13. Appendix 1SOP FA0117, Version 1.0

STANDARD OPERATING PROCEDURE (SOP)

Version 1.0, August, 2013

STANDARD OPERATING PROCEDURE FOR DEVELOPMENT OF A TWO-STEP METHODOLOGY TO DETERMINE VEGETABLE OIL SPECIES IN VEGETABLE OIL MIXTURES

Prepared byDr Tassos Koidis, Queen’s University Belfast, Date 31/8/13

Approved byAuthenticity Methods Working GroupDate 27/11/2013

Page 1 of 38

13. Appendix 1SOP FA0117, Version 1.0

CONTENTS

1.HISTORY / BACKGROUND...... 4

1.1Background...... 4

2.PURPOSE...... 4

3.SCOPE...... 4

4.DEFINITIONS AND ABBREVIATIONS...... 4

5.Principle of the method...... 5

6.MATERIALS AND EQUIPMENT...... 6

6.1Chemicals...... 6

6.2Water...... 6

6.3Solutions, standards and reference materials...... 6

6.4Commercial kits...... 6

6.5Plasticware...... 6

6.6Glassware...... 6

6.7Equipment...... 7

7.PROCEDURES...... 8

7.1Sample preparation...... 8

7.1.1 Preparation of binary/ternary admixtures...... 8

7.2FTIR spectra acquisition...... 8

7.3Raman spectra acquisition...... 11

7.4Analysis of fatty acids...... 13

7.4.1 Methylation of fatty acids...... 13

7.4.1.1 Preparation of reagents...... 13

7.4.1.2 Preparation of fatty acid methyl esters (FAMEs)...... 14

7.4.2 Chromatographic analysis of FAMEs...... 14

7.5 Quality assurance...... 15

7.5.1 Sample preparation...... 15

7.5.2 Spectroscopic analysis...... 15

7.5.3 Fatty acid analysis...... 15

8.CALCULATIONS AND DATA ANALYSIS...... 16

8.1 Data analysis of spectroscopic data (FTIR and Raman)...... 16

8.1.1 Spectral data handling and spectral pre-processing...... 16

8.1.1.1 Introduction of raw spectral data into an excel file...... 16

8.1.1.2 Application of spectral filters...... 16

8.1.2 Building and validating calibration models...... 20

8.1.2.1 Principal Component Analysis (PCA) (optional)...... 20

8.1.2.2 Soft Independent Modeling of Class Analogy (SIMCA)...... 21

8.1.2.2.1 Building of SIMCA calibration models...... 21

8.1.2.2.2 SIMCA models- prediction and validation...... 22

8.1.2.3 Partial Least Square-Discriminant Analysis (PLS-DA)...... 26

8.1.2.3.1 Building of PLS-DA calibration models...... 26

8.1.2.3.2 PLS-DA models- Prediction and validation...... 27

8.1.3 Quantification of oil species in oil admixtures...... 28

8.1.3.1 Building of quantification models...... 28

8.1.3.2 Quantifying the oil species of unknown oil admixtures...... 30

8.2 Fatty acid calculations...... 33

8.3 Stage procedure and criteria for decision making of an unknown sample...... 34

8.3.1 Screening step: FTIR spectrum of an unknown sample...... 35

8.3.1.1 FTIR spectral acquisition...... 35

8.3.1.2 Spectral data handling...... 35

8.3.1.3 Classification in existing models and decision making...... 35

8.3.1.4 Referral procedure...... 35

8.3.2 Confirmation stage: Chromatographic analyisis of an unknown sample...... 36

8.3.2.1 Sample preparation and analysis...... 36

8.3.2.2 Results and calculations...... 36

8.3.2.3 Fatty acid classification criteria of an unknown sample...... 37

8.3.3 Quantification of oil species in certain oil admixtures...... 38

9.RELATED PROCEDURES...... 38

10.ESSENTIAL REFERENCES...... 38

11.APPENDICES...... 38

Page 1 of 38

13. Appendix 1SOP FA0117, Version 1.0

1.HISTORY / BACKGROUND

1.1Background

It is common practice for food manufacturers to use refined vegetable oil mixtures as ingredients in confectionary, pastry, bakery and other food products. These mixtures are mostly composed of refined palm oil, sunflower oil and to lesser extents rapeseed, corn, coconut, cottonseed oils. Palm oil, the largest volume oil imported into UK, is used in high amounts. Until very recently there was no requirement for manufacturers to state the composition of the mixture and it was labelled under the generic term “vegetable oil”. With the recent EC1169/2011 regulation regarding vegetable oil labelling, the composition of vegetable oil must be declared in the label (European Commission, 2011). Food manufacturers must comply with this new requirement, although normally this is not a challenge for the industry as information on composition should be available via the product specification provided from their oil suppliers. It presents, however, a challenge for the legislation and enforcing authorities such as DEFRA amongst others, to monitor compliance of EU legislation and correctly labelled foodstuffs.

2.PURPOSE

The purpose of this SOP is to provide with a methodology that will allow identifying the oils species present in a refined vegetable oil blend.

3.SCOPE

This method is suitable for the qualitative identification of specific oil species (palm species consisting of palm oil, palm kernel oil, palm stearin, palm olein and seed oils: rapeseed oil and sunflower oil) in an oil blend consisting of a maximum of three different oil species. It is not suitable for the identification of oil species in oil blends containingfouror more different oils.

4.DEFINITIONS AND ABBREVIATIONS

  • FT-IR: Fourier Transform Infrared (spectroscopy)
  • FA: Fatty acid
  • FAME: Fatty acid methyl ester
  • GC-FID: Gas Chromatography-Flame Ionisation Detector
  • PCA: Principal Component Analysis
  • SIMCA: Soft Independent Modelling of Class Analogy
  • PLS-DA: Partial Least Square-Discriminant Analysis
  • PLS-R: Partial Least Square-Regression

5.Principle of the method

The methodology employed is a staged procedure that consists of a combination of FT-IR spectroscopy that is used to screen and classify the oils and the well adopted fatty acid methyl esters analysis using gas chromatography to confirm the composition of the oils when required.

These two techniques, when performed serially on the basis of the developed decision making system, exploit the small differences of the chemical composition between different oil species in different type of oil blends to classify the unknown sample in one of the 6 oil classes studied. In that way, both untargeted or fingerprint analysis (spectroscopic screening) and targeted approaches (fatty acid quantification by chromatography) are applied to increase result’s certainty. The system is designed to target the following 6 oil classes: palm oil and palm derivatives (whole palm oil, palm stearin and olein – P class), palm kernel, (PKO class), seed oil with high polyunsaturate content (rapeseed and sunflower oil and their mixtures, RS class), palm oil mixture with palm kernel oil (PPKO class) and palm oil mixture with the polyunsaturated seed oils such as rapeseed and sunflower oil (RSPO).

6.MATERIALS AND EQUIPMENT

6.1Chemicals

For spectroscopic measurements:

6.1.1Ethanol, analytical grade

For analysis of fatty acids:

6.1.2Methanol, HPLC grade.

6.1.3Potassium hydroxide, AR grade, (≥85% KOH basis, pellets, white), Sigma P1767

6.1.4Sodium sulphate anhydrous, (ACS reagent, ≥99.0%, anhydrous, granular), Sigma 239313.

6.1.5Hexane, HPLC grade

All chemicals were purchased from Sigma-Aldrich ( No special storage requirements were essential or particular hazards identified.

6.2Water

No water is used.

6.3Solutions, standards and reference materials

6.3.1Fatty acid methyl ester standard mixture:Provided by Sigma-Aldrich, Product code 47885-U

6.3.2Internal standard:Methyl tridecanoate. Provided by Sigma-Aldrich. Product code 91558-5ml.

6.4Commercial kits

Not applicable.

6.5Plasticware

6.5.1Pipettes tips: 1mL, 5mL (polypropylene, Thermo Fisher Scientific, Dublin, Ireland)

6.5.2Safety pipette filler (polypropylene, Thermo Fisher Scientific, Dublin, Ireland)

6.6Glassware

For spectroscopic measurements:

6.6.1Bottles glass amber 125 mL with 28mm white polypropylene caps, height 114mm, diameter 49mm supplied by VWR International Ltd. (Product code 215-2111)

6.6.21mL shell vials, 40 x 8.2mm, clear glass 8mm PE plug, transparent. Provided by VWR International Ltd. (Product code 548-0352).

For methylation of fatty acids:

6.6.3Measuring cylinder, glass, 100ml

6.6.4Measuring cylinder, glass, 1000ml

6.6.5Reagent bottle, glass, 1000ml

6.6.6Reagent bottle, glass, 100ml

6.6.7Graduated pipette, glass, 1ml

6.6.8GC vials with screw cap and septum, 2ml

6.6.9Pasteur pipettes, glass, 250mm with filler

No special cleaning produce is applied.

6.7Equipment

For spectroscopic measurements:

6.7.1Fourier Transform Infrared spectroscopic equipment: Nicolet iS5 Thermo Scientific (Thermo Fisher Scientific, Dublin, Ireland). ATR iD5 accessory-diamond. Detector: DTGS KBr. Beamsplitter: KBr. Product code: not available

6.7.2Raman spectroscopic equipment: Advantage 1064 Raman Spectrometer (DeltaNu Inc., Laramie, Wyoming, USA). Product code: not available

6.7.3Dry block heater (model: 25H heated and stirring ambient +5°C to 150°C temperature range purchased from Thermo Fisher Scientific (Dublin, Ireland, product code: 11767519).

6.7.4Automatic pipettes ‘eppendorf’: 1mL, 5mL (Eppendorf Research® plus)

No special calibration details were required.

For chromatographic analysis of fatty acids:

6.7.5Oven set at 100°C.

6.7.6Desiccator.

6.7.7Gas Chromatography-Mass spectrometry:Varian CP3800 Gas chromatograph fitted with Flame Ionisation Detector. Supplied by JVA Analytical, Dublin.

6.7.8GC analytical column CP-88-SIL for FAME, 100m x 0.25mm id, 0.2µm film thickness. Supplied by Agilent Technologies, Product number CP7489.

7.PROCEDURES

All oil samples should be stored in amber containers protected from the light at room temperature at used within 6 months to minimise variation introduced from potential autooxidation of the oils.

7.1Sample preparation

7.1.1Preparation of binary / ternaryadmixtures

  • The percentages in volume of each of the oils present in an oil mixture should be defined beforehand.
  • “Hard oils” (i.e. solid at room temperature) such as palm oil, palm olein, palm stearin and palm kernel oil, have to be melted to their liquid form prior to the making of the admixtures.
  • Melt the palm oil and the palm stearin in an air oven at 50-60°C and the palm kernel and palm olein at 30-40°C overnight.
  • Label the bottles that will contain the oil mixtures.
  • Gently shake the pure oils before a representative aliquot is taken. 100mL of each binary mixture may be prepared in 125 mL amber glass bottles. Measure volumetric aliquots of different oils using volumetric flasks. A different volumetric flask should be used for each type of oil.
  • Add the oils one by one in their liquid form to the glass bottles.
  • Flush the headspace of each bottle with N2 liquid to avoid oxidation and store them at room temperature protected from direct light.

7.2FTIR spectra acquisition

  • Melt the oil admixtures in an air oven at 50°C until clear.
  • Take aliquots of 1mL into small glass vials using a pipette.
  • Place the vials in a block heater at 50°C prior measurement in FTIR.
  • Clean the surface of the ATR diamond of the FTIR twice with ethanol or isopropanoland a soft tissue and let it dry.
  • Take the sample from the block heater and gently shake it manually.
  • Place 2-3 drops of oil on the surface of the ATR diamond using a pipette. Check there are no air bubbles.
  • Record the FTIR spectra at room temperature using the OMNIC software:
  • Open the OMNIC software (comes bundled with the Thermo FTIR instrument).

- Click ‘Collect’ and then ‘Experimentsetup’. Establish the operational conditions of the FTIR prior to analyse the samples.

  • Click ‘Collect’ and then ‘Collect sample’. Enter the spectrum title and click ‘OK’.

  • Confirm the collection of the background spectra clicking ‘OK’. And click ‘OK’ once again to confirm the collection of the sample spectrum.
  • Click ‘Yes’ to confirm that data collection has stopped and you want to add the collected spectrum to a particular window (e.g. window 1).
  • Click ‘File’ – ‘Save as’ and introduce the name of the file (e.g. the same name given before to the spectrum tittle)

  • Clean the oil from the ATR with ethanol or isopropanol after each measurement.
  • At least three spectra should be taken from each sample.

7.3Raman spectra acquisition

  • Melt the oil admixtures in an air oven at 50°C until clear.
  • Take aliquots of 1mL into small glass vials using a 1mL pipette.
  • Place the vials in a block heater at 50°C prior measurement in Raman.
  • Record the Raman spectra using the NuSpec software:
  • Double click on the NuSpec to start the program.
  • Before a sample is analysed on the spectrometer, it should be calibrated to ensure the best accuracy possible is achieved. To do this, place the polystyrene standard into the sample holder and click calibrate on the left hand side of the screen. This will inform you of the previous calibration value and the new calibration value, allowing you to select which is best. The calibration value should be as low as possible and should not exceed 1.
  • The first stage in analyzing a sample is to ensure that the laser beam is properly onto the sample. To do this, insert the sample into the sample holder, ensure the integration time is set at 1 and select continuous. You then must use the focus knob on the instrument to ensure the maximum intensity is obtained for the peaks. Once the maximum peak intensity is obtained click Abort on the left hand side of the screen to cancel the continuous laser. You are now ready to analyse your samples.
  • There are many different settings which can be altered on the homepage of the NuSpec software as shown below:
  • Laser power: Select ‘High’.
  • Integration time: Type ‘10 seconds’
  • No. of spectra to acquire: Type ‘2’
  • Display: This allows you to select what is shown on the display screen. You can either use Spectrum which shows the present spectrum or Average which will display the average of the spectra since two spectra was selected on the No. of spectra to acquire option.
  • Acquire/Continuous/Abort: Acquire allows you to acquire the spectrum using the selected settings. Continuous allows you to constantly direct the laser onto your sample which allows you to focus the beam onto the sample. Abort allows you to cancel the current acquisition or continuous exposure.
  • Save spectrum: This allows you save the spectrum which has just been acquired. The spectrum can be saved as Display, Multifile or Raw Data and as either a .dnu, .prn or .spc file type.
  • Once the settings have been established, take the sample from the block heater and gently shake it manually.
  • Place the vial into the sample holder and close the sample holder lid.
  • Click ‘Acquire’ to acquire the Raman spectrum.
  • Once the spectrum of the sample has been obtained, select the file format for the spectrum to be saved in (.spc file type) and click ‘Save Spectrum’ (first the Spectrum 1 and then the Spectrum 2). Spectrum 1 and 2 can be selected in the lower ‘Display’ window on the upper right hand of the screen.
  • Once you have saved the file, you will be returned to the NuSpec homepage which will allow you to analyse another sample as necessary.
  • At least two spectra should be taken from each sample.

7.4Analysis of fatty acids

Fatty acid methyl esters are prepared according to BS684-2.34:2001 part 5. Briefly, oil blends are heated to 60oC to ensure complete melting of the solid fat component before being thoroughly mixed prior to sampling. Subsamples (300mg) are taken in duplicate and dissolved in 10mls of hexane. An aliquot of the fatty acid methyl esters in hexane should be transferred to a vial prior to analysis by gas chromatography. Individual fatty acid methyl esters were detected by flame ionisation detection, identified by comparison with external fatty acid methyl ester standards and quantified by the use of an internal standard. For detailed calculations go to the procedure 8.2

7.4.1Methylation of fatty acids

7.4.1.1Preparation of reagents

7.4.1.1.1Sodium sulphate anhydrous:

Weigh approximately 50 ±0.01g of sodium sulphate into a clean dry silica basin, place in an oven (6.7.5) for 2 hours ±10.0min. Remove from the oven and cool to ambient temperature in a desiccator (6.7.6) before use.

7.4.1.1.2Anhydrous-methanol:

Using a clean dry measuring cylinder (6.6.4) measure out 1000ml of methanol (6.1.2) and transfer to a clean dry reagent bottle (6.6.5). The reagent is stable for three months.

7.4.1.1.3Methanolic Potassium hydroxide, 2N solution:

Weigh out 11.2 ±0.01g of potassium hydroxide (6.1.3) and transfer to a reagent bottle (6.6.6); using a clean dry measuring cylinder (6.6.3). Add 100ml of anhydrous methanol reagent (7.4.1.1.2) and dissolve. The reagent is stable for one month.

7.4.1.2Preparation of fatty acid methyl esters (FAMEs)

  • Allow the samples to reach ambient temperature before use.
  • Using a clean, dry graduated pipette (6.6.7) and safety pipette filler (6.7.7), add 0.5ml methanolic potassium hydroxide reagent (7.4.1.1.3).
  • Cap the vial and thoroughly mix the contents for 30 seconds.
  • Allow the mixture to stand undisturbed until two clear layers are formed.
  • Using a clean dry disposable Pasteur pipette, transfer approximately 2 ml of the upper layer to a clean dry labelled GC vial (6.6.8).
  • Close the GC vial with a cap and septum, store at –20°C in a spark-proof freezer, and analyse within one week.
  • Chromatographic analysis of FAMEs
  • Heat the oil blends to 60oC to ensure complete melting of the solid fat component before being thoroughly mixing prior to sampling.
  • Take subsamples (300mg) in duplicate and dissolve them in 10ml of hexane (6.1.5)
  • Transfer an aliquot of the fatty acid methyl esters in hexane to a vial prior to analysis by gas chromatography.
  • Place the vial in the autosample of the GC-FID.
  • Adjust the GC-FID operating conditions as follow:

Injector

  • Injector temperature 225°C
  • Injection volume 2.0 µl
  • Split ratio 50:1

Carrier gas

  • Carrier gas flow rate 1.0 ml/min (constant flow).
  • Carrier gas helium.

Detector

  • Detector Flame ionisation detector.
  • Detector temperature 225°C.
  • Range 12

Column oven

  • Initial temperature: 70°C
  • 8.0°C/min to 110°C, hold for 0.0 minutes.
  • 5.0°C /min to 170°C, hold for 10.0 minutes.
  • 2.0°C /min to 225°C, hold for 10.0 minutes.
  • 20.0°C /min to 240°C, hold for 5.0 minutes.

7.5Quality Assurance

7.5.1Sample preparation

  • In the preparation of in-house mixtures, binary and ternary mixtures have to be prepared with oils from mixed origin to minimise influence of compositional variation between the same oil specie.
  • The oils to be tested were preheated at 50°C before the spectroscopic measurements. Temperature check on the heat blocker has to be performed to ensure that oils are not over or under heated which will introduce variation in the measurements.
  • The sample preparation procedure for fatty acid analysis is based on a BS method.
  • Storage of oils and in-house admixtures: All obtained samples and resulting admixtures were stored individually in glass vials in the dark in room temperature with a headspace of 5% to avoid auto-oxidation and photo-oxidation.
  • Spectroscopic analysis
  • Spectra are acquired in triplicate for FTIR and in duplicate for Raman.
  • Both instruments are calibrated before the measurements.
  • Equipment must be maintained according to manufacturer’s guidelines.
  • The spectral acquisition itself should not introduce any variation in the measurements if done in a well maintained and calibrated spectrometer.
  • Fatty acid analysis

Fatty acid analysis with gas chromatography of fatty acids methyl esters (FAMEs) is performed according to the official British Standards method (BS EN ISO 5509:2001; BS 684-2.34:2001)

Blanks are included within each batch of samples to establish base line stability and instrument readiness. External standards are used to determine fatty acid retention times and individual fatty acid response factors but not for instrument calibration. An internal standard (methyl tridecanoate) is added to each sample prior to preparation and determination of the fatty acid methyl esters. All analyses should be carried out in duplicate.

8.CALCULATIONS AND DATA ANALYSIS

8.1Data analysis of spectroscopicdata (FTIR and Raman)

8.1.1Spectral data handling and spectral pre-processing

8.1.1.1Introduction of raw spectral data into an Excel file

  • Open TQ Analyst 8.
  • Under tab ‘Standards’ click ‘Open Standard…’ to upload all the spectra into the program. Choose ‘Spectra/Groups (*.SPA, *.SPG)’ in‘Files of type’in the ‘Open’ window. Select all the FTIR spectra and press ‘Open’. All the spectra will appear in the Standard Table. Click on ‘Show spectrum file names’ and ‘Show spectrum titles’ (optional) to show that information from the spectra in the Standard Table.
  • To save the spectral data in a .csv, go to ‘File’, and then click ‘Standards to text file’. A ‘Save as’ window will appear. Choose the destination folder and the file name. Save as type ‘Text (*.csv)’ and click ‘Save’.
  • Open Microsoft Excel. Go to ‘Data’ tab and click on ‘Get external data from text’. Find the saved .csv file with the spectral data and click ‘Open’. Select all data and paste them as transposed in a new sheet (sheet 2), so that the variables (wavenumbers) will be in columns and the samples will be in rows.
  • Back in TQ analysis,copy the column entitled ‘Spectrum title’ from the Standards Table under the ‘Standards’ tab and paste it in a new column of Excel at the beginning of the data in sheet 2.
  • Click ‘File’ and ‘Save as’ to save the final dataset. Choose the destination folder and the file name. Save as type ‘Excel Workbook (*.xlsx)’ and press ‘Save’.
  • Along with the data, the class identity of the samples (class abbreviation) has to be introduced in the Excel DataSheet.
  • All the samples (pure, binary mixtures) are categorised in 6 classes according to the following simple scheme: