Scientia Vitis: Decanting the Chemistry of Wine Flavor

Through experiments and the application of new technologies, scientists at UC Davis are working to determine the molecular makeup of a good glass of vino.

Scientia Vitis_0.jpg

Pasteur studying the diseases of wine in 1863

Pasteur studying the diseases of wine in 1863.

Institut Pasteur

The University of California, Davis (UC Davis) wine cellar is lined with rows of wooden barrels and old-fashioned wine bottles—some dating back to the end of Prohibition, when America’s wine industry had to start from scratch. Look inside the newest bottles, however, and you’ll see and taste the results of four decades’ worth of modern research on what makes a great wine. Continuing a century-old tradition, researchers at the Department of Viticulture and Enology at UC Davis are investigating the complex dance between science, art, and nature that creates flavor and aroma in wine.

The Tasting Booth

In Hildegarde Heymann’s lab tasting booths are illuminated by the deep glow of a dark red lamp. “We give people wine in dark glasses and change the room lighting to make the color of the samples harder to discern,” says the UC Davis professor of enology. “We want them to focus on what they taste and smell, not what they see.” A world-renowned expert on the molecular basis of aroma, Heymann says that wine flavor is a subjective, blurry experience that results from complex interactions between many different classes of compounds. Molecules intermingle and coalesce, just like the lingering notes of raspberry and blackberry that remain on the palate after sampling a characteristic young Shiraz.

‘The chemistry that sets a merlot apart from a pinot noir is nearly impossible to identify.’

Acids, sugars, and tannins are the most obvious contributors to wine flavor, but these three classes of molecules are accompanied by a remarkably varied cast of organic molecules—often aromatics—that, in combination, can produce an astonishing variety of flavors. Generally speaking, fruity flavors are attributed to interactions among esters, alcohols, and acids. Tannins, or phenol compounds, give wine an astringent mouthfeel, and sugars determine the sweetness of the wine. Yet to make things even more complicated, the interaction of these chemicals seems to depend on growing conditions and fermentation practices.

“The chemistry that sets a merlot apart from a pinot noir is nearly impossible to identify,” says Heymann. “There may be over 500 different flavor compounds unique to each variety.” Nevertheless, Heymann and her colleagues are attempting to connect specific combinations of molecules with familiar flavors. Since two different people can taste two very different things when sampling the same bottle of wine, Heymann passes cups of apple peel and soy sauce through tasting booth windows—definitions of flavor that everyone can identify. Instead of chemical reagents and noxious gases, the cabinets in Heymann’s lab contain bottles of soy sauce and parcels of chocolate—all used as a basis for comparison. This approach has allowed scientists to trace some of the more notable wine characteristics back to their chemical roots.

Molecules with a Single Flavor

One level below Heymann’s tasting room, Roger Boulton, a professor of viticulture, runs experiments on the sulfides produced during fermentation, surrounded by a laboratory full of spectrometers, chromatographs, and other traditional analytical equipment. “After 2,000 years of winemaking, only a few molecules have been correlated with a specific flavor,” says Boulton. One instance where a direct link has been established, he explains, involves methoxypyrazines, a family of molecules that makes wine taste like bell peppers.

Methoxypyrazines were initially found to play a role in wine flavor in 1975, says Boulton. They are now understood to be particularly prominent in cabernet grapes. While trace amounts of the molecule are considered acceptable, too much can overwhelm wine, producing a strong vegetable flavor. Heymann and her colleagues have since shown that the molecule breaks down under light, and viticulturists are now experimenting with growing practices that expose grapes to more sunshine in an attempt to minimize the chemical’s presence. Leaves are pulled off of the plants, which are then compared to control groups that grow with leaf cover. So far taste tests have shown that pepper-juice flavor can be altered by modifying growing conditions. “People can tell the difference,” says Heymann. “The way you grow the grapes absolutely matters.”

Yet, according to Boulton, the connection of taste to a particular molecule is rare. To show that the methoxypyrazine was involved in flavor, it first had to be isolated, then a receptor in the nose had to be identified. Finally, panels of tasters had to demonstrate that wines with higher levels of the methoxypyrazine smelled differently from those without. Scientists in search of flavor molecules in wine are rarely able to pass all these tests.

One of the most difficult problems in identifying the molecular source of flavor is that many of the suspected compounds have astonishingly low detection thresholds. In the case of methoxypyrazine, for example, the flavor is noticeable at 2 parts per trillion. As Heymann puts it, just a few drops of methoxypyrazine in a swimming pool would be enough to make you think you were swimming in bell pepper juice. “All of the flavors we are dealing with are very small in concentration, and analytically we don’t always even know their identity,” says Boulton. “This is especially true with red wines.”

One of the most difficult problems in identifying the molecular source of flavor is that many of the suspected compounds have astonishingly low detection thresholds.

As many of the flavor molecules in wine are quite potent, the nose can detect very small amounts. Unfortunately the chemistry equipment used in the lab isn’t as sensitive as the human nose.

Only a handful of other molecules have been tied to a distinct flavor. Short-chain volatile aldehydes like hexanal, pentanal, and nonanal have been shown to contribute grassy, nut-like, and orange-rose flavors, respectively. Specific terpenes have been shown to give Riesling its unique aroma. Glycosides from cabernet sauvignon and merlot grapes are thought to smell like fig, tobacco, and chocolate, but the flavors haven’t been correlated with a specific compound. Sometimes a molecule is associated with a specific place, as is the case with 3-mercapto-hexan-1-ol, a thiol that produces a rich citrus flavor in Sonya Blanc wines from New Zealand: “You can make this style in other countries using the same grapes,” Heymann explains, “but it’s much more difficult and doesn’t have the same flavor.”

The Basics: Tannins, Acids, and Sugars

Since the molecules that account for specific flavors are elusive, much research on wine and flavor has focused on the role of acids, tannins, and sugars. Experts have long advised diners to consider tannin levels when pairing wine with food. At Pure Food and Wine, a raw-foods restaurant in New York City, sommelier Joey Repice seeks organic, handcrafted wines with a well-balanced tannin structure to accompany the delicate flavors of the fresh vegetarian entrees. “There are too many subtleties with our cuisine, and wines with lots of muscle and tannin structure go better with meat meals,” says Repice. For meat dishes he recommends Italian wines known for having a good tannin structure—perhaps a Barbaresco.

Tannins are polymers of phenols that seep from grape skins during fermentation. They are sometimes thought to provide clarity, but too many can cause a bitterness sometimes known as the “pucker factor.” While tannins have always played a key role in the vocabulary of connoisseurs, scientists are not convinced that the molecules are related to flavor. “Flavor can be traced back to a molecule interacting with a smell receptor, and tannins don’t do this,” says Boulton. Instead, tannins temporarily bind to generic proteins on the surface of cells in the mouth. They also change the texture and viscosity of proteins in the layer of saliva that coats the tongue, making it less fluid and slippery. Although tannins wash off when you rinse with water, there is an immediate impact on the entire mouth when you sip a glass of tannin-rich wine. With the exception of catechin and epicatechin—two tannins which happen to behave like flavor molecules—phenols generally don’t target a specific receptor. The mouth-puckering, dry sensation associated with tannins—sometimes compared to pomegranate or lemon pith—is more closely related to an astringent mouthfeel than a bitter flavor.

Similarly, acids are thought to contribute to tartness, but their role in creating flavor remains debatable. “When you change the acidity of a wine, the tartness often changes,” Boulton says, “but there are also acidic wines that are not tart.” Boulton suggests that the flavor molecules are likely separate from the acids. While he admits this hypothesis must still be tested, it’s likely that the acidity creates a climate in which tart molecules can work their magic, he says.

Sugars influence almost every aspect of a wine.

Like tannins, acids don’t directly add to flavor, but they influence how the wine feels in the mouth. Much like tannins contribute to the overall clarity of a wine, acids help balance out the sweetness and give wine a more rounded feel. The key—scientists and sommeliers agree—is to keep both tannins and acids in check. “While a subtle acidity is important, it has to be balanced,” Repice says.

Sugars, on the other hand, influence almost every aspect of a wine. At the most basic level fermentation is the process by which yeast converts the glucose and fructose in crushed grape juice into alcohol and carbon dioxide. Controlling the amount of sugar in wine has historically been difficult because yeast has a complex metabolic cycle and sugar levels vary depending on the ripeness of the grape. Early in the growing cycle berries expand because the cells inside are multiplying and dividing. But as the fruit matures, the individual cells begin to grow, as water and carbon dioxide are converted into glucose, fructose, and other sugars. Glucose levels start out much higher than fructose, but as the grape ripens the ratio begins to change. Eventually, if the grapes are allowed to overripen, fructose levels outpace glucose, and the grapes start to turn into raisins.

While raisining is generally a bad sign for winemakers, Sauternes from France, Trockenbeerenauslesen from Germany, and a few late-harvest table wines make use of shriveling to achieve sugar levels so high that the wines remain sweet even after fermentation. More typically, however, too much sugar makes wine taste like rotten candy, overpowering the subtle flavors that give it character and body.

Fermentation and Tannin Control

Winemakers have been tinkering with wine flavor for millennia by varying the grape varieties, growing conditions, and fermentation processes, but the modern practices associated with the sciences of viticulture and enology date to only the late 19th century. Around 1860 the French physiologist Louis Pasteur firmly established that alcoholic fermentation is caused by yeast. The realization that fermentation was a biological process that might be controlled and might yield predictable results opened up an entirely new way of thinking about beermaking and winemaking.

Since then fermentation scientists have made profound contributions to a wide range of other scientific disciplines. Many of the early studies that broke ground for the fields of molecular biology and biochemistry relied on yeast as a model organism, and many were motivated by questions about the fermentation of wine, says Boulton. From mating assays that test genetic theories to cell-signaling experiments, simple brewer’s yeast (Saccharomyces cerevisiae) remains one of the most commonly used organisms in the field of molecular biology.

The realization that fermentation was a biological process that might be controlled and might yield predictable results opened up an entirely new way of thinking about beermaking and winemaking.

An important early scientist was Eugene Hilgard (1833–1916) who would eventually found the viticulture department at UC Davis. Bavarian-born and raised in the United States, Hilgard studied in Germany with such leading chemical thinkers as Carl Friedrich Plattner, Johann Joseph Scherer, and Robert Bunsen. He returned to the United States where his deteriorating health motivated a career change: he became an advocate for state-sponsored sciences that brought him to work outdoors in fresh air, particularly geological and agricultural surveys. With Hilgard, UC Davis found an outspoken advocate for practical, applied scientific research that would benefit the state’s growing wine industry.

Like most grape producers of his day, Hilgard believed that color was a marker of the fermentation process. In an ingenious but little-known experiment, Hilgard used a stereoscope—a popular Victorian device that creates the illusion of depth in a photograph by presenting a slightly different image to each eye—to track wine’s aging process. Hilgard’s stereoscope was designed by Michel Eugène Chevreul, a chemist whose work with dyes and pigments influenced the Impressionist and Neo-Impressionist movements. Chevreul observed that the eye naturally fused colors of slightly different shades, allowing contrasting hues to lend depth and intensity to an image. Whereas Chevreul used the stereoscope to observe distinctions between objects in a painting or textile, Hilgard used it to study the change in a wine’s color during fermentation. Spots of wine were applied to paper at increments during the aging process. “You could compare the paper to a fabric of a known color,” explains Boulton. As the green grape juice fermented, it changed to pink, red, and then finally to purple—a process that had been observed for thousands of years. “[Hilgard’s experiments] let people know when the color transitions had peaked, and the wine could be transitioned into aging barrels,” adds Boulton. As later work would show, tannins leach out of grape skins early in the vatting process. Allowing the juice to sit beyond this peak might result in too many tannins, with no additional color. Boulton credits Hilgard as “the first to quantify this process.”

Hilgard’s methods are now being automated. In 2001 Boulton and other UC Davis colleagues launched the Hilgard Project—a network of pressure transducers that monitor fermentation in vats around the world. “As only one crop of grapes can be grown each year, it can take decades for a vineyard to collect enough data to make any real conclusions,” says Boulton, “[but] with the Hilgard Project we are compiling enough data for real analysis to be performed.” The data are made available for public use, and Boulton says the scope will soon be expanded to include other sampling methods. Plans are in the works to install colorimeters and other sensors that can be used to monitor tannin levels and alcohol concentration directly. Identifying the threshold at which tannins stop contributing to color but continue to affect mouthfeel is a future goal.

State of the Art

The Hilgard Project is introducing contemporary quantitative measures to other aspects of wine production as well. Boulton installed pressure transducer sensors in three large, metallic vats in the UC Davis teaching winery to demonstrate that the devices can replace the standard rudimentary observations used during large-scale winemaking. “They allow us to monitor the sugar consumption that occurs during the fermentation process and to diagnose problems,” he explains. The device is intended to replace more traditional hydrometers—floating devices used to monitor the density of fermenting juices. Because liquid exerts a buoyancy force equal to the weight of the displaced volume, the device floats higher in dense fluid. Juice is usually denser before fermentation, when there are more dissolved solids. As the fermentation process completes, the dissolved sugars are burned up, and the hydrometer begins to sink.

“We get more accurate readings with sensors,” says Boulton. Because pressure transducer sensors are installed at the bottom of the vat, they deliver an overall average reading of the juice weight. The readings from hydrometers, however, are more local and therefore often cause sampling errors—especially when the juices are not well mixed. Boulton admits that on a vat-by-vat basis, the detailed measurements probably don’t lead to production of a higher quality wine—the end product isn’t likely to be any better than would come from a process that uses a hydrometer. “The idea is to collect data that can be used to understand wide-scale patterns in fermentation chemistry,” he says.

The data collection process supported by the Hilgard Project is part of a long trend toward automation and computation. Biosensors are being developed in research labs to help measure esters and alcohols at the molecular level, and remote sensing data are used to study the impacts of climate change on vineyards.

For all of the efforts toward systematic measurement methods, however, the structure of wine flavor hasn’t become any more lucid. New molecules are discovered in wine every year, but very few are shown to play a direct role in flavor or aroma. “Fifty years ago people believed there was a molecule that made Riesling or Pinot Noir unique, but now we realize it’s infinitely more complicated,” says Heymann. Even as future research correlates core aspects of wine back to flavor molecules, the synergistic interactions between key compounds will have to be analyzed.

In the meantime, Boulton and Heymann encourage people to take science into their own hands—perhaps by turning the living room or kitchen table into a wine tasting laboratory, and implementing some of Heymann’s taste-testing methods at home. “We recommend people form groups, taste wines, and look for descriptors,” says Boulton. “Start with your favorite fruit wines and cups of different fruit jellies. Anything that gets people thinking about wine from an analytic perspective helps the field move forward.”