Rapid Analysis of Sudan and Other Prohibited Dyes in Chilli Powder Using Ultra Performance Liquid Chromatography and Tandem Mass Spectrometry

RAPID ANALYSIS OF SUDAN AND OTHER PROHIBITED DYES IN CHILLI POWDER USING ULTRA PERFORMANCE LIQUID CHROMATOGRAPHY AND TANDEM MASS SPECTROMETRY

Dimple Shah1, Evelyn Goh2 and Jennifer A. Burgess1 1Waters Corporation, 34 Maple Street, Milford, MA, USA 2Waters Pacific PTE, Singapore

Recovery RT Precursor Cone Product Collision INTRODUCTION Name (min) (m/z) (V) (m/z) (V) To calculate method recovery, sample M was spiked (n=3) Sudan dyes are synthetic dyes frequently used prior to extraction at 10 µg/kg with 10 dyes and 50 µg/kg for for colouring plastics and other synthetic Dimethyl 77 30 Para red. The pre-spiked samples were quantified against the 4.12 226.1 42 MMS curves and recoveries were calculated. The percentage materials. Due to their intense colour, these Yellow 121 25 dyes have been fraudulently used to enhance recoveries (Figure 4) ranged from 60 to 95%. In order to 156.1 18 account for the high matrix complexity and improve recoveries, the appearance of various food products such as Para Red 4.32 294.1 34 one or more internal standards could be added to the method. spices, mixtures of spices and sauces. According 128 34 to the International Agency for Research on Rhodamine 399.2 42 % Recovery in chilli powder 3.35 443.3 80 Cancer (IARC), the Sudan and azo-dyes both B 355.2 60 120 have been classified as potential genotoxic and/ 100 Sudan 246.3 26 80 or carcinogenic substances. Their degradation 6.53 457.1 56 Black B products are also considered to be carcinogenic. 142.1 38 60 40

The European Union (EU) Commission Decision 128 28 20

87 87 77 69 64 85 62 93 85 84 Sudan I 4.71 249.1 30 73 2003/460/EC mandates testing for Sudan I in 156 20 0

chilli powder, or products containing chilli 121 20 powder. In 2004, the decision was expanded to Sudan II 5.64 277.1 30 156 20 include Sudan II, III and IV (2004/92/EC). Despite the prohibition on using Sudan dyes as 197.1 19 Sudan III 6.07 353.2 50 Chilli powder 3 (Sample M) food additives, RASFF (Rapid Alert System for 156 25 Figure 4. % Recovery of 10 dyes at 10 µg/kg and Para red at Food and Feed) found a number of positive 106.1 44 identifications in 2011-2012 with detection of Sudan IV 6.74 381.2 50 50 µg/kg in chilli powder 3 (sample M). 224 22 Sudan dyes in different food products.1 According to food testing laboratories in Europe, Sudan 93 25 Sample Analysis 3.16 215.1 34 the typical reporting limits are between 10 to 50 Orange G 122 20 A total of 14 samples of various spices (Table 1) were µg/kg for the Sudan and azo-dyes in spices. In extracted and analyzed in triplicate. All samples were Sudan Red 183.1 20 this study, a fast and sensitive method utilizing 6.54 380.1 34 quantified against the MMS calibration curve of sample M. Out 7B 169 30 of the 14 samples, six samples were found to contain Sudan a QuEChERS extraction procedure along with and azo-dyes. Sample A contained 5 banned dyes, out of which UPLC separation with MS/MS for the detection Sudan Red 123 18 4.65 279.0 28 Rhodamine B (13.9 µg/kg) was detected above the action limit. and quantification of 11 Sudan and azo-dyes is G 108 34 Sudan orange G was found at a moderately high level (276 µg/ presented. Table 2. List of sudan and azo-dyes with their retention times, kg) in curry powder (sample H). Rhodamine B was detected MRM transitions and tuning parameters. above the action limit in samples I, J and L. Sudan I was found in sample A (4.6 µg/kg) and L (14.03 µg/kg). These detection METHODS results demonstrate the feasibility of using this method for Standards RESULTS & DISCUSSION various spices. Figure 5 shows chromatograms of all the dyes Sudan I, III, IV, Sudan red G, Sudan red 7B, Sudan black B, detected in sample A. The chemical structure of all 11 dyes are shown in Figure 1. Dimethyl Yellow and Rhodamine B were purchased from Sigma The separation of these dyes was achieved using an ACQUITY -Aldrich. Sudan orange G and Para Red were purchased from UPLC H-Class system in 12 minutes. Figure 2 shows an overlay AccuStandard while Sudan II was purchased from Fisher of the MRMs of each dye in a solvent standard at 2 µg/L Scientific. The stock solutions for each dye were prepared at 1 (equivalent to 10 µg/kg in spice sample) for all dyes except mg/mL in acetonitrile. The working solutions were also Para red which is at 10 µg/L (equivalent to 50 µg/kg in spice prepared in acetonitrile. sample). Samples The samples analyzed in this application include contaminated and non-contaminated chilli and curry powders. The sample type for each of the 14 samples is listed in Table 1. Sample preparation The Waters® DisQUE™ QuEChERS (CEN method 15662)2 was used to prepare all samples. Briefly, two grams of powder Figure 5. Chromatograms of all detected dyes in sample A (chilli/curry/turmeric) was weighed into a centrifuge tube. The sample was mixed with 8 mL of water and vortexed for 30 TargetLynx offers the ability to quickly identify samples that seconds. The mixture was extracted with 10 mL of acetonitrile contain analytes of interest and automatically highlight any followed by the addition of QuEChERS CEN material (4 g parameters that fall outside the set analytical quality control MgSO4, 1 g NaCl and 1.5 g sodium citrate). The resulting criteria. Figure 6 shows results from sample I, where mixture was shaken for one minute. The tube was then Rhodamine B was found above 10 µg/kg (red box, Figure 6). centrifuged at 4000 rpm for 5 minutes. The supernatant was Targetlynx can also check that the ion ratio is within the placed into vials for analysis. required tolerance. If the ion ratios of the samples were outside the tolerance limit specified by EU commission Decision Sample ID Name Sample ID Name 2002/657/EC, they would be automatically flagged.

Sample A Chilli powder 1 Sample H Turmeric powder Figure 1. Structure of the dyes Sample B Paprika 1 Sample I Garam masala 1

Sample C Paprika 2 Sample J Curry powder 2 Dimethyl yellow Sample D Crushed chilli Sample K Curry powder 3 Sample E Paprika 3 Sample L Garam masala 2 Sample F Chilli powder 2 Sample M Chilli powder 3

Sample G Curry powder 1 Sample N Chilli powder 4 Rhodamine B

Sudan Red 7B Table 1. List of sample identifier codes and descriptions Sudan Red G Sudan II Sudan Black Para B Sudan Orange G Sudan I red Sudan III Sudan IV UPLC Conditions Figure 6. TargetLynx flagging of dyes that exceeded the re- ® quired action limit in sample I. LC: ACQUITY UPLC H-class System Figure 2. Overlay of MRMs for each of 10 dyes at 2 µg/L Column: ACQUITY BEH C18 2.1 x 100 mm, 1.7 µm (equivalent to 10 µg/kg in spice sample) and Para red at 10 µg/L (equivalent to 50 µg/kg in spice sample) in solvent. Matrix effects Mobile phase A: Water + 0.1% formic acid Matrix effects were investigated by comparing the slopes of the Mobile phase B: Methanol + 0.1% formic acid MMS calibration curves of sample M with the slopes of solvent Mobile phase C: Acetonitrile + 0.1% formic acid Linearity calibration curves. Sudan I, IV and Rhodamine B showed very o little or no matrix effect, while ion suppression was observed Column Temp: 45 C Injection vol: 5 µL Sample M was screened and no dyes were found above the for Sudan orange G, Dimethyl Yellow, Sudan II, Sudan red G, Flow rate: 0.4 mL/min Total run time: 12.0 min detection limits. Hence it was selected to evaluate method Sudan red 7B and Para red. Sudan black B and Sudan III recovery. Solvent and matrix match spiked calibration (MMS) Wash solvent: Acetonitrile Purge solvent: 10% Methanol showed an ion enhancement effect in sample M. Despite these curves were prepared for 10 dyes at concentrations of 0.5, 1, Gradient: Initial 80:10:10 (A:B:C); 0.5 min 40:30:30; 5.0 min matrix effects, the method can detect and quantify all the 2, 4, 8, 16, 32, 64 and 128 µg/L (equivalent to 2.5 to 640 µg/ 0:50:50; 9.0 min 0:50:50; 9.1 min 80:10:10; 12.0 min above listed dyes in the various spices tested. In order to kg in spice sample). Para red was spiked five times higher than 80:10:10 further reduce matrix effects, additional sample preparation the other dyes. The co-efficient of determination (r2) was could be employed. The addition of internal standards would above 0.99 for all dyes in sample M and the solvent calibration MS Conditions curve. Figure 3 shows an example of MMS curve for Sudan III MS: Xevo® TQD in sample M. CONCLUSIONS Ionisation mode: ESI positive  This method combines an easy sample Capillary voltage: 1.5 kV Source temp: 150oC Compound name: Sudan III preparation procedure using QuEChERS and UPLC Correlation coefficient: r = 0.999658, r^2 = 0.999316 o Calibration curve: 102.481 * x + -44.9251 -MS/MS detection for the analysis of 11 Sudan Desolvation temp: 500 C Desolvation Gas: 1000 L/hr Response type: External Std, Area Curve type: Linear, Origin: Exclude, Weighting: 1/x, Axis trans: None and azo-dyes in various spices Acquisition mode: Multiple Reaction Monitoring (MRM)  Even with significant matrix effects, the ACQUITY

60000 UPLC H-Class system with the Xevo TQD MS Acquisition and Processing Methods provided the detection levels at the regulatory 40000 For all compounds, two MRM transitions were obtained from requirements for the analysis of Sudan and azo-

the Quanpedia™ database. The most intense transition was Response 20000 dyes in contaminated samples used for quantification (Quantifier ion), and the second most  Automated peak processing, quantification and intense transition was used for confirmation (Qualifier ion). The -0 ug/kg dwell times for the transitions were automatically optimized to -0 50 100 150 200 250 300 350 400 450 500 550 600 flagging using the TargetLynx application give a minimum of 12 points across each chromatographic manager allows users to quickly identify non- peak. The MRM transitions, cone voltages and collision energies Figure 3. Matrix match spiked calibration curve of Sudan III compliant samples for all compounds along with expected retention times are from 0.5 ppb to 128 µg/L ppb (equivalent to 2.5 to 640 µg/kg shown in Table 2. The data were acquired using MassLynx® in sample M). References software, v. 4.1, and processed using the TargetLynx™ 1. https://webgate.ec.europa.eu/rasff-window/portal/ 2. http://www.waters.com/webassets/cms/library/docs/720003048en.pdf Application Manager.