From farm to fork: Top photonics technologies in food safety

The systems that keep our food and drink safe make good use of spectroscopy
01 September 2020
By Valerie C. Coffey
From farm to fork
The Linea SWIR camera uses an InGaAs sensor and a 40 kHz line rate (left) to detect subtle water content differences in similar looking items. In a quickly moving production line, sticks and stones can be selectively sorted out from other similar looking food items such as raisins or coffee beans. Credit: Teledyne DALSA

MACHINE VISION. Sorting and inspection of food during processing has historically been a job for humans. But machines are better than humans at numerous tasks, such as focusing for long periods, and sorting and verifying labels at superhuman speeds.

"Machine vision systems see subtle details that human sorters can't, like depleted moisture levels," says Mike Grodzki, product manager at machine vision provider Teledyne DALSA in Waterloo, Ontario. "And the cameras don't get tired, bored, distracted, or sick."

Vision systems involve multiple photonics technologies: lasers and LEDs, mirrors and lenses, high-resolution cameras, hyperspectral sensors, and spectrometers, all of which combine with smart software to efficiently detect foreign material and sort food based on color, size, shape, or chemistry. In most cases, red-green-blue or visible-wavelength cameras are adequate for quickly distinguishing good versus bad items as they fly down a conveyor belt or chute. For other products, sensors need more specific spectral cues to do the job.

The selection of wavelength band depends on the application. Ultraviolet (UV) spectra can detect surface aflatoxin-an invisible, tasteless, toxic fungus produced by Aspergillus mold-in food such as corn, chilies, peanuts, rice, and tree nuts. Infrared (IR) wavelengths are useful for sorting foreign matter from food, such as shells and husks of nuts.

In June, Teledyne DALSA introduced its first short-wave IR (SWIR) line-scan camera for machine vision with an indium-gallium-arsenide (InGaAs) sensor that can distinguish subtle variants in water content in a variety of foods. Water is highly absorbent in the SWIR region of the spectrum between approximately 1.4 and 3 µ. Thus, the Linea SWIR camera can more efficiently discern foreign contaminants and bruising during sorting than is possible at other wavelengths.

"Such vision systems could also someday play a very compelling role in reducing the spread of COVID-19 or future contagions in meat packaging plants, where employees must work in very close quarters to cut and package meat," said Grodzki. "Automating certain processes could allow proper social distancing and ensure that meat is free of contamination."

RAMAN SPECTROSCOPY. A powerful method for identifying ingredients and chemical composition, Raman spectroscopy involves projecting a laser onto a sample to measure the frequency shifts of inelastic scattering that correspond to the energies of specific molecular vibrations. Used with a spectrometer, and in some cases a microscope with a mapping stage, the resulting Raman spectra can reveal physical and chemical properties of a sample. For example, the technique can verify authenticity and contamination of food sources, or whether plastic packaging has the advertised layers, thickness, and chemical makeup.

"Raman microscopy is a powerful tool for characterizing food packaging," says Gary Johnson, an expert Raman spectroscopist at Intertek, an analytical lab in Allentown, Pennsylvania. Intertek uses various lasers for different applications. A green doubled Nd:YAG 532 nm laser coupled with a CCD detector is a common tool for Raman microscopy, often useful for identifying aromatic polymer additive compounds or complex organic molecules found in polymers and synthetics.

Sources at other wavelengths can stimulate Raman scattering depending on the application, such as UV lasers for resonance Raman spectroscopy of biomolecules like proteins, 785-nm lasers for surface- enhanced Raman spectroscopy detection of food contaminants like formaldehyde, and 1064-nm lasers for Fourier-transform IR Raman spectroscopy of dyes and other photoluminescent samples.

"We've done Raman analysis using 532 nm excitation to test the plastic wrap for frozen beef patties to ensure it had the specified ethylene vinyl alcohol copolymer barrier against oxygenation, and to diagnose problems with the seal on fruit cup lids," says Ellen Link, senior material scientist at Intertek.

NIR SPECTROSCOPY. The near-IR (NIR) part of the spectrum from 750 to 1400 nm (which varies, depending on whom you ask) plays an important role in food safety, providing information on the food chain, from soil conditions in the field to the ripeness of fruits and vegetables at harvest. NIR spectra can also measure pathogens, toxins, and adulterants in water and other drinks.

Damon Lenski, general manager at spectroscopic instruments developer Avantes in Colorado, said, "Milk adulteration is common in some Asian countries, where occasionally an unscrupulous provider will attempt to spike the levels of protein and fat with detergent or edible oils that can be really toxic. Noncontact NIR spectroscopy uses a common tungsten halogen incandescent lamp shined through a milk sample, which is simple, fast, and very sanitary. NIR can measure an abundance of parameters in milk, such as protein, moisture content, solids, fats, as well as harmful adulterants."

In March, Avantes launched a new handheld InGaAs spectrometer, the AvaSpec-Mini-NIR, for rapid nondestructive testing of numerous parameters of ripeness and quality in the field. The device, which is the size of a deck of cards, can be used in the dairy industry, or on produce still on the vine for precision detection of water content, soluble sugars content, acidity, pH, and aromatic compounds.

HYPERSPECTRAL IMAGING. Integrating multiple sensors across a wide range of wavelengths (for example, UV, VIS, NIR, and SWIR) in one system enables hyperspectral imaging that can be useful in microscopy to provide nanometer-scale identification of chemical structure and contaminants, or in machine vision systems to scan products inside and out for visible and nonvisible defects. Hyperspectral systems acquire a large data cube with precise spectral details that require intensive real-time computing power. So, choosing smaller targeted spectral bands across the breadth of the spectrum keeps the data constrained. Advanced machine learning software and neural networks offer promise to define custom accept/reject thresholds and learn over time to improve identification and sorting accuracy and efficiency.

In June, Horiba Scientific (Kyoto, Japan) launched the LabRAM Soleil Raman microscope, a laboratory solution with an impressive variety of hyperspectral imaging modes: reflectance and transmission imaging, bright field/dark field, epifluorescence, phase contrast, and differential interference contrast microscopy. An interchangeable grating turret and compatibility with up to four internal laser sources and six different filters enable users to quickly change the scanning wavelength from near UV to NIR spectral bands, and image-compression software allows fast hyperspectral imaging. The patented confocal QScan optical system can automatically scan and image a sample with an excitation laser to generate a confocal 3D map in x, y, and z directions. Just about any kind of sample is fair game for high-resolution imaging, spectral analysis, and measurement, including micro-organisms, trapped inclusions, or particles within powders.

"Spectroscopy methods are involved in a lot of advanced food security applications," says Lenski. "I think we're going to see a lot more of it going forward."

Valerie C. Coffey is a freelance science and technology journalist based in Palm Springs, California.

Enjoy this article?
Get similar news in your inbox
Get more stories from SPIE
Recent News
PREMIUM CONTENT
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?
close_icon_gray