【文章內容簡介】 in a single experiment, where it was necessaryto obtain the concentrated product of the process. The nano?ltra tion module is equipped with a NF90 membrane (Osmonics, Minnetonka, USA) which is posed of polyamide and polysulphone, with m2 of ?ltration area and 98% rejection of MgSO4 in a test performed by manufacturing with a spiral module at 20 176。C and bar. Approximately L of each solution permeated through the membrane over 30 min, this being the time necessary to plete the concentration in an open system, which means that the permeate was removed from the process. In the trials the permeate was removed and the retentate recirculated until a concentration factor of around four. The concentration factor is calculated according to Eq. (1):Fig. 1. Schematic diagram of the nano?ltration unitwhere Vf is the total volume used in the feed, Vc is the volume collected in the concentrate fraction and Fc is the concentration factor. Other experiments were carried out at different temperatures (20–45 176。C) and pressures (– bar), in order to evaluate the in?uence of these parameters on the permeate ?ux and the concentrated product quality. In these experiments, both the permeate and retentate were maintained under recirculation in closed systems. The permeate ?ux was calculated according to the following equation: J=Vp/t*Ap (2)where Vp is the permeate volume collected during the time intervalt and Ap is the membrane surface area of permeation. The quality of the ?ltration process was measured according to the quantity of ?avonoids and phenolic pounds present in permeate, evaluated as described in Sections –, and the ef?ciency was measured according to the ?ux permeate rate and retention index. This index measures the relation between the amounts of the pound of interest in permeate and in the concentrated solution, which demonstrates the ability of the membrane to retain this pound under the experimental conditions. The index is calculated according to Eq. (3), in which R is the retention index, Cp is the concentration of the pound of interest in the permeate, and Cr is the concentration of the same pound in the retentate:R=1Cp/Cr (3)It is important to know the rate of fouling that occurs in the membrane process, and one way of measuring this is to pare the permeate ?ux of the solution under study with the permeate ?ux when water is used as feed solution, under different pressures. Usually a variation in system pressure will cause a change directly proportional to the permeate ?ux. The fouling in?uence was measured by parison of the permeate ?ux of the aqueous propolis extract with the ?ux of distilled water only, increasing the pressurefrom to bar.3. Results and discussion The membrane process was carried out with the aqueous and ethanolic solutions in a closed system, in which the retentate and permeate streams being conducted back and mixed in a feed tank isolated from the environment, to evaluate the variation in permeate ?ux with time. The temperature was maintained at 20 176。C and the pressure at bar. The results are shown in Fig. 2.After stabilization of the process, the permeate ?ux began to decrease, after around 15 min of processing. The rate of decrease was higher for the alcoholic extract than for the aqueous extract, evidencing a greater rate of fouling with the alcohol solution. After 20 min of processing, the permeate ?ux tended to stabilize, that is, concentration polarization already occurred and fouling did not increase with time. The permeate ?ux in the stable region was about L/h m2 and L/h m2 for alcoholic and aqueous solutions, respectively. The difference between the permeate ?ux of these solutions can be explained by their different positions: the alcohol extract contains more pounds of low molecular weight, thus its concentration is more dif?cult to achieve, and this reduces the ?ux rate. Some of these pounds form a kind of wax which can cause more fouling in membrane.Tsui and Cheryan (2007) used nano?ltration to purify alcoholic corn extracts in the production of xanthophylls, and obtained a permeate ?ux of around L/h m2 when working at 27 bar and 50 176。C. Hossain (2003) studied the membrane concentration of anthocyanins from blackcurrant pomace extracts using ultra?ltration, obtaining a maximum permeate ?ux of L/h m2 at bar and 18 176。C. Using nano?ltration a permeate ?ux of 20 L/h m2 was obtained at 20 bar and 50 176。C in the concentration of red wine (Banvolgyi et al., 2006). The red wine concentration process is important since it can be considered a similar process to the concentration of alcoholic propolis, considering that they have similar pounds in solution and use alcohol as the solvent. The similarities between the processes allow a parison between results. Low pressures (around 6 bar) were used in the propolis concentration process, when pared to other processes cited in the literature, but even so the values obtained for the permeate ?ux were similar to those obtained in the other processes. Therefore this process can be assumed to be viable, mainly because of the reduced energy requirements necessary to generate the lower pressure adopted was not characteristic of nano?ltration processes, but was suf?cient to carry out this concentration procedure.Fig. 3 shows the difference between the curve of the permeate ?ux for the aqueous propolis extract and the curve of the permeate ?ux for distilled water to measure the degree of fouling in the process with the aqueous propolis difference between the permeate ?uxes of water and the propolis solution shows the amount of fouling in the process, under the same conditions of temperature and pressure. This parameter increased, reaching 32% at bar. The procedure also provided information on how the ?ux was affected by a pressure variation, showing that the ?ux changed linearly with the pressure in the region studied. In this pressure range