Extracts were obtained from commercial yerba mate (Ilex paraguariensis) samples (“La Merced, de campo”, Las Marías, Corrientes, Argentina); 2 g of yerba mate with 100 ml of distilled water were placed in a thermostatic bath (Haake, Germany) at 100˚C for 40 min. Once obtained, the extracts were filtered, kept in dark flasks and immediately cooled in an ice bath until analyzed. Liquid samples were frozen at −20˚C during 24 h, transferred to a −80˚C freezer for 24 h, and finally, freeze-dried (Heto FD4, Denmark) for 48 h at −50˚C under vacuum. The powders were stored in hermetic flasks in a dessecator. The yield of the whole process was calculated as the weight of lyophilized sample (g) obtained from the extracted liquor, per gram of initial dried yerba mate. The UV spectra of liquid extract thoroughly diluted and lyophilized, dissolved in distilled water to reach an equivalent concentration, were determined from 200 to 800 nm in a spectrophotometer (Shimadzu, UV-mini 1240, Japan). The solubility of lyophilized yerba mate extracts was determined dissolving 100 mg of freeze-dried extract in 1 ml of distilled water. The suspensions were stirred, left to rest for 24 h and centrifuged at 1090 g for 10 min at 25˚C (Beckman Coulter, Avanti J-25, USA).

Chromatographic analysis was performed in an HP 1100 liquid chromatograph (Hewlett Packard, US) equipped with a binary pump, thermostated column compartment, auto injector, degasser and diode array detector (DAD) connected to an HP workstation. A Zorbax 300 SB-C18 column (250 × 4.6 mm, i.d.), packed with 5 mm particles and connected to a guard column, was utilized. The mobile phases “A” and “B” consisted of a mixture of water, methanol and formic acid (79.7/20/0.3) and a mixture of methanol and formic acid (99.7/0.3), respectively. A staggered gradient elution program at 0.9 ml/min prepared as follows: 0% B/15 min, 10% B/15 min; 30% B/10 min; 60% B/10 min; 80% B/2 min, was employed. Finally, the mobile phase composition returned to 0% B in 5 min, and this composition was maintained for 10 min to equilibrate the original solvent composition of the stationary phase. Rutin, quercetin, kaempherol, caffeine, chlorogenic acid, gallic acid and caffeic acid were used as standards for identification. Their retention times, DAD spectra stored in the library and the extract samples spiked with each standard were used for identification purposes. Stock solutions of each standard (0.25 mg/ml) were prepared in 50% methanol-water, bubbled with nitrogen and stored in the refrigerator until use. Calibration curves at four different concentration levels were performed. Each level was tested by triplicate. Based on the absorption maxima, the wavelengths selected for the calibration curves were 280 nm for caffeine, 330 nm for chlorogenic and caffeic acids and 360 for rutin. Fresh samples of liquid extract were transferred to a syringe-driven filter (MillexGS, 0.22 um) and then were analyzed; for comparison purposes, equivalent amounts of lyophilized extract were solubilized in water. Results were expressed as mg of polyphenol compound/g of dried yerba mate.

Total polyphenol content was determined by the FolinCiocalteau method (TPC_FC). This test is based on the oxidation of phenolic groups with phosphormolybdic and phosphotungstic acids. A green-blue complex with absorption between 725 and 750 nm is obtained after a given reaction time. Two milliliters of Na₂CO₃ (2% w/v) (Anedra, Argentina) were mixed with 200 ml of the yerba mate liquid extract, left 2 for min in darkness and finally 200 ml of Folin-Ciocalteau reagent (Anedra, Argentina, 1:1) were added. The absorbance of these samples was measured at 725 nm in a spectrophotometer (Beckman DU 650, USA) after 30 min time reaction. TPC_FC was also determined on the reconstituted samples of lyophilized yerba mate extracts, to analyze the effect of the drying process. Gallic acid (Sigma-Aldrich, US) and chlorogenic acid (Fluka, US) were used as standards. Results were expressed as mg standard equivalent/g yerba d.b. (dried basis).

The photochemiluminescence (PCL) inhibition capacity of samples and pure compounds (chlorogenic and caffeic acids, rutin and caffeine) was determined as described by Popov and Lewin [9]. An automated PCL inhibition capacity analyzer system (Photochem, Analytik Jena AG, Jena, Germany) was used. Liquid extract, lyophilized extract and pure compounds, either diluted or dissolved in mili-Q water, were analyzed using the kit for integral Antioxidative Capacity of Water-soluble substances (ACW). Fresh samples of liquid extract were transferred to a syringe-driven filter (Millex-GS, 0.22 um) and then were analyzed; for comparison purposes equivalent amounts of lyophilized extract were solubilized in water. Results were expressed as mg of ascorbic acid equivalents per g of dried yerba mate, in the case of extracts, or per g of compound, in the case of standards.

Antiradical activity was determined by using DPPH· (Sigma-Aldrich, US) as a free radical. The method was adapted from Brand-Williams, Cuvelier and Berset [10] and is based on the reaction of specific compounds or vegetal extracts with the radical in an ethanolic solution. DPPH· reduction is followed by measuring the decrease of absorbance at 517 nm while the reaction occurs. Different concentrations of yerba mate extract were tested: 0.47 - 15.0 and 0.65 - 21.0 mg yerba mate/ml for liquid and lyophilized samples, respectively. A volume of 100 µl of each sample was added to 3.9 ml of DPPH· ethanol solution (25 mg DPPH·/ml ethanol). The decrease in absorbance was determined every 0.5 min for 10 min, and then every 15 min until the reaction reached a plateau. Gallic and chlorogenic acids were used as standard compounds for antiradical activity determination, as well.

EC₅₀, the amount of yerba mate extract needed to decrease the initial DPPH· concentration to 50%, and T_EC50, the time necessary to reduce the radical to this concentration, were determined according to Sánchez-Moreno, Larrauri and Saura-Calixto [11].

Analysis of antioxidant activity with DPPH· radical involves a higher economic cost and assay time. Thus, to find an easy way to characterize the antiradical activity of extracts during production, a possible linear relationship between Folin-Ciocalteau and DPPH· methods was studied. Linear models were fit to experimental data using a statistic program SYSTAT INC (Evanston, US).

Water was selected as the solvent for obtaining the yerba mate extracts, mainly to facilitate their use as food additive, although other solvents may be more efficient [12]. Besides being a green solvent, no further evaporation step is necessary. Process yield, including the lyophilization step, was 32.7% on dry basis of yerba mate. Yield values between 25% and 33% were also reported [8]. Depending on the industrial processing step (green leaves, zapecado, drying, and forced aging) of the vegetal material, yield values between 31% and 36% were obtained [13]. Powder yerba mate extracts showed a water solubility of 96% and a moisture content of 3.8%. Similarly, Sinija, Mishra and Bal [14] reported a moisture content of 3% - 5% for freeze-dried tea extracts in an instant beverage product.