Most compounds affecting aroma, flavor, color and mouthfeel are small enough to to pass a sterile filtration membrane element (above) unchallenged.
- Filtration opponents claim the practice strips wine of important characteristics; defenders insist filtration is essential to stability.
- All of the standard categories of compounds that contribute to the aroma, flavor, color and texture of wine--volatiles, acids, phenolics, polymerized tannins--are orders of magnitude smaller than the pores in a .45 micron sterile filtration membrane.
- Research literature is unfortunately silent on rigorous sensory evaluation of wine before and after filtration, leaving the subject open to personal, anecdotal accounts.
Recently I received an e-mail from Hamlet, winemaker for Prince of Denmark Vineyards. Hamlet wrote to address the question of filtration:To filter, or not to filter--that is the question: Whether 'tis nobler in the mind (of the winemaker) to sufferThe slings and arrows of outrageous particulatesOr to take arms against a sea of microbesAnd by removal end them.
OK, I made up the e-mail, but the Dane does raise an important concern. Filtration, a completely routine practice throughout the world's wine industry, remains the subject of intense debate. Indeed, that's the very first point the Oxford Companion to Wine
makes in its filtration
entry--"filtration of fine wines is a controversial issue"--before even explaining what the practice involves.
Critics of filtration claim that it strips wine of its character, though exactly what has been lost is often hard to pin down. Practitioners and proponents argue that filtration does nothing of the sort, but that it certainly does remove particulates, debris and other sludge, increase clarity and reduce the potential for microbial spoilage in the bottle.
After decades of experience with wine filtration, surely we know which side of this argument is right, right? Fans of filtration can cite a lot of basic chemistry to bolster their points, and opponents can charge a premium for wines labeled "unfiltered." But once again, the volume of hard research fails to match the volume level of the dispute.
Particulate Science 101
Officials at VA Filtration, which has mobile services in California (above) believe a lot of "stripping" is due to the excessive use of the wine as a cleaner during the microfiltration process.
Filtration comes, of course, in many flavors--pads, membranes, crossflow and variations on those themes. It is quite possible that these different methods have different effects, and it is possible as well that the timing of filtration--shortly after fermentation versus just before bottling--could make a difference. The debate, however, is rarely joined in such precise terms, and in any case, the starting point of understanding has to be the question of the relative size of key wine compounds and filter openings.
Asserting that filtration can strip wine suggests that the process takes something desirable out, some substance that enhances wine quality. This in turn suggests that these lost goodies are larger than the pores in a tight pad filtration--with only nominal, average sizing--or a .45 micron sterile--absolute sizing--membrane filtration. Yet virtually every compound we normally associate with wine aroma, flavor, color and mouthfeel is small enough to sail right through any of these barriers unchallenged.
The standard unit for measuring porosity in pad and membrane filters is the micron, a millionth of a meter, equivalent to roughly 0.00004 inch. A pencil dot is about 40 microns in diameter; a yeast cell is about 3. Sterile filtration--tight enough to catch the smallest bacteria normally found in wine--is defined as a porosity of .45 micron.
That's pretty small. But the largest molecule with a flavor impact that Roger Boulton of the University of California, Davis, could think of is thaumatin, a synthetic sugar substitute, which comes in at about one twentieth of a micron--.05, well under .45. If you wanted to doctor your wine with thaumatin, filtration wouldn't take it out.
When chemists get down to the molecular level, they stop talking about length and focus on molecular weight (MW), and the units morph from microns to Daltons. This is the realm where the usual suspects in wine chemistry reside. According to Boulton, molecules of sugars and acids come in at about 150-180 MW units; aromatic terpenes at about 300; most phenolics around 400. Tannins can range from 1,000 to 2,000, maybe 5,000; proteins in white wines can get up to 40,000, and polysaccharides to 100,000. But that .45-micron opening is the equivalent of about 600,000 MW units--an almost gargantuan opening for the comparatively teensy wine flavor molecules to breeze through.
At the same time, a certain amount of the good stuff does stick to the filter media--adsorption, in chemical parlance, as opposed to absorption. This is far more true of plate and frame pad filters than of membranes, and it's why filter pads turn red when they're put to work.
But that adsorption only continues to a point of saturation, and then nothing more sticks. That point is reached early in the game for almost any commercial-size filtration run. Ken Fugelsang at California State University, Fresno, for years a lecturer on filtratio n around the country, says that unless you're filtering volumes down around the single barrel level, this adsorption loss isn't likely to affect your wine--though he does know winemakers who make a point of blending the first part of the run back in with the rest in tank before bottling, just in case. Disappearing colloids?
Most winemakers I have heard complain about stripping don't or can't specify precisely what went wrong--it's just that the filtered wine is different and, they say, not as good. Vinovation's Clark Smith, who has made a good living removing alcohol from wine, is nonetheless a vocal filtration nay-sayer, claiming on his Grapecrafter blog and elsewhere that what gets taken out is "soulfulness"--which seems to refer to some combination of structure, texture and aromatic complexity. In this view, filtration may be fine for white wines, where fresh fruit and microbial stability are essential, but not for serious reds.
Soulfulness, of course, is hard to measure in a laboratory setting, but Smith and others go on to identify the physical material that undergird it: colloids, large, macro-molecular clumps of protein and polysaccharides. Colloids can play a role in mouthfeel, and may also have tannins and color compounds (anthocyanins) piggybacked onto them. If filtration knocks out colloids in large numbers, then it might be taking out essential goodies at the same time, which is exactly the argument made in the colloid entry in the Oxford Companion:
"Colloids are major contributors to a wine's viscosity, the extent to which the wine resists movement. Colloids can be removed by fining and by filtration. The fewer colloids are removed, the more body a wine will seem to have, the more astringent it may taste in youth, the slower it will mature, but the more complex it should taste when at its peak."
So does sterile filtration really yank out a lot of tasty colloids? Alas, we don't really know. Roger Boulton has his doubts. The best data he knows of, admittedly 20 years old, suggests that .45-micron filtration does not remove any significant amount of colloidal material, and he knows of no alternate data showing that it does.
Ken Fugelsang is similarly unaware of any conclusive evidence that filtration traps essential textural components. Both agree that there might conceivably be a colloidal effect; it just hasn't been shown.
"Any time something is not known," Boulton says, "it's possible."
Smith says that the significance of these macromolecules gets much more attention from European researchers than from their U.S. counterparts, who are much more focused on just counting things. His take on filtration is that even if colloids and their co-compounds get through the filter pores, they may get badly beaten up in the process, disrupting the structure and encouraging unpleasant, vegetative aromas--even if all the measurable phenolics are still there.
He uses an architectural analogy: "The number of bricks that go into a house has nothing to do with how livable the house is. So, how the colloids house the aromas and flavors may depend more on the architecture than on the number of bricks."
The missing sensory studies
Mirassou Winery has been promoting its unfiltered varietals for more than 30 years.
The argument against filtration is ultimately a sensory one--foes say the wine smelled and tasted better before filtration than it did afterward. (In some quarters, resorting to filtration is also portrayed as some kind of crime against nature, but that moral realm is beyond the scope of this publication.) So before we spend too much time and research money worrying over how many colloids can dance on the head of a .45-micron pin, shouldn't we consult the relevant sensory studies? Shouldn't we make sure there's a real, live, identifiable sensory effect to be explained?
Alas, according to Hildegarde Heymann at UC Davis, who has been looking into just this question, there's really not much literature to consult. Reams of anecdotal testimonials, yes--but virtually no rigorous sensory research. Heymann is currently shopping around for funding for a project by one of her graduate students to do a modest controlled study: evaluating wine before and after filtration, including immediately after, one month, six months, etc.; looking at oxygen levels, and also comparing wine that is simply pumped, not filtered. Why in the world hasn't someone already done this?
Heymann has some theories about why filtration sensory research hasn't gotten attention. It requires relatively large volumes of wine--remember that adsorption problem--and thus it becomes expensive. Credible research would have to be independent--that is, not funded by a company with a stake in the outcome. Since advocates on both sides of the issue have staked their reputations on firmly held beliefs, there's little motivation to fund research that might prove otherwise. Transport and oxygen
This research design in search of support brings up another variable in the filtration discussion. When people filter wine, they also do several other things at the same time: move the wine, pump it through pipes and tubes, run the risk of oxygen uptake, mix the wine in a way that may differ from what got sampled in the original tank, and possibly toss in a dose of sulfur dioxide somewhere along the way. Maybe one or more of these things is what "strips" the wine, not the filtration itself.
Tank samples taken without thorough mixing beforehand are notoriously undependable for chemical analysis as well as sensory evaluation. The wine four feet higher up may be something else altogether, and so comparing a final mixed wine with an initial unmixed wine can get us directly to apples and oranges.
More important is the chance of oxygen pickup during the filtration process. With proper care, equipment and attention, oxygen can be held at bay, and conscientious bottlers make a point of monitoring dissolved oxygen levels. But before holding the pad or the membrane responsible, we'd need to know for sure that oxygen wasn't in play.
The widely reported and vaguely defined phenomenon of "bottle shock" may also have some relevance here. There are testimonials aplenty that terrific wines get really dumb right after bottling--whether or not they have been filtered--and then come back to their earlier form after a period of rest. If a filtered wine can go away and then come back, that would imply the filtration didn't strip any goodies--they were just hiding. Final advice to Hamlet
Pending a lot of research that may or may not happen, winemakers are left to exercise their own judgment. Ken Fugelsang's advice is, when in doubt, sterile filter if there's any chance that what's going into the bottle is not chemically or microbially stable. On the other hand, if you're sure you have a stable wine, and you and your customers like the sensory properties the way they are--colloids, particulates and all--then skip the filtration.
The companies that make and sell filter rigs, pads and membranes offer the same advice. The filter folks did research years ago to demonstrate that filtration didn't hurt wine, and more recently demonstrated that membrane filters were just as benign as pad filters. Nicole Madrid, marketing manager for Pall Filters, puts it this way:
"Honestly, it's something the winemaker has to look at. Sometimes we find we filter a wine that may have not needed it. Maybe we're taking out colloidals. On the other hand, sometimes we take out a contaminant and actually open up the flavor. A white that hasn't gone through malolactic and has residual sugar, that's taking a big risk. But for a stable red, well analyzed, a wine you can be confident about, it's the winemaker's choice."