Acta Chromatographica
Volume/Issue:
Accepted Manuscript / Online First
Headspace – Solid phase microextraction vs liquid injection GC-MS analysis of essential oils: Prediction of linear retention indices by multiple linear regression
Authors:
Asya Hristozova Faculty of Chemistry, University of Plovdiv Paisii Hilendarski, 4000 Plovdiv, Bulgaria

Search for other papers by Asya Hristozova in
Current site
Google Scholar
PubMed Close
https://orcid.org/0000-0001-5348-9805
,
Margarita Batmazyan Faculty of Chemistry, University of Plovdiv Paisii Hilendarski, 4000 Plovdiv, Bulgaria

Search for other papers by Margarita Batmazyan in
Current site
Google Scholar
PubMed Close
,
Kiril Simitchiev Faculty of Chemistry, University of Plovdiv Paisii Hilendarski, 4000 Plovdiv, Bulgaria

Search for other papers by Kiril Simitchiev in
Current site
Google Scholar
PubMed Close
,
Slava Tsoneva Faculty of Chemistry, University of Plovdiv Paisii Hilendarski, 4000 Plovdiv, Bulgaria

Search for other papers by Slava Tsoneva in
Current site
Google Scholar
PubMed Close
,
Veselin Kmetov Faculty of Chemistry, University of Plovdiv Paisii Hilendarski, 4000 Plovdiv, Bulgaria

Search for other papers by Veselin Kmetov in
Current site
Google Scholar
PubMed Close
, and
Erwin Rosenberg Institute of Chemical Technologies and Analytics, Vienna University of Technology, 1060 Vienna, Austria

Search for other papers by Erwin Rosenberg in
Current site
Google Scholar
PubMed Close
Online Publication Date:
23 Feb 2024
Publication Date:
23 Feb 2024
Article Category:
Research Article
DOI:
https://doi.org/10.1556/1326.2024.01207
Keywords:
essential oils; gas chromatography (GC); gas chromatography-mass spectrometry (GC-MS); gas chromatography-tandem mass spectrometry (GC-MS/MS); linear retention indices prediction (LRI); headspace solid phase microextraction (HS-SPME); multiple linear regression (MLR)
Open access

Abstract

Due to the relative independence from the operational parameters, the linear retention indices (LRIs) are useful tool in gas chromatography-mass spectrometry (GC-MS) qualitative analysis. The aim of the current study was to develop a multiple linear regression (MLR) model for the prediction of LRIs as a function of selected molecular descriptors. Liquid injection GC-MS was used for the analysis of Essential oils (Rose, Lavender and Peppermint) separating the ingredients by a semi-standard non-polar stationary phase. As a result, a sum of 103 compounds were identified and their experimental LRIs were derived relying on reference measurements of a standard mixture of n-alkanes (from C8 to C20). As a next step, a set of molecular descriptors was generated for the distinguished chemical structures. Further, a stepwise MLR was applied for the selection of the significant descriptors (variables) which can be used to predict the LRIs. From an exploit set of over 2000 molecular descriptors, it was found that only 16 can be regarded as significant and independent variables. At this point split validation was applied: the identified compounds were randomly divided into training (85%) and validation (15%) sets. The training set (87 compounds) was used to derive two MLR models by applying i) the ‘enter’ algorithm (R2 = 0.9960, RMSЕ = 17) and ii) the ‘stepwise’ one (R2 = 0.9958, RMSЕ = 17). The predictive power was assessed by the validation set (16 compounds) as follows i) q2F1 = 0.9896, RMSE = 25 and ii) q2F1 = 0.9886, RMSE = 26, respectively. The adequateness of both regression approaches was further evaluated. Newly developed headspace-solid phase microextraction (HS-SPME) procedures in combination with GC-MS were used for an alternative analysis of the studied Essential oils. Twelve additional compounds, not previously detected by the liquid sample introduction mode of analysis, were identified for which the values of the significant descriptors were within the working range of the developed MLRs. For the last compounds, the LRIs were calculated and the experimental data was used as an external set for assessment of the regression models. The predictive power for both regression approaches was assessed as follows: Enter RMSE = 41, q2F2 = 0.9503 and Stepwise RMSE = 40, q2F2 = 0.9521.

Abstract

Due to the relative independence from the operational parameters, the linear retention indices (LRIs) are useful tool in gas chromatography-mass spectrometry (GC-MS) qualitative analysis. The aim of the current study was to develop a multiple linear regression (MLR) model for the prediction of LRIs as a function of selected molecular descriptors. Liquid injection GC-MS was used for the analysis of Essential oils (Rose, Lavender and Peppermint) separating the ingredients by a semi-standard non-polar stationary phase. As a result, a sum of 103 compounds were identified and their experimental LRIs were derived relying on reference measurements of a standard mixture of n-alkanes (from C8 to C20). As a next step, a set of molecular descriptors was generated for the distinguished chemical structures. Further, a stepwise MLR was applied for the selection of the significant descriptors (variables) which can be used to predict the LRIs. From an exploit set of over 2000 molecular descriptors, it was found that only 16 can be regarded as significant and independent variables. At this point split validation was applied: the identified compounds were randomly divided into training (85%) and validation (15%) sets. The training set (87 compounds) was used to derive two MLR models by applying i) the ‘enter’ algorithm (R2 = 0.9960, RMSЕ = 17) and ii) the ‘stepwise’ one (R2 = 0.9958, RMSЕ = 17). The predictive power was assessed by the validation set (16 compounds) as follows i) q2F1 = 0.9896, RMSE = 25 and ii) q2F1 = 0.9886, RMSE = 26, respectively. The adequateness of both regression approaches was further evaluated. Newly developed headspace-solid phase microextraction (HS-SPME) procedures in combination with GC-MS were used for an alternative analysis of the studied Essential oils. Twelve additional compounds, not previously detected by the liquid sample introduction mode of analysis, were identified for which the values of the significant descriptors were within the working range of the developed MLRs. For the last compounds, the LRIs were calculated and the experimental data was used as an external set for assessment of the regression models. The predictive power for both regression approaches was assessed as follows: Enter RMSE = 41, q2F2 = 0.9503 and Stepwise RMSE = 40, q2F2 = 0.9521.

1 Introduction

Usually, Essential oils are complex systems of multiple chemical components obtained from plant material through steam distillation. They are mainly composed of lipophilic and highly volatile metabolites such as mono and sesquiterpenes. Other classes of compounds discovered are oxides, alcohols, aldehydes, ketones, esters, and heterocycles [1]. However, their composition varies significantly depending on the place of growth, seasonal fluctuations and weather conditions [2, 3]. Due to the complexity of the composition of the oils and the danger of adulteration with other oils or synthetic materials, a need for selective technique with the capability to characterize the variety of compounds in the composition of the oils is required. Due to the volatile and relatively volatile nature of essential oil components, they can be analyzed using gas chromatography in combination with mass spectrometry [4, 5]. The gas chromatographic method is used for the qualitative analysis of volatile compounds, providing information on retention times as the criterion for peak identification. The mass spectrometer, used as a detector, offers additional information for the identification of individual compounds by comparing the spectra of each component with those in the spectral library. Since essential oils are mixtures of terpene or phenylpropane derivatives, their spectra are similar and the correct identification of the peaks is highly sophisticated, sometimes impossible [1]. To increase the reliability of analytical results for evaluating the composition of essential oil components, it is necessary to use a combination of criteria and different identification approaches, including methods of separation and concentration, comparison with spectra from databases and calculation of retention indices (RI) [1, 6, 7]. Retention indices RI denote the retention behavior of the compounds of interest according to a uniform scale determined by a series of closely related standard substances (n-alkane scale) for isothermal conditions. The linear retention indices (LRIs), used for programmed gas chromatographic runs, can be calculated by applying the equations proposed by van Den Dool and Kratz [1]. Many efforts can be focused on developing predictive models for LRIs in the analysis of essential oils by GC-MS to facilitate and reduce the identification process [8, 10, 11].

The development of the prediction models includes a selection of descriptors (physical-chemistry properties of the compounds), validating the model using an internal set of compounds and testing using an external set of compounds for the developed model. The prediction coefficient q2 is the main criterion used to assess the actual predictive power of the models [9, 12–14] as well as calculating the root mean square error (RMSE), etc.

In the present study linear regression models as a simple approach for predicting the LRIs in gas chromatographic analysis of volatile components of essential oils for the semi-standard non-polar stationary phase were developed. Two different techniques for analysis i) direct liquid injection with appropriate dilution and ii) alternatively HS-SPME [15] were used for the analysis of different essential oils bought from the local market. Newly developed HS-SPME were proposed. The data collected from the real samples were used to calculate the LRIs. Regression models were developed by the selection of significant descriptors (independent variables, x), correlated with LRI (dependent variable, y), evaluated by applying split validation and tested by an external set of compounds.

2 Experimental

2.1 Chemicals and reagents

Dichloromethane (GC grade) was purchased from Honeywell (Riedel-de Haën GmbH, Seelze, Germany). For the solid-phase microextraction procedure, a conventional Carboxene (CAR)/Polydimethylsiloxane (PDMS) SPME fiber with a coating thickness of 85 μm and an SPME holder from Supelco (Bellefonte, PA, USA) were used. Chromatography vials with a volume of 10 mL were purchased from Supelco (Bellefonte, PA, USA) and used for HS-SPME. An alkane mixture C8–C20 in hexane at a concentration of 40 mg L−1 was purchased from Supelco (Bellefonte, PA, USA).

2.2 Instrumentation

The study was carried out using a GC-MS/MS chromatographic system TSQ 9000 (Thermo Scientific, USA) with electron impact ionization (EI) at 70 eV and a PTV injector. A volume of 1 µL of each liquid essential oil sample, after appropriate dilution was injected using an autosampler AI 1300 equipped with a 10 µL glass syringe. HS-SPME injections were performed manually. The PTV injector was operated in Split mode (split ratio 10:1), at a constant inlet temperature of 280 °C. Metal liner (PTV Siltek Metal Liner, 2 mm ID, 2.75 mm OD, 120 mm Length, Thermo Fisher Scientific, USA) was used as suitable for both analyzes. GC column TG SQC MS (15 m × 0.25 mm × 0.25 µm film thickness, Thermo Fisher Scientific, USA) was used for the chromatographic separation. The temperature program of the column oven was optimized to obtain adequate chromatographic separation as follows: initial temperature of 40°C min −1 held for 2 min; at a rate of 1 °C min−1 to 200 °C; at a rate of 30 °C min−1 to a final temperature of 280 °C, which was held for 3 min. The solvent delay time and total analysis time were 2 and 168 min, respectively. Helium (purity 99.9999%) at a flow rate of 1.0 mL min−1 was used as a carrier gas. The transfer line and Ion source temperatures were set at 250 °C and 230 °C, respectively. Full scan mode in the range 35–500 amu was used, with a dwell time of 0.2 s. The data analysis was performed using Xcalibur 4.1 software, and initial identification of volatile organic compounds was performed by comparing their spectra with the spectra from the NIST Mass Spectral Library (NIST MS Search 2.3). Regarding the use of RI and the similarity of the mass spectra, the following acceptance criteria were applied respectively – a matching interval of ±20 units of the retention indices and a similarity value of at least 80%.

2.3 Sample type

Essential oils from Rose oil (Rosa damascena), Peppermint oil (Mentha piperita L.) and Lavender oil (Lavandula angustifolia) were purchased from the local market. All samples were stored in the dark at 4–8 °C temperature before analysis.

2.4 Liquide sample analysis

The essential oils were appropriately diluted in dichloromethane (Lavender oil DF = 100, Rose oil DF = 200, and Peppermint oil DF = 400). The diluted samples were prepared in triplicate and measured according to the instrumental conditions listed in section 2.2. by liquid injection GC-MS/MS.

2.5 HS-SPME of VOCs

The SPME fiber was preconditioned once in the injection port of the gas chromatograph for 30 min at 280 °C, before HS-SPME, according to the manufacturer's recommendations. A blank fiber injection was measured daily. On the bottom of the 10 mL glass (headspace) vial it was placed a piece of paper (dimensions: 0.5 × 0.5 cm). 5 µl Peppermint oil, 10 µl Lavender oil or Rose oil were transferred onto the paper surface and the vials were closed with polytetrafluoroethylene (PTFE) coated silicone rubber septum and aluminium caps. The vials were placed for specified conditioning time (Rose oil, 45 min; Lavender oil, 45 min; Peppermint oil, 30 min) at room temperature 24 °C ± 2 °C and no stirring. After the prescribed time, the SPME fiber was introduced into the headspace for experimentally determined sorption time (Rose oil, 10 min, Lavender oil, 10 min, Peppermint oil, 5 min) and injected into the GC injection port for desorption of 3 min. A schematic illustration of the optimized HS-SPME procedure is presented in Fig. 1.

Fig. 1.
Fig. 1.

HS-SPME sample preparation procedure for the analysis of volatile compounds in Essential oils

Citation: Acta Chromatographica 2025; 10.1556/1326.2024.01207

2.6 Prediction of LRIs by multiple linear regression (MLR)

The sum of all unique essential oil compounds detected by liquid injection GC-MS, identified according to accepted criteria in section 2.2., were used to develop a prediction model for LRIs by multiple linear regression (MLR) [13, 16–19]. The experimentally obtained retention times (tR) of the sample compounds and the tR of the n-alkanes from the standard solution C8–C20 were used for the calculation of the LRIs values, according to the equation of H. van Den Dool and D. J. Kratz (Eq. (1)) [7].
LRI=100[n+tx−tntn+1−tn]
The SMILES formulas of the studied compounds were obtained from the PubChem database [20]. A set of 1 D, 2 D descriptors and PubChem Fingerprint were calculated by PaDEL Descriptor software product 2.21 [21, 22] using the SMILES notations (.smi file format). A stepwise MLR technique was used for the selection of the significant and independent molecular descriptors, correlating with the experimentally calculated LRIs. The adequateness of each developed regression model was assessed by the corresponding coefficient of determination (R2) (Eq. (2)), adjusted coefficient of determination (Adj. R2) (Eq. (3)), and the root mean square error (RMSE) (Eq. (4)).
R2=1−RSSTSS
where:

  • TSS – total sum of squares

  • RSS – residual sum of squares

Adj. R2=1−R2(n−1)n−m−1
where:

  • R2 – coefficient of determination;

  • n – number of observations

  • m – number of independent variables in the data set

RMSE=∑i=1n(yi−ŷi)2n
The collinearity among the dependent variables (molecular descriptors) was evaluated by calculating the corresponding variance inflation factors (Eq. (5)):
VIFi=11−Ri2
where:

  • Ri2 – coefficient of determination in linear modelling of the i-th studied molecular descriptor from all other dependent variables j (i ≠ j).

  • Values of VIF less than 15 were accepted to indicate no variable collinearity [16].

Model development and validation were performed using split validation. The total set of compounds was divided randomly (taking into account tR) into a training set (85%) and a validation set (15%) [23]. The training set was used to develop a regression model with the significant and independent descriptors by two alternative algorithms: Enter and Stepwise, giving model equations as follows:
LRI=β0+β1 x1+β2 x2+…+βn xn
where:

  • xi: the value of the i-th molecular descriptor;

  • βi: the i-th regression coefficient.

The model equations were used to predict the LRIs of the validation set. For assessment of the predictive power of each regression model, the root mean square error (RMSE) (Eq. (4)) and a prediction coefficient q2F1 (Eq. (7)) were calculated:
qF12=1−∑i=1n(yi−ŷi)2∑i=1n(yi−y¯tr)2
The validated multiple regression models were further tested by an external set of compounds, detected via HS-SPME analysis of essential oils for which experimental LRIs were also calculated. The SMILES formulas of the tested compounds were obtained from the PubChem database (.smi file format). As a next step, were calculated 1D and 2D descriptors as well as PubChem Fingerprint by the software product PaDEL Descriptor 2.21 [22]. The corresponding theoretical LRIs were predicted using the model equations using the values of the molecular descriptors for each compound as independent variables. To assess the ability of the models to accurately predict the LRI, the RMSE and the prediction coefficient q2F2 (Eq. (8)) were calculated:
qF22=1−∑i=1n (yi−ŷi)2∑i=1n(yi−y¯ext)2
Where for Eqs (7) and (8):

  • yi – observed values of linear retention indices

  • ŷi – predicted values of linear retention indices

  • ȳtr – mean of the observed linear retention indices of the training set

  • ȳextt – mean value of the observed linear retention indices of the external set

3 Results and discussion

3.1 Analysis of essential oils by liquid injection GC-MS

The diluted Essential oils were analyzed in triplicate and according to the prescribed acceptance criteria (section 2.2.) were identified 49 components in the Lavender oil, 51 compounds in the analysis of two commercial brands of Rose oil and 32 compounds were identified in the Peppermint oil. A total of 103 unique compounds were selected and SMILES notations for each compound were derived from the PubChem database.

3.2 Multiple linear regression for predicting linear retention indices

3.2.1 Selection of significant independent variables (descriptors)

The LRI of each compound from the set of 103 compounds identified by liquid injection GC-MS were calculated according to the equation of H. van Den Dool and D. J. Kratz (Eq. (1)) using the average retention time determined from three replicates and the average tR of the n-alkanes from the standard solution (n = 3). The obtained LRI values and the SMILES notations for the compounds are presented in Table 1.

Table 1.

Calculated LRIs and SMILES notations for the identified by liquid injection GC-MS compounds in Rose, Lavender and Peppermint oils

CompoundLRISMILES notation
Ether, hexyl methyl822CCCCCCOC
3-Methylcyclopentanone840CC1CCC(=O)C1
3-hexenol855CCC=CCCO
Hexyl alcohol871CCCCCCO
Heptanal903CCCCCCC=O
α-Thujene920CC1=CCC2(C1C2)C(C)C
α-Pinene924CC1=CCC2CC1C2(C)C
Camphene936CC1(C2CCC(C2)C1=C)C
3-Methylcyclohexanone941CC1CCCC(=O)C1
Sabinene964CC(C)C12CCC(=C)C1C2
3-p-Menthene975CC1CCC(=CC1)C(C)C
n-Octanone-3987CCCCCC(=O)CC
β-Pinene990CC1(C2CCC(=C)C1C2)C
β-Myrcene990CC(=CCCC(=C)C=C)C
n-Octan-3-ol1002CCCCCC(CC)O
δ-3-Carene1004CC1=CCC2C(C1)C2(C)C
α-Terpinene1010CC1=CC=C(CC1)C(C)C
n-Hexyl acetate1016CCCCCCOC(=O)C
β-Cymene1018CC1=CC(=CC=C1)C(C)C
Limonene1020CC1=CCC(CC1)C(=C)C
Eucalyptol1022CC1(C2CCC(O1)(CC2)C)C
trans-β-Ocimene1040CC(=CCC=C(C)C=C)C
cis-β-Ocimene1044CC(=CCC=C(C)C=C)C
γ-Terpinene1050CC1=CCC(=CC1)C(C)C
4-Thujanol1060CC(C)C12CCC(C1C2)(C)O
Linalool oxide1066CC(=CCCC(C)(C1CO1)O)C
Linalool oxide furanoide1073CC1(CCC(O1)C(C)(C)O)C=C
α-Terpinolene1080CC1=CCC(=C(C)C)CC1
Linalool1104CC(=CCCC(C)(C=C)O)C
cis-Rose oxide1108CC1CCOC(C1)C=C(C)C
Phenylethyl Alcohol1115C1=CC=C(C=C1)CCO
1-Octen-3-yl-acetate1119CCCCCC(C=C)OC(=O)C
trans-Rose oxide1122CC1CCOC(C1)C=C(C)C
2,4,6-Octatriene, 3,4-dimethyl-1127CC=CC=C(C)C(=CC)C
3-Octanol, acetate1129CCCCCC(CC)OC(=O)C
Camphor1132CC1(C2CCC1(C(=O)C2)C)C
Isopulegol1137CC1CCC(C(C1)O)C(=C)C
Menthone1146CC1CCC(C(=O)C1)C(C)C
Nerol oxide1151CC1=CCOC(C1)C=C(C)C
n-Hexyl isobutyrate1151CCCCCCOC(=O)C(C)C
Isomenthone1154CC1CCC(C(=O)C1)C(C)C
Borneol1159CC1(C2CCC1(C(C2)O)C)C
Neo menthol1160CC1CCC(C(C1)O)C(C)C
Terpinen-4-ol1168CC1=CCC(CC1)(C(C)C)O
Menthol1175CC1CCC(C(C1)O)C(C)C
Cryptone1176CC(C)C1CCC(=O)C=C1
α-Terpineol1185CC1=CCC(CC1)C(C)(C)O
Butyric acid, hexyl ester1197CCCCCCOC(=O)CCC
γ-Terpineol1205CC(=C1CCC(CC1)(C)O)C
Borneol formate1216CC1(C2CCC1(C(C2)OC=O)C)C
Pulegone1229CC1CCC(=C(C)C)C(=O)C1
Nerol1232CC(=CCCC(=CCO)C)C
Carvone1236CC1=CCC(CC1=O)C(=C)C
Citronellol1241CC(CCC=C(C)C)CCO
Piperitone1244CC1=CC(=O)C(CC1)C(C)C
Isogeraniol1249CC(=CCC=C(C)CCO)C
β-Phenylethyl acetate1256CC(=O)OCCC1=CC=CC=C1
Linalyl acetate1262CC(=CCCC(C)(C=C)OC(=O)C)C
Geraniol1265CC(=CCCC(=CCO)C)C
Citral1271CC(=CCCC(=CC=O)C)C
Citronellyl formate1278CC(CCC=C(C)C)CCOC=O
Bornyl acetate1278CC(=O)OC1CC2CCC1(C2(C)C)C
Menthyl acetate1291CC1CCC(C(C1)OC(=O)C)C(C)C
Geranic acid methyl ester1325CC(=CCCC(=CC(=O)OC)C)C
Hexyl tiglate1333CCCCCCOC(=O)C(=CC)C
Eugenol1357COC1=C(C=CC(=C1)CC=C)O
Copaene1358CC1=CCC2C3C1C2(CCC3C(C)C)C
β-Bourbonene1365CC(C)C1CCC2(C1C3C2CCC3=C)C
Nerol acetate1368CC(=CCCC(=CCOC(=O)C)C)C
β-Elemene1378CC(=C)C1CCC(C(C1)C(=C)C)(C)C=C
Geranyl acetate1388CC(=CCCC(=CCOC(=O)C)C)C
Hexyl hexanoate1390CCCCCCOC(=O)CCCCC
β-Caryophyllene1397CC1=CCCC(=C)C2CC(C2CC1)(C)C
β-Cubebene1399CC1CCC(C2C13C2C(=C)CC3)C(C)C
α-Santalene1406CC(=CCCC1(C2CC3C1(C3C2)C)C)C
Methyleugenol1410COC1=C(C=C(C=C1)CC=C)OC
α-Guaiene1423CC1CCC(CC2=C1CCC2C)C(=C)C
trans-α-Bergamotene1424CC1=CCC2CC1C2(C)CCC=C(C)C
α-Caryophyllene1430CC1=CCC(C=CCC(=CCC1)C)(C)C
Germacrene1459CC1=CCCC(=C)C=CC(CC1)C(C)C
δ-Guaiene1488CC1CCC2=C(CCC(CC12)C(=C)C)C
γ-Cadinene1495CC1=CC2C(CC1)C(=C)CCC2C(C)C
Pentadecane1500CCCCCCCCCCCCCCC
β-Cadinene1509CC1=CCC2C(C1)C(CC=C2C)C(C)C
Elemol1538CC(=C)C1CC(CCC1(C)C=C)C(C)(C)O
β-Caryophyllene oxide1556CC1(CC2C1CCC3(C(O3)CCC2=C)C)C
Nerolidol1562CC(=CCCC(=CCCC(C)(C=C)O)C)C
Globulol1568CC1CCC2C1C3C(C3(C)C)CCC2(C)O
2-Phenylethyl tiglate1580CC=C(C)C(=O)OCCC1=CC=CC=C1
γ-Eudesmol1613CC1=C2CC(CCC2(CCC1)C)C(C)(C)O
α-epi-Cadinol1624CC1=CC2C(CCC(C2CC1)(C)O)C(C)C
β-Eudesmol1628CC12CCCC(=C)C1CC(CC2)C(C)(C)O
α-Eudesmol1632CC1=CCCC2(C1CC(CC2)C(C)(C)O)C
3-Heptadecene1674CCCCCCCCCCCCCC=CCC
Heptadecane1701CCCCCCCCCCCCCCCCC
Farnesol1723CC(=CCCC(=CCCC(=CCO)C)C)C
Farnesal1737CC(=CCCC(=CCCC(=CC=O)C)C)C
Benzyl Benzoate1748C1=CC=C(C=C1)COC(=O)C2=CC=CC=C2
n-Octadecane1800CCCCCCCCCCCCCCCCCC
β-Phenylethyl benzoate1834C1=CC=C(C=C1)CCOC(=O)C2=CC=CC=C2
1-Nonadecene1869CCCCCCCCCCCCCCCCCC=C
Nonadecane1901CCCCCCCCCCCCCCCCCCC
Eicosane1999CCCCCCCCCCCCCCCCCCCC

PaDEL Descriptor software 2.21 was applied to calculate a set of 1 D, 2 D descriptors and PubChem Fingerprint (total number of 2,325 molecular descriptors) using the SMILES notations. A stepwise linear regression algorithm was used to distinguish the significant and independent variables (xi) correlating with the calculated LRI as the dependent variable (y) (Eq. (5)). Only 16 significant descriptors (from a pool of 2,325) remain in the regression equation (Table 2).

Table 2.

A list of selected significant descriptors and their meaning with details about descriptors, taken from the PaDEL descriptor list

DescriptorClassDescription
MLFER_LMolecular linear free energy relation.Solute gas-hexadecane partition coefficient
MLFER_SMolecular linear free energy relation.Combined dipolarity/polarizability
ATSC5cAutocorrelation descriptor. 2DCentered Broto-Moreau autocorrelation – lag 5/weighted by charges
ATS5iAutocorrelation descriptor. 2DBroto-Moreau autocorrelation - lag 5/weighted by first ionization potential
n3HeteroRingRing Count Descriptor.

2D		Number of 3-membered rings containing heteroatoms (N, O, P, S, or halogens)
n4RingRing Coun tDescriptor.

2D		Number of 4-membered rings
n11RingRingCountDescriptor.

2D		Number of 11-membered rings
GATS3cAutocorrelation descriptor. 2DGeary autocorrelation – lag 3/weighted by charges
TIC3Information content descriptor. 2DTotal information content index (neighborhood symmetry of 3 rd order)
PubchemFP143Fingerprint≥ 1 any ring size 5
PubchemFP147Fingerprint≥ 1 unsaturated non-aromatic carbon-only ring size 5
PubchemFP553FingerprintO=CC=C
PubchemFP582FingerprintCCCCC
PubchemFP639FingerprintOCCCO
PubchemFP672FingerprintO=CC=C-[#1]
PubchemFP688FingerprintCCCCCCC

From Table 3 it can be seen that the following descriptors showing a linear relationship when determining the linear retention indices of compounds with 95% statistical confidence (R2 = 0.9960, Adj. R2 = 0.9951): MLFER_L, MLFER_S, ATSC5c, ATS5i, n3HeteroRing, n4Ring, n11Ring, GATS3c, PubChem Fingerprint descriptors: PubchemFP143, 147, 553, 582, 639, 672, 688 and TIC3. Additional information about the listed descriptors can be found in [24, 25]. From the VIF values presented also in Table 3, it can be concluded that there is no significant collinearity among the listed variables.

Table 3.

Regression model coefficients and their statistical assessment obtained by stepwise multiple linear regression for selected significant descriptors (experimental set of 103 compounds)

Model DescriptorCoefficientStandard deviationP-valueVIF
Intercept24118<0.001
MLFER_L1773<0.0013.84
MLFER_S14012<0.0012.61
PubchemFP639−27927<0.0012.21
ATSC5c−121550.0301.64
PubchemFP67226130.0483.91
PubchemFP582−4710<0.0011.18
PubchemFP147−5314<0.0011.16
n3HeteroRing9720<0.0012.39
n4Ring−558<0.0011.32
GATS3c−448<0.0011.35
ATS5i0.0060.001<0.00111.37
n11Ring−7419<0.0011.08
PubchemFP143−357<0.0011.77
PubchemFP688−265<0.0011.80
PubchemFP55326110.023.75
TIC3−0.270.10.045.79

It should be emphasized that the number of significant and independent variables is relatively small – only 16 molecular descriptors. The last fact is encouraging in view of developing a simple regression model with a limited number of included variables.

3.2.2 Development and validation of mathematical models

The set of 103 compounds was randomly divided into a training set and a validation set. 85% of compounds (n = 87) were included in a training set and the rest 15% were included in the validation set (n = 16).

Two multiple linear regression models were developed by Enter or Stepwise algorithm, using the calculated LRI of the 87 compounds in the training set as dependent variable (y), and the 16 significant and independent molecular descriptors as variables (xi).

The coefficients of the two developed regression models and their statistical assessment are presented in Tables 4 and 5, respectively.

Table 4.

Regression model coefficients and their statistical assessment obtained by Enter algorithm (training set, n = 87)

DescriptorCoefficientStandard deviationP-valueVIF
Intercept23319<0.001
MLFER_L1793<0.0013.74
MLFER_S13914<0.0012.98
PubchemFP639−27628<0.0012.21
ATSC5c−178680.0111.71
PubchemFP67242160.0105.20
PubchemFP582−5012<0.0011.12
PubchemFP147−55200.0071.11
n3HeteroRing9921<0.0012.41
n4Ring−549<0.0011.38
GATS3c−388<0.0011.29
ATS5i0.0060.001<0.00111.80
n11Ring−71200.0011.09
PubchemFP143−307<0.0011.54
PubchemFP688−256<0.0011.98
PubchemFP55310140.474.99
TIC3−0.290.150.055.88
Table 5.

Regression model coefficients and their statistical assessment obtained by Stepwise algorithm (training set, n = 87)

DescriptorCoefficientStandard deviationP-valueVIF
Intercept22619<0.001
MLFER_L1793<0.0013.73
MLFER_S12612<0.0012.27
PubchemFP639−27629<0.0012.21
ATSC5c−205670.0031.60
PubchemFP672529<0.0011.61
PubchemFP582−5312<0.0011.10
n4Ring−589<0.0011.29
n3HeteroRing10121<0.0012.41
GATS3c−408<0.0011.28
PubchemFP147−58200.0061.10
ATS5i0.0040.001<0.0015.06
PubchemFP143−297<0.0011.53
PubchemFP688−1950.011.45
n11Ring−70200.0011.09

The range of the predicted LRIs using both algorithms was from 833 to 1993.

The adequateness of the regression model using the Enter algorithm was assessed as follows: R2 = 0.9960, Adj. R2 = 0.9951, and RMSE = 17.

It can be seen that when using a stepwise algorithm, the PubChemFP 553 and TIC3 descriptors were excluded from the model equation as non-significant. The adequateness of the regression model using the Stepwise algorithm was assessed as follows: R2 = 0.9958, Adj. R2 = 0.9949, and RMSE = 17.

The obtained values of R2 and Adj. R2 close to one as well as the obtained low values of RMSE for both algorithms (Enter and Stepwise) show that the developed regression models can be regarded as adequate. This fact is further confirmed by analysis of the experimentally determined (observed) vs predicted plots (Figs 2 and 3). When using the training set, for both algorithms the established slope was statistically identical to one and the calculated intercept practically equals to zero.

The mathematical models were further validated by comparing the observed and calculated LRIs of the compounds from the validation set (16 compounds). The obtained results (Figs 2 and 3) are equivalent to the ones found for the training set i.e. for both algorithms the slope can be regarded as equal to one and meanwhile the intercept is statistically identical to zero.

In addition, for the validation set it were calculated the corresponding RMSE and the prediction coefficient q2F1. For the Enter algorithm, it was found that RMSE = 25 and q2F1 = 0.9896. For the Stepwise algorithm, it was determined that RMSE = 26 and q2F1 = 0.9886. The acceptably low RMSE values in combination with q2F1 values close to one show that the developed MLR models are still adequate even for the prediction of LRIs of compounds not used to build the regression function.

Fig. 2.
Fig. 2.

Distribution plot of the observed versus predicted values of the dependent variable LRI using the Enter algorithm (the uncertainty of the slope and the intercept is presented as 95% confidence interval)

Citation: Acta Chromatographica 2025; 10.1556/1326.2024.01207

Fig. 3.
Fig. 3.

Distribution plot of the Observed versus Predicted values of the dependent variable LRI using the Stepwise algorithm (the uncertainty of the slope and the intercept is presented as 95% confidence interval)

Citation: Acta Chromatographica 2025; 10.1556/1326.2024.01207

3.2.3 Development and optimization of HS-SPME

The validated MLR models were tested by an external set of compounds, experimentally identified by HS-SPME analysis of the Essential oils. To perform the HS-SPME a semi-polar CAR/PDMS coated fiber was selected, based on the chemical properties of the target compounds, their volatility, and as a good option for the extraction of a wide range of volatile and semi-volatile compounds reported in previously proposed methods [26, 27]. A better signal-to-noise ratio and elimination of matrix components can be achieved by SPME due to the affinity of the volatile compounds to the fiber coating and the realization of preconcentration.

The effects of the conditioning time, extraction time and sample volume on the HS-SPME extraction efficiency were studied using a one-variable-at-a-time approach. For the optimization study nine analytes from the analysis of Rose oil (α-Pinene, β-Pinene, Linalyl alcohol, cis-Rose oxide, Phenethyl alcohol, Nerol, Citronellol, Geraniol, Methyleugenol), seven analytes from Lavender oil analysis (α-Thujene, α-Pinene, Camphene, Sabinene, β-Cymene, Eucalyptol and γ-Terpinene) and nine from Peppermint oil (α-Pinene, Sabinene, Eucalyptol, Menthone, Isomentone, Menthol, Pulegone, Menthyl acetate and β-Caryophyllene) were monitored. The compounds were selected based on belonging to different chemical classes with different properties.

The effect of sample volume was investigated using 5 and 10 µL volumes of Rose, Lavender and Peppermint oils using a conditioning time of 30 min, 10 min sorption time, room temperature and no stirring. It was found that in the analysis of Lavender and Rose oil, 10 µL of the sample is sufficient to obtain high peak intensities under the prescribed chromatographic conditions. A 5 µL volume was found to be sufficient to obtain high intensities in the analysis of the Peppermint oil sample.

The conditioning time was studied at 15, 30, 45, and 60 min using 10 min sorption time, room temperature and no stirring. The obtained results from the optimization of the conditioning time showed that maximum peak areas for the selected volatile compounds in Rose oil and Lavender oil can be achieved after 45 min of conditioning, except α-Pinene, β-Pinene and Camphene. The signals of α-Pinene and β-Pinene decrease with increasing time when Rose oil is analyzed, and β-Pinene and Camphene have maximum achieved peak area at 60 min conditioning time if the analysis is conducted for Lavender oil. Despite the fact the signals achieved at 45 min were high enough for successful identification by GC-MS. Peppermint oil compounds show that the highest peak areas are achieved at 30 min of conditioning time for all selected analytes. Therefore, the conditioning time for the analysis of Rose and Lavender oil was 45 min, while 30 min was used for the analysis of Peppermint oil.

The optimization of the sorption time was performed at 2.5, 5, 7.5, and 10 min using 45 min conditioning time for the analysis of Rose and Lavender oil, and 30 min for the analysis of Peppermint oil, room temperature and no stirring. Due to the higher signals in the analysis of Peppermint oil, the effect of sorption time was carried out only at 2.5 and 5 min. From the achieved results it can be concluded that the sorption efficiency for some of the volatile compounds such as α-Pinene is different in the analysis of the three different matrices. In the analysis of Rose oil and Peppermint oil, the sorption time needs to be shorter to avoid the loss of α-Pinene. For all other volatile compounds, except α-Pinene, a longer sorption time is preferred. For the analysis of Rose and Peppermint oil, 5 min were chosen for HS-SPME, while for Lavender oil the time was set to 10 min.

3.2.4 Testing of the LRI prediction models

According to the prescribed criteria in section 2.2. a sum of 19 additional compounds were identified using the HS-SPME which were not registered (were below the methodological limit of detection) when liquid injection GC-MS analysis was accomplished. However, from the newly detected 19 compounds, only a reduced number of 12 have been selected as a test set since the values of their molecular descriptors were within the corresponding ranges established by the training set (Table 6).

Table 6.

Range of the values of the molecular descriptors in the training and test sets

DescriptorTraining set (n = 87)Test set (n = 12)
MinimumMaximumMinimumMaximum
MLFER_L3.529.754.427.10
MLFER_S0.131.410.160.73
ATSC5c−0.120.09−0.040.06
ATS5i166527156745822638
n3HeteroRing0101
n4Ring0101
n11Ring0100
GATS3c0.632.460.691.90
TIC35918677161
PubchemFP1430101
PubchemFP1470101
PubchemFP5530101
PubchemFP5820101
PubchemFP6390100
PubchemFP6720101
PubchemFP6880101

The experimentally determined LRIs for the test set of compounds as well as their SMILES notations are presented in Table 7.

Table 7.

Experimentally determined LRIs and SMILES for 12 additionally identified compounds using HS-SPME

SetDesignationLRISMILES
Test setDehydrosabinene942CC(C)C12CC1C(=C)C=C2
1,3,5-Cycloheptatriene, 3,7,7-trimethyl959CC1=CC=CC(C=C1)(C)C
p-Menthane979CC(C1)CCC(C1)C(C)C
δ-2-Carene1009CC1=CC2C(C2(C)C)CC1
Rosefuran1098CC1=C(OC=C1)CC=C(C)C
2,4,6-Octatriene, 2,6-dimethyl1128CC=C(C)C=CC=C(C)C
β-Ocimene epoxide1143CC(=CCC1C(O1)(C)C)C=C
Butyric acid, 2-methyl-, hexyl ester1239CCCCCCOC(=O)C(C)CC
Neryl formate1304CC(=CCCC(=CCOC=O)C)C
1,7-epi-Sesquithujene1382CC1=CCC2(C1C2)C(C)CCC=C(C)C
cis-α-Bergamotene1403CC1=CCC2CC1C2(C)CCC=C(C)C
δ-Cadinene1509CC1=CC2C(CCC(=C2CC1)C)C(C)C

The LRIs of the compounds from the test set were predicted based on the two MLR models. The root mean square error and the coefficient of prediction were calculated and for both algorithms, were Stepwise RMSE = 40, q2F2 = 0.9521 and Enter Stepwise RMSE = 41, q2F2 = 0.9503. Table 8 presents the observed values and predicted values derived from the model equations using the Enter or the Stepwise algorithm. The obtained results for the test set are worsen compared to the ones for the validation set but still informative having in mind that a simple linear regression with only 14 (or 16) descriptors is used.

Table 8.

The observed and predicted LRIs by the Enter and Stepwise MLR models for the test set

CompoundObserved LRI (y)EnterStepwise
Predicted LRI (ŷ)y-ŷPredicted LRI (ŷ)y-ŷ
Dehydrosabinene9428974589844
1,3,5-Cycloheptatriene, 3,7,7-trimethyl9591063−1031061−102
p-Menthane979981−29754
δ-2-Carene10099882199217
Rosefuran10981113−161111−13
2,4,6-Octatriene, 2,6-dimethyl1128105474105771
β-Ocimene epoxide1143111825111726
Butyric acid, 2-methyl-, hexyl ester12391246−71254−15
Neryl formate13041316−121314−10
1, 7-epi-Sesquithujene1382138111387−5
cis-α-Bergamotene14031420−171420−17
δ-Cadinene15091520−101521−12

In recent studies using MLR for the development of a method for the prediction of Kovats retention indices of Essential Oils in Gas Chromatography analysis Teuku et al. used 30 molecular descriptors [10]. For the training set the authors obtained R2 0.970, RMSE 57 and for the testing set R2 0.970, RMSE 57 [10]. Yan et al. developed a regression model incorporating 10 descriptors for a diverse set of flavour compounds (n = 107) achieving R2 = 0.9741, RMSEtraining = 54, RMSEcv = 66 and Q2 = 0.9619 [28]. In another study focused on the prediction of the Kovats retention indices of a large set of terpenes (n = 523), Hemmateenejad et al. proposed two MLR equations each including 8 descriptors (selected from a pool of 55 descriptors) with the following characteristics: i) R2 = 0.915, Q2 = 0.910, RMSCV (root mean square error of cross-validation) = 60; ii) R2 = 0.925, Q2 = 0.921, RMSCV = 56 [29]. Yan et al. built three MRL models for the prediction of the retention indices of compounds separated from plant Essential oils by different types of chromatographic columns i.e. column-NP (nonpolar), column-SP (slightly polar) and column-P (polar) [30]. A pool of 600 molecular descriptors was used to derive 8, 7, and 12 significant and independent descriptors which were included in the mathematical equation, respectively. The authors have declared the following results: i) column-NP model (n = 468) - R2 = 0.9522, RMSEtraining = 67, RMSEvalidation = 70, RMSETest = 69, Q2 = 0.9491; ii) column-SP model (n = 469) - R2 = 0.9528, RMSEtraining = 67, RMSEvalidation = 68, RMSETest = 68, Q2 = 0.9508; iii) column-P model (n = 457) - R2 = 0.9415, RMSEtraining = 94, RMSEvalidation = 102, RMSETest = 105, Q2 = 0.9307 [30].

The discussed results from other researchers show that the developed MLR models in the current work lead to comparable or better results in view of the obtained R2, RSME and q2 values.

4 Conclusions

It can be summarized that two approaches (direct liquid injection and HS-SPME) have been successfully developed for qualitative analysis of volatile compounds in rose, lavender and peppermint oil by gas chromatography with mass spectrometry. The proposed multiple linear regression models for prediction of the LRIs values for semi-standard non-polar stationary phase were evolved using experimentally obtained data from analysis of real samples of different Essential oils. The developed two regression models are relatively simple (include only 16 or 14 independent variables, respectively), and both lead to comparable and adequate results for predicting linear retention indices of volatile components in Essential oils. An important requirement for the adequacy of the mathematical models is that the values of the molecular descriptors for a given compound entered in the regression equations should be within the ranges we have studied and specified in the current work.

Author contributions

“Conceptualization, Asya Hristozova, methodology, Asya Hristozova and Slava Tsoneva; software, Asya Hristozova; validation, Kiril Simitchiev, Asya Hristozova and Margarita Batmazyan; formal analysis, Asya Hristozova and Margarita Batmazyan; investigation, Asya Hristozova and Margarita Batmazyan; resources, Asya Hristozova; data curation, Asya Hristozova, Kiril Simitchiev; writing—original draft preparation, Asya Hristozova and Margarita Batmazyan; writing—review and editing, Kiril Simitchiev; visualization, Asya Hristozova and Margarita Batmazyan; supervision, Erwin Rosenberg and Kiril Simitchiev; project administration, Veselin Kmetov; funding acquisition, Veselin Kmetov. All authors have read and agreed to the published version of the manuscript.”

Acknowledgments

Bulgarian National Science Fund, Project KP-06-Austria/3 (2021). Challenges and development of GC-MS/MS analytical methods – academic partnership.

Assoc. Prof. Nikolay Kochev and Assist. Prof. Veselina Paskaleva (University of Plovdiv "Paisii Hilendarski", Department of Analytical Chemistry and Computer Chemistry) are gratefully acknowledged for their support in this work.

References

  • 1.↑

    Trovato, E.; Micalizzi, G.; Dugo, P.; Utczás, M.; Mondello, L. In Handbook of Essential Oils: Science, Technology, and Applications; Baser, K. H. C., Buchbauer, G., Eds. CRC Press, 3rd ed., 2021; pp 229–251.

    • Search Google Scholar
    • Export Citation
  • 2.↑

    Nedeltcheva-Antonova, D.; Stoicheva, P.; Antonov, L. Ind. Crops Prod. 2017, 108, 36–43.

  • 3.↑

    Koksall, N.; Aslancan, H.; Sadighazadi, S.; Kafkas, E. Acta Sci. Pol. Hortorum Cultus 2015, 14, 105–114.

  • 4.↑

    Sparkman, O. D.; Penton, Z. E.; Kitson, F. G. Gas Chromatography and Mass Spectrometry: a Practical Guide; Elsevier, 2011; pp 15.

  • 5.↑

    Poole, C. F. Gas Chromatography; Elsevier, 2012; pp 519.

  • 6.↑

    Dool, V. D. H.; Kratz, D. P. J. Chromatogr. A. 1963, 11, 463–471.

  • 7.↑

    Rigano, F.; Arigò, A.; Oteri, M.; La Tella, R.; Dugo, P.; Mondello, L. J. Chromatogr. A. 2021, 1640, 461963.

  • 8.↑

    Liapikos, T.; Zisi, C.; Kodra, D.; Kademoglou, K.; Diamantidou, D.; Begou, O.; Pappa-Louisi, A.; Theodoridis, G. J. Chromatogr. B 2022, 1191, 123132.

    • Search Google Scholar
    • Export Citation
  • 9.↑

    Alexander, D. L. J.; Tropsha, A.; Winkler, D. A. J. Chem. Inf. Model. 2015, 55, 1316–1322.

  • 10.↑

    Noviandy, T. R.; Maulana, A.; Sasmita, N. R.; Suhendra, R.; Irvanizam, I.; Muslem, M.; Idroes, G. M.; Yusuf, M.; Sofyan, H.; Abidin, T. F.; Idroes, R. J. Eng. Sci. Technol. 2022, 17, 0306–0326.

    • Search Google Scholar
    • Export Citation
  • 11.↑

    Lucero, M.; Estell, R.; Tellez, M.; Fredrickson, E. Phytochem. Anal. 2009, 20, 378–384.

  • 12.↑

    Golbraikh, A.; Tropsha, A. J. Mol. Graph. Model. 2002, 20, 269–276.

  • 13.↑

    Qin, L.-T.; Liu, S.-S.; Liu, H.-L.; Tong, J. J. Chromatogr. A. 2009, 1216, 5302–5312.

  • 14.

    Chirico, N.; Gramatica, P. J. Chem. Inf. Model. 2012, 52, 2044–2058.

  • 15.↑

    Lancioni, C.; Castells, C.; Candal, R.; Tascon, M. Adv. Sample Preparation 2022, 3, 100035.

  • 16.↑

    Yan, J.; Cao, D.-S.; Guo, F.-Q.; Zhang, L.-X.; He, M.; Huang, J.-H.; Xu, Q.-S.; Liang, Y.-Z. J. Chromatogr. A 2012, 1223, 118–125.

  • 17.

    Seber, G. A. F.; Lee, A. L. Linear Regression Analysis; John Wiley & Sons: New Jersey, 2003; pp 35.

  • 18.

    Garkani-Nejad, Z.; Karlovits, M.; Demuth, W.; Stimpfl, T.; Vycudilik, W.; Jalali-Heravi, M.; Varmuza, K. J. Chromatogr. A. 2004, 1028, 287–295.

    • Search Google Scholar
    • Export Citation
  • 19.

    Idroes, R.; Noviandy, T. R.; Maulana, A.; Suhendra, R.; Sasmita, N. R.; Muslem, M.; Idroes, G. M.; Irvanizam, I. Int. Rev. Model. Simul 2019, 12, 373.

    • Search Google Scholar
    • Export Citation
  • 20.↑

    National Library of Medicine. PubChem on-line available on https://pubchem.ncbi.nlm.nih.gov (accessed Jan 24, 2024).

  • 21.↑

    PaDEL-Descriptor on-line available on http://www.yapcwsoft.com/dd/padeldescriptor (accessed Jan 24, 2024).

  • 22.↑

    Yap, C. W. J. Comput. Chem. 2011, 32, 1466–1474.

  • 23.↑

    Roy, P. P.; Leonard, J. T.; Roy, K. Chemometr Intell. Lab Syst. 2008, 90, 31–42.

  • 24.↑

    Platts, J. A.; Butina, D.; Abraham, M. H.; Hersey, A. J. Chem. Inf. Comput. Sci. 1999, 39, 835–845.

  • 25.↑

    Antanasijević, J.; Antanasijević, D.; Pocajt, V.; Trišović, N.; Fodor-Csorba, K. RSC Adv. 2016, 6, 18452–18464.

  • 26.↑

    Kalogiouri, N. P.; Manousi, N.; Rosenberg, E.; Zachariadis, G. A.; Paraskevopoulou, A.; Samanidou, V. Food Chem. 2021, 363, 130331.

  • 27.↑

    Lo, M.-M.; Benfodda, Z.; Bénimélis, D.; Fontaine, J.-X.; Molinié, R.; Meffre, P. ACS Omega 2021, 6, 12691–12698.

  • 28.↑

    Yan, J.; Cao, D.-S.; Guo, F.-Q.; Zhang, L.-X.; He, M.; Huang, J.-H.; Xu, Q.-S.; Liang, Y.-Z. J. Chromatogr. A. 2012, 1223, 118–125.

  • 29.↑

    Hemmateenejad, B.; Javadnia, K.; Elyasi, M. Anal. Chim. Acta 2007, 592, 72–81.

  • 30.↑

    Yan, J.; Huang, J.; He, M.; Lu, H.; Yang, R.; Kong, B.; Xu, Q.; Liang, Y. J. Sep. Sci. 2013, 36, 2464–2471.

  • 1.

    Trovato, E.; Micalizzi, G.; Dugo, P.; Utczás, M.; Mondello, L. In Handbook of Essential Oils: Science, Technology, and Applications; Baser, K. H. C., Buchbauer, G., Eds. CRC Press, 3rd ed., 2021; pp 229–251.

    • Search Google Scholar
    • Export Citation
  • 2.

    Nedeltcheva-Antonova, D.; Stoicheva, P.; Antonov, L. Ind. Crops Prod. 2017, 108, 36–43.

  • 3.

    Koksall, N.; Aslancan, H.; Sadighazadi, S.; Kafkas, E. Acta Sci. Pol. Hortorum Cultus 2015, 14, 105–114.

  • 4.

    Sparkman, O. D.; Penton, Z. E.; Kitson, F. G. Gas Chromatography and Mass Spectrometry: a Practical Guide; Elsevier, 2011; pp 15.

  • 5.

    Poole, C. F. Gas Chromatography; Elsevier, 2012; pp 519.

  • 6.

    Dool, V. D. H.; Kratz, D. P. J. Chromatogr. A. 1963, 11, 463–471.

  • 7.

    Rigano, F.; Arigò, A.; Oteri, M.; La Tella, R.; Dugo, P.; Mondello, L. J. Chromatogr. A. 2021, 1640, 461963.

  • 8.

    Liapikos, T.; Zisi, C.; Kodra, D.; Kademoglou, K.; Diamantidou, D.; Begou, O.; Pappa-Louisi, A.; Theodoridis, G. J. Chromatogr. B 2022, 1191, 123132.

    • Search Google Scholar
    • Export Citation
  • 9.

    Alexander, D. L. J.; Tropsha, A.; Winkler, D. A. J. Chem. Inf. Model. 2015, 55, 1316–1322.

  • 10.

    Noviandy, T. R.; Maulana, A.; Sasmita, N. R.; Suhendra, R.; Irvanizam, I.; Muslem, M.; Idroes, G. M.; Yusuf, M.; Sofyan, H.; Abidin, T. F.; Idroes, R. J. Eng. Sci. Technol. 2022, 17, 0306–0326.

    • Search Google Scholar
    • Export Citation
  • 11.

    Lucero, M.; Estell, R.; Tellez, M.; Fredrickson, E. Phytochem. Anal. 2009, 20, 378–384.

  • 12.

    Golbraikh, A.; Tropsha, A. J. Mol. Graph. Model. 2002, 20, 269–276.

  • 13.

    Qin, L.-T.; Liu, S.-S.; Liu, H.-L.; Tong, J. J. Chromatogr. A. 2009, 1216, 5302–5312.

  • 14.

    Chirico, N.; Gramatica, P. J. Chem. Inf. Model. 2012, 52, 2044–2058.

  • 15.

    Lancioni, C.; Castells, C.; Candal, R.; Tascon, M. Adv. Sample Preparation 2022, 3, 100035.

  • 16.

    Yan, J.; Cao, D.-S.; Guo, F.-Q.; Zhang, L.-X.; He, M.; Huang, J.-H.; Xu, Q.-S.; Liang, Y.-Z. J. Chromatogr. A 2012, 1223, 118–125.

  • 17.

    Seber, G. A. F.; Lee, A. L. Linear Regression Analysis; John Wiley & Sons: New Jersey, 2003; pp 35.

  • 18.

    Garkani-Nejad, Z.; Karlovits, M.; Demuth, W.; Stimpfl, T.; Vycudilik, W.; Jalali-Heravi, M.; Varmuza, K. J. Chromatogr. A. 2004, 1028, 287–295.

    • Search Google Scholar
    • Export Citation
  • 19.

    Idroes, R.; Noviandy, T. R.; Maulana, A.; Suhendra, R.; Sasmita, N. R.; Muslem, M.; Idroes, G. M.; Irvanizam, I. Int. Rev. Model. Simul 2019, 12, 373.

    • Search Google Scholar
    • Export Citation
  • 20.

    National Library of Medicine. PubChem on-line available on https://pubchem.ncbi.nlm.nih.gov (accessed Jan 24, 2024).

  • 21.

    PaDEL-Descriptor on-line available on http://www.yapcwsoft.com/dd/padeldescriptor (accessed Jan 24, 2024).

  • 22.

    Yap, C. W. J. Comput. Chem. 2011, 32, 1466–1474.

  • 23.

    Roy, P. P.; Leonard, J. T.; Roy, K. Chemometr Intell. Lab Syst. 2008, 90, 31–42.

  • 24.

    Platts, J. A.; Butina, D.; Abraham, M. H.; Hersey, A. J. Chem. Inf. Comput. Sci. 1999, 39, 835–845.

  • 25.

    Antanasijević, J.; Antanasijević, D.; Pocajt, V.; Trišović, N.; Fodor-Csorba, K. RSC Adv. 2016, 6, 18452–18464.

  • 26.

    Kalogiouri, N. P.; Manousi, N.; Rosenberg, E.; Zachariadis, G. A.; Paraskevopoulou, A.; Samanidou, V. Food Chem. 2021, 363, 130331.

  • 27.

    Lo, M.-M.; Benfodda, Z.; Bénimélis, D.; Fontaine, J.-X.; Molinié, R.; Meffre, P. ACS Omega 2021, 6, 12691–12698.

  • 28.

    Yan, J.; Cao, D.-S.; Guo, F.-Q.; Zhang, L.-X.; He, M.; Huang, J.-H.; Xu, Q.-S.; Liang, Y.-Z. J. Chromatogr. A. 2012, 1223, 118–125.

  • 29.

    Hemmateenejad, B.; Javadnia, K.; Elyasi, M. Anal. Chim. Acta 2007, 592, 72–81.

  • 30.

    Yan, J.; Huang, J.; He, M.; Lu, H.; Yang, R.; Kong, B.; Xu, Q.; Liang, Y. J. Sep. Sci. 2013, 36, 2464–2471.

  • Collapse
  • Expand
Cover Acta Chromatographica
Acta Chromatographica
Online ISSN:
2083-5736

Senior editors

Editor(s)-in-Chief: Sajewicz, Mieczyslaw, University of Silesia, Katowice, Poland

Editors(s)

  • Danica Agbaba, University of Belgrade, Belgrade, Serbia (1953-2024)
  • Łukasz Komsta, Medical University of Lublin, Lublin, Poland
  • Ivana Stanimirova-Daszykowska, University of Silesia, Katowice, Poland
  • Monika Waksmundzka-Hajnos, Medical University of Lublin, Lublin, Poland

Editorial Board

  • Ravi Bhushan, The Indian Institute of Technology, Roorkee, India
  • Jacek Bojarski, Jagiellonian University, Kraków, Poland
  • Bezhan Chankvetadze, State University of Tbilisi, Tbilisi, Georgia
  • Michał Daszykowski, University of Silesia, Katowice, Poland
  • Tadeusz H. Dzido, Medical University of Lublin, Lublin, Poland
  • Attila Felinger, University of Pécs, Pécs, Hungary
  • Kazimierz Glowniak, Medical University of Lublin, Lublin, Poland
  • Bronisław Glód, Siedlce University of Natural Sciences and Humanities, Siedlce, Poland
  • Anna Gumieniczek, Medical University of Lublin, Lublin, Poland
  • Urszula Hubicka, Jagiellonian University, Kraków, Poland
  • Krzysztof Kaczmarski, Rzeszow University of Technology, Rzeszów, Poland
  • Huba Kalász, Semmelweis University, Budapest, Hungary
  • Katarina Karljiković Rajić, University of Belgrade, Belgrade, Serbia
  • Imre Klebovich, Semmelweis University, Budapest, Hungary
  • Angelika Koch, Private Pharmacy, Hamburg, Germany
  • Piotr Kus, Univerity of Silesia, Katowice, Poland
  • Debby Mangelings, Free University of Brussels, Brussels, Belgium
  • Emil Mincsovics, Corvinus University of Budapest, Budapest, Hungary
  • Ágnes M. Móricz, Centre for Agricultural Research, Budapest, Hungary
  • Gertrud Morlock, Giessen University, Giessen, Germany
  • Anna Petruczynik, Medical University of Lublin, Lublin, Poland
  • Robert Skibiński, Medical University of Lublin, Lublin, Poland
  • Bernd Spangenberg, Offenburg University of Applied Sciences, Germany
  • Tomasz Tuzimski, Medical University of Lublin, Lublin, Poland
  • Adam Voelkel, Poznań University of Technology, Poznań, Poland
  • Beata Walczak, University of Silesia, Katowice, Poland
  • Wiesław Wasiak, Adam Mickiewicz University, Poznań, Poland
  • Igor G. Zenkevich, St. Petersburg State University, St. Petersburg, Russian Federation

 

SAJEWICZ, MIECZYSLAW
E-mail:mieczyslaw.sajewicz@us.edu.pl

Indexing and Abstracting Services:

  • Science Citation Index
  • Sci Search
  • Research Alert
  • Chemistry Citation Index and Current Content/Physical
  • Chemical and Earth Sciences
  • SCOPUS
  • GoogleScholar
  • Index Copernicus
  • CABI
  • CABELLS Journalytics

2023  
Web of Science  
Journal Impact Factor 1.7
Rank by Impact Factor Q3 (Chemistry, Analytical)
Journal Citation Indicator 0.43
Scopus  
CiteScore 4.0
CiteScore rank Q2 (General Chemistry)
SNIP 0.706
Scimago  
SJR index 0.344
SJR Q rank Q3

Acta Chromatographica
Publication Model Online only
Gold Open Access
Submission Fee none
Article Processing Charge 400 EUR/article
Effective from  1st Feb 2025:
700 EUR/article
Regional discounts on country of the funding agency World Bank Lower-middle-income economies: 50%
World Bank Low-income economies: 100%
Further Discounts Editorial Board / Advisory Board members: 50%
Corresponding authors, affiliated to an EISZ member institution subscribing to the journal package of Akadémiai Kiadó: 100%
Subscription Information Gold Open Access
Purchase per Title  

Acta Chromatographica
Language English
Size A4
Year of
Foundation		 1988
Volumes
per Year		 1
Issues
per Year		 4
Founder Institute of Chemistry, University of Silesia
Founder's
Address		 PL-40-007 Katowice, Poland, Bankowa 12
Publisher Akadémiai Kiadó
Publisher's
Address		 H-1117 Budapest, Hungary 1516 Budapest, PO Box 245.
Responsible
Publisher		 Chief Executive Officer, Akadémiai Kiadó
ISSN 2083-5736 (Online)

Monthly Content Usage

Abstract Views Full Text Views PDF Downloads
Jul 2024 0 61 16
Aug 2024 0 131 19
Sep 2024 0 168 14
Oct 2024 0 255 31
Nov 2024 0 122 18
Dec 2024 0 82 38
Jan 2025 0 32 16