Profile of a most unusual scientist entrepreneur with wine in her sights. A version of this article is published by the Financial Times.
September’s board meeting of VeriVin had to be delayed because its Mexican wine taster, ballet dancer founder was busy judging classic cars at Salon Privé at Blenheim Palace.
VeriVin is a tech startup based just outside Oxford, part-funded by the university. Masterminded by Dr Cecilia Muldoon (whose entry in our wine writing competition we published last year), it aims to compile a unique database of wine identity tags without ever opening a bottle – essential for seriously expensive wines. Via spectroscopy, using light to study matter, she hopes eventually to help the wine world stamp out counterfeiting, profile and control wine quality, monitor the condition of individual bottles, and possibly test for faults.
VeriVin is just one string to Muldoon’s bow. Born in Mexico – with a father whose classic car collection is kept mainly in the United States, for security – she started dancing before she was three years old, according to her mother. When she was 11, she was good enough to spend a year at the Paris Opera ballet school and then to be taken under the wing of the San Francisco ballet school. Eventually, her father put his foot down and she adopted a more conventional educational path, so now she continues to dance, but just as a hobby.
She was classmates with Mark Zuckerberg at a boarding school in New Hampshire and later majored in physics and finance at Princeton. On graduation in 2006, she was offered a job at Goldman Sachs but chose to study physics at Oxford instead. A PhD was followed by a stint with an instrumentation company in Switzerland, where it dawned on her that her academic specialism, quantum computing, might be applied to wine. (She is also a Wine & Spirit Education Trust graduate and was president of the Oxford University Wine Circle and captain of the Oxford university blind wine tasting team. Of course.)
With the encouragement of her PhD supervisor Professor Axel Kuhn, she has set up a research lab with a multinational staff of seven dedicated to analysing wine through the glass. An important mentor has been Steve Davies, Waynflete Professor of Chemistry, another wine-loving Oxford scientist (and part-time local restaurateur), with a famous cellar and a proven record in financially successful projects. A role model, one might say.
In their first few months they have learnt a lot. For a start, wine turns out to be much harder to analyse into its constituent parts by spectroscopy than many other liquids. So much of wine is water, then ethanol, that the complex array of minor ingredients, so important for identification, can be difficult to assess through a bottle – especially if that bottle is anything other than clear glass. If the glass has any yellow elements to it, for example, as most burgundy bottles do, it plays havoc with the spectra. One of her staff, perforce, is a glass scientist – and Dr Edoardo Ceci Ginistrelli has come to realise just how many different sorts of glass are used by wine bottlers. (A further complication is that recycled glass is often a mixture of them.)
Added to that, red wine has its own problems beyond the bottle it comes in – its fluorescence tends to drown everything else out. 'Once you can test red wine, you can test anything', Muldoon told me as she ferried me from Oxford Parkway to a meeting with some of her key associates in her racing green 1955 TR3. (Her next car, just delivered, is the 1974 Dino Ferrari below, and she is learning to race in a 1994 Mazda MX5.) For the moment, the VeriVin team have found that the perfect liquid to analyse with a spectrometer is olive oil in a clear glass bottle. 'You should see the beautiful spectra we get from olive oil. For a while we were kidding that we should become VeriOil', she said.
Their aim – and they have an Innovate UK quantum technologies grant to assist them – is to come up with ways that different wines can be analysed, characterised and possibly checked for faults by using a hand-held, 'idiot-proof', non-invasive spectrometer. One obvious skill that would be welcomed warmly by the wine community would be the ability to spot whether a wine is affected by TCA (trichloroanisole), the compound responsible for cork taint. 'But TCA may be just a relatively small quantity of molecules per bottle, so it could be difficult to spot', warns Muldoon. A more likely application would be for authenticating wines, especially fine wines.
So far, the team’s attention has been focused on the practicalities of getting the equipment to work. Their portable spectrometer is designed to be much quicker, cheaper, smaller and more versatile than any static bit of lab apparatus and much more sensitive than anything currently on the market. But Muldoon admits, 'we have yet to understand exactly how producers would want to use it'.
Their work has already attracted the attention of several major wine companies and the widely admired Australian Wine Research Institute. I for one would like to see more communication between the scientists and those on the ground in the wine business.
VeriVin’s present target is to develop their database of molecular identity tags for as many wines as they can get their hands on, ideally when they are bottled. These could be compared with a subsequent scan of the same wine to check authenticity and the condition of the wine. For instance, wine producers could check whether their wine has been damaged by excessive heat during shipment. The molecular composition of a wine could reveal which grape varieties it was made from.
They would like to assemble and analyse millions of these molecular fingerprints and even correlate them with subjective descriptions of the same wines (tasting notes in particular) using machine learning 'to correctly predict the taste and structural profile of each wine, as expressed in terms of keywords used by wine critics'. Such a database would certainly be unique, but they probably need to communicate more closely with wine professionals to work out precisely how it would be most useful.
Presumably there are also further applications beyond wine. There is no shortage of similarly luxurious products such as Manuka honey (glucose concentrations present no problem to spectroscopy, apparently) and balsamic vinegar that sorely need some sort of non-invasive authentication process. Quite possibly the perfume business could benefit from this sort of technology, too.
Analysing and authenticating wine
Most current ways of analysing wine today involve sending liquid samples to a professional laboratory that is equipped with various bits of expensive, heavyweight kit and, often, converting those samples to a gas.
As fine-wine prices have soared, so has the temptation to sell counterfeit wine. Examples are legion. In 2014 Rudy Kurniawan, a young Asian based in southern California who had infiltrated many a high-flown tasting group, was sentenced to 10 years for selling millions of dollars worth of fake wine, including combinations of vintage, producer and appellation that did not exist. It is not known how many of his creations are still in circulation, or in private cellars.
Specialists in fine-wine authentication have to depend on visual inspection of every last detail of a bottle of wine, including the foil, the glass type, the exact shape and weight of the bottle, what can be seen of the cork, and of course every tiny aspect of the label.