NMR Analytics

Detecting honey adulteration with NMR spectroscopy

Adulteration of honey is a growing problem for the entire industry. The conventional way to confirm the authenticity of honey is a quite tedious and complex one: Every type of adulteration has to be detected individually. By using NMR spectroscopy (Nuclear Magnetic Resonance), Quality Services International GmbH (QSI – Bremen, Germany), part of a consortium comprising Bruker BioSpin GmbH and ALNuMed GmbH, established a completely new performance standard in honey analysis. Already demonstrating a range of advantages compared to conventional methods, honey-profiling has the potential to make analysis even more reliable and exact in the future.

 

The problem with conventional honey testing methods

Adding syrup is a common way of honey adulteration, and a serious problem in times of the world market’s increasing demand for the natural product. Conventional testing methods mostly focus on the detection of certain marker compounds. As already mentioned, the problem is that an analysis for each type of adulteration is mandatory: E.g. you need 13C-analysis to detect C4-sugars that do not commonly occur in honey.

Unfortunately, the absence of marker compounds does not necessarily mean that the honey is free from adulteration. Some producers are able to create syrups used for adulteration that do not show the classical marker compounds which can be detected by conventional methods. As the conventional procedure may get bypassed, it got more than necessary to develop another way to prove honey authenticity.

 

The new approach: NMR profiling of honey

For QSI, the solution was NMR spectroscopy. Used over decades for structural analysis and quantification, it has become a reliable method to confirm the authenticity of food. As NMR spectroscopy has already worked out for juice and wine in the past, it was the project’s aim to use it as well for honey profiling. This includes botanical as well as geographical aspects and of course detecting adulteration – and with this food fraud.

The Bruker Ascend 400 FoodScreenerTM, developed by Bruker BioSpin GmbH (Rheinstetten, Germany), a standardized platform for food analysis, collects all samples that have been taken by QSI and ALNuMed GmbH (Bayreuth, Germany) from 2013 up to now. By today, over 4.000 samples from more than 45 countries have been analysed and provide a comprehensive base for reliable honey profiling. With each further analysis the data base evolves and gets expanded.

The reference database contains samples as well from polyfloral honeys as well as from honey dew and several monofloral honeys, for example: acacia, chestnut, linden, lychee, manuka, dandelion, lavender, rape and buckwheat. Most of them have their own spectroscopic fingerprint.

 

Confirming honey authenticity with NMR – how it works

In order to prove the authenticity of the references, the database was filled with a range of results from conventional methods. These included pollen analysis, organoleptic tests, sugar spectrums, analysis of 13C-ratios, detection of foreign enzymes and others. Furthermore, a variety of syrup samples was collected and adulteration experiments with different syrups were conducted.

With the evaluation of several thousand samples and variables, it was possible to visualize a quantile plot. By superposing it with the test sample, potential changes from normality can be detected almost on the first sight. Fig. 1 shows the spectrum of an authentic honey: The spectrum of the sample (bold black line) lies within or nearby the red band – no peculiarities are shown.

 

Figure 1: NMR spectra of authentic honeys (quantile plot) overlaid with an authentic honey (bold black line). First row: aliphatic region, second row: sugar region, third row: aromatic region.
Figure 1: NMR spectra of authentic honeys (quantile plot) overlaid with an authentic honey (bold black line). First row: aliphatic region, second row: sugar region, third row: aromatic region.

 

While the red band shows the most frequently detected signal intensities in authentic samples, the blue band represents the less commonly ones. If the honey is adulterated, it usually shows untypical features in the NMR spectrum, as seen in Fig. 2: Strong deviations from the reference database can be constituted, which declares the sample as quite suspicious. 10% rice syrup QSI added for demonstration purposes is to be made responsible for this obvious result.

 

Figure 2: Detail of NMR spectrum (5-5.5 ppm) of a honey spiked with 10% rice syrup (bold black line) compared to the reference database of authentic samples (coloured area).
Figure 2: Detail of NMR spectrum (5-5.5 ppm) of a honey spiked with 10% rice syrup (bold black line) compared to the reference database of authentic samples (coloured area).

 

The automated analysis is not only very efficient but also reproducible (see Fig.3), which is an important property of honey profiling.

However, extensive experience in the field of honey spectral analysis is mandatory. As well as a review of the results by using conventional methods is highly recommended for borderline cases – but this is also necessary in classical analytics.

 

Figure 3: Comparison of NMR spectra of the same honey sample measured repeatedly over a several days.
Figure 3: Comparison of NMR spectra of the same honey sample measured repeatedly over a several days.

 

Subsets of samples for proving honey authenticity

The capability of NMR spectroscopy to detect deviations is quite remarkable already. But in order to get even more precise results, statistical analysis is a powerful tool. Therefore, different models for a number of botanical and geographical origins have been created and are constantly extended. One example is a model for acacia honey. This model refers only to authentic acacia honeys in the database. Then again the geographical origin is verified by choosing a model of the origin in question. Thus, the declaration of origin made by the supplier, e.g. Mexico, can be verified by choosing the appropriate model, which is in this example the model for the geographical origin Mexico.

While in the mentioned cases the statistical models already exist, development of many other models from botanical and geographical origins is currently in process.

This is important as some spectra of monofloral honeys may lie outside the spectra of polyfloral honeys, although they are not adulterated. This is the case with linden honey for example, as shown in Fig. 4. As the flora in different countries may vary considerably, it is also important to maximize the number of models for geographical origins.

 

Fig. 4: Linden honey has a profile clearly different from all other types of mono- and polyfloral honeys (as can be seen in the overlay). This explains the need of having a normality model for linden honey (already part of the first release), so the fully automatic analysis can identify Linden honey and does not deliver false positive results due to the distinct deviation from spectra of polyfloral honeys. In this case the experienced user can identify deviations, however there are other examples, where only statistical classification can find the differences.
Fig. 4: Linden honey has a profile clearly different from all other types of mono- and polyfloral honeys (as can be seen in the overlay). This explains the need of having a normality model for linden honey (already part of the first release), so the fully automatic analysis can identify Linden honey and does not deliver false positive results due to the distinct deviation from spectra of polyfloral honeys. In this case the experienced user can identify deviations, however there are other examples, where only statistical classification can find the differences.

 

Honey profiling in the future

It is possible that further improvements of NMR spectroscopy will concern the ripeness of honey. The fact that especially Chinese honey is usually harvested very unripe, provokes deviations from the spectrum of authentic, ripe honeys. Therefore, it is likely that with NMR spectroscopy and statistical analysis we will be able to distinguish a lack of ripeness from adulteration in the future.

Meanwhile, QSI plans to extend the analyses with NMR spectroscopy from honey up to other substances like oils and essential oils and to detect 16-O-Methylcafestol.

 

Advantages of honey profiling with NMR

The advantages of honey profiling at a glance:

  • a variety of information with just one measurement
  • reproducibility and comparability
  • short measurement times
  • comprehensive statistical analysis
    • target analysis (quantification)
    • non target analysis (statistical classification and verification over the entire spectrum)
  • retrospective identification and quantification
  • determination of:
    • Adulteration
    • Processing steps
  • Geographical and botanical origin
  • detection of current and future methods of adulteration
  • high statistical confidence level owed to the development of a comprehensive and rapidly growing database consisting of already several thousand authentic samples analyzed by a multitude of conventional methods
  • simultaneous quantification of a broad set of parameters (most comprehensive collection of data worldwide)
  • retrospective evaluation/quantification of past data sets possible, as always the entire spectrum is acquired and stored
 

qsi_phone

Personal Consultation

+49 – (0)421 – 5966070

Mon. – Fri. 8.00 a.m. – 5.00 p.m. (CET)

Fax: +49 – (0)421 – 59 66 07 19

info@qsi-q3.de

 

© QSI - Quality Services International GmbH All Rights Reserved