of relative humidity, at room temperature controlled by air conditioning.
The evaluation of the moisture gain by weighing the capsules in an analytical balance (Mark 210A, Bel engineering, Monza, Otaly) was carried out daily until the samples reached a hygroscopic equilibrium, that is, until a constant mass.The equilibrium moisture contents were determined based on the mass difference of the samples.
For the mathematical modelling of the isotherms, theoretical and empirical models of Guggenheim−Anderson−de Boer (GAB), Brunauer−Emmett−Teller (BET), Dswin, Halsey and Henderson (Table 1) fitted the experimental data.The models were adjusted using an optimisation based on the Gauss−Newton method and convergence criteria of 10 -6 , resolved by mathematical processor Statistica 7  (Statsoft Onc., 2004).
The coefficient of determination (R 2 ) (Equation 1), the error (ε) between the experimental and predicted moisture values (Equation 2) and the mean of errors (|ε m |) (Equation 3) were used to evaluate the fit of the models to the experimental data.Mean values of errors less than 10% were considered as indicators of fine adjustments for the sorption isotherms (Peng et al., 2007).
where SQ = sum of the square; M expi = equilibrium moisture obtained experimentally; M teori = equilibrium moisture predicted by the model; ε i = relative error, and N = number of samples.
The colour characterisation was performed in quadruplicate for each natural dye (0.2 g) solubilised in a buffer solution (20 mL).For anthocyanins, the solution pH was 3, whereas, for betalains and curcumin the solution pH was 4, given that anthocyanin pigments are more stable at the lower pH (Pedro et al., 2016) and the others at the higher pH (Herbach et al., 2006;Suresh et al., 2007).The colour of the solutions was determined using COELab parameters, with the aid of a ColorQuest instrument (HunterLab., Reston, Virginia, USA).The readings of the chromaticity coordinates L* (brightness; 0 = black, 100 = white), a* (positive = red, negative = green) and b* (positive = yellow, negative = blue) were used to calculate the hue angle (h*) (Equation 4) and saturation index (C*) (Equation 5).
Table 1.Models used to fit the moisture adsorption isotherms of natural dyes.

Results and discussion
The moisture, a w and hygroscopicity values (Table 2), confirmed the ease by which the pigments are extracted and effectively dried by the manufacturing industry.
The low moisture values (<3%), indicated that the water content of the dyes is low, which contributes to their stability and hygroscopic behaviour.Tonon et al. (2009) found 2.57−2.88%moisture for spray-dried açaí, with maltodextrin as the carrier, whereas Silva et al. (2013) evaluated various carrier agents and temperatures during the spray drying of jabuticaba extracts, and observed a final humidity of the powders between 2.11 and 6.73%.For beet juice powders prepared by several different drying methods, between 2.3 and 8.4% humidity was recorded (Nemzer et al., 2011;Pitalua et al., 2010).The literature data verify that the moisture content of a powder depends on the matrix used and the drying conditions to which it was subjected.
The a w of the dyes can be regarded as favourable to stability, considering that under these conditions, there is no microbial growth and reactions, such as non-enzymatic browning, occur relatively slowly (Reid & Fennema, 2010).Compared to the values obtained in the current study, Duangmal et al. (2008) obtained lower a w (0.11 to 0.14) for anthocyanin roselle extracts, produced with maltodextrin as the carrier.The authors observed that there was greater stability for anthocyanins at a lower a w .For açaí powder, the a w ranged from 0.19-0.24,depending on the drying aid used, which, according to its chemical structure, causes differences in this parameter (Tonon et al., 2009).
Hygroscopicity, often determined in powdered products, is one of the most important properties in determining shelf life and stability during storage of food powders.The higher the hygroscopic value, the greater the susceptibility of the food to deterioration (Reid & Fennema, 2010).Caparino et al. (2012) observed hygroscopicity between 16.5 and 20.1 g•100 g -1 in mango powder produced by numerous drying methods.On jaboticaba extract powders, hygroscopicity ranged from 13.29−17.96g•100 g -1 and was dependent on the spray drying temperature and carrier agent (Silva et al., 2013).Hence, the type and concentration of carrier agent, temperature and drying method influence the hygroscopicity of powdered products, which may explain the data variation among the dye samples, which were processed with different carrier agents (glucose syrup for the anthocyanin dye, and maltodextrin for the betalain and curcumin dyes).Moreover, the results demonstrate the stability of these natural dyes under proper storage conditions and their high tolerance to microbial deterioration and degradation reactions.Such findings may imply a broader and safer use of these compounds, given that one limitation in the use of natural dyes is their susceptibility to microbial contamination, as they are materials of animal or vegetable origin.
The R 2 and |ε m | of the estimated parameters of the adsorption isotherm models are shown in Table 3.All models, except Henderson for curcumin, fitted appropriately to the experimental data, with R 2 values close to 1.The |ε m | values for the BET and Halsey models were lower than 10% for all the natural dyes.Mean deviations above this value were obtained with the Dswin  The BET model is one of the most used in the literature because it is related to the water sorption process.Based on the R 2 and |ε m | values (Table 3) and the theoretical basis of the BET model, it was selected to represent the adsorption isotherms of the natural dyes.The BET monolayer moisture (X m ) is an important parameter as it indicates the amount of water strongly adsorbed to specific sites on the surface of the material.Ot provides a first estimate of the moisture content at which a dehydrated product illustrates maximum stability (Reid & Fennema, 2010).
The X m values (Table 3) correspond to the moisture content considered safe to ensure the dyes stability during storage.For the anthocyanin and betalain dyes, these values agreed with the literature data for Bordo grape (3.53-4.17%),açaí (3.1-5.8%),raspberries (5.9%) and beetroot (3.27-6.04%)powder, but proved to be lower than tomato pulp (18.96%), and blueberry (18.9%) powders (Chranioti et al., 2015;Souza et al., 2015;Goula et al., 2008;Jiménez-Aguilar et al., 2011;Syamaladevi et al., 2009;Tonon et al., 2009).However, the X m for curcumin was shown to be superior to those cited above, suggesting that during storage, this dye is more stable compared to the powdered anthocyanins and betalains.Therefore, among the dyes evaluated, curcumin was the most stable, as it showed the lowest a w and hygroscopicity values and, also, the highest X m .
The BET model fitted the experimental data (Figure 1).All samples showed a sigmoidal sorption isotherm, characteristic of most foods (Reid & Fennema, 2010).Although the samples had similar characteristics, the curves of the natural dyes differed among themselves.According to Mathlouthi & Rogé (2003) and Szulc & Lenart (2012), powdered products have a complex structure, including soluble substances able to crystallise during the adsorption of water and, depending on the nature (amorphous or crystalline), the shape of the isotherm is different.
At low a w values, the samples showed a less hygroscopic behaviour, while at higher a w , there was a greater adsorption (Figure 1).Such a trend indicates that the higher the RH of the environment, the more hygroscopic and, consequently, the more susceptible the dye to undesirable reactions, and the greater care required for handling and storage.This same tendency was observed by Tonon et al. (2009) in açai juice powder, by Jiménez-Aguilar et al. ( 2011) in blueberry anthocyanin extract powders and by (Mosquera et al., 2012) in strawberry powder.Among the samples, the natural curcumin dye was shown to be the less hygroscopic, followed by betalain and anthocyanin, respectively (Tables 3 and 4).This difference is expected, given the dyes are produced from different sources and carriers and have distinct compositions.Tonon et al. (2009) documented that the particle size produced by the carrier agent may be associated with differences in water adsorption.The smaller the particle size, the greater the exposed surface and, consequently, the greater the adsorption of water from the environment.
The adsorption isotherms constructed, illustrate the hygroscopic behaviour of commercially available natural dyes for use in food, providing valuable insight into these compounds.The graphs (Figure 1) establish that when stored in low RH environments, anthocyanin and betalain dyes are more stable than curcumin dye.
The colour of natural dyes is affected by the concentration of the colouring principle, extraction methods, concentration and type of carrier agent, pH, and storage conditions (Cavalcanti et al., 2011;Herbach et al., 2006;Lin et al., 2009).The chromaticity coordinates L*, a*, b*, C* and h* of the three natural dyes evaluated, are shown in Table 4.The L* values can be considered low, while the positive a* and b* values indicate the presence of red, particularly regarding the anthocyanin and betalain dyes and yellow, predominantly in the curcumin dye.These values were confirmed by the h* data, which ranged from 0 to 25° for the anthocyanin and betalain dyes, indicating red tonality, and between 70 and 100° for curcumin, corresponding to a yellow tone.Moreover, these results concurred with those found in the literature for dyes and extracts produced with anthocyanins, betalains and curcumin, from various sources (Rosso & Mercadante, 2007;Herrera-Hernández et al., 2011;  Values represent the mean (n = 4) ± standard deviation.Jiménez-Aguilar et al., 2011;Lim et al., 2011;Pitalua et al., 2010).A higher C* value is related to a higher purity or colour intensity of the sample evaluated.The anthocyanin dye had a C* of 8.58, which indicates a less saturated, less pure colour.The C* values for the remaining two dyes were relatively high, which means that the colour provided was pure and intense.
Besides their colouring function, natural dyes are comprised of certain compounds that display antioxidant properties.Anthocyanins have been linked to the prevention of some diseases, and their antioxidant action is well-established (Kruger et al., 2014;Yang & Zhai, 2010).The antioxidant capacity of the anthocyanin dye (1166.59± 18.97 μM•L -1 Trolox•g -1 ) was much higher than that found by Souza et al. (2014), for Bordo grape extracts produced with various concentrations of maltodextrin as the carrier.On that study, the 2,2-diphenyl-1-picrylhydrazyl (DPPH•) antioxidant activity ranged from 25.4 ± 0.8-74.9± 0.3 μmol•L -1 Trolox g -1 , and 66.2 ± 0.3−231.5 ± 0.9 μmol•L -1 Trolox g -1 , based on the ferric reducing antioxidant power (FRAP) assay.According to the same authors, the antioxidant capacity of the samples decreased with increasing carrier content, due to the dilution effect of the extract.Conversely, the antioxidant capability of the evaluated anthocyanin dye was similar to that reported by Tonon et al. (2010) for açaí powder, using the DPPH method, which varied from 1101.73 ± 36.11-1165.84 ± 35.29 depending on the carrier agent used.
The pulp, juice and dye of beets are recommended as sources of bioactive compounds and as functional ingredients, due to the betalains contained in these vegetables, which possess antioxidant capacity (Vulić et al., 2012;Wootton-Beard & Ryan, 2011;Azeredo, 2009).The TEAC of the betalain dye (22.57± 0.26 μmol•L -1 Trolox•g -1 ) was higher than that mentioned in the literature for betalains extracted from the same source.Figiel (2010) analysed beets dehydrated by three different methods and determined between 0.49 and 0.66 μmol•L -1 Trolox•g -1 , by the FRAP assay.Gonçalves et al. (2013) measured the TEAC of purified betanin as 4.9 ± 0.7 μmol•L -1 Trolox•g -1 .Commercial dyes typically have a more complex composition than non-commercial dyes, which hinders the interpretation of the data and may explain the great differences between the antioxidant capacity of the analysed dye and those in the literature.For example, betalain dye also contains another antioxidant in its composition (citric acid), added to increase pigment stability (Gonçalves et al., 2013;Herbach et al., 2006).
The antioxidant capacity of the natural dyes seems to be affected by factors, such as pigment source, pH, anthocyanin, betalain or curcuminoid contents, the concentration of the carrier agents used in drying, the evaluation method and sample presentation (fruit, powder).This behaviour may explain the differences found between the experimental values and those reported in the literature.

Conclusions
The water content, a w and hygroscopicity of the natural dyes evaluated indicate their storage stability from a physical, enzymatic and microbiological perspective.The isotherms obtained for the three dyes showed a sigmoidal shape, characteristic of most powdered food products, and the BET model was considered the most representative.The dyes displayed intense colours and a high colouring power, indicating that these compounds can be used successfully to colour food products.Moreover, the high antioxidant capacities demonstrated that when used as a colour source in foods, natural dyes may exert an additional function with health benefits.

Table 2 .
Moisture, water activity and hygroscopicity of natural dyes.

Table 3 .
Estimated parameters of adsorption isotherms for natural dyes.

Table 4 .
Chromaticity coordinates of natural dyes.