728 Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... DOI: 10.17344/acsi.2021.6817 Scientific paper Quantification of Hydroperoxides by Gas Chromatography with Flame Ionisation Detection Damjan Jan Pavlica, Črtomir Podlipnik and Matevž Pompe* Faculty of Chemistry and Chemical Technology, University of Ljubljana, Ljubljana, 1001, Slovenia * Corresponding author: E-mail: Matevz.Pompe@fkkt.uni-lj.si Received: 03-12-2021 Abstract Hydroperoxides are of great importance in the fields of atmospheric and biological chemistry. However, there are several analytical challenges in their analysis: unknown and usually low UV absorption coefficients, high reactivity, thermal instability, and a lack of available reference standards. To overcome these limitations, we propose a GC-FID approach involving pre-column silylation and quantification via the effective carbon number approach. Four hydroperoxides of α-pinene were synthesized in the liquid phase with singlet oxygen and identified using literature data on isomer yield distribution, MS spectra, estimated boiling temperatures of each isomer (retention time), their thermal stability and deri- vatisation rate. The developed procedure was used for the determination of hydroperoxides in bottled and autooxidised turpentine. We anticipate that this method could also be applied in atmospheric chemistry, where the reactivity of singlet oxygen could help explain the high formation rates of secondary organic aerosols. Keywords: hydroperoxides, α-pinene, photooxidation, singlet oxygen, gas chromatography 1 Introduction Organic hydroperoxides are used industrially as rad- ical initiators, bleaching agents, and disinfectants. They are formed in the process of oxidative ageing, which they si- multaneously promote by radical chain reactions. In ethe- real solvents, they can be stable at low concentrations but become explosive at higher concentrations. Degradation by peroxidation decomposes all organic matter and is haz- ardous to health because hydroperoxides are irritating to skin, eyes, and mucous membranes and are potent aller- gens.1 In rats, they induce progressive oxidative damage and cell death when inhaled.2 Hydroperoxides (HPs) are formed in nature as prima- ry oxidation products of volatile organic compounds, for example, α-pinene, which is emitted from coniferous trees. This compound is the most abundant monoterpene in the air and plays an essential role in the growth of atmospheric particles.3 It is present in essential oils and thus in various types of cosmetic and cleaning products. It is also the main component of turpentine, which is used as a paint thinner and as an ingredient in paints, polishes, adhesives, topical remedies and household chemicals. It has been found that 3.1% of the German population is allergic to turpentine.4 The most likely major haptens in turpentine are Δ3-carene hydroperoxide and oxidation products of α- and β-pinene.5 Despite the need to monitor and quantify HPs in various matrices, their analysis is complicated due to low UV absorption, thermal instability, catalyzed decomposi- tion, and lack of available reference standards. Quantifica- tion is mainly performed by chemical assays, such as the iodometric6 or triphenylphosphine assay7 or assays with other reducing agents, followed by an analysis of the reac- tion products.8 However, these methods only provide in- formation on the total amount of HPs present, and inter- ference by other compounds cannot be excluded. For the monitoring and quantification of specific HPs, chromato- graphic and NMR methods can be used. Some authors reported using gas chromatography (GC) methods without derivatisation, but only for HPs with low molecular masses.9 HPs with higher molecular masses are partially decomposed at high oven elution tem- peratures and therefore often derivatised to more thermo- stable species. Most methods involve silylation11,12 or re- duction of HPs to alcohols with sodium sulfite,9,13 sodium borohydride,14 triphenylphosphine9,14 or trimethyl phos- phine.15 Derivatisation to alcohols can be used if the re- sulting alcohols were not previously present in the sample. HPs in the gas phase can be analysed directly by chemical ionization mass spectrometry.16 High-pressure liquid chromatography (HPLC) for HP quantification is very convenient because separation 729Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... occurs at lower temperatures. However, due to lack of chromophores, HPs must be detected by post-column re- actions or by MS. Post-column reactions include a method using phosphine (the fluorescent product phosphine oxide is formed)17 or a chemiluminescence reaction using lumi- nol.18 The preferred MS ionisation techniques for detect- ing terpene HPs are electrospray ionisation (ESI)19,20 and atmospheric pressure chemical ionisation (APCI).19,21 Post-column reactions are specific for the peroxy func- tional group, whereas in MS, specific fragment loss of 34 Da (loss of H2O2) is observed sporadically.21 Identification of the peroxy functional group can be confirmed by dual injection, with and without iodometric sample pretreat- ment, which reduces HP species to alcohols.20 Quantification of α-pinene HPs is very demanding because reference standards are nonexistent. Additionally, HPs have limited stability, so reliable quantitative methods are needed to assess purity, such as GC-FID with predicted relative response factors or NMR.11 Quantitative NMR spectrometry is a universal, non-destructive, absolute de- tection technique and provides a quantitative reference for other analytical methods. Analytes in the μM concentra- tion range can be detected, with precision and accuracy of around 1%.22 The authenticity of individual spectra can be assessed by generating various one-dimensional and mul- tidimensional experiments. The major hurdles are sensi- tivity, spectral overlap, dynamic range, selection of the in- ternal standard, interpretation and processing of the spectra, and the use of expensive equipment and deuterat- ed solvents. Therefore, when performing routine targeted analysis, optimized molecule-specific chromatographic methods are preferred. GC-FID has a dynamic range of 107 and the analysis time depends only on the mixture composition and not on the concentration as in NMR. In our case, the separation of isomers took 30 minutes. In the absence of calibration standards, the relative concentra- tions of the organic peroxides can be estimated from the GC-FID peak intensities by peak area normalization ap- proach, application of the effective carbon number (ECN) concept, or by some other algorithm based on the chemi- cal structure of the analytes.23 To date, only two HPs have been synthesised in the re- action of α-pinene with singlet oxygen.13,14 Electrophilic sin- glet oxygen (1O2) reacts with a double bond in the ene addi- tion reactions, where allylic hydrogen is abstracted to give allyl-HPs in which the double bond has migrated. The reac- tion of singlet oxygen with α-pinene in this manner gener- ates pinocarvyl-hydroperoxide and 4-hydroperoxy-4,6,6-tri- methylbicyclo[3.1.1]hept-2-ene (Fig.1.). The 1O2 attack on the double bond occurs on the sterically less congested π face. The two methyl groups on the methylene bridge are dis- tinctively anti-directing; therefore, the HPs resulting from the syn attack are formed only in trace amounts.14,24 Upon storage in solution, the OOH group can migrate to the other side of the double bond,25 which has already been observed as the rearrangement of pinocarvyl-hydroperoxide to myrte- nyl-hydroperoxide.9 In this work, we observe for the first time the rearrangement of 4-hydroperoxy-4,6,6-trimethylbi- cyclo[3.1.1]hept-2-ene (HP2) to verbenyl-hydroperoxide. In the absence of isolated reference standards, the identification of separate peaks in the GC chromatogram was based on literature data on isomer yield distribution, MS spectra, estimated boiling temperatures of individual isomers (retention time), their thermal stability, and rate of derivatisation. Trimethylsilylation increased the ther- mostability and allowed us to validate linearity, selectivity and repeatability of the GC-FID method. The concept of the effective carbon number allowed determination with- out standards of known purity. 2 Experimental Section 2. 1. Chemicals, Synthesis of HPs and Air Exposure Procedure For the synthesis of the HPs, we have used: α–pinene, >97% purity, Fluka (Buchs, Switzerland), methylene blue, Merck (Darmstadt, Germany) and HPLC grade acetoni- trile, ≥99.9% purity, Fischer (Zürich, Switzerland). HPs of α-pinene were synthesised by a modified photochemical procedure.13,14 Photooxidation of α-pinene Figure 1. Structures of the hydroperoxides studied: pinocarvyl-hydroperoxide 1, 4-hydroperoxy-4,6,6-trimethylbicyclo[3.1.1]hept-2-ene 2, myrte- nyl-hydroperoxide 3 and verbenyl-hydroperoxide 4. 730 Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... was carried out in a flask at room temperature in acetoni- trile using methylene blue as a sensitiser and a 60 W household daylight lamp as a light source. The flask was opened to allow oxygenation and mixed manually every 12 h for 14 days, followed by analysis by GC-MS and GC- FID. The structures of four resulting HPs of α-pinene are shown in Fig. 1. Derivatisation reagent N-Methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA) was purchased from Fluka (Buchs, Switzerland), toluene from Sigma-Aldrich (Taufkirchen, Germany), cumene-hydroperoxide, 80% purity from Sigma-Aldrich (Taufkirchen, Germany), tet- radecane of >99% purity from Merck (Schuchardt, Ger- many). Turpentine was purchased from HGtrade (Ljubljana, Slovenia). A sample of turpentine was exposed to air in an Erlenmeyer flask at room temperature and under a 60-watt household daylight lamp. The neck of the flask was cov- ered with aluminium foil to prevent contamination. The flask was stirred daily. After 20 days, the sample was deri- vatised, and the specific HPs were determined by GC-FID. 2. 2. Derivatisation Procedure For the analysis of turpentine oil ≈ 200 mg of sample was weighed into a vial, then ≈ 200 mg internal standard solution (3 mg/g cumene-hydroperoxide in toluene) and ≈ 200 mg MSTFA (250 µL) were precisely weighted. The vial was closed, mixed by hand, and kept at room temperature for 2 h. 1 μL of the resulting solution was injected into the GC-FID. For calibration, the following procedure was used: A stock solution of cumene-hydroperoxide at 2.5 mg/mL was prepared in acetonitrile and stored at 5 °C, calibration solutions (0.6, 1, 6, 25, 50, 90 μg/mL) were further diluted in acetonitrile. From each calibration solution, an aliquot of 0.4 mL was transferred to a vial, to which 0.4 mL of in- ternal standard tetradecane (40 mg/kg in toluene) and 0.4 mL of the derivatisation reagent MSTFA (50 mg/g in tolu- ene) were added. The vial was closed, mixed by hand, and kept at room temperature for 2 h. 1 μL of the resulting solution was injected into the GC-FID. The derivatised HP solutions were found to be stable in the refrigerator for at least three days. 2. 3. Instrumentation and Analysis The GC separation was performed on GC Trace 1300, Thermo Scientific (Waltham, USA), equipped with a Rxi–5Sil MS column from Restek (Bellefonte, USA), 30 m x 0.32 mm x 0.25 μm. The carrier gas was helium under a constant flow of 2 mL/min and a split ratio of 50:1. The injector and FID temperatures were 250 and 280 °C, re- spectively. The oven was held at 60 °C for 0.3 min; then the temperature was raised to 80 °C at a rate of 5 °C/min and held for 3 min, then the temperature was raised to 160 °C at a rate of 5 °C/min and to 275 °C at a rate of 40 °C/min and held for 4 min. The GC-MS separation was performed on GC Trace 1310 and MS TSQ 9000 from Thermo Scientific (Waltham, USA). A Restek (Bellefonte, USA) 5-MS column with 0.25 μm film thickness (30 m x 0.25 mm i.d.) was used for sep- aration. The temperature programme was translated from GC-FID with the help of EZGC, an online freely available method translator tool from Restek (Bellefonte, USA). The carrier gas was helium under a constant flow of 1.56 mL/ min. The injector and transfer line temperatures were 250 and 280 °C, respectively. The oven was held at 60 °C for 0.1 min; then the temperature was raised to 80 °C at a rate of 5.6 °C/min and held for 2.95 min, then the temperature was raised to 160 °C at a rate of 5.1 °C/min and raised to 275 °C at a rate of 38.4 °C/min and held for 4.15 min. The temperature of the ion source was 250 °C. 2. 4. Quantification Due to the lack of commercially available standards for the HPs, we used the concept of effective carbon num- ber (ECN) to calculate the response factors. The ECN is calculated using the contributions of different molecular structures with the error of predicting about 3% RSD.26 Since there are no recommendations for calculating the ECN of trimethylsilyl peroxides, we treated these com- pounds as the corresponding trimethylsilyl oxides with ECN for the H-C-O-TMS group = 3.69. The relative mass response factors of silylated peroxides were calculated us- ing the following equation: (1) where r = reference compound (cumene HP); x = uncali- brated compound and Mr = molecular mass. 3 Results and Discussion 3. 1. Qualitative Analysis Irradiation of α-pinene in acetonitrile solution with methylene blue as sensitizer resulted in four HPs. Initially, pinocarvyl-hydroperoxide and later 4-hydroper- oxy-4,6,6-trimethylbicyclo[3.1.1]hept-2-ene were formed. When methylene blue was replaced by rose bengal, no change in the reaction products was observed. Further- more, the same products were obtained by chemically pre- pared 1O2 in the reaction between NaOCl and H2O2, all confirming the involvement of 1O2 in the product forma- tion. Continuing the synthesis, two more HPs were formed, probably not only by rearrangement reactions25 but also by radical mechanisms,15 with H abstraction from α-pinene by peroxyl radicals and 3O2 addition. A typical chromatogram of the optimised separation of the four isomers is shown in Fig. 2. In the absence of 731Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... standards, the assignment of separation order was based on literature data on isomer yield distribution and estimat- ed boiling temperatures (retention time). The identifica- tion was later confirmed with MS spectra, thermal stability and rate of derivatisation. The most abundant HP in the reaction of 1O2 with α-pinene is HP1, with an absolute yield of 99%.14 It is reasonable to assume that the structur- al variations between the isomers do not affect their FID detector response; if so, the chromatogram’s largest peak belongs to pinocarvyl-hydroperoxide (HP1). The remain- ing three isomers can be compared in order of elution be- cause chromatographic retention time depends on chemi- cal structure (size, shape, charge, and composition). For the isomers, the more branched the chain, the lower the boiling point tends to be. Therefore, the tertiary HP 4-hy- droperoxy-4,6,6-trimethylbicyclo[3.1.1]hept-2-ene (HP2) elutes first, and the primary HP myrtenyl-hydroperoxide (HP3) elutes last. The remaining peak belongs to the ver- benyl-hydroperoxide (HP4). Figure 2. GC-FID chromatogram of four HP isomers obtained by photooxygenation of α-pinene: Cumene-hydroperoxide (IS, 14.5 min), 4-hydroperoxy-4,6,6-trimethylbicyclo[3.1.1]hept-2-ene (2, 16.4 min), pinocarvyl-hydroperoxide (1, 16.7 min), verbenyl-hy- droperoxide (4, 16.8 min) and myrtenyl-hydroperoxide (3, 17.1 min). Retention times are given in parentheses. 3. 2. Derivatisation Ideally, one would prefer to detect HPs directly, without derivatisation.9 To test this possibility, different injector temperatures were compared (from 70 °C to 250 °C), and significant decomposition of HPs was observed. Primary HPs are known to be the most thermolabile,12 and indeed, 20% of HP3 was degraded with temperature. HP1 was the least decomposed at 10%. To test the effect of deg- radation on the column, the analysis was performed under a fast and slow temperature gradient. The HPs elute at about 130 °C, and at this temperature partial decomposi- tion has already been observed in the injector. However, since the compounds spend most of their retention time dissolved in the liquid stationary phase, this could stabilize them. Therefore, we additionally tested the decomposition on the column with fast and slow temperature gradient. Under a fast temperature gradient, we quantified 3 to 9% more specific HPs, confirming the decomposition in the column. This rules out the possibility of avoiding thermal degradation by cool-on-column injection, so α-pinene HPs require derivatisation for quantitative determination. Derivatisation to alcohols requires that the resulting alcohol was not previously present in the quantified prod- uct mixture or that its concentration was known before- hand. Essential oils of conifers and hence our sample, tur- pentine, contain some proportion of corresponding alcohols. Alcohols are also formed after the degradation of hydroperoxides. Neuenschwander et al.15 determined HPs via double injection, with and without reduction. The HP yield was quantified from the increase in alcohol content obtained, and no difference in yields was observed be- tween split injection at 250 °C and cool-on-column injec- tion at 50 °C. Since thermal degradation of HPs was ob- served in our experiments, they would be underestimated by this reduction method. We opted for silylation with MSTFA, in which the active hydrogens in the HPs are re- placed by a TMS group. Silylation has a shortcoming: it cannot be applied to consumer product matrices with high water or alcohol content (e.g. eau de toilette, detergents). After derivatisation, the positional isomers could be separated chromatographically with even better resolu- tion, while retention times increased by only 1-1.5 min (Fig. 3). A reversal in elution order was observed for com- pounds HP1 and HP4. This was confirmed by comparing their derivatised/underivatised MS spectra and by com- paring their GC-FID peak areas, as FID responses in- Figure 3. GC-FID chromatogram of TMS derivatives of α-pinene HPs obtained from the reaction of α-pinene with singlet oxygen: Cumene-hydroperoxide (IS, 15.8 min), 4-Hydroperoxy-4,6,6-tri- methylbicyclo[3.1.1]hept-2-ene (2, 17.0 min), verbenyl-hydroper- oxide (4, 17.8 min), pinocarvyl-hydroperoxide (1, 18.3 min) and myrtenyl-hydroperoxide (3, 18.8 min). Retention times are given in parentheses. 732 Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... creased proportionally to the addition of three carbon at- oms. The thermal stability of the TMS derivatives of HPs was investigated under different injector temperatures ranging from 70 °C to 270 °C. No adsorption on the col- umn was observed at low temperatures, and no thermal decomposition was observed up to 250 °C. The repeatabil- ity of derivatisation at LOD (1 ppm, n=6) showed an RSD of 4.6%; thus, the method allows accurate determination. Since HPs decompose at higher temperatures, we derivatised HPs at room temperature. The stability of HPs at room temperature was examined for 4 hours to exclude possible decomposition during the derivatisation process. Derivatisation was considered complete when chromato- graphic peaks for TMS derivatives stopped increasing and no peaks corresponding to unreacted HPs remained in the GC-FID chromatogram. Tertiary hydroperoxides (HP2 and IS) were derivatised in 25 min, primary HP (HP3) in 5 min, after only brief mixing. This difference can be ex- plained by steric hindrance. We opted for a derivatization time of 2 h to give some extra time for samples with high concentrations of HPs. 3. 3. EI Fragmentation Identification was made by classical mass spectra in- terpretation and by comparison with an authentic refer- ence standard, 80% cumene-HP. The TMS derivative of cumene-HP and the internal standard tetradecane were the only chromatographic peaks in calibration solutions. Their identity was confirmed by a NIST mass spectra li- brary search. The mass spectrum of the TMS derivative of cumene-HP is characterized by a large fragment peak at [M-105]+ and a smaller peak at m/z 105 (Fig. 4). The ions at m/z 135 and m/z 151 apparently correspond to [M-OSi- (CH3)3]+ and [M-Si(CH3)3]+, respectively. The molecular ion cannot be observed. The second most abundant peak is the tropylium cation, which is characteristic of aromatic compounds. Figure 4. Mass spectra of the TMS derivative of cumene-HP. Figure 5. Mass spectra of the TMS derivatives of α-pinene HPs: 4-Hydroperoxy-4,6,6-trimethylbicyclo[3.1.1]hept-2-ene (HP2), verbenyl-hydroperoxide (HP4), pinocarvyl-hydroperoxide (HP1) and myrtenyl-hydroperoxide (HP3). 733Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... The tropylium cation is also observed in the mass spectra of α-pinene and its derivatives from the NIST mass spectral library as well as in the mass spectra of our TMS derivatives of α-pinene-HPs (Fig. 5). Again, the molecular ions are not observed and the fragmentation is extensive. The extensive fragmentation into a large number of low- mass ions makes selected-reaction monitoring less profita- ble, but on the other hand, the spectra are more informa- tive and allow discrimination between the different positional isomers. Comparison of the mass spectra of derivatised and underivatised HPs confirmed the reversal of the elution order for HP1 and HP4 after derivatisation. Common to all spectra is both a signal at m/z 135, due to the loss of the TMS-peroxy radical (-105 Da) and a specific ion series of terpenes with the molecular formula CnH2n-5: 65, 79, 93, 107, 121, and 135 (Fig. 5). The base peaks are typical hydrocarbon fragments: in the spectra of HP2 and HP4 m/z 93 (C7H9+) and for HP3 m/z 91. The base peak of HP1 is m/z 89, corresponding to [OSi(CH3)3]+. Other TMS fragments are also observed: m/z 73, corresponding to [Si(CH3)3]+ and m/z 105, corre- sponding to [OOSi(CH3)3]+. This is to be expected since most ionisation occurs at the silicon (ionisation potential 8.1 versus 13.6 eV for oxygen).10,12 Even when there are similarities between isomers in their EI spectra, the ions’ relative intensities vary consider- ably. The relative abundance of high-molecular-mass ions decreases in the order primary HP > secondary HPs > ter- tiary HP (Fig. 5). This trend can be explained by a greater distance of the ionized atoms from the strained bicyclic skeletal structure in primary HP and by fragmentation mechanisms. We propose an H-rearrangement mecha- nism for the stabilization of m/z 151, which would help explain its high abundance in primary HP (Fig. 6). Figure 6. The mechanism for the formation of the fragment m/z 151, which is formed in higher amount in myrtenyl-hydroperoxide (HP3). 3. 4. Method Validation A validation procedure was carried out i.e. linear re- gression range, precision and limit of quantification/detec- tion were determined. Quantification was based on the peak area for cumene-HP relative to the peak area of the internal standard tetradecane. The linearity of the GC method was evaluated from 0.6 to 90 μg/mL of cumene-HP using five concentration levels, 0.6, 1, 6, 25, 50, 90 μg/mL. The R2 value was greater than 0.999, LOD was 0.6 μg/mL, and LOQ was 1 μg/mL. The LOD was determined as the concentration giving a signal to noise ratio (S/N ratio) of at least 3, and LOQ as the lowest point of the calibration curve subject to linearity. Injection repeatability was eval- uated using six injections of a standard solution, and the percentage of relative standard deviation (%RSD) in the peak area was 0.15%. Sample repeatability was evaluated by preparing six replicates of the same sample (with deri- vatisation for GC), and the %RSD in the peak area was 4.6%. The validation proved that the developed GC meth- od was suitable for monitoring the α-pinene reaction with singlet oxygen. The selectivity of the method was verified by analysing turpentine samples, and all four HPs could be identified in autooxidised turpentine (Fig. 7). 3. 5. Analysis of Real Samples To investigate the applicability of the proposed method for the determination of HPs in real samples, tur- pentine was analysed before and after autoxidation. The sample of turpentine contained 72% α–pinene and 9% β-pinene. During exposure to air, HPs concentrations in- creased with time (Fig. 7). Turpentine autooxidation also increased the mixture’s complexity; new peaks were formed as the hydroperoxides were degraded to secondary oxidation products, e.g. aldehydes, alcohols, epoxides. The concept of the effective carbon number allowed us to quantify the responses without standards of known purity. The calculated value of the relative mass response factor for α-pinene HP with IS cumene-peroxide was 0.987. Due to a poor evaluation of the chemical structure in the ECN calculation, a bias could enter the quantification. In our case, the ECN could be overestimated by about 2% be- cause we used an aromatic internal standard and aliphatic analytes.27 HPs in the turpentine sample were confirmed by four points of identification, retention times of HPs and HPs TMS derivatives, and by MS spectra of HPs and HPs TMS derivatives. The method’s selectivity was verified by analysing samples of turpentine and screening for peaks that might interfere with α-pinene HPs. HP3 coeluted with a compound with a normalised concentration of 150 ppm (chromatogram A in Fig. 7, the right part of the double peak). With increasing concentration after prolonged au- tooxidation, the concentration of HP3 increased (chroma- togram B in Fig. 7). Therefore, in an oxidised turpentine sample, an overestimation of 2% HP3 is to be expected at a concentration of 7.57 mg HP3/g. The turpentine sample data show a high presence of HPs. The total mass fraction of HPs in bottled turpentine was 0.1% and increased to 5.1% after 20 days of air expo- sure. HP2 had the highest yield, which is expected for a radical reaction in which the most stable, tertiary radical is 734 Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... formed. HP2 represents 43% of all radically sensitized HPs, and HP1 represents 64% of all HPs synthesized with singlet oxygen (Table 1). With this difference in yields, it would be possible to assess the importance of singlet oxy- gen as an atmospheric oxidant based on measurements of the concentrations of individual α-pinene HPs in the air. 4. Conclusions The manuscript addresses the problem of quantify- ing reactive unstable organic species for which no stand- ard reference material is available. We present the first GC- FID method for the quantification of all four α-pinene hydroperoxides formed in a reaction with α-pinene. The hydroperoxides were prepared by a simple photochemical synthesis in a laboratory flask. Pre-column silylation im- proved their stability, and the concept of effective carbon number allowed quantification despite the standards’ poor stability. We believe that this new synthesis and analysis approach could be used for other unstable hydroperoxides as well. The applicability of the proposed method was demonstrated on samples of bottled and oxidised turpen- tine. Each analysis was performed within 200 min with a quantification limit in the μg/mL range. After 20 days of air exposure, the mass fraction of hydroperoxides in tur- pentine increased 35-fold to 5.1%. This level is likely capa- ble of causing oxidative damage to the skin and lungs. For more complex matrices, such as hydroalcoholic products and atmospheric particles, an extraction step could be added. To further improve accuracy, isolation of individual α-pinene HPs and their purity determination by NMR would allow calibration and full validation of our GC-FID method. GC-MS or LC-MS could provide addi- tional selectivity and better robustness, especially if iso- tope-labelled internal standards were available. In addition to demonstrated importance of hydrop- eroxides in the analysis of essential oils, hydroperoxides of α-pinene are also important in atmospheric chemistry, where photoreactions of α-pinene with singlet oxygen could help explain high formation rates of secondary or- ganic aerosols.3,27 The formation of hydroperoxides with singlet oxygen is, in contrast to the radical formation, in- dependent of the NOx concentration. As NOx levels de- crease due to emission control measures, photochemical HPs will become even more important for atmospheric chemistry. Acknowledgements The study was carried out with financial support from the Slovenian Research Agency (P1-0153) and the World Federation of Scientists. The authors would like to thank Aleksandra Kuljanin and Dr. Ida Kraševec for ad- vice and help during manuscript preparation. 5. References 1. A.-T. Karlberg, M. A. Bergström, A. Börje, K. Luthman, J. L. G. Nilsson, Chem. Res. Toxicol. 2008, 21, 53-69. DOI:10.1021/tx7002239 2. L. A. Morio, K. A. Hooper, J. Brittingham, T.-H. Li, R. E. Gordon, B. J. Turpin, D. L. Laskin, Toxicol. Appl. Pharmacol. 2001, 177, 188–199. DOI:10.1006/taap.2001.9316 3. M. Ehn, J. A. Thornton, E. Kleist, M. Sipilä, H. Junninen, I. Pullinen, M. Springer, F. Rubach, R. Tillmann, B. Lee, F. Lopez-Hilfiker, S. Andres, I.-H. Acir, M. Rissanen, T. Jokinen, S. Schobesberger, J. Kangasluoma, J. Kontkanen, T. Niemi- nen, T. Kurtén, L. B. Nielsen, S. Jørgensen, H. G. Kjaergaard, Figure 7. The chromatogram of turpentine before (A) and after 20 days of autooxidation (B). Table 1. Concentrations of α-pinene hydroperoxides in turpentine before and after 20 days of air-exposure compared to concentrations of hydroperoxides synthesised photochemically with singlet oxygen (data in mg/g). HP2 HP4 HP1 HP3 Σ Turpentine 0.416 0.207 0.186 0.626* 1.44 oxidised 21.7 12.5 8.84 7.57 50.6 turpentine photooxidised 3.55 2.14 17.6 4.17 27.4 α-pinene *double peak 735Acta Chim. Slov. 2021, 68, 728–735 Pavlica et al.: Quantification of Hydroperoxides by Gas Chromatography ... M. Canagaratna, M. D. Maso, T. Berndt, T. Petäjä, A. Wahner, V.-M. Kerminen, M. Kulmala, D. R. Worsnop, J. Wildt, T. F. Mentel, Nature 2014, 506, 476–479. DOI:10.1038/nature13032 4. R. Treudler, G. Richter, J. Geier, A. Schnuch, C. E. Orfanos, B. Tebbe, Contact Dermatitis 2000, 42, 68–73. DOI:10.1034/j.1600-0536.2000.042002068.x 5. S. Hellerström, N. Thyresson, S.-G. Blohm, G. Widmark, J. Invest. Dermatol. 1955, 24, 217–224. DOI:10.1038/jid.1955.35 6. K. S. Docherty, W. Wu, Y. B. Lim, P. J. Ziemann, Environ. Sci. Technol. 2005, 39, 4049–4059. DOI:10.1021/es050228s 7. T. Nakamura, H. Maeda, Lipids 1991, 26, 765–768. DOI:10.1007/BF02535628 8. G. L. Beutner, S. Ayers, T. Chen, S. W. Leung, H. C. Tai, Q. Wang, Org. Process Res. Dev. 2020, 24, 1321–1327. DOI:10.1021/acs.oprd.0c00251 9. W. F. Brill, J. Chem. Soc. Perkin Trans. 2 1984, 0, 621–627. DOI:10.1039/p29840000621 10. J. Polzer, K. Bächmann, J. Chromatogr. A 1993, 653, 283–291. DOI:10.1016/0021-9673(93)83186-V 11. S. Leocata, S. Frank, Y. Wang, M. J. Calandra, A. Chaintreau, Flavour Fragr. J. 2016, 31, 329–335. DOI:10.1002/ffj.3324 12. J. Rudbäck, A. Ramzy, A.-T. Karlberg, U. Nilsson, J. Sep. Sci. 2014, 37, 982–989. DOI:10.1002/jssc.201300843 13. G. O. Schenck, H. Eggert, W. Denk, Justus Liebigs Ann. Chem. 1953, 584, 177–198. DOI:10.1002/jlac.19535840112 14. C. W. Jefford, A. F. Boschung, R. M. Moriarty, C. G. Rimbault, M. H. Laffer, Helv. Chim. Acta 1973, 56, 2649–2659. DOI:10.1002/hlca.19730560748 15. U. Neuenschwander, F. Guignard, I. Hermans, ChemSus- Chem 2010, 3, 75–84. DOI:10.1002/cssc.200900228 16. F. Bianchi, T. Kurtén, M. Riva, C. Mohr, M. P. Rissanen, P. Roldin, T. Berndt, J. D. Crounse, P. O. Wennberg, T. F. Men- tel, J. Wildt, H. Junninen, T. Jokinen, M. Kulmala, D. R. Worsnop, J. A. Thornton, N. Donahue, H. G. Kjaergaard, M. Ehn, Chem. Rev. 2019, 119, 3472–3509. DOI:10.1021/acs.chemrev.8b00395 17. K. Akasaka, H. Ohrui, J. Chromatogr. A 2000, 881, 159–170. DOI:10.1016/S0021-9673(00)00330-7 18. M. J. Calandra, J. Impellizzeri, Y. Wang, Flavour Fragr. J. 2015, 30, 121–130. DOI:10.1002/ffj.3232 19. J. Nilsson, J. Carlberg, P. Abrahamsson, G. Hulthe, B.-A. Persson, A.-T. Karlberg, Rapid Commun. Mass Spectrom. 2008, 22, 3593–3598. DOI:10.1002/rcm.3770 20. R. Zhao, C. M. Kenseth, Y. Huang, N. F. Dalleska, J. H. Sein- feld, Environ. Sci. Technol. 2018, 52, 2108–2117. DOI:10.1021/acs.est.7b04863 21. M.-C. Reinnig, J. Warnke, T. Hoffmann, Rapid Commun. Mass Spectrom. 2009, 23, 1735–1741. DOI:10.1002/rcm.4065 22. S. K. Bharti, R. Roy, TrAC Trends Anal. Chem. 2012, 35, 5–26. DOI:10.1016/j.trac.2012.02.007 23. T. Cachet, H. Brevard, A. Chaintreau, J. Demyttenaere, L. French, K. Gassenmeier, D. Joulain, T. Koenig, H. Leijs, P. Liddle, G. Loesing, M. Marchant, Ph. Merle, K. Saito, C. Schippa, F. Sekiya, T. Smith, Flavour Fragr. J. 2016, 31, 191– 194. DOI:10.1002/ffj.3311 24. M. Prein, W. Adam, Angew. Chem. Int. Ed. Engl. 1996, 35, 477–494. DOI:10.1002/anie.199604771 25. I. A. Yaremenko, V. A. Vil’, D. V. Demchuk, A. O. Terent’ev, Beilstein J. Org. Chem. 2016, 12, 1647–1748. DOI:10.3762/bjoc.12.162 26. J. T. Scanlon, D. E. Willis, J. Chromatogr. Sci. 1985, 23, 333– 340. DOI:10.1093/chromsci/23.8.333 27. M. Kállai, J. Balla, Chromatographia 2002, 56, 357–360. DOI:10.1007/BF02491945 28. A. Manfrin, S. A. Nizkorodov, K. T. Malecha, G. J. Getzinger, K. McNeill, N. Borduas-Dedekind, Environ. Sci. Technol. 2019, DOI:10.1021/acs.est.9b01609. 29. Z. Tan, K. Lu, A. Hofzumahaus, H. Fuchs, B. Bohn, F. Hol- land, Y. Liu, F. Rohrer, M. Shao, K. Sun, Y. Wu, L. Zeng, Y. Zhang, Q. Zou, A. Kiendler-Scharr, A. Wahner, Y. Zhang, At- mospheric Chem. Phys. 2019, 19, 7129–7150. DOI:10.5194/acp-19-7129-2019 Except when otherwise noted, articles in this journal are published under the terms and conditions of the  Creative Commons Attribution 4.0 International License Povzetek Hidroperoksidi so zelo pomembni na področju atmosferske in biološke kemije. Vendar pa pri njihovi analizi obstaja več analitičnih izzivov: neznani in običajno nizki absorpcijski koeficienti, visoka reaktivnost, toplotna nestabilnost in pomanjkanje razpoložljivih referenčnih standardov. Da bi odpravili te omejitve, predlagamo pristop GC-FID, ki vkl- jučuje predkolonsko silacijo in kvantifikacijo s pristopom na podlagi efektivnega števila ogljikov (angl. Effective Carbon Number). V tekoči fazi smo s singletnim kisikom sintetizirali štiri hidroperokside α-pinena in jih identificirali na podlagi literarnih podatkov o izkoristku posameznega izomera, MS spektrov, ocenjenih temperaturah vrelišča vsakega izomera (retencijski čas), njihovi toplotni stabilnosti in stopnji derivatizacije. Razviti postopek smo uporabili za določanje hi- droperoksidov v ustekleničenem in avtooksidiranem terpentinu. Predvidevamo, da bi se ta metoda lahko uporabila tudi v atmosferski kemiji, kjer bi reaktivnost singletnega kisika lahko pomagala razložiti visoke stopnje tvorbe sekundarnih organskih aerosolov.