New Technologies for Detecting Defects in Coffee
Defects in green beans plague the coffee industry, and finding defects continues to prove challenging. However, new methods to identify defects such as hyperspectral imaging and electromagnetic noses are emerging, but are they viable solutions?
By Anne-Marie Hardie
The identification and isolation of defects in the green bean is an issue that continues to challenge the speciality coffee industry. If left alone, defects, like black beans, potato defect, and dehydrated beans, can negatively impact the quality of the cup. Well-trained roasters can manually isolate the defective beans, however, this process requires a significant time investment, and is not a viable solution for mass-market production. This has led to some roasters choosing not to purchase beans from particular regions for fear of a tainted crop. Meanwhile other roasters are actively looking towards new research and advanced technology, such as hyperspectral imaging and electromagnetic noses to see if they could be viable solutions.
Potato defect is exactly this type of issue resulting in some roasters dismissing coffee from Rwanda, Burundi, the Democratic Republic of Congo and Western Uganda, fearing that their batch could potentially have this defect. “If buyers approve a coffee based on a small set of data with the expectation that this coffee will be defect-free, they will be let down later with that high expectation,” said Chelsea Thoumsin, coffee quality specialist, Counter Culture Coffee, Durham, North Carolina. “On the flip side, if a cupper encounters potato defect in a cupping of 20 cups, the likelihood of them finding another in the next 50+ cups is very low.”
Concerned about potato defect and how it may affect the speciality coffee in this region, Counter Culture is invested in conducting research and collecting data to better understand what makes this defect come to the surface and to find efficient tools to isolate the defect. “How buyers and roasters purchase (or not purchase) coffees due to this has a direct impact on the specialty coffee industry in these areas,” said Thoumsin.
For the past ten years, Counter Culture has been manually cataloguing and analysing green beans for potato defect. To date, they’ve catalogued approximately 140 coffees, with 99 percent of the beans from these African regions displaying some extent of the defect.
“Our goal is to better understand when this defect displays itself, we are looking at what we can learn from high instance rates and low instance rates,” said Timothy Hill, coffee buyer and quality manager, Counter Culture. “We hope to use this information to efficiently program technology to identify and isolate this defect.”
Technology is fast and efficient. If it is effective in isolating defects, it could alleviate buyers’ concerns of purchasing coffee in these regions.
Testing Hyperspectral Imaging
Researchers at the University of Genoa have been studying the capabilities of NIR (Near-Infrared Reflectance) hyperspectral imaging in isolating defective green beans. This technology can break light into dozens, sometimes even hundreds of narrow bands creating a spectral signature, which could be used to identify the variances within both the green and roasted bean. Their goal was to test a multi-variate pattern recognition to not only see if the defective beans could be identified, but if the equipment could then be used to classify the different defects.
According to lead researcher, Cristina Malegori, PhD, department of pharmacy, University of Genoa, Genoa, Italy, the most challenging of the defects to isolate is identifying beans with an irregular visual appearance. In order for this technology to be an effective tool for the coffee industry, the equipment would need to be able to both isolate and identify the individual types of defects. To test if this theory was possible, Malegori implemented a pattern recognition approach to isolate and classify dehydrated beans, black beans, and dried cherries from the non-defective beans. The results of this research were particularly promising, with the NIR hyperspectral imaging able to isolate each separate defect.
However, due to the cost and complexity of hyperspectral equipment it has been rarely used in the manufacturing process. Yet, advances in technology are now paving the way for this technology to be used on a wide spread basis. One example of this is Imec’s RGB-NIR multi-spectral platform which integrates standard RGB colour filters, NIR-cut filters, NIR narrowband-pass filters, and on chip micro-lenses technology. The benefit is that the filter can be produced using chip process technology making this piece of equipment much more cost effective.
Imec’s innovation program manager, Kris Van De Voorde, has also been investigating how hyperspectral imaging can be used to alleviate some of the challenges with identifying variances in roasted coffee. Currently, the most frequently used method to ensure consistency in the roast is manually preparing and measuring frequent samples. Hearing this plight of the roaster, Van De Voorde decided to see if hyperspectral imaging equipment could obtain more accurate results. The results were positive, allowing the roaster to accurately determine the level of roasting on individual beans, eliminating the laborious preparation that was previously used to determine consistent roast levels. “With a hyperspectral camera, you can make things visible that are invisible to our own eyes,” said Van De Voorde. “You can identify materials, such as foreign objects that do not belong in a product, or fungi. You can screen the composition of products to find their content in moisture, sugar, fat or protein. You can also check the quality and composition of surfaces and packaging.”
But could this type of equipment be used to detect a defect as complex as the potato defect?
“We looked at whether the electromagnetic spectrum of the potato defect could be identified, and then used it to help sort out the defective beans from the healthy ones,” said Hill. Currently, Counter Culture has received two sets of results using hyperspectral imaging to isolate the defect. The first result was positive, with the equipment positively identifying the defect, however, the second set was less promising. “It’s easier when you’re sorting for metal and glass in fruit, but at the moment there are limitations with the system being able to detect the specific chemical compound that is in this defect,” said Hill.
Exploring Other Options
Detecting the molecule that is responsible for potato defect is extremely challenging as there are several variables that impact how the defect is displayed, Susan Jackels, PhD, professor emerita of chemistry, Seattle University, Seattle, Washington, explained. “One of the key challenges with using technology is that the potato defect is in such small quantities that it is like trying to find one needle in several different haystacks,” said Jackels. “The other complicating factor is that the defect may show in different areas of the bean, sometimes on the surface, and other times within the bean itself.”
Counter Culture is continuing to look at how hyperspectral imaging can become a viable solution for the future. In the meantime, it is also considering whether other forms of technology, primarily an electromagnetic nose, could be used to detect the distinct scent of the chemical compound.
The current process of isolating the potato defect at Counter Culture consists of filling small glass jars with one-third water and then placing them on a germination mat, as the defect is easier to identify when the coffee is warmed up. After 6-12 hours, the jars are smelled to see if there is any potential defect.
Thoumsin noted that the human nose is extraordinarily sensitive to the compound that is called potato defect – Isopropyl Methoxypyrazine (IPMP: 3-isopropyl 2-methoxypyrazine) – with science supporting that less than one part per billion will tip our noses off to the compound. It is for this reason that Counter Culture developed a protocol for isolating green beans for GCMS readings (Gas Chromatography–Mass Spectrometry) in efforts to build a system which can recognize IPMP on its own. “As you can imagine, sniffing through jars and breaking jars down into smaller jars is very time consuming, and at such a small scale, not the most efficient. But for now, it’s important work so we can build these signatures,” she said.
The current limitations with an electronic nose, said Jackels, is that they lack the sensitivity that the human nose has with detecting this defect.
By manually establishing these signatures of potato defect, researchers hope to program technology with specific information that will help consistently identify the defect. The goal at Counter Culture is to establish the groundwork, by collecting additional data and single green bean samples, to further understand the defect. “We’re learning a lot about IPMP as a chemical as well as other chemicals that are apparently correlating to IPMP. Mapping the instance rate of potato defect on hundreds of thousands of grams of coffee has, to our knowledge, never been done before,” said Thoumsin.
Anne-Marie Hardie is a freelancer writer, professor and speaker based in Barrie, Ontario. She may be reached at: [email protected].