Influence de l'adsorption d'alginates sur les propriétés de membranes organiques d'ultra et de microfiltration

Les applications potentielles des cultures de microalgues et cyanobactéries en dépollution d'effluents dans des photobioréacteurs à membrane souffrent de performances limitées par un colmatage de l'élément filtrant dû en grande partie aux exopolysaccharides sécrétés par ces micro-organismes. Cette étude du laboratoire quantifie les effets de l'adsorption de ces polysaccharides sur des membranes organiques d'ultra et microfiltration tangentielle de matériaux et charges de surface différents. L'alginate de sodium est utilisé comme adsorbat modèle. Les membranes propres sont d'abord testées par une mesure de flux à l'eau pure, puis mises en contact avec une solution d'alginate durant un temps choisi. Le flux à l'eau pure des membranes après adsorption est ensuite à nouveau mesuré.La réduction relative du rayon de pore (ZEMAN, 1983) met en évidence l'effet de la mouillabilité et des charges superficielles. L'étude comparée de membranes d'ultra et microfiltration montre que cette réduction relative du rayon de pore augmente avec le seuil de coupure ou le diamètre de pore. L'effet de la concentration révèle aussi que la résistance hydraulique d'adsorption (MATTHIASSON, 1983) à l'équilibre évolue selon l'isotherme de LANGMUIR. Le modèle cinétique traduisant l'évolution de la résistance d'une membrane d'ultrafiltration proposé par BAKLOUTI et al.(1984), amélioré par AIMAR et al. (1988) puis discuté par RUIZ-BEVIÁ et al. (1997), est complété par un nouvel exposant agissant sur le facteur temps.La comparaison des résistances à l'écoulement de membranes de microfiltration avec celle d'une membrane d'ultrafiltration hydrophile neutre permet de dégager des critères de choix pour l'optimisation du fonctionnement d'un photobioréacteur à membrane utilisable en dépollution d'effluents.Potential applications of microalgae and cyanobacteria for treatment of wastewater effluents using membrane-photobioreactors suffer from limited performance due to fouling effects, mainly attributable to exocellular polysaccharides secreted by these micro-organisms. A membrane photobioreactor is defined as a process associating the culture of photosynthetic micro-organisms with a continuous separation by membrane filtration of the biomass and the water treated. The goal of the present laboratory-scale study was to quantify polysaccharide adsorption effects on organic membranes (ultra and microfiltration) characterised by different materials and surface charges. Sodium alginate was used as the "model adsorbate".Seven plane organic membranes were tested. The influence of membrane cut-off (or of pore diameters) as well as that of the material polyethersuphone (PES), polyacrylonitrile (PAN), polyvinilidene fluoride (PVDF) and of its properties (hydrophobicity, surface charges, …) were assessed. The study consisted of two parts :1. the first part was concerned with the kinetics of alginate adsorption and the influence contact time and solute concentrations on the reduction of pore diameter (ZEMAN, 1983) or on the increase of hydraulic resistance (MATTHIASSON, 1983);2. the second part dealt with adsorption equilibrium (formulations of LANGMUIR and FREUNDLICH).The study constituted the first step of a research program aimed at developing membrane photobioreactors for the treatment of specific industrial effluents. The fluid used to test the membranes was quality II pure water (ISO 3696 norm). Tangential velocities were set to 2.5 m.s-1, corresponding to a Reynolds number of 2500. To represent exopolysaccharides, we used alginic acid at concentrations of 1, 10 and 50 g, neutralised with sodium hydroxide at pH 9. New (or clean) membranes were first characterised through pure water flux measurements. J0, the flux of pure water for a new membrane, was obtained (flowrate / unit of surface area), and then the membrane was kept in contact, for a definite duration, with the alginate solution. After adsorption and rinsing, the pure water flux was measured again. Ja, the pure water flux, was measured through the membrane after adsorption.Adsorption model at equilibrium:The effect of adsorption is quantified under the form of the relative pore size reduction as described by ZEMAN (1983) and included in the relation : ∆r / r=1 - (Ja / Jo)1/4. A variation of this quantification is that of the MATTHIASSON model (1983) applied to the pure water flux, based on DARCY's law expressing the relative value of the hydraulic resistance of the adsorbed layer Ra in relation to the intrinsic resistance of the membrane Rm : Ra / Rm=(Jo / Ja) - 1.To express adsorption phenomena at the solid/liquid interface of membranes, we used LANGMUIR's law together with MATTHIASSON's experimental observation (1983): the relative resistance Ra / Rm due to adsorbed compounds is proportional to the mass "x" of solute adsorbed per unit of membrane surface area, x=Kx.Ra. If one assumes that the mass m of a homogeneous plane membrane per unit of membrane surface area is proportional to its adsorbing surface area Ω per unit of membrane surface area (m=Km.Ω), and if one combines the flux equations expressed by DARCY's and POISEUILLE's laws, then the result is m=K'm.Rm in a homogeneous membrane. Substituting x and m in LANGMUIR's law results in the equilibrium model Rae / Rm=(Jo / Ja) - 1=a.c / (1 + bc) in which c=concentration of adsorbing solute; a and b are coefficients; and Rae is the resistance due to compounds adsorbed at equilibrium. Kinetic model: To show the evolution of membrane resistance with time, we suggest the introduction of an empirical exponent j over the time parameter in the AIMAR et al. model (1988).Results: The effect of changing the alginate concentration reveals that the hydraulic resistance of adsorption, at equilibrium, (MATTHIASSON, 1983) evolves according to LANGMUIR's isotherm. The relative decrease of pore radius ∆r / r in the presence of l g.l-1 of sodium alginate shows that a quasi-plateau is obtained after two hours using the most hydrophobic membrane. The curves ∆r / r=f (t) for five membranes made of different materials, monitored during the transition phase before the plateau with common 1 g.l-1 concentrations, reveal similar adsorption behaviour, characterised by the limiting common value ∆r / r=0.06 ± 0.005. However, the uncharged hydrophilic membrane PAN 3038 stands out owing to a much lower ∆r / r value of 0.09. This peculiar behaviour can also be observed in the influence of the alginate concentration: hydrophobic and charged hydrophilic membranes display a saturation effect with ∆r / r little affected by the increase of alginate concentration, whereas the uncharged hydrophilic membrane PAN 3038 displays a ∆r / r value three to six times lower with great sensitivity to concentration effects at concentrations below 10 g.l-1. The model Rae / Rm=(Jo / Ja) - 1=a.c / (1 + bc) is in agreement with the experimental results obtained with hydrophobic and hydrophilic membranes. The proposed kinetic model shows that time dependence of R (t) does not seem to be linked to the nature of membranes. However, compared with concentration, R (c) is very sensitive to the nature of membranes. A comparative study of ultra and microfiltration membranes shows that the reduction in ∆r / r values increases with molecular weight cut-off (or pore diameter).Criteria for the choice of membranes: A comparative study of three polyacrylonitrile membranes reveals that membrane 3038 PAN (neutral) displays a very interesting, peculiar behaviour: its adsorption, expressed by ∆r / r or Rae/(Rae+Rm) is four to six times weaker than that of the other two. The surface charge of membranes seems to influence the intensity of adsorption in a significant way. Wetability also has a strong influence on adsorption. The sum of resistances Rae + Rm of ultrafiltration membrane 3038 PAN is only four times as great as those of hydrophobic microfiltration membranes. Experimentation already showed that, in the presence of microparticles, interactions between the layer of adsorbed alginate and microparticles will increase the likelihood of fouling of microfiltration membranes, decreasing their resistance down to the level of very little adsorbing ultrafiltration membrane IRIS 3038 (ROSSIGNOL et al., 1999).A culture system of marine microalgae in a membrane photobioreactor using ultrafiltration membrane IRIS 3038 PAN displayed a stable permeation flux during 6 weeks and easy regeneration, which meant adsorption was almost nil. The ability of some microalgae to assimilate ammonia nitrogen, nitrates and phosphates contained in waste water with excellent efficiencies (e.g., Phormidium bohneri: SYLVESTRE et al., 1996) allows one to consider using membrane photobioreactors in the treatment of home or industrial effluents. Other microalgae such as Chlorella salina (GARNHAM et al., 1992) are capable of fixing large amounts of heavy metals (Co, Mn, Zn, etc…); grown in membrane photobioreactors, they could depollute industrial effluents