Combined qualitative and quantitative analysis of food packaging materials using the ZenoTOF 7600 system


Daqiang Pan1, Jack Steed2, Jianru Stahl-Zeng1 and Clemens Bidmon3

1SCIEX, Germany; 2SCIEX, UK; 3TÜV Rheinland, Germany

Abstract


In this technical note, a comprehensive, accurate mass spectrometry method for the characterization of extractable and leachable (E&L) compounds in food packaging was developed. Using the ZenoTOF 7600 system, 3 workflows were explored: suspect screening, unknown screening and MRMHR quantitation (Figure 1). The combined workflows demonstrated the benefits of each acquisition mode to fully characterize and quantify compounds in a food contact material. Both NIST and SCIEX MS/MS libraries were used to streamline the data processing workflow, reducing the number of compounds that required manual review. For unknown screening, the Formula Finder and ChemSpider tools within SCIEX OS software reduced the amount of time required for compound identification. Further, using the optimized compound-specific parameters in MRMHR acquisition, limits of quantitation (LOQs) as low as 0.1 µg/L were achieved for some compounds. These workflows highlight the ability to perform quantitative and qualitative analyses on the ZenoTOF 7600 system.

Key benefits of the SCIEX ZenoTOF 7600 system for the analysis of E&L compounds
 

  • Confident compound identification using MS/MS library searching: The SCIEX E&L MS/MS library and an E&L subset of the NIST library were used to provide MS/MS spectra for more than 2400 compounds. Libraries can easily be created and added to, improving their efficacy and allowing for retention time matching.

  • Sensitive analytical quantitation: The MRMHR acquisition mode was used to obtain the highest levels of sensitivity and specificity for both MS1- and MS2-level quantitation

  • One software platform, 3 workflows: Both qualitative and quantitative data were processed and reported within the SCIEX OS software analytics workspace, removing the need for data exportation or familiarization with other software packages

Figure 1. The combined workflows used with the ZenoTOF 7600 system for E&L applications. 

Introduction


E&L compounds have long been a safety concern in the EU and commission regulation 10/2011 has been in place for more than a decade to regulate food contact materials and ensure their safety to the public.1 This regulation defines the respective limits or prohibitions of many compounds that must be managed to ensure safety. However, this list is not exhaustive and therefore unexpected or unknown compounds in food contact materials must also be analyzed. These components are considered non-intentionally added substances (NIAS) and should be identified in food contact materials to ensure that they are not harmful and are controlled to an acceptable level.2

Methods


Sample preparation:

Sample series 1: The sample consisted of a red polypropylene (PP) plastic, which was stored in 3% acetic acid for 2 hours at 100°C and for 10 days at 40°C. After incubation, 50 µL of the sample was transferred to an HPLC vial for analysis. The sample was then spiked at a concentration of 1 µg/L and 10 µg/L with a mixture of 68 different E&L compounds.

Sample series 2: The sample was a polyvinylidene chloride (PVDC) film stored in 95% ethanol at 60°C for 10 days. After incubation, 50 µL of the sample was transferred to an HPLC vial for analysis. The sample was then spiked at a concentration of 1 µg/L and 10 µg/L with a mixture of 68 different E&L compounds.
For these sample preparations, 3% acetic acid and ethanol were used based on their definitions as food simulants in EU regulation 10/2011.1

Chromatography: The ExionLC AD system was used for separation with a Phenomenex Kinetex™ EVO C18 column (1.7 µm, 100 x 2.1 mm). Mobile phase A was water with 0.1% formic acid and mobile phase B was methanol with 0.1% formic acid. The injection volume was 2 µL, the column oven was set to 40°C and the flow rate was 0.4 mL/min. Additional gradient conditions are listed in Table 1. 

Table 1. Gradient conditions.

Mass spectrometry: Data was acquired using the ZenoTOF 7600 system, operated in both positive and negative ionization modes using electrospray ionization (ESI). MRMHR and data-dependent acquisition (DDA) with the Zeno trap enabled (Zeno DDA) were used. Tables 2, 3 and 4 define the source and MS parameters used for both methods.

Data processing: All data was processed using SCIEX OS software.

Table 2. Source conditions used for both MRMHR and Zeno DDA.

Table 3. Zeno DDA MS parameters.

Table 4. MRM table for both positive and negative methods. Exact m/z values were used for processing.

Suspect screening


For E&L applications, using a curated suspect screening list combined with MS/MS spectral library searching for compound confirmation significantly improves the ease and efficiency of data processing due to the vast number of compounds of interest. Therefore, a suspect list containing 369 compounds provided by TÜV Rheinland was used in the analysis. From those 369 compounds, 102 were identified in positive ion mode and 69 were identified in negative ion mode, all with <5 ppm mass error of the precursor mass for the 10 µg/L spiked sample (sample series 1). In addition to the suspect list, the SCIEX E&L library and an E&L subset of the NIST library were used in tandem to provide the maximum number of identified targets. In total, the mass spectral libraries contained more than 2400 compounds that included MS/MS spectra, which were used to improve compound identification confidence compared to precursor (TOF MS or MS1) data alone. In total, 27 compounds in positive ion mode and 7 compounds in negative ion mode provided a library match score >70 % for the 10 µg/L spiked sample (sample series 1).

Figure 2 shows the results table generated in SCIEX OS software for sample series 1 spiked at 10 µg/L, highlighting the top hits achieved from the combined MS/MS libraries. This results table demonstrates the ease of data review possible with SCIEX OS software, as 3 levels of confidence were used to determine a significant result (mass error confidence, isotope confidence and library hit confidence). Each parameter has a user-defined set of tolerances within the software to provide a visual cue that indicates whether the criteria were met: a green tick for acceptable, orange triangle for marginal and a red circle for unacceptable. 

Once the initial screening was performed to assess potential hits, the extracted ion chromatograms (XICs) were reviewed to ensure that the analyte peak was detected in the spiked sample (10 µg/L) but not in the solvent blank or unspiked sample. Figure 3 shows example XICs for an analyte identified in SCIEX OS software.

Figure 2. Top library hits achieved for sample series 1. The top library hits are listed alongside the 3 criteria used to determine a confident hit (mass error, isotope confidence and library confidence). This table is easily generated by the software and visually identifies acceptable hits (green tick), marginal hits (orange triangle) and unacceptable hits (red circle). These criteria can be used to rapidly identify which compounds must be verified by the user to confirm or reject the initial identification. 

Figure 3. Unspiked (P1-A) and spiked (P1-C) sample series 1 XICs for 4,4-methylene-di-aniline and corresponding results table. The XICs show the absence of the analyte in an unspiked sample (left) alongside its corresponding peak in the 10 µg/L spiked sample (right). The results table (bottom) shows the 3 criteria used to determine a confident identification. The grey square shown for the unspiked sample indicates that no peak was integrated, whereas the green tick shown for the spiked sample indicates a confident identification.

Unknown screening


In addition to suspect screening, unknown screening can be performed to obtain a more comprehensive characterization of the compounds in an E&L sample. This screening is used to detect compounds that are not on the suspect screening target list but are identified as NIASs that might be present in final products. Due to their potential to leach into the food contained within the packaging, it is important to identify and assess their potential harm to the user.

To perform unknown screening, the same dataset acquired for the suspect screening was used. The data was processed using the unknown screening workflow in SCIEX OS software to identify the unknown features. Initially, the same mass spectral libraries were used to assess the putative matches outside of the suspect list. Of the 44 compounds identified in the positive ion acquisition mode, 17 were unique compounds that were not detected using the suspect screening. In addition, both the Formula Finder and non-targeted peaks features in SCIEX OS software were utilized. Figure 4 shows how to implement these tools in the processing method. A control (extraction solvent) was also used to compare against the sample. This comparison helped reduce false positives by only including sample peaks that were sufficiently larger than blank peaks, based on a user-specified area ratio threshold.

Figure 4. Formula Finder and non-targeted peaks criteria defined within the SCIEX OS software processing method. The Formula Finder criteria are used to determine the suspected chemical formula of the unknown compound by utilizing the elements chosen by the user alongside the maximum acceptable mass error (ppm) tolerance. The non-targeted peaks criteria first determine a range for peak searching based on minimum and maximum retention times. Peak searching is optimized based on the LC gradient used, peak detection sensitivity, area ratio threshold (how much larger the peak must be in the sample when compared to a representative blank) and grouping peaks by adduct or charge.

XICs were reviewed, as in the suspect screening workflow. XICs from both blank and spiked samples were reviewed to determine whether the unknown component was genuine prior to the use of ChemSpider in the software. The ChemSpider tool is built into the SCIEX OS software and can provide tentative identification of the unknown compound by matching the theoretical MS/MS fragmentation pattern to that obtained from non-targeted acquisition. Additional compound confirmation can be achieved by purchasing an authentic standard of the tentatively identified compound. An example of unknown compound identification using the ChemSpider tool is shown in Figure 5. Once this identification has been achieved, the newly identified compound can easily be added to existing MS/MS libraries to improve the efficacy and useability of the library. The growth of the MS/MS spectra list can help reduce the time required for and improve the efficiency of the analysis. 

Figure 5. Tentative identification of an unknown component using the ChemSpider tool in SCIEX OS software. Once the tentative chemical formula is identified, the MS/MS spectra can be matched against theoretical fragmentation patterns of potential compounds in the ChemSpider database. The MS/MS spectra (right) show a known ion in blue and an undetermined ion in red, based on the theoretical fragmentation performed. This example highlights a strong hit, as nearly all the ion fragments were matched by the software.

Quantitation


Finally, the components of interest or at risk of being harmful were quantified. Quantitation was performed using DDA (MS1 only), however, the maximum sensitivity and specificity were achieved using MRMHR acquisition. MRMHR uses the optimized transitions and compound-specific parameters of collision energy (CE) and declustering potential (DP). The transitions established by TÜV Rheinland in a previous triple quadrupole instrument method were used and the compound-specific CE and DP parameters were optimized for the ZenoTOF 7600 system.

A small subset of relevant compounds was analyzed using MRMHR and limits of quantitation (LOQ) as low as 0.1 µg/L in solvent and a linear range between 0.1 and 100 µg/L were achieved. Therefore, the MRMHR workflow, in combination with the non-targeted acquisition workflows, demonstrated the flexibility of the ZenoTOF 7600 system for both screening and low-level quantitation of E&L compounds. Figure 6 shows the calibration curve, precision, % accuracy and sample results for a representative compound. Figure 7 shows the low fragment mass error (mDa) observed for 4,4’-MDA across the calibration curve analyzed.

Figure 6. Standard and spiked sample data (sample series 1) using MRMHR for 4,4-methylene-di-aniline. XICs are shown for the matrix blank and the lowest concentration analyzed (0.1 µg/L). The calibration curve highlights the observed linearity between 0.1 and 100 µg/L. Three XICs are shown for 1 µg/L spike replicates. The precision and accuracy data highlight excellent precision with %CV values <10% and accuracy values ranging from 84% to 113%. An R2 value >0.99 was achieved for the calibration curve.

Figure 7. Fragment mass error (mDa) across the calibration curve for4,4-methylene-di-aniline. This data table shows the fragment mass error achieved for each injection at each concentration of the calibration curve. The mass error ranged from -1.426 to 0.352 mDa, highlighting the high level of mas

Conclusion
 

  • SCIEX OS software provides a single platform for suspect and unknown screening alongside a full suite of quantitative features

  • A suspect list of 369 compounds was used alongside an MS/MS spectral library containing more than 2400 compounds to screen for relevant E&L compounds

  • The sensitivity of a targeted compound list was assessed with LOQs as low as 0.1 µg/L and calibration ranges between 0.1 and 100 µg/L. These results showcase that low-level quantitation is easily achievable using the ZenoTOF 7600 system.

References
 

  1. Commission Regulation (EU) No 10/2011 of 14 January 2011 on plastic materials and articles intended to come into contact with food – August 2023

  2. Dossier - non-intentionally added substances (NIAS) – Food Packaging Forum – April 2013
Request information