Techniques - Stable Isotope Ratio Analysis (SIRA)

Biological and Geological Isotopes

Isotopes are different forms of the same element that contain equal numbers of protons but different numbers of neutrons.  Hence they differ in their atomic weight but not in chemical properties. This difference in atomic weight means they can be separated, and measured, using mass spectrometry.

Most elements have naturally occurring heavier isotopes in very small amounts e.g. carbon, consists of approximately 99% ¹²C, 1% stable ¹³C, a tiny amount of radioactive ¹⁴C. It is the change in proportion of the heavier minor isotopes that serves as authenticity markers.

Authenticity markers can be divided into biological (H, C, N, O, S) and geological isotopes (Sr). Isotope ratios are measured in relation to an international standard.

Hydrogen and Oxygen Isotope Ratios

The change in ratio of hydrogen (²H deuterium) and oxygen ¹⁸O isotopes can indicate geographical origin. Seawater contains a very small natural proportion of D₂O and H₂O¹⁸. As seawater evaporates and falls as rainfall, the D₂O and H₂O¹⁸ are slowly fractionated; because they are slightly heavier, the D₂O and H₂O¹⁸ rainfall tends to fall before H₂O¹⁶. Thus, a higher ratio of D/H and ¹⁸O/¹⁶O builds up where the rain falls first, and then is depleted as the rain moves further inland. The plants and animals then take these higher heavier ratios from the rain and ground water, which will distinguish them from plants and animals grown further inland.

Nitrogen Isotope Ratios

The nitrogen isotopic ratio in plants can indicate the type of fertiliser used. Synthetic fertilisers are produced from the nitrogen in the air, which only has 0.37% ¹⁵N.  Organic fertilisers (manure, fungal action etc) enrich the ¹⁵N content.

Carbon Isotope Ratios

Carbon ¹³C/¹²C ratios are dependent on the biosynthesis mechanism of the two main groups of plant – C₃ and C₄ plants. Most plants are C₃ and are lower in the heavier ¹³C isotope . Maize and sugar cane, which could be used to adulterate honey and fruit juice, are C₄ plants and have higher ¹³C/¹²C ratios. Sugar beet is a C₃ plant, and hence requires special methodology to distinguish it from the intrinsic sugars in grape must, honey or fruit juice. (Fortunately, the third group of plants - CAM - are rare and unlikely to be used as adulterants; pineapple sugar is too expensive, and agave is more profitably used for tequila production)

Abundance of heavier Carbon Isotope ¹³C (as % total carbon)

Stontium Isotope Ratios

The strontium isotopic ratio ⁸⁷Sr/⁸⁶Sr reflects the geological structure underneath and around the land used for growing crops or rearing animals, and so can be indicative of the geographical origin.

Using Stable Isotopes to Build an Authenticity Classification Database

A database of isotope ratio profiles is constructed from  "authentic" reference samples.  A mathematical classification model is then derived, which can involve both advanced statistics ("chemometrics") and Machine Learning.  These models can be based on a single isotope ratio or a combination of elements.  They can incorporate other multi-dimensional analytical data such as trace metal or mineral concentrations.

Test samples are then compared with this classification model to check if they are consistent with the reference samples.

The classification model can either be:

• One class (i.e. "Is the sample part of this authentic population, or is it not?")
• Multi-class (i.e. "Which of these population groups does the sample fall into?")

 Example - Two-class Model for Farmed vs Wild-Caught Sea Bass using SIRA combined with Fatty Acid Profile

Cabon Isotopes Image Copyright: Food Authenticity Network.

Sea Bass Image Copyrighti: Crown

 Isotopes Used in Authenticity Analysis

Each application (test) requires a bespoke database.  Databases, typically, are not transferrable between laboratories.  Thousands of databases have been constructed, covering everything from the verification of "UK" pork to the regional origin of premium Chinese teas, and from sugar addition to honey to the feeding regime of premium meat (e.g. "corn-fed" chicken or "grass-fed" beef).  There is an index of some databases on the FAN resources site.

Example Applications

1.  Honey Sugar-Addition (C₄ sugars) Using EA-IRMS (Elemental Analyser- Isotope Ratio MS)

Most honey contains complex sugars produced from bees foraging on C₃ plants, whereas two cheap industrial sugars derive from maize (corn) and sugar cane, which are C₄ plants. The AOAC 998.12 method compares the carbon isotope ratio in the whole honey (i.e. mainly driven by the sugars, vulnerable to adulteration) to that from the extracted protein fraction (i.e. likely to be independent of any adulteration).  The elemental analyser combusts the protein and whole honey samples to CO₂, which then enters the mass spectrometer, so that the δ¹³carbon isotope ratio can be calculated and compared. If the δ¹³C of the whole honey sample differs from the protein δ¹³C, then a C₄ sugar syrup has been added. The AOAC method can detect C₄ sugar syrups at levels ≥ 7%, but it cannot detect C₃ sugar addition such as rice syrups, invert beet sugar, or fructose-glucose syrups derived from wheat. The method has been modified such that there is a double extraction of proteins from the honey, the second one removes pollen as well. This improves the method and also indicates where sugar feeding of the bees has taken place.

2.   Honey Sugar-Addition (C₃ Sugars) Using EA/LC-IRMS (Liquid Chromatography-IRMS)

In this method, adopted by JRC, the honey is first diluted and separated into its component saccharides i.e., fructose, glucose, disaccharides and trisaccharides by high performance liquid chromatography (HPLC). The separated saccharides are then oxidised to convert them to CO₂ before entering the mass spectrometer. The carbon isotope ratio of each individual sugar componenet is measured. 

The JRC have proposed a complex matrix of isotope ratio limits as an acceptance boundary for the purity of honey.  These were based on a collaborative trial in 2020 on the LC-IRMS combined with the EA-IRMS.  This concluded that it was only necessary to measure the carbon isotopic ratio of the mono, di and tri-saccharides to determine whether a C₃ sugar had been added in amounts >10%.T here was not enough evidence that measuring % peak  area of the oligosaccharide peak is a useful parameter for sugar adulteration without further work, nor that more complex isotopic measurements were sufficient for this purpose.  Also, the levels of oligosaccharides in sugar syrups and even in honey itself are very variable, and hence it is not a consistent authenticity marker to be of universal use. Even in syrups derived from starch e.g., rice starch, the oligosaccharide peak only is measurable with high levels of addition of syrup.

The technique is universal and is based on fundamental principles and properties of food production systems.  It is difficult to fake.  It is particularly strong when testing an ingredient or raw material that is well-characterised and consistent and the question is "Is this batch different?".

The degree of confidence is always based on the size and relevance of the database and whether it is keeping pace with agronomic, seasonal, and other changes.  The database must be fully representative of the test sample, including all of the agricultural inputs.  If these inputs change the database may no longer be valid.

Isotopic signatures in the final product can change quickly as inputs change.  For example, changing an animal's feed ration for a few weeks will change the isotopic signature in meat, eggs or milk.  This makes results difficult to interpret for animals which switch between rations, or for short-lived animals such as broiler chickens.

Methods are not quantitative and tend to be better for major components of samples rather than mixtures or blends at low proportions.  For example, the JRC method for honey can only detect C₃ sugar addition above 10%.

"Geographic Origin" of isotopic signature relates to general topography.  This rarely coincides neatly with national boundaries.  So differentiating between a South American and Spanish orange is much easier than between a Dutch and French apple.

Results are rarely clear-cut.  Laboratories usually report in the format "Results were consistent with ..........our database of authentic samples" or "Results were not consistent with ............."  Decision boundaries are blurred.  Variables that determine isotope rations are complex and there may be alternate (innocent) explanations why the test sample does not match the database.

 Example - Wild-caught vs Farmed Sea Bass Using SIRA combined with Tryglyceride Profile

Circular datapoints are the reference database.  It can be seen that there is an overlap of the two populations. Sample results falling in this overlab could not be classified.

Triangular datapoints are retail test samples declared as "wild".  Most samples lie within the boundaries of the wild seabass. Two samples (GL205 and DS209) definitely are in the farmed population, and a third is in the overlap area on the very edge of the 95% confidence limit, and therefore would need confirmatory traceability evidence that it is not farmed.

There is one outlier of the wild sea bass population, which illustrates the importance of having a comprehensive database with fish caught in different areas and regions of the Atlantic and Mediterranean seas. Similarly, one of the market samples (BT204) is just outside the farmed population, again indicating that it is an outlier, but maybe the database needs augmenting. There are 8 samples well outside the database population of wild sea bass, but they are also well away from the farmed population. Again, this probably indicates that because of diet differences, there may well be other populations of wild sea bass outside of the wild database.

Image - Crown Copyright