Handbook of Vanilla Science and Technology

A method with potential for routine analysis was published by TTB (Tax, Trade Bureau -US government, then BATF) researchers (Jagerdeo et al. 2000). They prescribed a simple HPLC method with good sensitivity and linearity for the desired components, vanillin, etc., to be employed for product surveillance. Several gradient and isocratic profiles were studied but were not disclosed. Waliszewski et al. (2007) developed a very rapid HPLC method for determining the vanillin content in vanilla extracts. Through a very intensive evaluation of columns, mobile phases, and detection wavelengths, a method was developed that could be completed in approximately 2 minutes using a Nucleosil C18 column, 60:40 methanol: water mobile phase, and detection at 231 nm with a gradient elution, as compared to the 7 to 36 minutes for other known methods.

An extension of AOAC Official Method 990.25, for the HPLC determination of vanillin, vanillic acid, p-hydroxybenzaldehyde, and p-hydroxybenzoic acid in fortified and imitation vanilla flavors, was developed in order to include ethyl vanillin. From this activity, Method 990.25 was modified (Kahan etal.1997) to include the determination of ethyl vanillin in vanilla products using a C₈ column, methanol-acidified water (10 + 90), and UV detection at 254 nm.

Chemists continue to develop methods to measure the chemicals expected in natural vanilla extracts, and also try to improve methods to detect other added flavor chemicals and even banned substances. An example is the more sophisticated LC-MS method developed by de Jager et al. (2007) at CFSAN-FDA (Center for Food Safety and Nutrition, US Food & Drug Administration). Specifically ethyl vanillin and coumarin were measured in 24 vanilla extracts obtained principally from the United States and Mexico. With this optimized method, which was compared to the routine UV detection method, the authors claimed a more accurate assessment and concluded that the samples were devoid of contamination. This is in contrast to an earlier study where nearly 68% of the Mexican samples contained coumarin (Thompson and Hoffmann 1988).

Jurgens (1981) used the ratios of key vanilloid constituents, notably vanillin/ p-hydroxybenazaldehyde, calculated from the HPLC analysis, to verify the authenticity of vanilla extracts. Several papers have appeared that reviewed the value of these ratios and have presented results that have either challenged or verified the utility of the measured constituents and their calculated ratios. Ehlers etal. (1999) reported on an investigation of the effects of extraction solvents, time of extraction, as well as extraction equipment. They noted that the quantity of vanillin did vary considerably with extraction conditions. Interestingly, the ratios of the components were not affected. However, because regulations especially in France require specific vanillin content as well as the correct ratios, they suggested that the French regulation should be revised. German researchers (Scharrer and Mosandl 2001) analyzed a variety of vanilla bean samples of different origins and years of harvest with typical aqueous ethanol and diethyl ether extractions for the four vanilloid components, which were analyzed by HPLC and GC. Ethyl vanillin and veratraldehyde were used as internal standards when developing the two methods. Complete constituent tables were included for the GC and HPLC results. From these analyses they too believed that the ratios currently being used were too restrictive.

French researchers did an extensive study of the 2000 vanilla crop (Charvet and Derbesy 2001). They analyzed 265 vanilla bean samples using HPLC, including those from Madagascar and Comores. The data they obtained on the four key vanilla components were compared to data for other samples collected over the previous five years. They noted the decreasing content of the two aldehydes, vanillin and p-hydroxybenzaldehyde, while the content of the two acids seemed nearly constant. A statistical assessment was not presented beyond the charts used to depict the trends.

Further investigation into the applicability of the vanilloid ratios was extended to those vanilla components measured in dairy products (Littmann-Nienstedt and Ehlers 2005). Changes can occur in the ratios when vanilla is used in dairy products. Milk enzymes stoichiometrically convert vanillin and p-hydroxybenzaldehyde to their respective acids. Therefore, we need to take these potential conversions into account when calculating the vanilla constituent ratios in dairy products. The authors presented an adjusted formula that could be useful for these products.