Rudy Kurniawan considered himself a bon-vivant, ingratiating himself in the exclusive world of wine connoisseurs. The twenty-something collector graciously shared expensive vintages with other devotees, all the while selling millions of dollars of supposed rare wines to unsuspecting collectors.
By 2012, the bubble had burst. The FBI raided his wine cellar and discovered Kurniawan blended cheaper wines, old corks and aged labels to carry out his deception. In 2018, he achieved the dubious distinction of becoming the first person in the U.S. to be convicted of wine fraud, and received a 10-year sentence.[i]
In 2017, leading authority Maureen Downey estimated $3 billion worth of the wine market is usurped by fake wines globally.[ii] Labeling and packaging are usually the most obvious signs, but since some vintages can sell in the tens of thousands of dollars, fraudsters have become more sophisticated
Enter the world of food science, and a group of researchers working out of the University of Adelaide in Adelaide, South Australia. They set out to apply pioneering instrumentation and advanced chemometrics to identify wines by its chemical composition.
Ruchira Ranaweera, MSc, led the project as part of her doctoral curriculum, joined by researchers Dimitra L. Capone, Ph.D.; Susan E Bastian, Ph.D.; David W. Jeffery, Ph.D. and HORIBA Applications Scientist Adam M. Gilmore, Ph.D.
The aim of the study, titled Authentication of the geographical origin of Australian Cabernet Sauvignon wines using spectrofluorometric and multi-element analyses with multivariate statistical modelling[iii], was to see if fluorescence spectroscopy, using HORIBA’s proprietary A-TEEM technology, and a novel use of a multivariate algorithm could effectively and economically identify a number of wine samples produced from various regions in Australia.
“Wine is very susceptible to fraud because it has a very complex chemical composition,” Ranaweera said. “So it's so difficult to identify certain adulterations. If we can authenticate the geographical origin or the varietal origin of the wine, then we can give that product or the brand more value. That’s the economic side.”
“Wine is a luxury good,” David Jeffery, Associate Professor in Wine Science at the University of Adelaide said. “So there's a lot of value in wine. And usually any highly valuable food product is open to fraudulent activity because there's a lot of money that can be made by taking something that's much cheaper that maybe looks like the product and selling it for something that's much more expensive. It's a global issue.”
Elements are often studied in wine, along with isotopes, volatile compounds and phenolics.
“Those compositional aspects have an influence on the sensory properties of wines from different regions as well,” Jeffery said. “Phenolics drive color, mouthfeel and taste and aroma compounds can originate from the berry but yeast introduces some of its own, which relates to the amino acid profile of the grapes as well. Together, these aspects feed into characteristics that people are looking for in terms of wine sensory profiles. So regional wines, even if it's the same variety, can have different characteristics depending on where they come from, and we'd like to understand that. In fact, authenticity is one thing, but what we really want to get to is what's underpinning that in terms of the chemical markers.”
The researchers intended to use the underlying chemical markers, even if not identified as yet, to classify wines that are unique to different regions of Australia in comparison to wines from Bordeaux, the birthplace of Cabernet Sauvignon. Scientists have typically used inductively coupled plasma-mass spectrometry (ICP-MS) to analyze the elements in the wine. So Jeffery and the team chose that method to compare with A-TEEM fluorescence spectroscopy molecular fingerprinting.
A-TEEM can quickly and simultaneously acquire Absorbance, Transmittance and a fluorescence Excitation Emission Matrix, and automatically corrects for the Inner Filter Effect. A-TEEM is exclusively featured on HORIBA’s Aqualog and Duetta spectrofluorometers.
The aim of the project was to identify the unique characteristics that are inherent for Australian regions. The main idea was to identify a method that gave rapid results and was relatively simple to implement. To achieve that, the researchers were looking for some chemical markers using the two above-mentioned methods.
“I think it is really important, not only for the lab analysis, but if a method could be used in the supply chain,” Ranaweera said. “So an instrument which is very user friendly and gives rapid results with high sensitivity provides an advantage.”
The two analytical methods were used with commercial Cabernet Sauvignon wines from vintage 2015, originating from three wine regions of Australia, along with Bordeaux, France[iv], as a benchmark for the study. All told, the researchers sampled dozens of wines ― much of which was donated by Australian producers.Hundreds of EEMs were acquired with the Aqualog during the study. The A-TEEM method proved to be remarkably accurate for classification, meeting the researchers’ greatest expectations.
She believes the Aqualog with A-TEEM technology has many advantages over ICP-MS.
“When we compare the all the rapid methods that are available for wine analysis, the A-TEEM technique is a standout because it meshes both the absorbance and transmittance, as well as the excitation emission matrix of the sample at the same time,” she said. “So we can obtain the molecular fingerprint of the wine using this A-TEEM technique, which is very unique for each wine.”
The Aqualog and A-TEEM provided other advantages beyond accuracy. Sample preparation and speed of acquisition were at the top of the list.
“In the analysis process, there are less preparation steps we have to use compared to, for example, HPLC or GC methods,” Ranaweera said. “This is a very quick method. We get the spectral results within minutes and don't have to spend much time on sample preparation. We just need to do dilutions with a solvent and can then use the instrument to analyze the samples. So it's very user friendly and gets us rapid results. That's what's most important.”
Ranaweera and her colleagues used multivariate algorithm XGBoost, or Extreme gradient boosting discriminant analysis (XGBDA), which yielded 100% correct classification. As a machine learning technique, XGBDA was an inspired choice for statistical analysis, which delivers high performance and accuracy, especially being very effective in the classification of these types of multiclass unbalanced groups of data, she said.
The researchers believe that combining A-TEEM and XGBDA has great potential for accurate authentication of wines and could be used in the supply chain to verify the provenance of wine.
Ranaweera continues to work on the authentication project as she pursues her Ph.D. Hesitant to give too much information away, she said she’d be going to go further into identifying the chemical drivers behind this classification. Jeffery is also encouraging his other Ph.D. students, who work on an array of different projects unrelated to authenticity, to “throw every sample they have” at the Aqualog to further learn how to exploit its unique capabilities.
Overall, wine makers and consumers should become the great beneficiaries of Ranaweera’s research.
[i] Bosker, Bianca, A True-Crime Documentary About the Con That Shook the World of Wine, The New Yorker, Oct. 14, 2016
[ii] Cho Lee, Jeannie, Fake Wine Is A Billion Dollar Market And Here Are The Ways To Identify Them, Forbes, Feb. 17, 2017
[iii and iv] Ranaweera, Ruchira, MSc; Capone, Dimitra L., Ph.D.; Bastian, Susan E., Ph.D.; Jeffery, David W., Ph.D.; Gilmore, Adam M., Ph.D.; Authentication of the geographical origin of Australian Cabernet Sauvignon wines using spectrofluorometric and multi-element analyses with multivariate statistical modelling, Food Chemistry, July 2020
Do you have any questions or requests? Use this form to contact our specialists.