C&EN: COVER STORY - AN EYE ON FOOD

Most people in the developed world take food for granted. If you need groceries, you go to the supermarket. Don't feel like cooking? You might order a pizza or Chinese takeout. In a hurry? You're likely to hit a fast-food joint. Want something different? You might try a new restaurant. In many cases, the food you buy is grown, processed, and even prepared by someone else. Do the products you buy have a consistent quality and are they safe to eat? Are you actually getting what you pay for?

The answers to these questions are nearly always yes, thanks to numerous analytical measurements made behind the scenes at food-processing facilities to ensure that food ingredients and finished products meet federal specifications for different food categories and are free from pathogenic microorganisms. These tests range from simple to complex.

Stroll through any supermarket aisle, and it will be apparent that many of the items for sale are processed foods--dairy products, ready-to-eat meats, soups, condiments, and more. The ability of companies to produce large quantities of these foods requires highly automated production facilities. Companies known as system integrators help food processors implement automation equipment to monitor and control production. All this is done with an eye toward maximizing usage of raw materials and minimizing waste to help boost profit margins, as well as to ensure food quality and safety.

More than 1,000 companies serve as system integrators, notes Vernon J. Spaulding, vice president of sales and marketing at ESE Inc., a process engineering company based in Marshfield, Wis. These companies, known as automation solution providers, set up program logic controllers and human-machine interfaces to control all aspects of a production facility.

"Today, the components of the instruments are pretty standard, and the accuracy is pretty good," Spaulding observes. "But what makes the difference for in-line instrumentation is the software-user interface--that is, how easy it is for the operators to program and calibrate. It's all about having a complete turnkey, closed-loop automation system."

NIR spectroscopy is the workhorse of the process industry. It's a simple, inexpensive method that focuses on light absorption in the 650-1,050-nm region. This is a "sweet spot" for the food and beverage industry to monitor C-H bonds of fats and oils; O-H bonds for moisture, alcohol, and sugars; and N-H bonds of proteins.

Spectra can be acquired using in-line flow-through sample cells, or grab samples can be measured off-line, Spaulding notes. Most analyses are based on the spectrum profile over a range of wavelengths and are facilitated by process software that quickly compares the data to calibration files derived from standard spectra that are generated during more extensive lab tests.

ESE makes two NIR systems that use a halogen lamp source with a photodiode array detector. The Process Quantifier, designed for in-line use, has fiber-optic probes inside a flow-through cell that mounts into a product stream. The spectral signal from the cell is processed in the spectrometer placed adjacent to the process line and is part of the plant's automated control system.

"This type of system gives the customer the ability to quickly acquire a spectrum and analyze the necessary parameters, usually within about 30 seconds," Spaulding says. "Using traditional sampling methods, it would take an hour or several hours to do the same analysis in a lab. This faster response gives the producers the ability to optimize their process by adjusting a parameter, such as protein, fat, or moisture, generally to less than 0.5%, depending on the constituent." The software interface is bidirectional, so someone monitoring a system can change a recipe or load a whole new set of parameters on the fly, if needed.

The company's second NIR instrument, the Food Quantifier, is used for rapid lab testing. Samples are placed in a plastic bag and inserted into the instrument to obtain the spectrum. Both NIR instruments are used to analyze samples for fat, protein, carbohydrates, pH, salt, moisture, percent solids, and more.

ESE works with some of the largest milk processors and natural and process cheese manufacturers in the world, as well as manufacturers of butter and cultured milk products such as yogurt and sour cream. Other types of products the company handles include puddings, pasta sauces, soups, salad dressings, condiments such as ketchup and mustard, and nonfermented beverages.

There are about 100 dominant system integrators in the food and beverage sector, Spaulding notes, but only a few dozen companies focus on dairy products. "Of those companies, there's only one--and that's ESE--that manufactures NIR instrumentation and does the system integration," he says.

ESE has started a "Lean Enterprise" initiative, akin to Lean Manufacturing and related business management strategies such as Six Sigma, to promote maximizing efficiency and ingredient utilization in the food processing industry. "It's a very natural fit to go out into the marketplace and educate our customers on how we can apply these methodologies to their benefit," Spaulding says.

One example, from the dairy industry, involves the price of milk and other products, 60% of which is due to the cost of the raw milk, Spaulding notes. The situation is parallel to that of natural gas in the chemical industry. Manufacturer margins are sometimes squeezed between the price of the raw milk and the price at which a product can be sold. For a product like cheddar cheese, the whey by-product can sometimes be as valuable as the cheese itself, he says.

"That makes it important for the manufacturer to be able to manage the incoming raw milk and accurately know parameters such as percent butterfat and proteins as the milk is being processed," Spaulding says. "The key really comes down to instrumentation. If you can't measure it, you can't control it. For us, we do it with NIR."

The goal for ESE with Lean Enterprise is to be able to automatically measure, control, and report material mass balance information from incoming raw milk to end-product shipments and to convey the information into a financial system that can track trends and evaluate performance, he notes. For now, Spaulding believes ESE is the only company that can offer this type of complete package.

ESE is a midsized company with about a 2% market share in the diffuse $430 million automation and in-line instrumentation market in the U.S. But with time, the company would like to see its market share expand to more than 10%, he says.

The advantage of process NIR instruments is that they are rugged yet simple in design and generally cost-effective, according to Donald A. Hawker, vice president of marketing at Aspectrics, a process instrument manufacturer based in Pleasanton, Calif. The disadvantage is that analysis is limited to only a few compounds of interest, he says. If multiple components are to be measured, then multiple NIR units are required, thus reducing the cost advantage and adding complexity to plant engineering.

Fourier transform infrared (FTIR) products have better resolution and work well in the lab environment, Hawker adds. But because the moving mirror of the interferometer is sensitive to vibrations, it can be costly to employ in a production facility. Also, FTIR instruments scan at a fairly low speed--only one to three scans per second--reducing the usefulness of real-time, in-line measurements, he says.

EP-IR is similar to FTIR in that individual wavelength segments are modulated and the spectral intensity is computed via a Fourier transform, Hawker notes. Light from a broadband source passes through a sample and a diffraction grating, and then onto a rotating disk that is made by a photolithographic process. The disk contains up to 256 concentric encoding tracks, which effectively replace the detector elements of a conventional array detector, he explains.

Individual grating-dispersed wavelengths of the sample signal are reflected off the disk to a single lead sulfide or lead selenide detector. The detector signal can be combined at a rate of 100 scans per second, giving the instrument the ability to accurately measure low levels of compounds in the visible through mid-IR range. The cost of the EP-IR spectrometer is slightly more than single NIR units, but much less than FTIR, he says.

"The EP-IR spectrometer's inherent sensitivity, ruggedness, and reasonable cost should prove to be an excellent tool for food and beverage manufacturers to assist in quality control in the process environment, where real-time results are important," Hawker notes. In February, the company began shipping evaluation instruments to companies that manufacture analytical and process instrumentation. Aspectrics is working with system integrators and their suppliers to use EP-IR for food analysis of fats, proteins, and sugars, but the details are confidential for now, he says.

Polychromix, based in Wilmington, Mass., introduced another new type of compact process NIR instrument at Pittcon 2005: the Digital Transform Spectrometer (DTS), which is based on diffractive microelectromechanical system (MEMS) chip technology. It joins the MEMS-based chemical process NIR instrument made by Axsun Technologies, Billerica, Mass., which was introduced at Pittcon 2004.

Polychromix' original focus was on using MEMS technology to develop electro-optical devices that can manage multiple wavelengths on a single optical fiber for telecommunications, notes Yariv Geller, the company's director of marketing and business development. "The work in the telecommunications field really lends itself nicely to spectroscopy because the wavelength ranges are nearly the same," Geller says.

The hand-sized DTS is designed to receive a spectrum from an external sample cell via an optical fiber. The incoming signal is dispersed across the MEMS chip so that each part of the chip interacts with a different wavelength. "Essentially, we have a programmable filter with which we can control the MEMS chip to block or reflect any combination of wavelengths to a single InGaAs detector, with spectra acquired in less than a second," Geller notes. The detector signals are then processed using a digital transform method.

Like Aspectrics' EP-IR, the DTS system is essentially a cross between an FTIR instrument and a grating instrument with an array detector, except it has no moving parts. One instrument covers the 900-1,700-nm range, while a second instrument covers the 1,700-2,500-nm range. A third high-resolution version is designed for specific narrow-band applications in the 1,100-1,300-nm range.

For now, the DTS instruments depend on using a separate light source and sample cell or probe, but Polychromix is developing accessories, such as a portable light source. "Ultimately, by developing accessories, we will be able to have a complete portable solution," Geller adds.

In general, the devices can be used for any application in which IR is currently used, Geller says. In May, Polychromix announced a partnership with Ocean Optics, Dunedin, Fla., to distribute the DTS instruments. Polychromix also is looking to work directly with process solution providers to incorporate the instruments into their offerings.

Geller doesn't believe the new devices will necessarily replace existing NIR or mid-IR systems, but he expects the new instruments to sell more for new applications. One example is measuring trans-fat content of foods, an important parameter now that the Food & Drug Administration has issued rules for listing of trans fat on nutritional labels beginning next January. "When a new regulation like this comes along, a company like Polychromix with a low-cost solution could have a tremendous advantage in the food and beverage industry," Geller says.

A promising technique for the direct and more complete analysis of the thousands of compounds naturally found in foods is Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS), according to chemistry professor Alan G. Marshall of Florida State University, Tallahassee. Marshall is a coinventor of the technique and founding director of the ICR program at the National High Magnetic Field Laboratory in Tallahassee.

FT-ICR MS offers much higher resolution than a standard mass spectrometer, allowing Marshall and his coworkers to distinguish between ions that differ in mass by as little as 0.0005 dalton--less than the mass of an electron.

"In effect, we are about 200 times better at separating peaks than the best single-stage gas chromatograph, high-performance liquid chromatograph, or gel technique. Plus, we can identify the elemental formulas of the peaks," Marshall points out.

The lab's flagship instrument uses an electrospray ion source operating in either positive- or negative-ion mode. Ions are initially collected in a trap just outside the magnet and then injected into an 11-cm-diameter cyclotron that sits inside the 22-cm bore of a 9.4-tesla superconducting magnet. That's the same magnet strength as that in 400-MHz proton nuclear magnetic resonance spectrometers. Each ion circulates inside the cyclotron at a frequency that depends on its mass. Like NMR, ICR MS uses a radio-frequency signal for excitation and detection of molecules.

Over the years, Marshall's group has further developed the technique and used it to analyze drugs, oligosaccharides, phospholipids, and peptides and proteins. More recently, he has turned to the compositional analysis of complex chemical mixtures, such as coal extracts, crude oil, gasoline and diesel fuels, and even wine. The high mass resolution of FT-ICR MS allows the "fingerprints" of different classes of compounds in these samples to be used to identify their origin and properties.

In one example, Marshall and Ryan P. Rodgers used the technique to resolve some 20,000 chemical species in a single spectrum of crude oil without having to first use standard chemical pretreatments or chromatographic separation. "Chemists take it for granted that it's necessary to separate and purify a chemical in order to identify it uniquely," Marshall says.

Because petroleum has about as many components as there are human genes, Marshall and Rodgers have used a spin-off of the word "genomics" to dub the analysis of crude oil and its distillates "petroleomics" (Acc. Chem. Res. 2004, 37, 53). In a similar vein, Marshall, Rodgers, and graduate student Zhigang Wu have taken on the analysis of the complex composition of vegetable oils, an area they are calling "oleomics."

"There is considerable health-related interest in the identity and levels of saturated and unsaturated fatty acids and their glycerides in food, especially vegetable oils used for cooking," Marshall notes. Additionally, the authenticity of oils can be important to consumers because of cost. For example, relatively expensive olive oil can be diluted with cheap olive oil or another oil.

With regard to safety, one need is the ability to quickly identify the presence of a toxic adulterant. A famous case a few years ago involved an olive oil diluted with canola oil that had been denatured with aniline for industrial use, Marshall relates. The oil was later linked to several hundred deaths in Europe.

Although all vegetable oils have similar components, such as di- and triglycerides, free fatty acids, and sterols, the mix of components and their relative abundances vary quite a bit, according to Marshall. "A key point about vegetable oils is that one can't simply rely on single-mass-unit resolution for analysis because there can be a dozen or more chemically different species all at the same nominal mass, and only one of them is the compound you think it might be. You really need the high resolution available with FT-ICR MS to figure it all out."

In one set of experiments, Marshall, Rodgers, and Wu dissolved canola, olive, and soybean oils purchased at a local supermarket in a chloroform-methanol solution for analysis (J. Agric. Food Chem. 2004, 52, 5322). In negative-ion mode, the instrument detected about 3,000 distinct components in the oils, and in positive-ion mode, about 2,000 components. The differences between the two modes reflect the acid and base properties of the compounds, Marshall notes. Carboxylic acids and tocopherols tend to deprotonate to become negatively charged; di- and triglycerides tend to be protonated and become positively charged, he says.

The team identified several markers that could be used to routinely characterize vegetable oils. For example, canola and olive oils have a higher proportion of C₁₈fatty acids with a single double bond, while soybean oil has a higher proportion of C₁₈ fatty acids with two double bonds.

In a proof-of-concept experiment, the researchers diluted olive oil with various amounts of soybean oil and analyzed the samples. Soybean oil can be readily distinguished even at low concentration by the presence of a fatty acid that doesn't occur in olive oil and by overall differences in glyceride abundances, they report.

"The tremendous resolving power of FT-ICR MS shows us that these oils are more complex than previously believed," notes Gary R. Takeoka, a research chemist at the Department of Agriculture's Western Regional Research Center in Albany, Calif. Takeoka co-organized a symposium on the authentication of food and wine last March at the American Chemical Society's national meeting in San Diego, in which Marshall presented the vegetable oil research.

"This technique gives us the ability to drastically reduce the overlap and interferences that characterize other chemical and spectroscopic analyses," Takeoka adds. "I believe FT-ICR MS has great potential as a new tool in detecting food adulteration."

One drawback is the cost, which currently is between $500,000 and $1.5 million per instrument, Marshall says. By his count, there are about 600 FT-ICR MS instruments worldwide. They have been commercially available from Bruker Daltonics and IonSpec Corp. for several years and more recently from Thermo Electron Corp., he notes. These instruments are targeted mostly for the study of peptides and proteins in the proteomics arena. Mass spectrometry is the fastest growing segment of commercial spectrometers, Marshall points out, and FT-ICR MS is growing twice as fast as mass spectrometry overall.

MARSHALL BELIEVES the technique could be very useful in food science and other areas to identify the critical markers needed to analyze a material, as his group has shown with wines and vegetable oils. "Then you would only need to test for those markers using a more routine technique," he says.

Quality is only part of the story in food science. Much research is aimed at improving methods for killing microorganisms that cause illnesses and ensuring that foods treated by these methods are safe before they enter the supply chain.

The key disease-causing culprits in foods are Listeria monocytogenes, Escherichia coli O157:H7, Salmonella enteritidis, Campylobacter jejuni, and Norwalk virus (Norovirus), notes microbiologist Richard H. Linton, director of the Center for Food Safety Engineering (CFSE) at Purdue University. CFSE uses teams of biochemists, microbiologists, physicists, and engineers to develop better approaches for detecting and preventing food-borne hazards, Linton says.

"Our focus is to be able to bring in a food sample, run it through a detection platform, and find out if it contains a pathogen and how much is there," he notes.

One of the current hot topics in food safety is finding a quicker method to detect L. monocytogenes. This bacterium can contaminate almost any food, from vegetables to ready-to-eat meats. It's responsible for about 2,500 cases of listeriosis in the U.S. each year. Although that number may not seem high, listeriosis has a 25% mortality rate, the highest rate among food pathogens, Linton says.

Commercial sterilization technologies use heat, high pressure, irradiation, and other methods to destroy pathogens during food preparation, Linton explains. But food can later be contaminated during packaging. Cooking the food will kill the bacteria, but ready-to-eat products like sandwich meats often aren't recooked by consumers.

Manufacturers pull some samples for testing and ship out the remaining product while test results are pending, Linton says. The volume of dairy products and meats is so large, though, that these products can't be sequestered for up to a week while test results are awaited. An additional factor is that the number of viable bacterial cells that can contaminate food following heat treatment and before packaging may be low and difficult to detect. That's a problem with L. monocytogenes, because as few as 100 lingering cells are thought to be enough to cause illness.

"We would like to develop a detection system that will give good information to the industry very quickly on the relative safety of a product," Linton says. "We want to be able to do this in a matter of minutes, rather than in days, which it now takes."

One approach being pursued by Purdue scientists and engineers working on CFSE projects is to develop microfluidic devices to concentrate bacterial cells for detection. Agricultural and biological engineer Michael R. Ladisch, microbiologist Arun K. Bhunia, electrical and computer engineer Rashid Bashir, and their coworkers have focused on using antibodies that specifically bind L. monocytogenes. Bashir's group has developed silicon-based microfluidic devices that allow for rapid detection of live bacterial cells by culturing them on a chip. This work, combined with development of low- conductivity growth media from Bhunia's lab and sample concentration and recovery techniques from Ladisch's group, has led to formation of a start-up company to commercialize the technologies.

Current food testing, which is very effective, involves taking a product and preparing about 250 mL of extract solution. If contaminated, the solution might contain fewer than 10 L. monocytogenes cells per mL, Ladisch says. Standard bioassay test kits used by the industry to test the solution require a high concentration of cells--up to about 1 million per mL. An enrichment culture is used to increase the number of bacteria in the sample, but this causes a delay.

"If you had a concentrated population of microorganisms that can be probed at a small volume, less than 1 µL on a chip, the test could be much faster," Ladisch says.

The researchers achieve this by concentrating the usual 250-mL extract down to about 500 µL, then loading it onto an antibody-based fiber-optic probe or microfluidic chip. The antibody specifically binds L. monocytogenes, trapping it inside the probe or chip. With the probe, the bacteria are detected in just a few hours by measuring a change in the optical signal. On the chip, the bacteria also are quickly detected by the electrical conductivity caused by the metabolic activity of the bacterial cells, a technique developed and patented by Bashir's group.

As the work moved forward, Ladisch and graduate student Tom T. Huang came up with a quick way to make test chips in the lab without needing time-consuming photolithography or stamping fabrication methods. In their "press-fit" method, glass fibers are placed on a glass slide or silicon wafer in a desired pattern, and then a small square of flexible poly(dimethylsiloxane) is "pressed" over the top of the fiber (Anal. Chem. 2005, 77, 3671). The polymer adheres to the glass, creating a tight seal around the fibers but leaving a small channel on either side for fluid flow. The hydrophilic glass fibers sandwiched between the hydrophobic top and bottom layers direct the flow of liquids through the device, but flow can also be controlled by a syringe pump.

The fibers can be positioned to create linear channels, T-shaped junctions, and right angles, Ladisch explains, and two fibers laid across one another can create well-like junctions. In addition, the fiber surfaces can be modified ahead of time by dip-coating them into solutions containing commercially available ion-exchange microbeads. Antibodies that selectively bind L. monocytogenes or other bacteria can be subsequently attached to the microbeads.

Press-fit devices effectively function as microscale liquid chromatography columns, and detection of bacteria can be carried out using fluorescent labels. It takes only about 15 minutes to make one of the devices.

"The idea with these chips is that you wouldn't use them for production, but only for research purposes," Ladisch says. "They enable you to make microfluidic channels with different chemistries for quick tests. Once you have a system you want, you can go back to design and fabricate a complete microchip and detection system." This is a general lab tool, he adds, and his group has not pursued a patent on the process.

This type of research being carried out in food science helps to motivate and excite young people to enter chemistry and technical fields, Ladisch believes. "Basically, it's a combination of molecular biology, nanotechnology, and microfluidics--bionanotechnology. But chemistry is what makes it work," he says. "If we can package this technology and communicate the excitement we have for it, then it will encourage students to pursue science and engineering."

SPLITTING HAIRS FT-ICR MS can resolve dozens of peaks at the same nominal mass, as shown in this expanded view of a negative-ion spectrum of an olive oil (top). Complete analysis of vegetable oils has yielded markers, such as the relative abundance of C₁₈ fatty acids with zero, one, two, or three double bonds (bottom), that can be used to check the purity of different oils.

Latest News RSS Feed