Access features, news and views from the latest issue and from our chemistry archives.

June-August 2022

Ultrafiltration for wine protein stabilisation

In my column in the September–November 2020 issue (p. 40), I outlined several approaches to replace the use of bentonite to achieve protein stabilisation in white wine. In essence, the alternatives included the addition of a polysaccharide (carrageenan) or a polyphenol (grape seed powder), flash pasteurisation with aspergillopepsin enzymes and magnetic nanoparticles.

The so-called heat test is the industry standard for assessing protein stability ( This test involves heating a small volume of wine at 80°C for two hours and then cooling for three hours before assessing haze development by eye or by turbidity measurement: <2 NTU is required for a pass result. Any replacement for bentonite must satisfy the heat test. Chemical and sensory analyses are usually employed to check that there has been minimal impact on overall wine composition.

The bentonite replacements that I described earlier need some form of separation after the treatment to clarify the wine. Ultrafiltration (UF) could simplify protein stabilisation in a single step provided the UF process can be managed with minimal impact on the chemical and sensory properties of the wine. UF functions by size exclusion and can be regarded as sitting between microfiltration and nanofiltration. The size range for particle cut-off in UF is 0.01–0.1 μm, although this depends on membrane type to some extent. Blockage of pores during filtration also influences the success of size separation, so operating the filter in the cross-flow method is normal practice.

A study in 1987 explored the feasibility of achieving protein stability in white wine. While some success was achieved, there appeared to be ‘protein leakage’ into the permeate, suggesting an issue with membrane selectivity. The need to characterise the protein profiles in both the permeate and the retentate was seen to be essential to ensure that the membrane has the necessary selectivity to remove the proteins responsible for heat instability.

Significant advances in membrane technology have occurred over the last decade, especially with better definition of pore size and selectivity. Polyethersulfone membranes are now commonly used because they have good mechanical strength and stability. From a commercial perspective, it is also preferable that the extent of permeation is at least 90% or higher. The interaction of wine components, other than proteins, with the membrane leading to blockage needs to be considered when designing UF treatment.

These questions formed the basis of Yihe (Eva) Sui’s PhD project at the University of Adelaide with support from VAF Memstar and funding from the Australian Research Council Training Centre for Innovative Wine Production. Proof-of-concept experiments were performed with a benchtop UF unit and filtering five litres of white wine through a 10 kDa or a 20 kDa nominal molecular mass cut-off (NMMCO) filter. A permeate that satisfied the protein stability test was obtained and the wine proteins were concentrated in the retentate (Sui et al. Aust. J. Grape Wine Res. 2021, vol. 27(2), pp. 234–45).

One of the limitations of bentonite treatment is the loss of product due to difficulty in separating the treated wine from the bentonite residue (see September–November 2020 issue, p. 40). With UF, the retentate needs to be protein-stabilised before it can be back blended with the permeate. Scaling up the UF experiments yielded sufficient retentate for further examination. Significant protein removal could be achieved by heat treatment with added pectolytic enzyme. When the treated retentate was blended with the permeate, the resulting wine required only a small amount of bentonite to be heat stable (Sui et al. Aust. J. Grape Wine Res. 2021, vol. 27(2), pp. 234–45).

This initial study clearly demonstrated the ability of UF for white wine protein fractionation. The next phase of the Adelaide group’s work was to apply the proposed methodology on a commercial scale. Six wines, with volumes ranging between 5400 and 71 000 litres, were filtered using 5 or 10 kDa NMMCO filters (Sui et al. Sep. Purif. Technol. 2022, vol. 284, 120227). Permeation rates between 90% and 97% were achieved.

The heat test showed that all six permeates were heat-stable, even though not all protein had actually been removed from the permeate. Not all classes of protein found in wine can lead to heat haze or instability, implying that UF is successful in the removal of the haze-forming proteins.

The UF procedure will also retain other wine macromolecules. Thus, the permeate was found to be markedly lower in wine phenolics and this was linked to a lower brown colour. The added bonus from a winemaking perspective is that the reduced phenolic concentration will moderate the astringency and/or bitterness of the wine, thereby removing the need for a subsequent fining step with a protein (see September–November 2021 issue, p. 41). This, together with the elimination of the requirement for bentonite fining, are two major winemaking benefits of UF.

There were measurable differences in some basic wine parameters, such as titratable acidity, but the changes did not affect the sensory properties of the wine. In a study yet to be published, the Adelaide group found that the wine sensory profiles and quality scores of UF-treated protein-stable wines were comparable with traditionally fined wine. It is reasonable to expect more developments in the use of UF for wine adjustment, especially as advances in membrane technology may lead to the targeted removal of specific wine components.

Geoffrey R. Scollary FRACI CChem ( has been associated with the wine industry in production, teaching and research for the last 40 years. He now continues his wine research and writing at the University of Melbourne and Charles Sturt University.

Book and software reviews

To offer your services as a book or software reviewer for Chemistry in Australia, please contact Damien Blackwell at