It was very interesting to see sulphur dioxide appear in the news recently. No less a body than the LGC described the accurate detection of SO₂ at low levels in a difficult matrix as “the most challenging investigation” within its Government Chemist Annual Review for 2016. The work was carried out as part of the referee function of the LGC as it examined foods containing components such as garlic, known to present positive interferences to standard test methods.

When used as a food preservative, sulphur dioxide is added in the form of a sulphite or metabisulphite and there is no doubt that testing for sulphite residues can be markedly affected by interfering species. Since the official acceptance of the Monier-Williams procedure back in the 1920s, the story of sulphur dioxide testing is one littered with attempts to deal with this very problem. But perhaps, as someone might once have said, it would be best to start at the very beginning.

Sulphur dioxide itself is an unpleasantly toxic gas. It is produced within nature by volcanoes and forest fires, and also industrially by the burning of fossil fuels. It is a significant contributor to acid rain. All of this is probably very bad. However, it is a wonderfully effective anti-microbial, particularly against yeast and moulds, and is used extensively both as a food preservative and also during the production of wine and beer. All of this might be considered very good. As with so many chemicals, it is the dose that makes the poison – and I must give credit to Paracelsus for that one.

The other problem with sulphur dioxide is that it can provoke allergenic-type reactions in sensitive individuals, particularly those with asthmatic tendencies. This reaction can be quite severe even when the dose is quite low. Therefore, although one would struggle to categorise SO₂ as an allergen in the strict sense of the term, it is classified within EU food labelling law as a declarable allergen. Along with gluten, it is also exceptional as having a lower limit below which it can be considered as absent; in the case of sulphur dioxide that limit is 10 mg/kg.

This takes us to the hub of the problems associated with testing for sulphur dioxide. Firstly, there are legislative limits for its use within food as a useful food additive. These can range from a maximum of 2000 mg/kg in dried apricots, down to a maximum of 10 mg/kg in table grapes. Secondly, if there is any present it must be declared as an allergen, but only when the level is greater than 10 mg/kg. This gives a significant analytical requirement of accuracy and precision at higher levels, but an extreme analytical requirement at and around the 10 mg/kg level.

At this point it is only fair to recognise that much of the testing for sulphur dioxide in food is relatively routine, and certainly fit for purpose. There are many published procedures that can be used on many products without any apprehension. However, when presented with a new or difficult matrix then many standard approaches may be compromised, and the idea of a one-size-fits-all analytical approach may be flawed.

So how does the diligent and conscientious analytical chemist approach this challenge? The vast majority of analytical methods are, at least in part, based upon the afore-mentioned Monier-Williams method. This procedure usually requires a food sample to be gently boiled in acidic solution under a condenser and with a low flow of nitrogen gas as a carrier. This should allow the analyst to distil off the sulphur dioxide gas whilst retaining other volatiles. The sulphur dioxide is usually collected in a receiver solution of dilute hydrogen peroxide, thus forming sulphuric acid which can be easily titrated using standardised sodium hydroxide solution.

The obvious problem is the “whilst retaining other volatiles” part. If any other acidic volatiles are not retained then false positives will be achieved. There are many strategies that can be used to sidestep this issue. The final titrated solution can be treated with barium chloride, and the resulting barium sulphate determined gravimetrically – but this takes some time, and limits of detection will be an issue. The receiver solution can be replaced with standardised iodine, and the reducing power of sulphur dioxide can be measured by titration – but just as acidic volatiles can be a problem, so can volatiles with redox potential.

The selection of acidification reagent itself may come into play: hydrochloric acid releases sulphur dioxide very well, but if the distillation is too vigorous it may be encouraged into the gas phase itself and become an interferent; alternatives such as phosphoric acid may not enter the gas phase, but could also struggle to release the SO₂, particularly in high-sugar products. In addition, the distillation solution can be diluted with methanol in order to reduce the boiling point and so aid in the retention of the volatiles – but some of those volatiles, particularly those associated with onions and garlic, seem just as flighty as SO₂ itself. Consequently, it is clear that it is possible to adjust and optimise almost every part of the test procedure. This can even include sample handling; vigorous mechanical homogenisation of sulphated apricots can lead to low recoveries – the inescapable laws of thermodynamics releasing SO₂ within the blender.

In light of all this, perhaps it is not surprising that testing for sulphur dioxide residues can be such a challenge. It is certainly interesting that the LGC had to turn to the ultimate analytical tool in the form of mass spectrometry to resolve their analytical problem. There are methods using ion chromatography post-distillation (or even without distillation) that have been well documented, and these are certainly both specific and sensitive enough in most instances. As a chemist who enjoys the slightly black-art of IC then it would have been my first port of call. However, the fact that the LGC was required to move further into liquid chromatography-mass spectrometry not only highlights the analytical difficulty, but also the stimulating and interesting puzzles that can be presented to the food analyst.