The reduction and inactivation of viral pathogens in natural waters is a major goal to achieve, due to the close relationship between these organisms and disease outbreaks. Available treatments based on bacteriological criteria are not always effective since viruses are more resistant and difficult to remove.
Membrane processes, like ultrafiltration, are increasingly used in potabilization to remove viruses and are a good barrier at nanometer level. The removal of virus particles is controlled by different mechanisms, such as electrostatic repulsion, attachment and size exclusion.
Ultrafiltration membranes are not expected to achieve significant retention of viral particles by size exclusion, since their pore size is similar or larger than the virus diameter; electrostatic repulsion and aggregation are therefore the predominant mechanisms during virus filtration.
Particles exhibiting the same charge as the membrane surface are generally less likely to be allowed through the membrane, as they are repelled on approach back to the bulk feed solution. Particle size increases due to aggregation, which occurs when electrostatic repulsive forces can be effectively overcome by van der Waals attraction forces, allowing attachment following inter-particle collision events. Electrostatic repulsion and attachment are strongly dependent on surface chemistry, pH, ionic strength, and chemical nature of the dissolved solids.
Non-pathogen viruses have been extensively used as models in water research. Pseudomonas aeruginosa bacteriophage PP7 is a good surrogate for poliovirus in water treatment processes, since both are icosahedral and have similar diameter (25 - 30 nm).
Previous work has shown that virus retention by polymeric ultrafiltration membranes was highly dependent on the water matrix. In particular, removal of PP7 was increased with high concentration of Ca2+, low concentration of Mg2+, simultaneous high or low concentrations of NO3- and HCO3-.
The filtrations were performed at cross flow in a custom made membrane filtration unit, connected to a feed tank through a peristaltic pump, a permeate tank on a scales, and control instruments (two pressure gauges and a flowmeter).
The objective of this work was to elucidate the mechanisms of virus filtration in relation to membrane surface - viral particle interaction and aggregation processes in order to be able to explain the experimental observations. Membrane surface and virus charges were measured, as well as particle size, for a variety of water matrixes. The inter-virus and virus-membrane interaction energy was modeled by Derjaguin, Landau, Verwey and Overbeek (DLVO) theory of colloidal stability.
Materials and methods
A flat sheet ultrafiltration membrane was employed in this work, made of a polyethersulfone dense skin and an amorphous support of cellulose. The average size of pore is 0.067 mm and MWCO is 50 kDa. Pseudomonas aeruginosa phage PP7 (ATCC 15692-B2) was used as model virus. Reagent grade NaCl, NaNO3, NaHCO3, CaCl2, and MgCl2 and Type I water was used in all experiments.
Preparation of the virus stock started with incubation of host bacteria in nutrient broth, and then PP7 was inoculated and incubated. Afterwards, the virus suspension was centrifuged and the supernatant filtered through a PVDF membrane. Purification of this stock consisted of dialysis through a 100 kDa MWCO membrane (SpectraPor Biotech CE, Spectrum Laboratories, USA) twice: first, against Type I water, and secondly, against the appropriate solution (NaCl, NaNO3, NaHCO3, CaCl2, MgCl2 of 1 mM, 10 mM, 100 mM ionic strength). The final suspension was again filtered and kept at 4ºC overnight before measuring the zeta potential and hydrodynamic diameter.
The zeta potential of the bacteriophage and the membrane, as well as the hydrodynamic diameter of the bacteriophage were measured in a Zetasizer Nano ZS (Malvern, UK) at 25ºC, using disposable cuvettes and the appropriate surface zeta potential accessory. Size of the bacteriophage was measured by Dynamic Light Scattering (DLS) and zeta potential by Laser Doppler Micro-electrophoresis. The membrane zeta potential was derived from electro-osmosis measurements.
For the DLVO modeling, PP7 was considered a sphere due to its icosahedral shape and the membrane was regarded as an infinite plate compared to the virus. Attractive van der Waals forces and electrical double layer repulsion were considered as the fundamental present interactions. Classical equations proposed by Hamaker (The London - van der Waals attraction between spherical particles, Phys. 4 (1937) 1058-1072) or collected in the existing literature were used. Moreover, non-DLVO forces such as Born repulsion, Lewis acid–base interactions and hydration repulsion were considered and their importance determined.
Results and discussion
Characterization: The average hydrodynamic diameter of the bacteriophage ranged between 44 and 84 nm, which differs from the diameter of the isolate virus of 25 - 27 nm.
PP7 had negative zeta potential for all the background solutions over the studied range of pH (5 to 8). For each salt, with increasing ionic strength, zeta potential of the bacteriophage increases (turning less negative).
For the flat membrane, the zeta potential showed always negative values, with only one exception: when the solution was NaNO3 10 mM for the whole range of pH considered.
The membrane is less negative than the virus for the case of NaCl 1 and 10 mM, NaNO3 1; 10 and 100 mM, NaHCO3 1 mM, CaCl2 1; 10 and 100 mM ionic strength, MgCl2 1 and 100 mM ionic strength. The membrane is more negative than the virus for the case of NaHCO3 10 mM ionic strength. Both the membrane and the virus have approximately the same charge for NaCl 100 mM, NaHCO3 100 mM and MgCl2 10 mM ionic strength.
DLVO modeling: The interaction potential energy for a virus like particle of PP7 and the flat sheet of polyethersulfone membrane in different background solutions was modeled. In the case of solutions formed by NaCl, NaNO3, NaHCO3, CaCl2 and MgCl2 1 mM ionic strength, NaCl, NaHCO3 10 mM ionic strength an energy barrier can be predicted and a net repulsion potential between the primary aggregates and the membrane is expected. In the case of solutions formed by CaCl2 and MgCl2 1 mM and 10 mM ionic strength, NaNO3, NaCl, NaHCO3 100 mM ionic strength a secondary minimum is expected. And in the case of NaNO3 10 mM only a primary minimum is predicted, principally because of the positive charge of the membrane, and attachment to the membrane is more likely to happen.
When ionic strength was 10 mM, DLVO analysis was performed as function of pH; and it could be concluded that pH is not an important parameter when regarding attachment of the virus to the flat surface; which can be thought as the consequence of the almost constant zeta potentials for both the bacteriophage and the membrane. In all cases the total potential interaction are very close, and in particular it can be seen that the curve at pH=8 for NaCl, NaNO3, CaCl2 and MgCl2 separates from the others, indicating extra repulsion for NaCl, NaNO3 and MgCl2 and more attraction for CaCl2. For NaHCO3 this situation is seen at pH=6 and less repulsion is expected.
Filtration: DLVO theory predicts that attachment of the bacteriophage to the membrane will increase when increasing the concentration of Ca2+ and Mg2+ respectively. However, net repulsion is only predicted at very low ionic strengths, when no secondary minimum would arise. This fact is in concordance with laboratory results since though low, removal was achieved.
DLVO theory predicts that attachment of the bacteriophage to the membrane will increase when increasing HCO3-; expecting net attraction at high ionic strength.
DLVO theory predicts that attachment of the bacteriophage to the membrane will increase when increasing NO3- from 1 mM to 10 mM and then decrease when increasing NO3- from 10 mM to 100 mM; expecting net attraction at 10 and 100 mM.
We can derive that when increasing both concentrations of HCO3- and NO3-, the effect of HCO3- outweighs the effect of NO3-. However, when one concentration increases and the other one decreases, the same asseveration no longer stands. These divergences between DVLO and experimental observations must be attributed to non-DLVO interactions.
Another limitation to the application of DLVO theory arises from the fact that viruses can be thought as soft particles (covered with a ion-penetrable surface layer of polyelectrolyte), as opposed to hard particles where the surface is clearly defined and no ion can penetrate it.
DLVO predicts the interactions studied between bacteriophage PP7 and polyethersulfone membrane, though further refinement should be made. It would be convenient to take into account that viruses are biological particles and therefore not perfect, rigid spheres with homogeneous surface; that commercial polyethersulfone membranes are not smooth homogenous surfaces; and when particles are approaching the supposition of constant charge no longer stands.
For the viral model, a partial removal is achieved. Ultrafiltration procedures to disinfect waters prove to be a technology to produce safe water, though its efficiency is affected by the chemicals present in the aqueous matrix.