Flavor

by Bob Holmes

  The first step in making a flavor, Mullin tells me, is to be clear about what the client wants. Suppose, for example, that I had come to Givaudan saying I want a strawberry flavor. Well, fine. They already have thousands of strawberry flavors. Do I want a ripe one, a green one, an especially fruity one? Do I want a simple, inexpensive recipe or a costly but more realistic version? The client’s answers to those questions will help determine the right starting point. I think of the best strawberries I’ve ever eaten, the ones I used to buy at the farmer’s market near where I lived in coastal California. The strawberry fields were just down the road, and I’m convinced that the berries that were picked too perfectly ripe to ship ended up in our market. Their fragrance was powerful enough to seduce you from clear across the parking lot. That’s the strawberry I want.

  No matter, though. Mullin’s already picked the recipe we’ll use for our demo. He hands me a single sheet of paper with a short list of ingredients. “Mother Nature’s already decided what goes into a strawberry,” he tells me. Of course, no customer could afford to include all the hundreds of flavor compounds found in a real strawberry, nor would they need to. The trick is to pick the key ingredients that will bring our flavor close enough to the real thing, for a price we can live with. For many flavors, you’d start with the character compound: amyl acetate for banana, methyl benzoate for cherry, citral for lemon. But strawberry has no character compound—there is no single molecule that smells like it, so even the simplest strawberry flavor has to be built up from several components, each of which contributes one facet of what we’ll perceive as strawberry. Mullin’s recipe has just four ingredients—simple enough for me to make quickly, but sufficient to clearly sketch out the flavor of strawberry.

  Before we go into the lab, Mullin introduces the ingredients to me one by one in his office. The first is ethyl butyrate. He grabs a brown glass bottle off his desk, unscrews the cap, and dips a strip of filter paper into the liquid inside. Then he offers the strip—flavorists call it a blotter—to me to sniff. It has a bright, generically fruity aroma, and it supplies the essential top note of our flavor.

  Flavorists sniff a lot of blotters in the course of their work, enough that most of them carry books of blotters in their pockets like smokers carry matches. (Mullin’s still bear the logo of a previous employer, a flavor house he last worked at seven years ago.) Like almost every flavorist who ever passed me a blotter, he warns me not to let it touch my nose as I sniff—a drop of concentrated odorant on the nose is just as disabling to a flavorist as a sprained ankle is to an athlete. There’s also the question of what to do with dipped blotters after smelling, especially if you might want to sniff them again in a few moments. Most flavorists I visited just set the blotters on the corner of the desk, which risks a lot of aromatic contamination of the desk surface. Mullin, however, uses an old hand’s trick of the trade: he creases the blotter with his thumbnail just below the dipped part, so that when I set the blotter down, its moist tip sticks safely up from the desk.

  Item two in the recipe is cis-3-hexenol. Mullin dips another blotter and passes it to me. This one smells exactly like freshly mown grass, and adds a green note to the flavor. (Look for that grassy green note the next time you eat a strawberry. You may not have noticed it before, but it’s there.)

  Next up is furaneol, which supplies a brown, cotton-candy-like sweet smell characteristic of ripe strawberries. “If you make a strawberry without furaneol, in my opinion, you’ll never sell it,” says Mullin. “The more you put in, the better it is—but there comes a point where you can’t afford it.” Furaneol gives the lingering finish that helps define a good strawberry flavor. “It just carries and carries and carries,” he says.

  The fourth and final ingredient in our flavor is gamma-decalactone. On the blotter, it smells a bit peachy. It’s there, Mullin says, to fill a gap in timing: ethyl butyrate hits right up front, followed quickly by cis-3-hexenol, but furaneol’s contribution takes a while to develop. That could leave a hole in the flavor, which the gamma-decalactone fills.

  I now have four blotters on the desk in front of me, tips raised off the surface like a family of baby cobras. Following Mullin’s instructions, I gather up all four and waft the bundle under my nose. Presto: strawberry! Not the strawberry of my dreams, the farmer’s market berries from California, but a recognizable strawberry, nonetheless—and further proof, if any is needed, that a skilled flavorist can assemble a flavor that smells nothing like any of its individual components.

  Root beer is another good example of this. Once upon a time, as you might guess from its name, root beer was made from an extract of sassafras root. But safrole, the main aromatic oil from sassafras root, turns out to be carcinogenic, and in 1960, the United States banned its use in soft drinks. Root beer manufacturers had to concoct their flavor some other way, and Mullin shows me one option: a top note of methyl salicylate (which smells like wintergreen Life Savers), a middle note of anise-smelling anethol, and a lingering base note of vanillin. Put them together, and it’s unmistakably root beer. It surprised me to learn that root beer’s top note is wintergreen. I’d never noticed that before, and I doubt many others outside the flavor industry have, either. But once you know to look for it, it’s definitely there. (Actually, most Europeans—who didn’t grow up drinking root beer, and thus don’t instantly recognize that blend of flavors—get it right away. Many of them can’t imagine why we drink the stuff in North America, because to them it smells like the wintergreen-scented liniment you rub on sore muscles. “Why would you want to drink something that smells like a rugby locker room?” one Brit asked Bob Sobel on encountering root beer for the first time.)

  But enough sniffing. It’s time to hit the lab. Mullin grabs a lab coat off a hook behind his door and gives it to me, along with safety glasses and a handful of plastic eyedroppers. “Let’s go make a strawberry,” he says. We’re making a test-sized batch, the amount a flavorist might produce while tweaking a formula, so the process turns out to be a simple matter of measuring fluids into a beaker. Just a tiny bit of ethyl butyrate and cis-3-hexenol, 0.08 grams of each, which turns out to be somewhere between three and four drops. Mullin suggests measuring those first, so that if I accidentally put too big a squirt into the beaker, I won’t have to discard a large volume of ingredients. Then fifteen grams—about a tablespoon—of urine-yellow furaneol and a squirt of gamma-decalactone. Stir, then dilute with water.

  Now it’s time to see what we’ve made. You’ll recall that we sometimes perceive retronasal flavors rather differently from orthonasal smells. As a result, serious testing of a flavor formula almost always happens by actually drinking the concoction—no mere sniffing at this stage. Sipping the flavor, I find it a little disappointing. It doesn’t have the ripe, ripe oomph I was hoping for, and the green note—inconspicuous on sniffing—comes through too strongly in the mouth. The next step, says Mullin, would be to modify the recipe by using a little less cis-3-hexenol and a little more ethyl butyrate next time, to see if that gets closer to my target.

  In practice, this trial and error would go on again and again, with repeated taste testing until the client is finally happy with the result. It’s a slow process: Mullin’s assistant can mix up perhaps a dozen formulas in a day, particularly if they’re more complex, so settling on a final flavor could take days or weeks. That makes a custom flavor like this an expensive proposition.

  To speed things along, Givaudan has developed ways to automate the process somewhat. Earlier that day, another of their researchers, Andy Daniher, had showed me a suitcase-sized device they call the MiniVAS (for Virtual Aroma Synthesizer) that Givaudan flavorists can take out on house calls. The device has slots for thirty vials containing aroma “keys,” which can be single odorants or complex mixes such as lemon-peel extract or cola flavoring. By moving sliders on a touch screen, users can change the proportions of each key in the mix, then see how the aroma changes. (The Mini-VAS has three output ports shaped like the negative of a
nose, so that flavorist and clients can all sniff at the same time.)

  “Let’s talk spiced rum,” says Daniher. With a touch of his finger, he starts air bubbling through a rum base. Another touch adds a hint of strawberry to the rum—a terrible idea, we all agree. Two more quick touches and the strawberry is replaced by orange. Much better. “Now you can start to say ‘I like this, I don’t like that.’ You can just show what you like about a flavor,” says Daniher. “You can do a lot of flavor-creation work very quickly and zero in to a formula you can compound.” Best of all, the whole thing can be controlled remotely, so that a flavorist here in Cincinnati can collaborate with another flavorist in Asia and a client in London, and all can smell the same thing simultaneously.

  For many customers at the large flavor houses, all this analysis may be overkill. They may be able to bypass flavor development entirely—and save a lot of money in the process—by using an off-the-shelf flavor. At Givaudan, these clients end up talking to Laurence Roquet, who manages what the company calls its “portfolio”—a searchable library of previously created flavors. Roquet is French, and she looks it—tall and slender, with bobbed, black hair and a round face emphasized by her huge, round glasses. “Why re-creating the wheels?” she asks, in her fluent but slightly askew English. “Why make another strawberry when you have hundreds of strawberries on the shelves? We have so many flavors, good stuff inside. Why not do that?”

  Givaudan’s full portfolio might have one hundred thousand flavors in it, though the core portfolio that they use regularly is around three thousand flavors. Each one of the flavors in that core library is assigned a series of tags that describe the flavor itself (juicy, pithy, and so forth), its possible applications (sweet, savory, cold beverages), and its regulatory status (is it organic, natural, GMO-free, approved for alcoholic beverages). That lets Roquet and her staff quickly pull up a short list of flavors that meet a client’s requirements. Then it’s time for tasting. Often, the client finds a flavor that they’re happy with as is. If not, the portfolio at least suggests the right starting point for further tweaking. About 70–80 percent of Givaudan’s flavor projects start at Roquet’s desk, she says.

  Pulling ready-made flavors off the shelf is one end of what you can think of as the innovation spectrum. At the other end, Givaudan also puts a lot of effort—and plenty of cash—into discovering and re-creating new flavors. Often, this involves prospecting in the natural world, looking for fruits, flowers, or other plant parts that offer new flavor molecules to beguile the world’s palates. One favorite source is the botanic garden at the University of California, Riverside, which hosts the world’s largest collection of citrus trees. By sampling fruits in the Riverside collection, Givaudan’s flavorists have found several new citrus flavors, including a sweet lime with just a hint of pepper. “That was nature showing us something we would not have thought about,” says Daniher. Finding these gaps—white spaces on the flavor map, even in a region as well traveled as citrus—is always amazing, he says.

  Sometimes Givaudan’s explorers go further afield. Years ago, Peppet participated in an expedition to Gabon, Africa, where Givaudan chartered a blimp to float above the rain forest canopy so that technicians could collect scents from every flower and fruit they could find. Back home, flavorists sorted through it all, looking for elements that they could add to their arsenal of flavor chemicals.

  Other times, they don’t have to go far at all. “It occurred to us that we don’t have to go into jungles,” says Daniher. “We can go into restaurants.” Givaudan technicians order up an interesting dish, something with a flavor that they’re interested in replicating. This is what they call their “gold standard”—the real thing, the target flavor they will try to approximate in the lab. The whole order goes into a chamber that captures the aromas rising off the food. Then the technicians analyze this “headspace” to figure out what makes it tick, so that flavorists can find ways to re-create it in their lab.

  Daniher opens a vial labeled “kalbi flavor” and hands it to me. It smells just like the grilled meat, redolent with soy sauce and garlic, that’s so delicious in a good Korean restaurant. “What I like about this is you can really smell the fatty, grilled notes,” he says. But this wasn’t extracted from real Korean barbecue—Givaudan’s flavorists have re-created it from individual chemical components to match the headspace analysis. It’s basically the gold standard in a bottle—an almost perfect match, but probably too expensive to be practical as a commercial flavor. Now it’s up to their flavorists to develop a cheaper version that delivers nearly the same effect.

  Another project they’re working on is a flavor element that Daniher calls “richness.” “Richness is what you get from slowly cooked foods,” he explains. “We’ve all tasted a great stew that’s been cooking for a long time.” Daniher’s researchers think they now understand which flavor molecules are responsible for this long-cooked flavor. When pressed for more details, Daniher clams up. “Proprietary stuff,” he says. In essence, Givaudan may have isolated the flavors of time, care, and patience. If they’re right, they could be on to something really big.

  A job like re-creating Daniher’s kalbi flavor or “richness” is likely to end up on the desk of someone such as Mary Maier at Givaudan. The world of industrial flavor is so broad that flavorists tend to carve out particular corners to specialize in. One person I spoke with had spent a long, illustrious career specializing in sweet brown flavors: honey, maple syrup, cola, and the like. There are fruit flavorists and beverage flavorists, dairy flavorists and candy flavorists. One of the biggest divides tends to be between sweet flavorists and savory ones. Maier, a senior flavorist, is one of the latter. Working in a world of meat flavors is tougher than doing fruit flavors, she says, because the flavors themselves are more complex. “There’s not just one molecule that you smell and say ‘Aha!’” she says. Maier is a short, fit woman with straight brown hair down to her shoulder blades, held back by a thin headband. Remarkably, she’s a second-generation flavorist—as a college student, she used to mix up samples for her father, who also worked for Givaudan, and she ended up making a career there herself.

  Much of Maier’s work involves riding herd on the Maillard reaction, a complex network of chemical changes that happens during the browning of proteins and sugars. But where you and I typically start our Maillard reactions with a piece of beef or chicken, professional flavorists like Maier often start with protein extracts such as autolyzed yeast extract, or even pure amino acids and sugars, to give better control of their results. Start with the amino acid cysteine and the Maillard reaction proceeds toward chickeny-meaty flavors. Start with methionine and you get something potatoey-cabbagey; phenylalanine gives honeylike flavors or—in combination with the sugar fructose—a flavor that some describe as “dirty dog.” (There’s that paradox again: a little bit of something obnoxious adds interest to a complex flavor.)

  We head out into Maier’s lab to taste some flavors. First up is a chicken flavor that one of her customers wants to put into a powdered soup mix. Maier spoons some of the mix into a beaker, adds water, heats it on a hot plate, and offers me a spoonful. I taste onion, celery, and some doughy or grainy stuff from the noodles, but Maier says that’s all irrelevant to the task at hand. Those flavors are part of the customer’s base soup. Her job is the chicken flavor; from her perspective right now, everything else is just noise. Concentrating now on the chicken, I think I get some roasted-chicken notes, but Maier corrects me. What I’m tasting are the astringent, bony, fatty notes of a boiled chicken, not the caramelized brown, sulfury notes of a roasted one. The flavor is not bad, though: she’s getting close to the target she wants.

  Next project in her lab is a chicken patty, which will be breaded, prefried, and frozen, then baked at home by the consumer. There’s a version of these already on the market, but the manufacturer wants to change ingredients. Maier doesn’t know why they want to do that—to save money? To use more readily available ingredients? It
doesn’t matter. Her mission is to make the new version taste the same as the old one.

  So her technicians have whipped up a test batch. There’s a blank—the unflavored chicken nugget, which tastes generically chickeny. There’s the original version, the target. And there’s the test version, with the current version of Maier’s new flavoring. On tasting the test patty, one of the technicians immediately says “That’s extremely strong!” This is the first time they’ve tasted the new flavor on chicken, and it wasn’t as aggressive during earlier taste tests in water—further proof, if any were needed, that there’s no substitute for tasting a flavor in its final setting. Flavorists can’t do their jobs in the abstract.

  Maier takes a bite of the target next, and pauses to consider. “I get a precursor in there,” she says. That is, she tastes one of the ingredients in the Maillard reaction that hasn’t fully reacted—a sign that the client’s original Maillard starting point wasn’t quite right. She doesn’t say it aloud, but I can imagine her thinking that whoever designed the original flavor did a sloppy job.

  After some discussion, the team agrees that the target also has a more grilled, sulfury-meaty character, while the test nugget is more smoky and livery. The lack of grilled flavor in the test version disappoints Maier. “This is just softer. We’re not getting that impact,” she says. They decide to do another version next week to see if they can get closer to the target. Meanwhile, they’ll also send the target off for analysis, to try to identify the precursor that’s out of whack.