Vegetable oil speciation in processedfoods how legislation drives innovation

Transcription

1 Vegetable oil speciation in processedfoods how legislation drives innovation Dr Tassos Koidis Lecturer in Food Science and Nutrition 06/07/

2 legislation vsinnovation 06/07/

3 Food safety legislation is very strict 06/07/

4 Food labelling: the exemption The labels are complex and yet informative The EU has the stricter legislation in the world IN THERE THE REGULATOR PUTS SOME STRICT RULES THAT COMPANIES MUST ADHERE BUT NOT ALWAYS THE RESEARCH EXIST TO POLICE THE LEGISLATION PROPERLY 06/07/2016 4

5 New Regulation in place for vegetable oils sold as ingredients in processed foods EC Regulation 1169/2011, food businesses required to declare the composition of the oil mixture in the food product label. [Dec 2014: live] vegetable oils of interest: palm, sunflower, rapeseed; coconut oil Monitoring the compliance of the regulation is challenging for national bodies and other enforcers.

6 06/07/

7 06/07/

8 Can you find out what type of vegetableoil is present in processed foods? analytical method capable to identify oil species in the vegetable oil blends? 06/07/

9 What is the issue with palm oil? sustainable production? production raised 500% during the last 20 years Fair trade / fair wages? Deforestation In supermarkets,50% of all baked goods, confectionery, spreads, cosmetics, cleaning agents, air fresheners etc. the average first-world citizen consumes at least 10kg of palm oil each year! 06/07/

10 Fusion of spectroscopy and chromatography Staged procedure Fingerprinting Is to extract information about the whole of the sample product by analysing a wealth of variables that defines it which on their own they might not mean much but in systematically viewing them all together do supply a overview of the unique character (fingerprint) of the sample. 06/07/

11 Refined palm (12), sunflower (12), rapeseed (12) Pure oils (n=36) ADMIXTURES Pure PO, SUN and RO and admixtures on PO-RO, PO-SO oil (n=30 pure oils +48 admixtures) all oils used for baking (n=78) Different oils are mixed in clearly defined binary mixtures INDEPENDENT 1 Pure oils and admixtures (n=22 pure oils + 36 admixtures of these pure oils) Calibration set (n=58) INDEPENDENT 1 pure oils and admixtures (n=8 pure and 12 admixtures) Prediction set (n=20) Commercial biscuits with unique oil sources (21) and with blends (4) [as declared in the label] Validation set (n=25) market Osorio, M.T., Haughey, S., Elliott, C., Koidis, A.(2015).. Food Chemistry, 189, 67-73

12 IDENTIFICATION OF VEGETABLE OIL SPECIES IN BISCUITS

13 the use of vegetable oils in bakery products A crust that has a hard oil like palm stearin A creamy sauce that has a soft vegetable, fat, for example, palm olein, rapeseed oil or coconut oil pastry Crust: Hard fraction Palm Stearin filling: Soft fraction Palm Olein

14 Bakery/pastry products rich tea plain (no filling) digestive biscuits all butter other with filling cream filling cocoa filling bourbon 06/07/

15 Market study / Industry intelligence 9% In biscuits without fillings 12% 79% palm oil sunflower oil ropo In biscuits with fillings.. Common fillings include chocolate, bourbon and white cream (vanilla). Cream fillings Palm oil Occurance inproducts 100% 80% 60% 40% 20% 0% Oil species applied in chocolate biscuits 92% 42% 42% 17% 8% Palm cocoa butter coconut rapeseed 3% 13% 6% 19% 7% 52% palm oil, palm kernel oil, palm fat veg oil (Palm) veg oil (Palm); veg oil(palm olein, palm) palm kernel oil, palm oil, rapeseed oil vegetable oil(palm, coconut) 06/07/

16 Key features of the study 36 authentic vegetable oil samples 12 palm, 12 sunflower, 12 rapeseed oils trusted sources from the industry in house chemical analysis with official methods 48 admixtures of oils (ROPO, SOPO) 5 oil classes (PO, SO, ROPO, SOPO, RO) trained with supervised discriminative algorithms 3+ layers of validation 06/07/

17 Baking of biscuits FTIR Raman Extraction of fat from biscuits Spectra collection Data analysis RESULT: What oil your biscuit has? Classification algorithms

18 In-house baking

19 solvent extraction

20 spectra acquisitions raman ftir 1064 nm laser frequency cm -1 range 10 cm -1 resolution cm -1 range DTGS detector on d-atr number of scans: 32 resolution: /07/

21 cis-ch=chbending out of the plane -C-O stretch -C-O stretch; -CH 2 bending -C-O stretch -CH 2 bending -C=O stretch Symmetrical and asymmetrical stretching of CH 2 trans=c-h stretch Changes in the intensities of spectral bands of different oils are related to the composition of molecular bonds absorbing at those regions. -CH 3 bending FTIR spectra on pure oils cis-c=c stretch 6 pure oil samples: - 2 PKO - 2 PO - 1RO - 1 SO PKO: Palm Kernel Oil; PO: Palm oil; RO: Rapeseed oil; SO: Sunflower oil

22 Pre-processing steps FTIR Raman 06/07/

23 Model and class development and validation FTIR and RAMAN DATA TRAININGof the MODEL using calibration set VALIDATION SIMCA Soft Class Model of Class Analogy Creates a PCA space for each class and determines DtModel PLS-DA Patrial Least Square discriminant analysis supervised classification

24 PCA scores plot of FTIR data of different vegetable oils

25 PCA FTIR from oilsof in-house and commercial products CM:SO PORO RO PO CM:PO CM:PORO POSO SO

26 Classification results: for in-house samples SIMCA and PLS-DA model performance on prediction dataset using Raman and FTIR (26 samples from inhouse-baking biscuits for Raman and FTIR in separate) SIMCA PLS-DA TARGET GROUP SENSITIVITY (%) SPECIFICITY (%) OVERALL CLASSIFIC RATE (%) RAMAN FTIR RAMAN FTIR RAMAN FTIR PO PORO POSO SO RO TOTAL (%) 19/26 21/ PO PORO POSO 100 (a) SO RO TOTAL (%) 26/26 100% 21/ % for SIMCA we accept samples with Pconf>0.05 and for PLD-DA with Pconf>0.5; if no limits are applied. correct classification is at 100% at all cases

27 Gas Chromatography/Mass Spectroscopy

28 06/07/

29 Fatty acids profiles Oil species OIL SOURCES C14:0 C16:0 C18:0 C18:1c C18:2c C20:0 C22:0 In-house biscuits 1.3 ± ± ± ± ± ± ±0.0 PO Commercial bisc. 1.3 ± ± ± ± ± ± ±0.0 Untreated oils 1.1 ± ± ± ± ± ± ±0.0 SO In-house bisc. 0.1 ± ± ± ± ± ± ±0.0 Commercialbisc. 0.1 ± ± ± ± ± ± ±0.1 Untreated oils 0.1 ± ± ± ± ± ± ±0.0 RO Untreated oils 0.0 ± ± ± ± ± ± ±0.0 OO Untreated oil(lit.) 0.0 ± ± ± ± ± ± ±0.1

30 further challenge: to identify the type of oils used in confectionery products Fat extractionwith solvent (hexane, ethyl acetate) Palm components: -Confectionary fats -Cocoa butter replacers (CBR) -Cocoa butter equivalents (CBE) Analysis (screening, confirmation)

31 An interlaboratory trial with 14 participants having different brands of Fourier-transform infrared spectrometers for the identification of oil species in mixtures of oils 14 Participants P VALIDATION 9 oil samples (Pure oils and admixtures) PKO FTIR spectra Preprocessing RS P Classification model PKO PKO P RS an instrument agnostic FTIR method a different approach to model transfer 06/07/

32 Adulterants Reference samples by other instruments Other samples New reference sample Data augmentation in food authenticity: synthesizing spectroscopic data of vegetable oils for performance enhancement Blender Spectral Intensifier Shifting along x axis Adding Noise New samples In any grade (%) Any type of noise Absorbance Absorbance Absorbance Absorbance st sample 2nd sample New sample Wavenumber (cm -1 ) Initial spectrum New spectrum Wavenumber (cm -1 ) Initial spectrum New spectrum Wavenumber (cm -1 ) Initial spectrum New spectrum Wavenumber (cm -1 ) (a) (b) (c) (d) 06/07/

33 the team the group photo. Acknowledgements for funding most of the research shown today 06/07/