Analysis of mineral oil hydrocarbons in consumer products using silver-phase liquid chromatography-gas chromatography

(1)

1

Bachelor Thesis Chemistry

Analysis of mineral oil hydrocarbons in consumer

products using silver-phase liquid chromatography-gas

chromatography

author

Roxane Biersteker

30

th

_{of June 2017}

Studentnumber 10808272

Research institute Supervisors

Van ‘t Hoff Institute for Dr. W.T. Kok & Prof. dr. ir. J.G.M.

Molecular Sciences Janssen

Research group Daily Supervisor

(2)

2

Index

1. Abstract 3 2. Introduction 4 3. Experimental section 3.1 Equipment 9 3.2 Characterization 3.2A HPLC separation 9 3.2B Analysis by GC-FID 11 3.3 Quantification

3.3A Solid Phase Extraction 11

3.3B HPLC separation 11

3.3C Analysis by GC-FID 12

4. Results

4.1 Characterization

4.1A Separation between MOSH and MOAH internal standards 13

4.1B Characterization of the mineral oils 14

4.2 Quantification 18

5. Discussion

5.1 Characterization: separation between MOSH and MOAH

5.1A Ag/Silica column 19

5.1B Ag/Polymer column 19

5.1C Comparison columns 20

5.2 Characterization: classification aromatic compounds

5.2A Ag/Silica column 21

5.2B Ag/Polymer column 21 5.2C Comparison columns 22 5.3 Quantification 22 6. Conclusion 23 7. Outlook 24 8. References 25 9. Appendices 27

(3)

3

1. Abstract

Mineral oil hydrocarbons are present in almost all food as a consequence of contamination and numerous intentional uses in the production process. Due to the difference between the main components of mineral oil, Mineral Oil Saturated Hydrocarbons (MOSH) and Mineral Oil Aromatic Hydrocarbons (MOAH), with regards to their toxicity, an individual analysis is needed. In this research the first steps were taken towards the development of a new method, based on silver-phase liquid chromatography- gas chromatography, for the quantification and characterization of mineral oil hydrocarbons in consumer products. Analysis of internal standards using comprehensive LCxGC-FID, showed that a separation of around 1 minute between MOSH and MOAH markers was established. However, analysis of the LCxGC-FID chromatograms of three mineral oil samples with different grades of refinement showed that compounds elute in the timeframe between the MOSH and MOAH markers. Regarding the mineral oil characterization, a Ag/Polymer column was more suited to separate aromatics into groups than a Ag/Silica column, which is possibly a result of the length of the column. Finally, results showed that the LC-GC method is reliable for determining the amount of aromatics in mineral oils for both columns. Further research should be undertaken to determine whether a complete separation between MOSH and MOAH was established, for example by analyzing the fractions by GC-VUV.

In bijna elk voedsel zitten koolwaterstoffen die afkomstig zijn uit minerale olie. Deze kunnen slecht voor de gezondheid zijn en daarom is een analyse methode nodig die deze koolwaterstoffen detecteert. De koolwaterstoffen kunnen verdeeld worden in twee groepen, genaamd MOSH en MOAH. MOSH bevat onverzadigde koolwaterstoffen en MOAH bevat verbindingen met aromatische ringen. Aangezien de MOAH kankerverwekkend kunnen zijn, is het belangrijk dat deze apart van de MOSH gedetecteerd kunnen worden. In dit onderzoek zijn drie minerale oliën gekarakteriseerd en is het gehalte MOAH bepaald aan de hand van vloeistofchromatografie en gaschromatografie. Door het gebruiken van zilver in de stationaire fase van de vloeistofchromatografie kolom, was de verwachting dat een betere scheiding tussen MOSH en MOAH bereikt zou kunnen worden dan eerder mogelijk was. Een betere scheiding tussen simpele alkanen en aromatische verbindingen was mogelijk, maar bij het analyseren van de minerale oliën bleek dat er maar een kleine scheiding tussen de MOSH en de MOAH zat. De resultaten van de

(4)

4 drie methodes die gebruikt zijn voor het vaststellen van het gehalte MOAH in de minerale oliën, kwamen sterk overeen. Dit suggereert dat de methodes betrouwbaar zijn, maar meer onderzoek moet gedaan worden om vast te stellen dat de MOSH en MOAH compleet gescheiden zijn en om de scheidingsmethode te verbeteren. Zodra de methode geoptimaliseerd is, zullen voedsel en cosmetica producten geanalyseerd worden.

2. Introduction

Mineral oil hydrocarbons (MOH) are present in almost all foods as a consequence of contamination and numerous intentional uses in the production process.1 MOH are mixtures of hydrocarbons consisting of thousands of chemical compounds of varying size and structures. They are primarily derived by physical separation and chemical conversion processes such as hydrogenation and alkylation from crude oils, but also from synthetic products that are derived from liquefaction of natural gas, biomass and coal.MOH consists of three classes of compounds: paraffins, naphthenes and aromatics (Fig. 1), and they are usually highly alkylated. Two types of MOH are distinguished, namely mineral oil saturated hydrocarbons (MOSH) and mineral oil aromatic hydrocarbons (MOAH). The former consists of linear and branched alkanes as well as alkyl-substituted cycloalkanes (paraffins and naphthenes), the latter include alkyl-substituted polyaromatic hydrocarbons (aromatics).

(5)

5

Figure 1: Alkanes (paraffins), naphthenes and aromatics1

Foods that contain the highest mean concentration of MOH include fish products, oilseeds, confectionery and vegetable oil.2–5 To illustrate, in fresh- and sea-water fish MOSH concentrations up to 1200 mg/kg fat were found, with a mean value around 200 mg/kg fat.6 Analysis of edible oils showed that a large proportion were contaminated with more than 10 mg/kg mineral paraffins with a maximum concentration of around 400 mg/kg.7,8 For rice a mean MOSH concentration is 132 mg/kg and the expected percentages of MOAH in MOH 15–30%. Other possible sources for MOH in rice and the expected proportion affected are depicted in Figure 2.1

(6)

6 As shown in Figure 2, a source for MOH in food is packaging materials, especially those that are made from recycled board and paper. Moreover, printing inks and MOH that are used as additives in plastics contribute to MOSH levels, together with additives that are directly used in the processing of food. Further uses of MOH are as release agents for sugar products and bakery ware or surface treatment of foods such as confectionery.

The above mentioned sources of MOH result in an estimated dietary exposure of MOSH between 0.03 and 0.3 mg/kg b.w. per day.1 The background exposure to MOAH is estimated at 20%, with respect to MOSH, although there has been little quantitative analysis of this group of hydrocarbons. These estimations do not take into account that lip care products and lipsticks are also a possible source of exposure, since a large part of these products end up being ingested. A study that was conducted in 2015 showed that 68% of the products that were tested contained at least 5% MOSH and synthetic hydrocarbons (POSH), and 31% contained more than 32% of these compounds.9 As a consequence, it is likely that regular application of these products result in a significant increase in exposure to MOH.

There is little information available on the toxicity and absorbance of MOH in mammals, especially data on MOAH are scarce. Hydrocarbons from about ten to fifty carbon atoms are considered in toxicology studies, since hydrocarbons with less than 10 carbon atoms are highly volatile, and hydrocarbons with more than fifty carbon atoms are improbable to be absorbed following ingestion. Previously published studies on the effect of MOSH are not consistent, although there is concordance with the fact that they are mostly absorbed from the gut into the lymphatic system.10–

13 _{In rats, the estimated absorption of n- and cycloalkanes varies from 25% (C}

26-C29) to 90% (C14

-C18).14–18 In humans, mineral oil has been found in the spleen, liver and lymph nodes as well as in

body fat collected during Caesarian section.19,20 Surprisingly, the mineral paraffins of all body fat samples that were analyzed had a highly similar molecular mass distribution; they are centered on n-C23/n-C24 and range from n-C16 to n-C30. Since this pattern does not resemble a particular mineral

oil product, it suggests that selection takes place regarding absorbance of hydrocarbons. This could be by effective elimination of certain hydrocarbons and limited absorption of heavier compounds. Figure 3 shows the chromatogram of a typical sample, in which MOSH is mainly represented by the broad hump (unresolved peaks).

(7)

7

Figure 3: Chromatogram of MOSH from human abdominal body fat (85 mg/kg).20

Without specific removal of MOAH, all MOH are mutagenic. Especially three to seven ring MOAH cause the mutagenicity, even more so when they have short side chains.21,22 They can be activated by P450 enzymes into genotoxic carcinogens, which can be followed by the formation of DNA adducts.23,24 Although MOSH are not carcinogenic, long chains in high amounts could act as tumor promoters.25–29_{Furthermore, some uncomplicated MOAH, e.g. naphthalene, are}

carcinogenic by a non-genotoxic mode of action. Due to these differences between the MOSH and MOAH, with regards to their toxicity, individual analysis is needed.

The determination of MOSH and MOAH has been proven to be challenging, since they form irregular humps of unresolved peaks in gas chromatograms.30 The first on-line coupled high performance liquid chromatography–gas chromatography method was developed in 1991 and was improved throughout the years.31 A huge drawback of these methods was that they were not able to detect MOAH separately. The first simple method for determining MOSH and MOAH separately appeared in 2009.32_{GC with flame ionization detection (GC-FID) is used since it}

provides practically the same response per mass for components of structures that are similar. As a consequence, quantification is possible without pure standards. MOSH and MOAH are of the same mineral oil fraction and thus of the same volatility range, meaning that selectivity must come from preseparation. For this, high performance liquid chromatography (HPLC) with a silica gel column is used, since silica provides strong retention for aromatic hydrocarbons. Internal standards are used to establish and check the fraction windows. Analysis showed that a gap of thirty seconds

(8)

8 was established between MOSH and MOAH. One of the main obstacles is the limit of quantification of 1 mg/kg, since for polycyclic aromatic hydrocarbons more than 100 times lower legal limits are enforced. This means that if the MOAH consists of 1% carcinogenic material, a detection limit of 1 mg/kg is not satisfactory. Therefore, it is fundamental to develop a method to detect MOAH below this limit. Moreover, the current method presents some issues regarding to the separation between MOSH and MOAH. Due to the small separation window between MOSH and MOAH, the analysis becomes problematic when the content of the MOAH fraction is less than 1 %, in relation to the total amount of mineral oil hydrocarbons.

Using a comprehensive LCxGC method for the analysis of mineral oil solves the issues mentioned before. This method is based on the collection of many fractions during the LC separation and their subsequent GC analysis. This allows to confirm whether the separation window is sufficient for the MOSH-MOAH separation of real samples and, with the appropriate LC conditions, to do a characterization of the mineral oil hydrocarbons based on polarity and molecular weight distribution. This information is useful to know which type of aromatic compounds, regarding to the number of rings and alkylation level, are present in the sample. The standard quantitative separation of MOSH and MOAH using solid phase extraction uses silver ion-silica as a solid phase.33,34_{This is because silver has strong favorable interactions with double bonds, and thus}

aromatics are more retained that aliphatic compounds. However, no previous studies have been reported that apply this stationary phase in LC-GC-FID.

In this research the first steps are taken to develop a new method, based on silver-phase liquid chromatography- gas chromatography, for the quantification and characterization of mineral oil hydrocarbons in consumer products. The main objective of this study is to establish a separation between Mineral Oil Saturated Hydrocarbons and Mineral Oil Aromatic Hydrocarbons, and to classify aromatic compounds in terms of the number of rings. Furthermore, the aliphatic and aromatic content of the mineral oil samples is determined and compared to the standard method based on Solid Phase Extraction. Two different silver-phase LC columns, one silica bonded and one ion-exchange, are tested, due to their specific interaction with π-bonded compounds. The method is applied to the analysis of three mineral oils with different grades of refinement.

(9)

9

3. Experimental section

3.1 Equipment

- HPLC Waters 2696

- Photodiode Array Detector Waters 996

- Ag/Silica column Agilent Chromsep SS (length 10 cm, diameter 4.6 mm)

- Ag/Polymer column Chrompack Chromspher pi (length 25 cm, diameter 4.6 mm) - GC-FID Agilent 6890N

- DB- 5HT column J&W Scientific (length 15 m, diameter 0.32 mm, film 0.1 μm)

3.2A HPLC separation

In all samples internal standards (Sigma Aldrich) were used for establishing and checking the fraction windows (Fig. 4).30_{5-α-Cholestane was used to establish the end of the MOSH fraction}

and tri-tert-butyl benzene marked the beginning of the MOAH fraction. 1-Methylnapthalene and perylene were used to mark the end of the MOAH fraction. Perylene was checked by UV detection, but was not added to the samples that were analyzed due to solubility problems in hexane. Besides these markers, internal standards were added for the verification of adequate method performance and for the quantification (Appendix 3). The internal standards were injected in the LC to establish the separation between MOSH and MOAH.

(10)

10

Figure 4: Markers that were used to establish and check the fraction windows.30

For the characterization of and separation between MOSH and MOAH three mineral oils were analyzed, for which two different LC columns were used. The samples were dissolved in hexane and injected in the LC, and fractions of twenty seconds were collected. For both columns, Table 1 shows the concentrations of the samples, the method and injection volume as well as the collection time. The eluents that were used are hexane and dichloromethane (DCM) (Biosolve Chimie SARL); details of the LC methods are given in Appendix 1.

Table 1: LC methods and concentrations of samples analyzed for the characterization Column conc. MO (mg/mL) method injection volume (µl) collection time (min) Ag/Silica 50 9 40 4.00 – 25.00 Ag/Polymer 50 21 40 4.00 – 25.00

(11)

11 3.2B Analysis by GC-FID

After the fraction collection the samples were injected (1 μL) into the GC-FID (splitless technique, 2 min., split 100:1). The oven temperature was programmed at 15 °C /min from 60 °C (3 min) to 350 °C (3 min). The carrier gas was helium at a flow rate of 2 mL/min. The data was processed by the program Wolfram Mathematica.

3.3 Quantification

3.3A Solid Phase Extraction

Quantitative separation of aliphatic and aromatic hydrocarbons was carried out using silver ion-silica solid-phase extraction.34,35 Ag/Silica (0.5 g) (Sigma Aldrich) was loaded into an cartridge using two retaining frits and placed in an oven (120 °C, 2 h) for activiation. The Ag/Silica was washed using dichloromethane (10 mL) followed by a conditioning step with hexane (4 mL). The sample (0.5 mL, 20 mg/mL) including internal standards (0.006 mg/mL) was loaded onto the column. The aliphatic hydrocarbons are eluted with hexane (5 mL) and the aromatic hydrocarbons with hexane/dichloromethane (1:1, 6 mL) followed by dichloromethane (2 mL). The MOSH and MOAH fractions were analyzed using GC-FID (section 3.2B).

3.3B HPLC separation

For the quantification three mineral oils (MO) were analyzed, for which two LC columns were used. The samples were dissolved in hexane and injected in the LC, after which the MOSH and MOAH fractions were collected. Table 2 (Ag/Silica column) and Table 3 (Ag/Polymer column) show the concentrations of the samples and internal standards, the method and injection volume as well as the collection time of the fractions. The eluents that were used are hexane and dichloromethane (DCM); details about the LC methods are listed in Appendix 1.

(12)

12

Table 2: LC methods and concentrations of samples analyzed for the quantification - Ag/Silica column Sample conc. MO (mg/mL) conc. I.S. (mg/mL) method injection volume (µl) collection time MOSH (min) collection time MOAH (min) MO A 100 0.150 12 20 3.6 – 5.5 5.5 – 15.00 MO E 100 0.150 12 20 3.6 – 5.5 5.5 – 15.00 MO H 100 0.150 12 20 3.6 – 5.5 5.5 – 15.00

Table 3: LC methods and concentrations of samples analyzed for the quantification - Ag/Polymer column Sample conc. MO (mg/mL) conc. I.S. (mg/mL) method injection volume (µl) collection time MOSH (min) collection time MOAH I (min) collection time MOAH II (min) MO A 100 0.150 24 20 4.00 – 6.66 6.66 – 8.33 8.33 – 30.00 MO E 100 0.150 24 20 4.00 – 6.66 6.66 – 8.33 8.33 – 30.00 MO H 100 0.150 21 20 4.00 – 6.66 6.66 – 8.33 8.33 – 30.00 3.3C Analysis by GC-FID

After the MOSH and MOAH fraction collection the samples were evaporated to 0.5 mL and 1 μL was injected into the GC-FID. The same program was used for the GC-FID as described in section 3.2B. The quantification was done using the internal standards (‘I.S. method’) and by calculating the following ratio: area MOAH : area MOSH+MOAH (‘area method’).

(13)

13

4. Results

4.1A Separation between MOSH and MOAH internal standards

Table 4 shows the fractions in which the MOSH and MOAH internal standards elute as well as the separation time between these standards. The same methods are used as for the characterization. The corresponding LC chromatograms are shown in Figure 5 (Ag/Silica column) and Figure 6 (Ag/Polymer column). The fractions are each 20 seconds and starting from 4 minutes (section 3.2A).

Table 4: Separation MOSH and MOAH internal standards

Figure 5: LC-UV chromatogram Ag/Silica column, I.S. 0.150 mg/mL + perylene, method 9.

Column fractions MOSH

I.S. start fraction MOAH I.S. separation time (min) Ag/Silica 2 – 4 9 1.33 Ag/Polymer 5 – 8 12 1.00

(14)

14

Figure 6: LC-UV chromatogram Ag/Polymer column, I.S. 0.150 mg/mL , method 21.

4.1B Characterization of the mineral oils

In the Figures 7 – 12 the LCxGC-FID chromatograms of mineral oil A, E and H are shown. All mineral oils were analyzed using the Ag/Polymer column as well as the Ag/Silica column. In Appendix 2 the LC chromatograms of MO A, E and H are shown.

(15)

15

Figure 8: 2-D LCxGC-FID chromatogram of MO A (50 mg/mL) Ag/Polymer column

(16)

16

Figure 6: 2-D LCxGC-FID chromatogram of MO E (50 mg/mL) Ag/Silica column

Figure 7: 2-D LC-GC-FID spectrum of MO E (50 mg/mL) Ag/Polymer column Figure 9: 2-D LCxGC-FID chromatogram of MO E (50 mg/mL) Ag/Silica column

(17)

17

Figure 11: 2-D LCxGC-FID chromatogram of MO H (50 mg/mL) Ag/Silica column

(18)

18 4.2 Quantification

In Table 5 the results of the quantification of MOAH in MOH are set out. Solid phase extraction was used for the separation of MOSH and MOAH. The chromatograms and corresponding tables that were used for the calculations are shown in Appendix 3.

Table 5: Ratio MOAH in MOH (SPE)

MO area method I.S. method

A 0.26 0.23

E 0.46 0.38

H 0.31 0.20

In Table 6 the results of the quantification of MOAH in MOH are set out. The Ag/Silica column was used for the separation of MOSH and MOAH. The chromatograms and corresponding tables that were used for the calculations are shown in Appendix 3

Table 6: Ratio MOAH in MOH (Ag/Silica column)

A 0.26 0.3

E 0.47 0.48

H 0.34 0.35

In Table 7 the results of the quantification of MOAH in MOH is shown. The Ag/Polymer column was used for the separation of MOSH and MOAH. The chromatograms and corresponding tables that were used for the calculations are shown in Appendix 3.

Table 7: Ratio MOAH in MOH (Ag/Polymer column)

A 0.25 0.27

E 0.43 0.46

(19)

19

5. Discussion

5.1 Characterization: separation between MOSH and MOAH

5.1A Ag/Silica column

As shown in Table 4, the internal standards that establish the MOSH window elute until fraction 4 and the first MOAH marker starts at fraction 9. According to previous studies, all MOSH and MOAH should elute in these windows. 30,32 The LCxGC chromatograms of MO A, E and H (Fig. 7, 9, 11) show a gap after fraction 4. This possibly means that all MOSH elute as indicated by the marker. However, GC-VUV detection should be used to establish whether indeed no MOAH is present in the given MOSH window. Aromatics absorb at a higher wavelength and thus can be detected if they are present in the fractions. If analysis shows that the MOSH fractions only contain aliphatic compounds and the fractions after only aromatic compounds, 5-α-cholestane is effective as a marker to establish and check the end of the MOSH window. On the contrary, 1,3,5-tri-tert-butylbenzene would not be useful as a marker to establish the start of the MOAH window, since in all spectra MOAH seems to start at fraction 6 or 7 instead of fraction 9. A possible explanation of this could be that the analyzed mineral oils contain highly alkylated aromatic compounds, which due to steric hindrance do not interact as much with the silver stationary phase as less-alkylated aromatics. As a consequence, these compounds elute before than the MOAH marker, which is only moderately alkylated.

5.1B Ag/Polymer column

As shown in Table 4, the internal standards that establish the MOSH window elute until fraction 8. This should mean that all MOSH elute until this fraction as well, according to previous studies. The start of the MOAH standards is at fraction 12, meaning that all MOAH should elute starting from this fraction. The LCxGC chromatograms of MO A (Fig. 8) shows a gap at fraction 11. This could mean that in fraction 12 the first MOAH elutes, and in fraction 10 the last MOSH. Important to note, between fraction 6 and 10 a different kind of pattern can be observed. Possibly, these are aromatic compounds that are highly alkylated. To establish whether these fractions contain aromatic or aliphatic compounds or both, GC-VUV should be used.

(20)

20 The 2-D LCxGC chromatogram of MO E (Fig. 10) shows a small gap at fraction 10, which possible means that MOSH elutes before and MOAH after this fraction. The 2-D LCxGC chromatogram of MO H (Fig. 12) shows that possibly MOSH elutes before fraction 9 and MOAH after fraction 11, which is exactly as the internal standards indicated.

For the Ag/Polymer column, the 2-D LCxGC chromatograms of MO A and E show that between the expected MOSH and MOAH windows, which are established using markers, compounds elute. This probably is a consequence of the presence of highly alkylated aromatic rings. If analysis by GC-VUV shows that these compounds are aromatics, a higher alkylated benzene molecule needs to be used as a marker to establish the MOAH window.

5.1C Comparison columns

The main objective of this study was to establish a separation between Mineral Oil Saturated Hydrocarbons and Mineral Oil Aromatic Hydrocarbons. Analysis of the internal standards using comprehensive LCxGC-FID, showed that a separation of 1 minute (Ag/Polymer column) and 1 minute and 20 seconds (Ag/Silica column) between the MOSH and MOAH markers was established. However, analysis of the LCxGC chromatograms of the samples showed that compounds elute in the timeframe between the MOSH and MOAH markers. The findings in this study are contrary to previous studies, which have suggested that the markers should correctly establish the MOSH and MOAH windows. This result may be explained by the fact that in previous studies, a silica-phase column was used opposed to silver-phase columns in this study. However, in the previous study a separation of 30 seconds was established between the markers, whereas in this study this separation was (more than) 1 minute. This would suggest that the separation between MOSH and MOAH in this study is better. A possible explanation is that in the previous study also compounds eluted in between the two windows, but because the analyzed concentration of the mineral oil hydrocarbons is lower than in this study, these compounds cannot be detected in the GC-FID. Another possible explanation for the compounds eluting in between the windows is the high concentration of mineral oil that is analyzed in this study, which might result in an overloaded column. Furthermore, highly alkylated benzenes that might be present in the samples could elute at similar retention times as MOSH due to steric hindrance.

(21)

21 Although the expected separation between MOSH and MOAH around 1 minute was not established for the mineral oil samples, the separation seems to be slightly better when using the Ag/Silica column. This could possibly be explained by the favorable interactions between silica and aromatics. Moreover, the silica and silver are less tightly bound than the polymer and the silver, leaving more room for the strong π-interactions between silver and the aromatics.

5.2 Characterization: classification aromatic compounds

5.2A Ag/Silica column

The LC chromatogram of the internal standards (Fig. 5) on the Ag/Silica column show that the mono-ring aromatics elute together, and the di-ring aromatics separately. The LCxGC chromatograms also show, although not clearly, two bands in the aromatic part. This could be a slight separation between the mono and poly-ring aromatics, but to be certain GC-VUV could be used, since different absorbance spectra are expected for mono- and poly-ring aromatics.

5.2B Ag/Polymer column

The LC chromatogram of the internal standards (Fig. 6) on the Ag/Polymer column shows that the aromatics are divided into classes, according to the number of rings and degree of alkylation. The internal standards appear as individual peaks, in which the first peaks are single aromatic ring species that are sterically protected by a high degree of alkylation. These are followed by MOAH with more rings and/or less alkylation. This effect is also observed in the LCxGC chromatograms of the mineral oils (Fig. 8, 10, 12), in which bands of alternating intensity are visible. A higher intensity color responds to a higher responds in the GC-FID, and thus a higher concentration of MOH. According to the internal standards, the di-ring aromatics elute after fraction 30 (14 minutes), and the mono-ring aromatics before this fraction. The high intensity of fraction 31 is caused by the breakthrough of dichloromethane, since this eluent reduces the strong retention power of the column to retain MOAH.32_{Since mineral oil E has the highest intensity band after}

fraction 30, it probably has the most poly-ring aromatics. Mineral oil H has the lowest intensity band after fraction 30, meaning that it has the least multi-ring aromatics. To classify these compounds into sub-groups time-of-flight mass detection could be explored, since compounds can be distinguished by selecting their unique masses.35_{Another method of detection of aromatic}

(22)

22 Regarding the retention times of the mineral oil hydrocarbons in the GC-FID, a downwards shift can be observed for mineral oil H compared to the other samples. This means that the average boiling point of these hydrocarbons is lower, meaning that they have a lower average molecular mass.

5.2C Comparison columns

In this study, one of the objectives was to classify aromatic compounds in terms of the number of rings. For this, three mineral oils were analyzed using comprehensive LCxGC-FID. The results showed that the Ag/Polymer column is most suited to separate aromatics into groups. This might be explained by the length of the column, which is twice as long as the Ag/Silica column. An Ag/Silica column of similar length should be used to conclude which stationary phase is more suitable to separate aromatics into classes. Moreover, time-of-flight mass detection could be used as well as GC-VUV, to determine whether the separation into classes was established.

5.3 Quantification

One of the aims of this study was to determine the aliphatic and aromatic content of the mineral oil samples and to compare them to the standard method based on Solid Phase Extraction. Analysis of the samples that were separated using SPE, showed that MO A contains about 26% MOAH, MO E 46% and MO H 31% (Table 5). This was determined using the ratio area MOAH: area MOSH+MOAH, since the internal standard method was not as accurate due to overlapping peaks in the GC chromatograms. To determine the aromatic contents using the internal standards, analysis should be done with different internal standards, or a lower concentration mineral oil. The samples that were separated by LC were analyzed using a slightly shorter GC columns and had a lower concentration, which resulted in less or no overlapping of the internal standards. Hence, the area method as well as the internal standard method can be used to determine the amount of aromatics in the mineral oils.

The MOSH and MOAH fraction collection time for the Ag/Silica column was determined using the marker 5-α-cholestane. The LCxGC chromatograms show that a separation of MOSH and MOAH supposedly happens after the fourth fraction. Analysis of the MOSH and MOAH fractions showed that MO A contains about 28% aromatics, MO E 48% and MO H 35% (Table 6).

(23)

23 The fraction collection time for the Ag/Polymer column was determined by analysis of the 2-D chromatograms of the samples. As mentioned previously, the 2-D chromatograms show that between the expected MOSH and MOAH windows, compounds elute. Since these are most likely highly alkylated aromatics, this fraction was treated as MOAH. Analysis of the fractions by GC-FID showed that MO A contains around 26% aromatics, MO E 45% and MO H 31% (Table 7).

When these determined percentages of aromatics in mineral oils are compared, they appear highly similar for each method. This could mean that the LC-GC method is reliable for determining the amount of aromatics in mineral oils using both columns. However, these findings may be somewhat limited by the uncertainty of separation between MOSH and MOAH. The collected MOSH and MOAH fractions should be analyzed by GC-VUV to determine whether a complete separation was established. Moreover, since MOSH and MOAH appear as a huge hump in the GC chromatograms, a baseline uncertainty should be taken into account.

6. Conclusion

The main objective of this study was to establish a separation between Mineral Oil Saturated Hydrocarbons and Mineral Oil Aromatic Hydrocarbons. Analysis of the internal standards using comprehensive LCxGC-FID, showed that a separation of 1 minute (Ag/Polymer column) and 1 minute and 20 seconds (Ag/Silica column) between the MOSH and MOAH markers was established. However, analysis of the LCxGC-FID chromatograms of three mineral oil samples with different grades of refinement showed that compounds elute in the timeframe between the MOSH and MOAH markers. These compounds could be highly alkylated benzenes, which due to steric hindrance can interact less with the stationary phase and thus elute at similar retention times as MOSH.

The second aim of this study was to classify aromatic compounds in terms of the number of rings. The results showed that the Ag/Polymer column is most suited to separate aromatics into groups, which is possibly a result of the length of the column. To establish whether the difference in separation between the aromatics is due to the difference in stationary phase or the length, a silver-silica phase column of equal length should be used in the analysis of the mineral oils. Another objective of this study was to determine the aliphatic and aromatic content of the mineral

(24)

24 oil samples and to compare them to the standard method based on Solid Phase Extraction. When these determined percentages of aromatics in mineral oils were compared, they appeared highly similar for each method. This could mean that the LC-GC method is reliable for determining the amount of aromatics in mineral oils using both columns. However, a base-line uncertainty as well as the uncertainty in the separation between MOSH and MOAH should be taken into account. Further research should be undertaken to determine whether a separation between MOSH and MOAH was established, for example by analyzing the fractions by GC-VUV.

7. Outlook

Further research should be carried out to establish whether a separation was obtained between MOSH and MOAH. The fractions could be analyzed by GC-VUV to determine what kind of compounds elute between the expected MOSH and MOAH windows. Another method to determine this is to analyze the three mineral oil samples that are separated using solid phase extraction into MOSH and MOAH. These fractions have already been analyzed using the LC-GC-FID method. However, these results have yet to be analyzed. A further study could also classify the groups of aromatics regarding the number of rings or even into subgroups by using GC-MS-ToF. In order to establish a difference in retention time between MOSH and MOAH in the GC-FID, a 50% phenyl–50% dimethylpolysiloxane column could be used. Since this column has a high Kovat’s retention index for benzene, it is expected that the aromatic compounds will be retained longer than the aliphatic compounds. As a consequence, a shift upwards in the LCxGC chromatograms of the mineral oil might be visible for the aromatic part.

Ultimately, consumer products will be analyzed to characterize the mineral oil hydrocarbons and to quantify the mineral oil aromatic hydrocarbons.

(25)

25

8. References

1. EFSA (European Food Safety Authority). EFSA J. 2012, 10 , 1–185. 2. Tennant, D.R. Food Chem. Tox. 2004, 42, 481–492.

3. Moret, S.; Grob, K.; Conte, L.S. Lebensm. Unters. Forsch. 1997, 204, 241–246. 4. Biedermann, M.; Grob, K. Eur. J. Lipid Sci. Technol. 2009, 111, 313.

5. Neukom, H.P.; Grob, K.; Biedermann, M. Atmos. Environ. 2002, 36, 4839–4847. 6. Grob, K.; Huber, M.; Boderius, U. Food Addit. Contam. 1997, 14, 83–88.

7. Wagner, C.; Neukom, H.P.; Grob, K. Mitt. Lebensm. Hyg. 2001, 92, 499–514. 8. Moret, S.; Populin, T.; Conte, L.S. Food Addit. Contam. 2003, 20, 417–426. 9. Niederer, M.; Stebler, T.; Grob, K. Int. J. Cosmet. Sci. 2016, 38, 194–200.

10. Baxter, J. H.; Steinberg, D.; Mize, C.E. Biochimica et Biophysica Acta. 1967, 137, 277–290. 11. Albro, P. W.; Fishbein L.. Biochimica et Biophysica Acta. 1970, 219, 437–446.

12. Albro, P.W.; Thomas, R. Bull Environ Contam Toxicol. 1974, 12, 289–294. 13. Tulliez, J. E.; Bories, G. F. Lipids. 1979, 14, 292–297.

14. Firriolo, J.M.; Morris, C. F.; Trimmer, G. W. Toxicol. Pathol. 1995, 23, 26–33.

15. Griffis, L. C.; Twerdok, L. E.; Francke-Carroll, S. Food Chem. Toxicol. 2010, 48, 363–372. 16. Cnubben, N. H. P.; van Stee, L. L. P. TNO Quality of Life Report V8503 (unpublished study report), 2010. Submitted to EFSA in May 2011.

17. Smith, J.H.; Mallett, A. K.; Priston, R. A. Toxicol. Pathol. 1996, 24, 214–230. 18. Trimmer G.W.; Freeman, J. J.; Priston, R. A. Toxicol. Pathol.. 2004, 32, 439–447. 19. Boitnott, J. K.; Margolis, S. Johns Hopkins Med. J. 1970, 127, 65–78.

20. Concin, N.; Hofstetter, G.; Plattner, Food Chem. Toxicol. 2008, 46, 544–552. 21. Granella, M.; Clonfero, E. Int. Arch. Occup. Environ. Health, 1991, 63, 149–153. 22. Ingram, A.J.; Phillips, J. C.; Davies, S. J. Appl. Toxicol., 2000, 20, 165–174.

23. Roy, T. A.; Johnson, S. W.; Blackburn, G. R.; Mackerer, C. R. Fundam. Appl. Toxicol.

1988, 10, 466–476.

24. Mackerer, C. R.; Griffis, L. C.; Grabowski Jr, J. S.; Reitman, F. A. Appl. Occup. Environ Hyg. 2003, 18, 890–901.

25. Exxon Biomedical Sciences. Project 158830 (unpublished study report). 1991c, submitted to EFSA in August 2011.

(26)

26 26. Shell. Toxicology Report No. 98.1315 (unpublished study report). 1998a, submitted to EFSA in May 2012.

27. Shell. Toxicology Report No. 98.1317 (unpublished study report). 1998b, submitted to EFSA in May 2012.

28. Granella, M.; Clonfero, E. Int. Arch. Occup. Environ. Health. 1991, 63, 149–153. 29. Rivedal, E.; Mikalsen, S. O.; Roseng, L. E. Pharmacol. Toxicol. 1992, 71, 57–61. 30. Biedermann, M. Grob, K. J. Chromatogr. 2012, 1255, 56– 75.

31. Grob, K.; Lanfranchi, M.; Egli, J. AOAC Int. 1991, 74, 506.

32. Biedermann, M.; Fiselier, K.; Grob, K. J. Agric. Food Chem. 2009, 57, 8711. 33. Bennet, B.; Larter, S. R. Anal. Chem. 2000, 72, 1039–1044.

34. Moret, S.; Barp, L.; Grob, K. Food Chem. 2011, 129, 1898–1903.

(27)

27

9. Appendices

Appendix 1: LC methods

Table 8: LC method 21 (gradient) Time (min) hexane (%) DCM (%) Flow (mL/min) 0.00 100 0 0.5 5.90 100 0 0.5 6.00 50 50 0.5

Table 10: LC method 9 (isocratic) Time (min) hexane (%) DCM (%) Flow (mL/min) 0.00 100 0 0.3

(28)

28

Appendix 2: Characterization chromatograms

Figure 13: MO A 50 mg/mL Ag/Silica column

(29)

29

Figure 15: MO E 50 mg/mL Ag/Silica column

(30)

30

(31)

31

Figure 18: MO H 50 mg/mL Ag/Polymer column

Appendix 3: Quantification chromatograms

(32)

32

Figure 20: MO A 100 mg/mL Ag/Polymer column

Figure 21: MO E 100 mg/mL Ag/Silica column

(33)

33

Figure 22: MO E 100 mg/mL Ag/Polymer column

Figure 23: MO H 100 mg/mL Ag/Silica column

(34)

34

Figure 24: MO H 100 mg/mL Ag/Polymer column

Appendix 4: GC-FID chromatograms quantification

(35)

35

Table 12: GC-FID Ag/Silica column MOSH A2 100 mg/mL

Peak Retention time Area Ratio Concentration

(mg/mL) undecane 3.932883 1751903 bicyclohexyl 6.20558 1711231 1 0.150 tridecane 6.31775 2074898 MOSH 14.6053 554237859 323,88 48.58 5-α-cholestane 15.5762 2672122

Figure 26: Ag/Silica column MOAH A2 100 mg/mL

(36)

36

Table 13: GC-FID Ag/Silica column MOAH A2 100 mg/mL

Peak Retention

time

Area Ratio Concentration

(mg/mL) hexylbenzene 5.82717 1564508 1-methylnaphthalene 6.23725 1550019 1 0.150 biphenyl 6.90308 2861023 1,3,5-tri-tert-butylbenzene 7.50617 1734828 nonylbenzene 8.5305 1682918 MOAH 13.6331 184899958 119.29 17.89

(37)

37

Table 14: GC-FID Ag/Silica column MOSH E2 100 mg/mL

time

Area Ratio Concentration (mg/mL)

undecane 3.92533 1861746

bicyclohexyl 6.20183 1879537 1 0.150

tridecane 6.31675 1991460

MOSH 15.2425 741146168 394.32 59.15

5-α-cholestane 15.5865 1419054

(38)

38

Table 15: GC-FID Ag/Silica column MOAH E2 100 mg/mL

time

(39)

39

Table 16: GC-FID Ag/Silica column MOSH H2 100 mg/mL

Peak Retention time Area Ratio Concentration

(mg/mL)

bicyclohexyl 6.19742 1678621 1 0.150

tridecane 6.31183 2249088

MOSH 15.0675 1001065047 596.36 89.45

5-α-Cholestane 15.5638 1931779

(40)

40

Table 17: GC-FID Ag/Silica column MOAH H2 100 mg/mL

time

(41)

41

Table 18: GC-FID Ag/Polymer column MOSH A 100 mg/mL

(mg/mL) undecane 3.80975 168643 undecane 3.866 1337306 bicyclohexyl 6.16367 1555108 1 0.150 tridecane 6.28142 1673189 5-α-cholestane 15.5551 1508170 MOSH 15.9532 496911916 314.60 47.19

(42)

42

Table 19: GC-FID Ag/Polymer column MOAH I E 100 mg/mL

Peak Retention time Area Concentration

(mg/mL)

bicyclohexyl 6.16858 24414

MOAH 14,2743 33733097 2.244943712

(43)

43

Table 20: GC-FID Ag/Polymer column MOAH II A 100 mg/mL

(mg/mL) hexylbenzene 5.78875 1334276 1-methylnaphthalene 6.2045 1302254 biphenyl 6.8745 1323919 1 0.150 1,3,5-tri-tert-butylbenzene 7.47825 1399619 nonylbenzene 8.508 1292558 MOAH 8.75158 145253563 109.71 16.46

(44)

44

Table 21: GC-FID Ag/Polymer column MOSH E 100 mg/mL

(mg/mL) undecane 3.88175 1542899 bicyclohexyl 6.16758 1505222 1 0.150 tridecane 6.28467 1627921 5-α-cholestane 15.5582 1007296 MOSH 15.9235 318337985 211.49 31.72

Figure 34: Ag/Polymer column MOAH I E 100 mg/mL

Table 22: GC-FID Ag/Polymer column MOAH I E 100 mg/mL

(mg/mL)

(45)

45

Figure 35: Ag/Polymer column MOAH II E 100 mg/mL

Table 23: GC-FID Ag/Polymer column MOAH II E 100 mg/mL

(46)

46

Figure 36: Ag/Polymer column MOSH H 100 mg/mL

Table 24: GC-FID Ag/Polymer column MOSH H 100 mg/mL

(mg/mL) undecane 3.88225 1507808 bicyclohexyl 6.16608 1570748 1 0.150 tridecane 6.28392 1736569 5-α-Cholestane 15.5457 852628 MOSH 6.94292 525145915 334.33 50.15

(47)

47

Figure 37: Ag/Polymer column MOAH I H 100 mg/mL

Table 25: GC-FID Ag/Polymer column MOAH I H 100 mg/mL

(mg/mL)

5-α-cholestane 15.5452 35706

(48)

48

Figure 38: Ag/Polymer column MOAH II H 100 mg/mL

Table 26: GC-FID Ag/Polymer column MOAH II H 100 mg/mL