The cupuassu pulp was supplied by the Mixed Agricultural Cooperative of Tomé-Açu (CAMTA), located in the city of Tomé-Açu, Pará, Brazil (02° 25' 08” S and 48° 09' 08” W). The pulp was stored in polyethylene bags (1 kg), frozen at -18°C, and transported to the Federal University of Pará (UFPA) (01° 27' 21” S and 48° 30' 16” W). For the experiments, the pulp was thawed under refrigeration (≈ 4°C) and then brought to room temperature (≈ 25°C). Maltodextrin with a dextrose equivalent of 20 (Maltogill 20, Cargill Agrícola SA, Uberlândia, Brazil) was used as a drying agent.

The cupuassu powder was obtained in a mini spray dryer (Büchi B-290, Büchi Labortechnik AG, Flawil, Switzerland) at the drying laboratory of the Federal University of Pará (UFPA) (01° 27' 21” S and 48° 30' 16” W). The mixture (pulp:water, ratio 1:1 w/w) was homogenized in a colloidal mill (Brasil 56-RC-6332, São Paulo, Brazil) for 3 min and filtrated through 0.30 mm mesh. Then, 40% maltodextrin was added under the 6.84% total solids content of the filtered mixture. The spray dryer was fed with the formulation into the chamber by a peristaltic pump, using a feed flow rate of 7.5 mL/min and feed temperature of 25°C. The equipment was programmed to operate with nozzle internal diameter of 0.7 mm, parallel current flow, aspiration air flow in 100%, compressed air pressure of 0.8 MPa, drying air flow rate of 35 m³/h and inlet and outlet air temperatures of 150 and 93°C, respectively.

The physicochemical properties of cupuassu powder were determined according to the methods proposed by the AOAC [23AOAC. Association of Official Analytical Chemists Official methods of analysis 16. 1997.]: moisture in a vacuum oven at 70°C (Marconi MA030/12, São Paulo, Brazil) (method 920.151); ashes (method 940.26); protein (method 920.152) (nitrogen-protein conversion factor of 6.25); lipids (method 963.15); total and reducing sugars (method 920.183b); total titratable acidity (method 942.15A); pH in a potentiometer (Hanna HI 2221, Woonsocket, USA) (method 981.12); and Ascorbic Acid content (AA) (method 967.21). Other analyses were also performed, such as: total phenolic compounds (TPC) according to the methodology described by Singleton and Rossi [24Singleton VL, Rossi JA. Colorimetry of total phenolics with phosphomolybdic phosphotungstic acid reagents. Am J Enol Vitic 1965; 16: 144-58.], using a spectrophotometer (Pharmacia Biotech Ultrospec 2000, Michigan, USA) at a wave-length of 760 nm and the result was expressed as mg of gallic acid equivalents (GAE) per gram of sample; total soluble solids by direct reading in a digital refractometer (Quimis Q767BD, Diadema, São Paulo, Brazil); water activity (a_w) by direct reading in a digital thermohygrometer (Aqualab 4TEV, Decagon, Pullman, USA) at 25°C; and color parameter for L* (lightness), a* (redness\greenness), b* (yellowness\ blueness), C* (chroma – color intensity), and h° (hue angle – values of 0°, 90°, 180°, and 270° denote pure red, pure yellow, pure green, and pure blue colors, respectively) was measured using a digital colorimeter (Konica Minolta CR-400, Osaka, Japan) calibrated with white plate. All analyses were carried out in triplicate.

Powder morphology was evaluated by Scanning Electron Microscopy (SEM) in the Nanomanipulation Laboratory (PPGF/UFPA). The samples were fixated onto stubs with conventional conductive double-sided adhesive tape and metallized with a gold/palladium allow in a mini sputter coater (Quorum Technologies SC7620, Kent, UK) using 5 mA current for 120s. Next, the samples were analyzed in a scanning electron microscope (Tescan Vega3, Brno, Czech Republic) using electron beam current of 85-90 µA, 5 Kv acceleration voltage, approximately 15 mm working distance (WD), and magnification of 2000x.

The adsorption and desorption isotherms were obtained at 25°C in a Vapor Sorption Analyzer (VSA) equipment, using the DVS (Dynamic Vapor Sorption) method, which is based on the readings of sample mass and a_w successively until the pre-established equilibrium condition is reached. For the construction of the isotherms, a sample of cupuassu powder was initially stored in a desiccator with silica gel under vacuum and at 25°C for 24 hours to ensure a_w < 0.1 in the sample [25Souza TCL, Souza HA, Pena RS. Colorimetry of total phenolics with phosphomolybdic phosphotungstic acid reagents. Am J Enol Viticult 1965; 16: 144-58.
[http://dx.doi.org/10.1002/star.201200184] ]. After this step, approximately 1 g of the sample was weighed in a stainless steel capsule, in the analytical microbalance of the VSA. The data were obtained in an adsorption-desorption cycle, for a range of 0.1 to 0.9 a_w. The equilibrium condition was determined when the mass change and the time variation (Δm/Δt), between consecutive readings, reached a value lower than 0.05. After the analysis, the dry mass of the sample was determined in a vacuum oven at 70°C [23AOAC. Association of Official Analytical Chemists Official methods of analysis 16. 1997.].

The monolayer moisture content (m_o) was calculated using the BET linear equation (Eq. 1) [11Brunauer S, Emmet PH, Teller E. Adsorption of gases in multimolecular layers. J Am Chem Soc 1938; 60(2): 309-19.
[http://dx.doi.org/10.1021/ja01269a023] ].

(1)

where m = equilibrium moisture content (g H₂O/100 g d.b.); a_w = water activity (dimensionless), m_o = monolayer moisture content (g H₂O/100 g d.b.); and C = constant related to the sorption heat.

The mathematical models proposed by Halsey [13Halsey G. Physical desorption on non-uniforme surfaces. J Chem Phys 1948; 16(10): 931-7.
[http://dx.doi.org/10.1063/1.1746689] ], Henderson [14Henderson SM. A basic concept of equilibrium moisture. Agric Eng 1952; 33(1): 29-32.], Oswin [15Oswin CR. The kinetics of package life. III isotherm. J Chem Technol Biotechnol 1946; 65(12): 419-21.
[http://dx.doi.org/10.1002/jctb.5000651216] ], Peleg [16Peleg M. Assessment of a semi-empirical four parameter general model for sigmoid moisture sorption isotherms. J Food Process Eng 1993; 16: 21-37.
[http://dx.doi.org/10.1111/j.1745-4530.1993.tb00160.x] ] and GAB described by Maroulis et al. [12Maroulis ZB, Tsami E, Arinos-Kouris D, Saravacos GD. Application of the GAB model to the sorption isotherms for dried fruits. J Food Eng 1988; 7(1): 63-78.
[http://dx.doi.org/10.1016/0260-8774(88)90069-6] ] (Table 1) were fitted by nonlinear regression analysis to the experimental sorption data of the product, using the software Statistica 7.0.

The goodness of the fit of each model was evaluated in terms of maximum coefficient of determination (R²), minimum standard error of estimate (SEE) (Equation 2) and minimum mean absolute percentage error (P) (Equation 3). According to Lomauro et al. [22Lomauro CJ, Bakshi AS, Labuza TP. Evaluation of food moisture sorption isotherm equations. Part II: Milk, Coffee, Tea, Nuts, Oilseeds, Spices and Starchy foods. Lebensm Wiss Technol 1985; 18(2): 118-24.], if the P value is less than 10%, the model is acceptable.

(2)

(3)

where N = number of observations; m_e = experimental value of equilibrium moisture content (g H₂O/100 g d.b.); m_p = predicted value of equilibrium moisture content (g H₂O/100 g d.b.); and F = degree of freedom of the regression model.

The cupuassu powder composition, physicochemical characterization and color parameters are presented in Table 2.

Table 1
Mathematical models used to fit the sorption isotherms of cupuassu powder at 25°C.

The powder presented high acidity, low moisture and low a_w, which allows the product to present greater stability to degradation processes, since, the lower availability of water and the pH limit the microbial growth [26Castro TMN, Zamboni PV, Dovadoni S, Cunha Neto A, Rodrigues LJ. Parameters of quality of frozen fruit. Rev Inst Adolfo Lutz 2015; 74(4): 426-36.], and consequently, prevent the deterioration of the product during the storage. Additionally, the low fat and high protein contents; besides, the presence of vitamin C and phenolic compounds are important attributes in the product. The major constituent of the product was sugar, which is attributed to the addition of maltodextrin. In turn, despite the addition of maltodextrin, the characteristic yellowish-white color of the pulp was maintained in the powdered cupuassu. In other words, the white coloration of the added maltodextrin did not promote the depreciation of the characteristic color of the product. Similar behavior was observed for Gac fruit aril powder by Kha et al. [27Kha TC, Nguyen MH, Roach PD. Effects of spray drying conditions on the physicochemical and antioxidant properties of the Gac (Momordica cochinchinensis) fruit aril powder. J Food Eng 2010; 98(3): 385-92.
[http://dx.doi.org/10.1016/j.jfoodeng.2010.01.016] ].

The scanning electromicrograph of the cupuassu powder (Fig. 1) shows that the product presents spherical particles with some small orifices (pore). Porous microspheres affect protection of the active material and facilitate the moisture gain since the water can penetrate the pores more easily [28Malik SA, Ng WH, Bowen J, et al. Electrospray synthesis and properties of hierarchically structured PLGA TIPS microspheres for use as controlled release technologies. J Colloid Interface Sci 2016; 467: 220-9.
[http://dx.doi.org/10.1016/j.jcis.2016.01.021] [PMID: 26803601] ].

Table 2
Composition, physicochemical characterization and color parameters of cupuassu powder.

Fig. (1) Scanning electron microscopy images of cupuassu powder containing 40% maltodextrin.

Fig. (2) Sorption isotherms of cupuassu powder at 25°C and the models of Peleg and GAB fitted to the adsorption and desorption data.

Fig. (2) shows the adsorption and desorption data for the cupuassu powder at 25°C. According to the adsorption data, the product will exhibit microbiological stability (a_w < 0.6) [29Labuza TP. The effect of water activity on reaction kinetics of food deterioration. Food Technol 1980; 34: 36-59.] if its moisture content remains below 13 g H₂O/100 g d.b., which corresponds to product moisture of 11.5 g H₂O/100 g. According to the BET equation, the moisture values equivalent to the monolayer (m_o) were 4.3 g H₂O/100 g d.b. for adsorption and 7.6 g H₂O/100 g d.b. for desorption (R² > 0.98). The m_o adsorption value indicates the highest stability moisture level of the powder product. On the other hand, based on the m_o value for the desorption process, it is recommended not to dry the cupuassu pulp added of 40% maltodextrin at moisture levels below 7.6 g H₂O/100 g d.b, to avoid unnecessary energy loss. Since, at this moisture level, the a_w of the product will be less than 0.6, so its microbiological stability will be ensured.

The sorption isotherm presented type III behavior, which is characteristic of products rich in sugars, according to the classification of Brunauer et al. [11Brunauer S, Emmet PH, Teller E. Adsorption of gases in multimolecular layers. J Am Chem Soc 1938; 60(2): 309-19.
[http://dx.doi.org/10.1021/ja01269a023] ]. This is attributed to the presence of total sugars as the major constituent of the product, which is especially related to the addition of maltodextrin. Similar behavior was reported previously in atomized products as syrup powder [30Farahnaky A, Mansoorib N, Majzoobia M, Badii F. Physicochemical and sorption isotherm properties of date syrup powder: Antiplasticizing effect of maltodextrin. Food Bioprod Process 2016; 98: 133-41.
[http://dx.doi.org/10.1016/j.fbp.2016.01.003] ], strawberry powder [31Oliveira MIS, Tonon RV, Nogueira RI, Cabral LMC. Stability of spray-dryed strawberry pulp produced with different carrier agents. Braz J Food Technol 2013; 16(4): 310-8.
[http://dx.doi.org/10.1590/S1981-67232013005000037] ] and orange juice powder [32Islam MZ, Kitamura Y, Yamano Y, Kitamura M. Effect of vacuum spray drying on the physicochemical properties, water sorption and glass transition phenomenon of orange juice powder. J Food Eng 2016; 169: 131-40.
[http://dx.doi.org/10.1016/j.jfoodeng.2015.08.024] ], all using the maltodextrin as a drying agent. According to Ghorab et al. [33Ghorab MK, Toth SJ, Simpson GJ, Mauer LJ, Taylor LS. Water-solid interactions in amorphous maltodextrin-crystalline sucrose binary mixtures. Pharm Dev Technol 2014; 19(2): 247-56.
[http://dx.doi.org/10.3109/10837450.2013.775157] [PMID: 23477494] ], MSI of amorphous materials such as maltodextrin often show type II or type III profiles.

The adsorption isotherm showed a linear behavior until 0.5 a_w level, and from this point, it assumed an exponential behavior with a progressive increase of the product’s moisture content. According to this behavior, the product requires greater care, when stored and/or handled in environments with RH above 50%, since it is more susceptible to humidification, and therefore, more favorable to degradation processes (caused by undesirable reactions) and to microorganisms’ growth. Thus, to avoid excessive moisture gain, it is necessary the use of impermeable packaging or packaging with low water vapor permeability.

Another feature of the product is the presence of bioactive compounds such as vitamin C (105.84 mg/100 g) and total phenolics (133.8 mg GAE/100 g), which are very susceptible to oxidative processes. Therefore, to minimize such processes, it is strongly indicated that the packages also have impermeability to air and does not allow light to pass through. In this context, Polyethylene Terephthalate with Aluminum foil and Low-Density Polyethylene (PET/Al/LDPE) film presents as an ideal packaging material for the product [34Alves RMV, Ito D, Carvalho JLV, Melo WF, Godoy RLO. Stability of biofortified sweet potato flour. Braz J Food Technol 2012; 15(1): 59-71.
[http://dx.doi.org/10.1590/S1981-67232012000100007] ].

A type of packaging used to store powder juice presents 70 mm × 70 mm dimensions and holds 25 g of the product. Using these parameters, the equation proposed by Costa et al. [35Costa TS, Carmo JR, Pena RS. Powdered tucupi condiment: Sensory and hygroscopic evaluation. Food Sci Technol 2018; 38(1): 33-40.
[http://dx.doi.org/10.1590/1678-457x.36816] ] was used to estimate the storage time required for the product to reach a critical moisture level (13 g H₂O/100 g d.b. at a_w = 0.6). It was considered the minimum value of water vapor permeability of 0.01 g H₂O/m².day for a PET/Al/LDPE film with 86 µm thickness, being 11 µm of PET, 9 µm of Al and 66 µm of LDPE, when exposed to an environment at 38°C and RH of 90% [34Alves RMV, Ito D, Carvalho JLV, Melo WF, Godoy RLO. Stability of biofortified sweet potato flour. Braz J Food Technol 2012; 15(1): 59-71.
[http://dx.doi.org/10.1590/S1981-67232012000100007] ]. On the regarded conditions, the estimated storage time for the cupuassu powder with 3% moisture is over two years.

Fig. (2) also shows the beginning of hysteresis effect at the capillary condensation region (a_w ≈ 0.7), which extends to the monolayer region (a_w ≈ 0.1). However, the hysteresis loop did not close in this region, which can be attributed to loss of flowability and crystallinity of the product during the adsorption process. The addition of maltodextrin increases the glass transition temperature (T_g) of the product and generally tends to form amorphous solids. On the other hand, the increase in moisture content decreases the T_g of the product due to the water plasticizing effect. The amorphous solids can undergo physical changes such as stickiness, collapse, caking, tackiness, agglomeration, crystallization and compaction during processing, handling and storage. These physical changes in dehydrated products are related to their T_g. Below the T_g, the food is expected to be stable, whereas above this temperature, the rate of physical, chemical and biological changes of the product increases with the storage temperature [33Ghorab MK, Toth SJ, Simpson GJ, Mauer LJ, Taylor LS. Water-solid interactions in amorphous maltodextrin-crystalline sucrose binary mixtures. Pharm Dev Technol 2014; 19(2): 247-56.
[http://dx.doi.org/10.3109/10837450.2013.775157] [PMID: 23477494] , 36Tonon RV, Baroni AF, Brabet C, Gibert O, Pallet D, Hubinger MD. Water sorption and glass transition temperature of spray dried acai (Euterpe oleracea Mart.) juice. J Food Eng 2009; 94(3-4): 215-22.
[http://dx.doi.org/10.1016/j.jfoodeng.2009.03.009] , 37Fang Z, Bhandari B. Effect of spray drying and storage on the stability of bayberry polyphenols. Food Chem 2011; 129(3): 1139-47.
[http://dx.doi.org/10.1016/j.foodchem.2011.05.093] [PMID: 25212349] ].

Table 3
Estimated parameters and performance criteria of the models fitting to the moisture sorption data of cupuassu powder at 25°C.

The coefficient of determination (R²), standard error of estimate (SEE) and mean absolute percentage error (P), as well as the coefficients of the mathematical models fitted to the moisture sorption data of the powdered product at 25°C are shown in Table 3. According to the results, the Peleg and GAB models can predict with good precision the MSI of the cupuassu powder (Fig. 2), since the R² maximum, SEE minimum and P ≤ 10% indicated the best fit to both adsorption and desorption data of the product. Comparable results were observed by Farahnaky et al. [30Farahnaky A, Mansoorib N, Majzoobia M, Badii F. Physicochemical and sorption isotherm properties of date syrup powder: Antiplasticizing effect of maltodextrin. Food Bioprod Process 2016; 98: 133-41.
[http://dx.doi.org/10.1016/j.fbp.2016.01.003] ] for MSI of the powdered syrup. Several studies report the higher efficiency of the GAB model on predicting the MSI of powdered fruits, for example, a commercial cupuassu powder without drying agent [38Silva AE, Silva LHM, Pena RS. Hygroscopic behavior of açaí and cupuaçu powders. Food Sci Technol (Campinas) 2008; 28(4): 895-901.
[http://dx.doi.org/10.1590/S0101-20612008000400020] ], strawberry pulp powder [31Oliveira MIS, Tonon RV, Nogueira RI, Cabral LMC. Stability of spray-dryed strawberry pulp produced with different carrier agents. Braz J Food Technol 2013; 16(4): 310-8.
[http://dx.doi.org/10.1590/S1981-67232013005000037] ], mango pulp powder [39Moreira TB, Rocha EMFF, Afonso MRA, Costa JMC. Behavior of adsorption isotherms of freeze-dried mango pulp powder. Rev Bras Eng Agric Ambient 2013; 17(10): 1093-8.
[http://dx.doi.org/10.1590/S1415-43662013001000011] ], yellow mombin pulp powder [40Oliveira GS, Costa JMC, Afonso MRA. Characterization and hygroscopic behavior of lyophilized yellow mombin in pulp powder. Rev Bras Eng Agric Ambient 2014; 18(10): 1059-64.
[http://dx.doi.org/10.1590/1807-1929/agriambi.v18n10p1059-1064] ], tamarind pulp powder [9Muzaffar K, Kumar P. Moisture sorption isotherms and storage study of spray dried tamarind pulp powder. Powder Technol 2016; 291: 322-7.
[http://dx.doi.org/10.1016/j.powtec.2015.12.046] ], among others. In turn, there are few reports of the Peleg model application in the MSI prediction of powdered fruits. According to Silva et al. [41Silva RNG, Figueirêdo RMF, Queiroz AJM, Feitosa RM. Moisture adsorption isotherms of umbu-cajá powder. Rev Ed Agric Sup 2015; 30(1): 33-6.], the Peleg model resulted in better fits to the experimental data of umbu-cajá powder.