Food contamination has made the headlines in recent years. Over the past decade, 36 significant contaminations have been collated on Wikipedia with these prompting anything from an international recall, to share prices falling, to (in at least two recent examples) death sentences being handed out.

This month (September 2018) sees the 10th anniversary of what the World Health Organisation described as one of the largest food safety events it has had to deal with in recent times, with baby milk in China found to have been adulterated with the flame-retardant chemical melamine. 300,000 were affected, 54,000 were hospitalised, and six babies died. The contamination led to the execution of two and multiple prison sentences.

Here in the UK, it’s just 5 years since the horsemeat scandal, which affected multiple supermarket chains. Tesco, for example, had £300 million (c.400 million USD) wiped from its share price after horsemeat was used by its suppliers (without label) to bulk up beef in several products including economy burgers (29% horsemeat) and spaghetti bolognese (60%).

And while one could argue that horsemeat is a safe ingredient (it’s taboo in the UK but eaten throughout Europe), even at this level consumer fraud and commercial malpractice has significant consequences. The lack of labeling gives concerns for individuals allergic to particular meat species, or those with religious taboos or ethical aversions. Notably the same year (2013) also saw halal lamb burgers contaminated with pork served to school children in Leicester – a city with a very large Muslim population.

In America, government data shows recalls more than doubled between 2004 and 2014 (288 to 659), with a 2012 study by the US Food Marketing Institute and the Grocery Manufacturers Association stating that the average direct costs of a recall was $10 million plus the effect of reduced sales. And while, according to the insurance firm Swiss Re, three quarters (73%) of recalls in the US take place for microbiological contamination or labelling issues, 14% is for a physical / chemical contamination, or the use of unapproved ingredients.

Fig 1: Food contamination incidents per year since 2001. Note the rapid drop after 2008 / 9 and 2013 which saw major international food scandals reported in the news

With food prices being driven up by climate change, the gains for (and therefore the risk of) organisations using bulking ingredients can only increase.

Infrared sensors in food analysis

Several academic papers have examined the use of IR for food adulteration – notably Kamruzzaman et al.’s 2015 paper in the journal Food and Bioprocess Technology on detecting horsemeat adulteration in minced beef. This showed how the addition of horsemeat increased the absorbance at specific wavelengths between 400 and 1000 nm (see fig 2).

Fig 2 – Absorbance spectral features of minced beef meat adulterated with horsemeat. The arrow indicates the direction of increasing the level of adulteration. The most significant absorption wavelengths are highlighted – reproduced with permission from Kamruzzaman et al., 2015

With the vast majority of materials giving off a form of thermal radiation in the IR spectrum, and also absorbing different wavelengths of infrared energy, it is possible to capture a large amount of information about a food sample using IR sensors. IR energy excites the covalent bonds in a molecule, with the resulting vibrational modes being a unique identifier for a specific chemical structure – giving food inspectors a fingerprint for identification.

This is true for samples in all states, be it solid, liquid or gas. Food producers (as well as inspectors) can use IR to monitor in considerable detail for contaminants or additives, measuring not just the presence of a specific element, but a chemical’s atomic structure and molecular interactions, giving an overall sample composition and relative concentrations. And while several forms of IR can be used to identify food adulteration (Kamruzzaman et al.’s analysis, for example, uses visible near-infrared hyperspectral imaging), the interaction of mid-IR radiation with a given sample has been shown to provide the most useful spectral fingerprint.

It should be noted that near-IR spectroscopy can be useful for non-homogeneous samples and trace analysis, however in comparison with mid-IR spectroscopy, the spectra derived from the fundamental bands measured are lower in intensity and more convoluted than the overtone/harmonic bands derived from mid-IR radiation. Additionally, the range of applications for mid-IR sensors are increasing with mid-IR devices becoming available for some trace detection requirements.

Types of sensors

Sensors can be either passive or active. Active (or emitting) photodetectors (also known as quantum detectors) are typically used for near-IR sensing. They have faster response times and greater sensitivity than passive (detection only) thermal detectors that require cooling to cut thermal noise and by emitting, they influence the samples they are measuring. As such, passive pyroelectric sensors, which convert changes in temperature from IR radiation into an electric signal, are particularly useful in food analysis.

Considerations in choosing the sensors for IR spectrometry

The way food is inspected varies from country to country and even in the US it is based on which organisation is responsible. The USDA (Department of Agriculture) regulates meat, poultry, and processed egg products and has an agent in every processing plant in the U.S; the FDA (Food and Drug Administration) minds everything else but requires no testing of food from manufacturers.

As Fortune put it in 2016, “it is hard to conceive of a more haphazard regulatory regime than the one intended to protect Americans’ food,” and quoting Secretary of Agriculture (USDA) Tom Vilsack’s, example of why pizza highlights this problem: “If it’s a plain pizza with tomato sauce, mozzarella, crust, it’s the FDA. If it’s a slice of pepperoni, it’s mine. Sausage, it’s mine. Mushroom, it’s theirs.”

But there are some common principles for engineers to follow when specifying a system, especially given that for food inspection (and indeed many applications), portability is a key factor:

  • Use smaller sensor processes to achieve denser, higher-resolution arrays and modules.
  • Sensors need to be physically robust to cope with shock, vibration and high operating temperatures.
  • Systems should be easy to set up without calibration, giving stable, accurate operation, with no degradation
  • The sensor should have a high signal-to-noise ratio, responsivity, and sensitivity
  • Sensor technologies that require cooling should be avoided
  • Ensure low-power technologies are used for portable systems
  • Modular packaging options will simplify systems integration

Passive pyroelectric sensors made from piezoelectric materials (see Pyreos offering here) match these specifications and come with three additional benefits: they’re manufactured in a standard semiconductor foundry, so take advantage of competitive process options and are economical, they’re scalable for both high and low volumes, and the technology has significant R&D taking place, giving improved performance.

Additional elements in a system

Several additional components will be required when developing an IR sensor for food inspection. Firstly, an IR sensor measures small changes, so electrical voltages can be very small. An integrated amplifier (or pre-amplifier) with an internal voltage regulator is therefore typically required. Additionally, if output is an analogue signal (outputs can be either or both digital / analogue), the addition of an A2D converter will also be needed.

Most IR sensors will also require integration with an MCU and this can either be built into the sensor package, or an external one, interfaced via I2C, USB etc. Finally, to enable a sensor to focus on specific frequencies or wavelengths, an optical filter or lens can be placed in front of the sensor. And, an optical band-pass filter can be used to limit spectral response, block certain radiation wavelengths, to avoid false alarms, for example, or to detect specific substances.

Conclusion

Food adulteration will continue to be an issue as food chains get complex and the benefits of either lax practices or purposeful malfeasance outweigh the risks of getting caught. As climate change causes the cost of food to increase, the benefits are set to increase, and government bodies need to escalate the risks too.

Mid-IR sensors based on PZT pyroelectric materials gives the most cost-efficient way to deliver this in a small, rugged and low-power system.

References:

Kamruzzaman et al, 2015, Food and Bioprocess Technology: Assessment of Visible Near-Infrared Hyperspectral Imaging as a Tool for Detection of Horsemeat Adulteration in Minced Beef

DOI: 10.1007/s11947-015-1470-7