Advertisement

Analytical and Bioanalytical Chemistry

, Volume 410, Issue 19, pp 4657–4668 | Cite as

Ionic liquids as water-compatible GC stationary phases for the analysis of fragrances and essential oils

  • Cecilia Cagliero
  • Carlo Bicchi
  • Chiara Cordero
  • Erica Liberto
  • Patrizia Rubiolo
  • Barbara Sgorbini
Research Paper
Part of the following topical collections:
  1. Ionic Liquids as Tunable Materials in (Bio)Analytical Chemistry

Abstract

Fragrances and products deriving from essential oils are often formulated or diluted in aqueous media, usually ethanol/water. Gas chromatography (GC) is the technique of choice to analyze volatiles. However, when using columns coated with conventional stationary phases, its application to aqueous samples often requires time-consuming and/or discriminative sample preparation techniques to extract the target analytes from the aqueous medium, so as to avoid its direct injection. In GC with conventional columns, water produces peak asymmetry, poor sensitivity and efficiency, strong adsorption, stationary phase degradation, and, last but not least, it is not easy to detect reliably when present in high amounts. In 2012, Armstrong’s group introduced new fully water-compatible ionic-liquid (IL)-based GC capillary columns based on phosphonium and imidazolium derivative cations combined with trifluoromethanesulphonate. These columns were recently made available commercially by Supelco, under the trade name Watercol™. These derivatives maintain IL’s unique selectivity and chromatographic properties, and enable water to be used as injection solvent, thus avoiding the sample preparation procedures required by conventional columns. This study reports and critically discusses the results of commercially available water-compatible IL columns for direct analysis of aqueous samples in the fragrance and essential oil fields by GC with thermal conductivity (TCD) and/or flame ionization detectors (FID). The results showed that water-compatible IL-based stationary phases can successfully be adopted for qualitative and quantitative analysis of fragrances and essential oils directly diluted in aqueous solvents. On the other hand, the study also shows that their inertness needs to be further increased and (possibly) the range of operative temperature extended when water is the main solvent of the sample.

Keywords

Ionic liquids GC Aqueous samples Water-compatible stationary phases Essential oils Fragrances 

Introduction

Gas chromatography (GC) is the technique of choice for the analysis of fragrances and essential oils. However, these products are often formulated or diluted in aqueous media, usually ethanol/water in different ratios. Quali-quantitative analysis of their composition, or of one or more specific components, either as markers of quality or that are limited by regulatory authorities (e.g., suspected allergens in perfumes or cosmetics by the EU [1, 2]) is often required. When they are in an aqueous medium, this often entails the adoption of sample preparation techniques to extract target analytes that can operate a discrimination of the components of the sample, or the dilution with compatible solvents that affects sensitivity to avoid direct injection in GC of aqueous solutions. A similar approach is also necessary when water is an analyte to be quantified, since with columns coated with conventional stationary phases, it produces degradation of those phases, peak asymmetry, poor sensitivity, poor efficiency, and strong adsorption; further, it cannot be detected in high amounts, since it produces broad peaks with low peak area repeatability, and unsatisfactory limits of detection and quantitation [3]. In the past, these problems were solved, although not satisfactorily, by using packed columns filled or wall-coated (PoraPLOT) with molecular sieves as stationary phases.

Over the last two decades, ionic liquids (ILs) have been shown to have great potential as GC stationary phases [4]. In 2012, Armstrong et al. showed that some ionic-liquid (IL)-based GC capillary columns have not only good selectivity but also high stability and compatibility toward water as analyte, compared to traditional commercial columns [3]. Moreover, these stationary phases provide good peak symmetry, thus avoiding chromatographic interference with other analytes. One of IL’s main advantages is that their chemical structures can be custom-designed to add compatibility with specific compounds to their selectivity. The original ILs proposed by Armstrong’s group were based on phosphonium and imidazolium derivative cations, combined with anions consisting of 2 or 3 units of trifluoromethanesulphonate; the group proposed a number of applications to test the reliability of the above and other ILs derivatives, as GC stationary phases to measure the water content of matrices in different fields, including in active pharmaceutical ingredients [5, 6] and in honey [7], and also to measure the water/ethanol content of various consumer products. More recently, Supelco introduced water-compatible IL columns commercially under the trade name Watercol™, with different retention properties based on the above mentioned ILs (for details, see experimental [8]).

To the best of the authors’ knowledge, the adoption of water-compatible ILs as GC stationary phases when water is the main solvent, as is often the case in the fragrance and essential oil fields, has not yet been investigated. This study evaluates the quali-quantitative performance of water-compatible IL columns for direct analysis of aqueous or water/ethanol samples in the fragrance and essential oil field by GC, with thermal conductivity (TCD) and/or flame ionization detectors (FID). TCD was applied to detect the peak of water while FID was used to increase the response of target compounds with water as solvent.

Materials and methods

Samples

A mixture of ethanol and water in a 1:1 ratio was analyzed to test the columns’ performance when analyzing aqueous solutions. The Grob test mixture [9], consisting of a mixture of decane, dodecane, 1-octanol, methyl decanoate, methyl undecanoate, methyl dodecanoate, 2,6-dimethylphenol, 2,6-dimethylaniline, dicyclohexylamine, and 2-ethylcaproic acid in hexane and methylene chloride, was purchased from Sigma–Aldrich (Milan, Italy) and analyzed as such. The suspected allergen standard mixture included 29 compounds: (1) limonene (CAS: 138-86-3), (2) linalool (CAS: 78-70-6), (3) estragole (CAS: 140-67-0), (4) phenylacetaldehyde (CAS: 122-78-1), (5) methyl 2-octynoate (CAS: 111-12-6), (6) citronellol (CAS: 106-22-9), (7) geraniol (CAS: 106-24-1), (8) benzyl alcohol (CAS: 100-24-1), (9) neral (CAS: 106-26-3), (10) geranial (CAS: 141-27-5), (11) α-isomethyl ionone (CAS: 15789-90-9), (12) methyl eugenol (CAS: 93-15-2), (13) hydroxycitronellal (CAS: 107-75-5), (14) α-ionone (CAS: 127-41-3), (15) eugenol (CAS: 97-53-0), (16) lilial (CAS: 80-54-6), (17) cinnamaldehyde (CAS: 104-55-2), (18) anisyl alcohol (CAS: 1331-81-3), (19) farnesol isomers (CAS: 4602-84-0), (20) cinnamyl alcohol (CAS:104-54-1), (21) amyl cinnamaldehyde (CAS: 122-40-7), (22) hexyl cinnamaldehyde (CAS: 39350-49-5), (23) α-pentylcinnamyl alcohol (CAS: 14316-49-5), (24) vanillin (CAS: 121-33-5), (25) lyral isomers (CAS: 130066-44-3), (26) coumarin (CAS: 91-64-5), (27) benzyl benzoate (CAS: 120-51-4), (28) benzyl salicylate (CAS: 118-58-1), and (29) benzyl cinnamate (CAS: 103-41-3). They were solubilized in a mixture of water and ethanol 1:1 at a concentration of 500 mg L−1. A stock quantitation standard mixture consisting of linalool (2), linalyl acetate (30), and α-ionone (14) was prepared by adding the pure standards (purchased from Sigma–Aldrich) to an appropriate volume of ethanol at an initial concentration of 10 g L−1. Working solutions were prepared by diluting the stock solution in appropriate volumes of a mixture of water and ethanol 1:1. A set of five commercial perfumes was purchased from a local market and injected as such. 2-Methylbutanol (from Sigma–Aldrich) at a concentration of 1 g L−1 as internal standard was added to both calibration solution and commercial perfumes.

The essential oils of peppermint (Mentha x piperita L.) and lavender (Lavandula angustifolia Mill.) were obtained by hydrodistillation following the European Pharmacopeia [10], while tea-tree (Melaleuca alternifolia (Maiden & Betche) Cheel) essential oil was supplied by Erboristeria Magentina SrL (Poirino, TO, Italy). The essential oils were solubilized in a mixture of water and ethanol 1:1 at a concentration of 5 g L−1 before analysis. Deionized water (18.2 MΩ cm) was obtained from a Milli-Q water purification system (Millipore, Merck, Milan, Italy) while ethanol (99.9%) was purchased from Sigma–Aldrich (Milan, Italy).

Instrumental set-up

Analyses were carried out on a Shimadzu GC-FID-TCD system consisting of a Shimadzu GC 2010 equipped with FID in parallel with a TCD; the two detectors were alternately operated depending on the experiments; data were processed with GC Solution 2.53SU software (Shimadzu, Milan, Italy). Analyte identification was performed by GC-MS using a Shimadzu QP2010-PLUS GC-MS system equipped with Shimadzu GCMS Solution 2.51 software.

Columns

GC analyses were carried out with two 30 m × 0.25 mm dc × 0.20 μm df Watercol™ columns of different polarities, i.e., Watercol™ 1460 (non-bonded Tri(tripropylphosphoniumhexanamido)-trimethylamine trifluoromethanesulfonate) and Watercol™ 1910 (non-bonded 1,11-Di(3-hydroxyethylimidazolium)3,6,9-trioxaundecane trifluoromethanesulfonate). The results were compared with those from a 30 m × 0.25 mm dc × 0.25 μm df column coated with 14%-cyanopropylphenyl 86%-dimethylpolysiloxane (OV-1701) and from two other commercial ionic liquid columns, namely SLB-IL60 and SLB-IL60i (30 m × 0.25 mm dc × 0.20 μm df). All IL columns were from Supelco (Bellefonte, PA, USA) while OV-1701 column was from MEGA (Legnano, Mi, Italy).

Analysis conditions

The analysis conditions were as follows: injector temperature 220 °C, liner volume 1 mL, FID temperature 230 °C, FID sampling rate 40 ms, TCD temperature 250 °C, TCD sampling rate 40 ms, and TCD makeup gas He (flow, 1 ml min−1). All temperatures were reduced to 200 °C for 1910. The temperature program for all columns (with the exception of Watercol™ 1910) is as follows: (i) 40 °C//2 °C/min//230 °C (2 min) for analyses of water:ethanol mixture, allergens, quantitation standard mixture, and perfumes; (ii) 40 °C//3 °C/min//230 °C (2 min) for Grob test; and (iii) 70 °C//3 °C/min//230 °C (2 min) for essential oils. The final temperature was set at 180 °C for analyses with Watercol™ 1910, while the time of the final isotherm was increased to 30 min. Flow rate (He) was 1 mL min−1. Injection mode was split. Split ratio was 1:100 for perfumes and quantitation standard mixture, 1:20 for Grob test, allergens, and essential oils, while a split ratio of 1:5 was used for TCD analyses. The MS operated in electron impact ionization mode (EI) at 70 eV, scan rate of 666 μ/s, and mass range of 35–350 m/z.

Analytes identification and quantitation

Analytes were identified by comparing their mass spectra and linear retention indices (ITS) to those of authentic standards, or to those in commercial or in-house libraries, or to literature data. Retention indices were calculated vs. a C9-C30 hydrocarbon solution analyzed under the conditions reported above.

The external standard calibration method was applied to quantify the target components of the commercial perfumes with GC-FID, by building a calibration curve for each compound. The analytical performances of the quantitation methods were tested for analyte repeatability and intermediate precision, and linearity R2. Peak area repeatability was measured by analyzing each point of the calibration curves five times by GC-FID, while intermediate precision was determined by analyzing aliquots of the same samples every 3 weeks over a period of 3 months. The limit of detection (LOD) and limit of quantitation (LOQ) were determined by injecting the quantitation standard mixture at increasing dilutions until reaching a signal-to-noise ratio of 3:1 (LOD) and 10:1 (LOQ).

Head space SPME sampling conditions

A 1 cm PDMS/DVB OC (over coated) fiber from Supelco (Bellefonte, USA) was used. The sampling conditions are as follows: amount 10 μL, sampling temperature 50 °C, and sampling time 15 min.

Results and discussion

This study evaluates the performances of the two water-compatible columns, i.e., Watercol™ 1460 (hereafter 1460) and Watercol™ 1910 (hereafter 1910). The numbers, 1460 and 1910, indicate the respective water ITS on the two columns, calculated vs. a hydrocarbon mixture. The study tested these columns qualitatively in terms of stability, intermediate precision of ITS, peak width and symmetry (tailing factor), and analyte recovery, and quantitatively for their linearity, repeatability, LOD, and LOQ. Tests were carried out with a 1:1 ethanol/water mixture, the Grob test, a standard mixture of regulated suspected allergens included in the EU list [2], and on real-world samples consisting of ethanol/water solutions of three essential oils of interest in the cosmetic, food and pharmaceutical fields (lavender, peppermint, and tea-tree oil), and five commercially available perfumes. All results were compared to those obtained with a conventional OV-1701 column and/or a commercial inert IL column (SLB-IL60i). In particular, SLB-IL60i was chosen because it had been shown to give comparable or sometimes better chromatographic results than those of columns coated with conventional stationary phases (polydimethyl siloxane or PEG20M) [11]. Lastly, where necessary, the Watercol™ results were compared to those obtained for the same analytes dissolved in conventional solvents, or resulting from suitable sample preparation procedures.

Column performances and selectivity

The performances of the 1460 and 1910 columns were evaluated in terms of stability to water injection, inertness, and efficiency.

Column performance after water injection

The first series of experiments evaluated the stability of the columns investigated when water is injected as main solvent. The consistency of column performance when large amounts of water are injected was evaluated with a 1:1 water/ethanol solution. The sample was injected in both 1460 and 1910 columns installed in a GC-TCD system (Fig. 1), 20 times in the same day, and 3 consecutive times each day for the subsequent 10 days to measure water and ethanol peak areas, retention times, and linear retention indices (ITS). Retention of water with the two columns significantly varied because of their widely differing selectivity, as well as their relative retention vs. ethanol, which, conversely, was rather constant. In particular, with 1910, the ITS difference between water and ethanol was about 450 units (for analysis conditions, see experimental). The results of 50 injections of the water/ethanol solution showed a very stable performance, ITS RSD% being below 2% for both analytes analyzed on both columns, and peak abundance was highly consistent, with RSD% on their absolute areas never exceeding 4%.
Fig. 1

GC-TCD profiles of a 1:1 ethanol:water standard mixture analyzed with the two Watercol™ columns

Column performance with the Grob test

Simultaneously, the Grob test was also injected for ten consecutive days into the two investigated columns, to check their inertness and efficiency vs. a set of model analytes. Table 1 reports retention times, peak width, tailing factors, and recovery of the Grob test components, compared to a conventional SLB-IL60i. The recovery percentage vs. SLB-IL60i was measured using 1-octanol as internal reference for normalization.
Table 1

Retention times, tailing factors peak widths (σ), and recovery vs. IL60i of the Grob test components when analyzed on Watercol 1460, Watercol 1910, and SLB-IL60i

 

Ret. Time

Tailing factor

σ (min)

Recovery (%)

IL60i

Watercol 1460

Watercol 1910

IL60i

Watercol 1460

Watercol 1910

IL60i

Watercol 1460

Watercol 1910

Watercol 1460

Watercol 1910

1-Octanol

21.97

23.85

16.87

1.46

1.136

1.227

0.042

0.047

0.049

100.0

100.0

2,3-Butanediol

N.D.

28.22

32.72

N.D.

1.847

1.541

N.D.

0.060

0.134

N.M.

N.M.

2,3-Dimethylphenol

38.01

39.46

32.27

0.962

1.108

0.967

0.040

0.035

0.037

80.2

96.1

3,5-Dimethylaniline

34.85

34.58

31.25

0.986

1.034

0.995

0.038

0.035

0.039

123.3

115.8

Decane

3.80

3.01

2.15

1.043

N.M.

1.134

0.016

N.M.

0.015

N.M.

93.2

Dicyclohexylamine

59.42

21.48

N.D.

5.685

2.00

N.D.

0.818

0.151

N.D.

85.0

0.0

Dodecane

8.64

5.58

2.97

0.942

1.049

0.931

0.025

0.174

0.021

88.6

79.8

Hexanoic acid, 2-ethyl-

36.89

43.95

N.D.

1.305

2.61

N.D.

0.239

0.098

N.D.

564.5

N.M.

Methyl decanoate

29.27

20.03

12.75

0.937

1.07

0.937

0.033

0.056

0.058

101.2

64.7

Methyl undecanoate

34.33

23.09

15.08

0.908

1.032

0.839

0.034

0.045

0.060

97.1

66.3

Methyl dodecanoate

39.174

26.00

17.45

0.896

1.018

0.686

0.034

0.037

0.073

93.0

75.1

N.D. not detected, N.M. not measured

As expected, the two columns differed in terms of both retention and selectivity. Retention of 1910 was lower than that of 1460, the total analysis time for the components detected being around 20% less. Their selectivity vs. the Grob test components differed quite widely, making them complementary in particular for the analysis of complex mixtures.

With 1460, most peaks were narrow and with a good symmetry, with the exception of dicyclohexylamine and 2-ethylhexanoic acid, whose peak shapes were significantly distorted, their tailing factors being 2.00 and 2.61 respectively. The 2,3-butandiol peak width was acceptable (0.060) but it was strongly adsorbed (about 90%), explaining its high tailing factor (1.847).

With 1910, under the recommended analysis conditions, dicyclohexylamine and 2-ethylhexanoic acid did not elute after 75 min, probably indicating an irreversible interaction with the stationary phase, and possibly because the maximum allowable operative temperature (MAOT) of this stationary phase is 180 °C [8]. All other peaks eluted with good symmetry, with the exception of methyl dodecanoate (and to a lesser extent of methyl undecanoate). Their low tailing factor is due to peak-leading, probably related to a moderate overloading of long-chain esters on the investigated IL stationary phases, although their peak widths were in line with those of the other components.

Watercol™ performance with a suspected allergen standard mixture

A standard mixture of 29 compounds in the perfume field, 24 of them included in the list of EU-suspected allergens, was analyzed with both columns, to evaluate 1460 and 1910 performances vs. conventional and IL columns on analytes of routine interest in the field. Table 2 reports tailing factors and σ values, together with the area repeatability (n = 3) of each allergen, on the two columns compared to SLB-IL60i, while Fig. 2 shows the recovery of the analytes measured vs. that of OV 1701, taken as reference, and compared to that of commercial SLB-IL 60 and SLB-IL 60i. Figure S1 (see Electronic Supplementary Material, ESM) reports the GC-FID profiles of the allergen standard mixture analyzed on IL60i, 1460, and 1910 columns. As a preliminary consideration, the results indicate that efficiency and inertness of both columns are good, while component recovery is comparable to that of SLB-IL 60, indicating the adsorption of some components.
Table 2

Tailing factors and σ values together with area repeatability of each component of the investigated allergen standard mixture on 1460 and 1910 columns compared to SLB-IL60i

 

Ret. Time

Tailing factor

σ (min)

Repeatability (n = 3)

IL60i

W.C. 1460

W.C. 1910

IL60i

W.C. 1460

W.C. 1910

IL60i

W.C. 1460

W.C. 1910

W.C. 1460

W.C. 1910

1

Limonene

5.007

5.504

4.542—Coelution

1.071

1.06

Coelution

0.017

0.213

Coelution

1.25

0.89

2

Linalool

12.444

25.305

22.599

1.008

1.057

0.995

0.025

0.078

0.041

2.31

0.51

3

Estragole

18.473

27.839

25.68

0.999

1.03

0.926

0.028

0.056

0.040

4.59

1.45

4

Phenylacetaldehyde

19.64

31.764

Coelution 1

1.23

1.293

Coelution 1

0.040

0.060

Coelution 1

4.58

1.14

5

2-Octynoic acid, methyl ester

20.221

27.304

24.778

0.965

1.006

0.885

0.029

0.064

0.042

1.94

1.27

6

Citronellol

20.447

37.195

Coelution 2

1.071

0.997

Coelution 2

0.031

0.049

Coelution 2

4.36

1.46

7

Geraniol

22.31

40.303

Coelution 1

1.055

1.199

Coelution 1

0.032

0.046

Coelution 1

3.10

1.14

8

Benzyl alcohol

23.01

48.303

51.087

1.078

1.342

0.979

0.035

0.042

0.045

5.38

1.10

9

Neral

23.617

31.602

31.869

0.989

0.983

1.049

0.031

0.064

0.043

0.79

3.24

10

Geranial

25.053

33.689

34.005

0.997

1.047

1.003

0.031

0.057

0.041

0.89

1.66

11

α Iso-methyl-ionone

29.857

37.466

Coelution 2

0.97

1.018

Coelution 2

0.032

0.077

Coelution 2

2.19

1.46

12

Methyl eugenol

32.403

46.408

47.651

0.991

0.984

0.909

0.032

0.038

0.037

1.86

0.78

13

Hydroxy citronellal

32.55

51.205

51.953

0.995

1.033

0.85

0.033

0.040

0.036

1.96

1.58

14

α-Ionone

32.866

41.333

35.84

1.066

1.039

0.931

0.031

0.055

0.041

4.25

2.11

15

Eugenol

33.213

58.056

70.32

1.02

1.316

2.459

0.034

0.042

0.184

1.65

1.94

16

Lilial

33.972

45.967

43.398

1.004

0.973

1.132

0.032

0.048

0.098

4.92

1.49

17

Cinnamaldehyde

34.606

52.437

55.646

0.997

1.048

0.951

0.038

0.043

0.045

4.45

0.86

18

p-Anisyl alcohol

37.148

N.D.

74.003

1.013

N.D.

0.997

0.041

N.D.

0.064

N.D.

1.15

19a

Farnesol isomer a

36.814

56.923

50.506

0.999

1.31

0.955

0.031

0.043

0.035

5.31

0.69

19b

Farnesol isomer b

36.981

57.765

51.322

0.989

1.27

0.934

0.032

0.045

0.034

4.89

0.34

20

Cinnamyl alcohol

37.759

N.D.

71.474

1.021

N.D.

0.96

0.048

N.D.

0.062

N.D.

0.60

21

Amyl-cinnamaldehyde

40.922

56.463

50.838

1.282

0.894

0.797

0.051

0.040

0.041

1.89

0.97

22

α-Hexyl-cinnamaldehyde

43.728

59.533

52.741

0.973

0.961

0.81

0.033

0.038

0.041

2.34

1.45

23

α-Pentyl-cinnamyl alcohol

44.049

64.191

66.651

1.228

1.322

0.873

0.038

0.039

0.050

1.49

0.87

24

Vanillin

45.853

N.D.

N.D.

1.395

N.D.

N.D.

0.040

N.D.

N.D.

N.D.

N.D.

25a

Lyral isomer a

46.424

68.792

Coelution 3

0.879

1.148

Coelution 3

0.038

0.040

Coelution 3

4.37

1.13

25b

Lyral isomer b

46.817

69.2

Coelution 3

0.979

1.241

Coelution 3

0.036

0.041

Coelution 3

3.25

1.13

26

Coumarin

49.251

71.153

82.705

0.984

0.932

0.90

0.047

0.043

0.078

5.12

1.05

27

Benzyl benzoate

49.639

70.157

71.474

0.982

0.956

0.96

0.037

0.037

0.062

0.78

1.59

28

Benzyl salicylate

51.894

74.287

80.034

1.017

1.602

1.033

0.041

0.051

0.153

1.02

0.67

29

Benzyl cinnamate

61.776

86.691

85.85

0.998

1.041

2.637

0.044

0.038

0.289

1.13

3.27

N.D. not detected

Fig. 2

Recovery of suspected allergens, calculated from the normalized absolute area of each compound with each investigated IL vs. OV-1701 columns, taken as reference

Watercol™ 1460

With this column, three sample components (p-anisyl alcohol (18), cinnamyl alcohol (20), and vanillin (24)) were not detected, probably because of either irreversible adsorption and/or retention outside the time range of the analysis (120 min). Moderate peak distortion occurred for all substance with a free hydroxyl group in their structure (mainly alcohols and phenolic compounds) with tailing factors ranging between about 1.20 (eugenol (15)) and 1.34 (benzyl alcohol (8)). Only benzyl salycilate (28) showed a tailing factor of 1.6. Peak width was in general higher than that of the same components analyzed with IL60i, in particular for the early-eluting peaks, indicating that the columns have lower efficiency. A further factor affecting efficiency is the minimum operative temperature of 1460, which is slightly higher than the initial temperature adopted for this analysis (40 vs. 60 °C), as is clear from the σ value of limonene (1) (0.213). The same analysis carried out starting from 70 °C resulted in a correct peak shape for limonene (σ, 0.041; tailing factor, 1.043). The initial temperature of 40 °C was chosen so as to run all analyses under the same conditions, and to obtain comparable data. This consideration is also confirmed by the σ values of the late-eluting peaks, which are comparable and in some cases better than those of the corresponding peaks with IL60i. Area repeatability was very good, and RSD% ranged from 0.78 for benzyl benzoate (27) to 5.38 for benzyl alcohol (8).

Lastly, the recovery percentage vs. OV-1701 was measured using methyl-eugenol (5) as internal reference for normalization, its peaks with both 1460 and 1910 being of comparable intensity, and of similar symmetry and width (Fig. 2). Twelve components presented an adsorption above 35%; their normalized area was reduced to below 65% compared to their normalized area when analyzed with OV-1701, and the most strongly adsorbed were benzyl alcohol (8), whose recovery was 22.1%, and the farnesols (19a and 19b) at 8.7 and 17.6%, respectively. The 1460 column had inertness similar to that of IL 60, where several compounds were significantly adsorbed.

Watercol™ 1910

This column showed different selectivity from 1460 and had operative temperatures in the range of 40 to 180 °C. With 1910, only one compound was not detected (vanillin (24)) and six coeluted (see Table 2). Most peaks have satisfactory symmetry, falling in the range of 0.8–1.2, without any apparent relationship to their specific structural characteristic(s) or function(s). Only eugenol (15) and benzyl cinnamate (29) presented highly asymmetrical peaks, with tailing factors of 2.46 and 2.64 respectively. Peak widths (σ values) were in many cases comparable or only slightly higher than those obtained with IL60i. Eugenol (15), benzyl salycylate (28), and benzyl cinnamate (29) showed very broad peaks (see above) and had σ values of 0.184, 0.153, and 0.289, respectively, probably because the operative temperature was too low, due to the thermal limits of the stationary phase. Area repeatability was very good: RSD% ranged from 0.34 for farnesol (19b) to 3.27 for benzyl cinnamate (29).

As for 1460, the percent recovery vs. OV 1701 was determined using methyl-eugenol (5) as internal reference for normalization (Fig. 2). The 1910 column inertness with the allergen standard mixture was slightly better than that of IL60. Fourteen components presented relative adsorption above 35%, meaning that their recovery vs. OV 1701is less than 65%. The lowest value was that of eugenol (15), whose recovery was 30.2%, in spite of its good shape and peak width.

Quali-quantitative analysis of real-world aqueous samples

Direct analysis of essential oil aqueous solutions with Watercol™ columns

Many cosmetic preparations (lotions, tonics, perfumes, etc.) require essential oils to be solubilized in aqueous media. Essential oils are in general lipophilic, meaning that solubilization discriminates their components by polarity, which alters, among others, the relative ratios between hydrocarbons and oxygenated compounds in the final product, since the former are poorly soluble in water or ethanol/water solvents. In some cases, quality control of specific markers or biologically active components is thus mandatory. Water-compatible IL stationary phases emphasize the general behavior of all IL in GC, i.e., they have a peculiar selectivity that affords very good discrimination between light hydrocarbons and oxygenated compounds; with the columns investigated here, hydrocarbons are poorly retained, mainly eluting in the region of the solvent, and not being well separated from one another, unlike what occurs with oxygenated compounds. Watercol™ columns can therefore be very useful for the direct analysis of essential oil aqueous solutions. Figure 3 reports the GC patterns obtained with 1460 (a) and 1910 (b) of the oxygenated fractions of lavender and peppermint essential oils, and of total tea-tree essential oil, obtained after direct injection of their 1:1 ethanol/water solutions.
Fig. 3

1460 (a) and 1910 (b) GC patterns of the oxygenated fractions of lavender and peppermint essential oils, and that of total tea-tree essential oil

Watercol™ 1460

All oxygenated monoterpenoids of the three essential oils were very well separated. Sesquiterpene hydrocarbons also eluted in the same region of the chromatogram as the oxygenated compounds, as shown in the lavender (caryophyllene) and peppermint (caryophyllene, germacrene D) essential oil patterns (Fig. 3a). However, all oxygenated markers of both essential oils were base-line separated, affording correct characterization of the essential oil aqueous solutions; specifically, linalyl and lavandulyl acetates and linalool and lavandulol for lavender essential oil, and the four menthol isomers, the menthone isomers, menthyl acetate, pulegone, and menthofurane for peppermint essential oil. Tea-tree essential oil is characterized by 1,8-cineole, terpinen-4-ol, and α-terpineol. The latter two markers were very well separated, while 1,8-cineole could only partially be separated when applying an initial temperature of 40 °C, since it elutes in the cluster of peaks of monoterpene hydrocarbons and solvent(s). However, the quantitation of terpinen-4-ol and α-terpineol is of particular interest, since these compounds contribute to defining the origin of this essential oil, which is often the object of frauds, because of the higher quality and consequent higher cost of essential oils originating from Australia.

Watercol™ 1910

The 1910 column was slightly less effective, although all oxygenated monoterpenoids of lavender and peppermint essential oils were base-line separated (Fig. 3b). In the tea-tree essential oil, 1,8-cineol coelutes with the solvent(s), but terpinen-4-ol and α-terpineol were very well separated, making correct quality and origin control possible.

Direct identification of suspected allergens in commercial perfumes

Five commercial perfumes of different brands were purchased in a supermarket and analyzed directly as such, to detect and identify characterizing components and suspected allergens (see previous paragraph). Analyte identification was achieved by crossed comparison of the results obtained with the two Watercol™ columns investigated (i) through their linear retention indices (ITS) calculated vs. a standard mixture of C9-C30 hydrocarbons by GC-FID analysis, in combination with (ii) their mass spectra obtained by GC-MS after HS-SPME sampling.

Figures 4 and 5 report the GC-FID patterns of the five commercial perfumes and of the reference allergen standard mixture, together with the components identified in them, with 1460 and 1910 columns, respectively.
Fig. 4

GC-FID profiles of commercially-available perfumes and allergen standard mixture with Watercol 1460 column. Peak identification: 1: limonene, 2: linalool, 3: estragole, 4: phenylacetaldehyde, 5: methyl 2-octynoate, 6: citronellol, 7: geraniol, 8: benzyl alcohol, 9: neral, 10: geranial, 11: α-isomethyl ionone, 12: methyl eugenol, 13: hydroxycitronellal, 14: α-ionone, 15: eugenol, 16: lilial, 17: cinnamaldehyde, 18: anisyl alcohol, 19: farnesol isomers, 20: cinnamyl alcohol, 21: amyl cinnamaldehyde, 22: hexyl cinnamaldehyde, 23: α-pentylcinnamyl alcohol, 24: vanillin, 25: lyral isomers, 26: coumarin, 27: benzyl benzoate, 28: benzyl salicylate, 29: benzyl cinnamate, 30: linalyl acetate

Fig. 5

GC-FID profiles of commercially available perfumes and allergen standard mixture with Watercol 1910 column. Peak identification: see Fig. 4

Direct quantitation of suspected allergens and markers in commercial perfumes

The same perfumes were also submitted to true quantitation by external calibration with the pure standards, to quantify one characteristic component (linalyl acetate (30)) and two suspected allergens (linalool (2) and α-ionone (14)) taken as markers. These compounds were chosen because they were present in most of the perfumes investigated in different amounts. These analyses were also used to evaluate analytical performance of Watercol™ in terms of analyte linearity (R2), area repeatability, and intermediate precision, again to assess consistency of column behavior over time. The R2 values show that the linearity of both columns in the concentration range investigated was very good (i.e., always above 0.997 for 1460 and above 0.991 for 1910). Area repeatability was also very good with both IL columns, the maximum RSD% being 5.6% for linalyl acetate (30) with 1460 and 5.3% for linalool (2) with 1910. Intermediate precision for both columns was slightly lower (maximum RSD%, 8.4% for linalyl acetate (30) on 1460 and 9.3% for linalool on 1910). LOD and LOQ with both columns were in agreement with the leave-on suspected allergens limits, the highest LOD being 3 ppm (linalool (2) and α-ionone (14)) for 1460 and 5 ppm (linalyl acetate (30)) for 1910, while LOQ was 7 ppm (linalool(2)) for 1460 and 9 ppm (linalyl acetate (30)) for 1910. Table 3 reports the figures of merit of the method applied. With 1910 in the analysis of perfume 4, linalyl acetate (30) coeluted with another component, thus altering the quantitative results (Fig. 5 and Table 4).
Table 3

Figures of merits of the quantitative method applied to Linalool (2), Linalyl acetate (30), and α-Ionone (14)

Col.

Analyte

Linear range investigated (mg L−1)

Slope ± error

Linearity (R2)

LOD (mg L−1)

LOQ (mg L−1)

Repeatability % RSD 50 mg L−1 (n = 5)

Intermediate precision % RSD 50 mg L−1 (n = 4)

1460

Linalool (2)

50–5000

172.73 ± 6.93

0.997

3

7

5.2

7.9

Linalyl acetate (30)

50–5000

146.29 ± 5.54

0.997

2

5

5.6

8.4

α-Ionone (14)

50–2000

146.98 ± 4.30

0.999

3

6

3.4

6.1

1910

Linalool (2)

50–5000

183.41 ± 7.01

0.994

4

7

5.3

9.3

Linalyl acetate (30)

50–5000

161.15 ± 4.87

0.991

3

9

4.9

8.7

α-Ionone (14)

50–2000

137.58 ± 5.26

0.998

5

8

4.3

7.6

Table 4

Concentration of Linalool (2), Linalyl acetate (30), and α-Ionone (14) in five commercial perfumes after analysis on Watercol 1460, Watercol 1910, and OV 1701 columns

 

Analyte

Absolute amount (mg L−1)

Mean (SD)

Column 1460

Column 1910

Official method [12, 13]

Perf. 1

Linalool (2)

4314 (40)

4745 (15)

4413 (22)

Linalyl acetate (30)

3772 (28)

3971 (32)

3312 (35)

α-Ionone (14)

912 (6)

869 (5)

841 (8)

Perf. 2

Linalool (2)

419 (3)

450 (0.04)

439 (3)

Linalyl acetate (30)

751 (7)

767 (0.09)

738 (6)

α-Ionone (14)

125 (1)

94 (0.01)

103 (2)

Perf. 3

Linalool (2)

1947 (12)

1989 (23)

2012 (19)

Linalyl acetate (30)

N.D.

N.D.

N.D.

α-Ionone (14)

404 (2)

Coelution

380 (4)

Perf. 4

Linalool (2)

2798 (16)

Coelution

2846 (13)

Linalyl acetate (30)

4653 (39)

4782 (41)

4568 (38)

α-Ionone (14)

N.D.

N.D.

N.D.

Perf. 5

Linalool (2)

3893 (37)

4124 (41)

4073 (36)

Linalyl acetate (30)

4667 (51)

4918 (49)

4505 (52)

α-Ionone (14)

N.D.

N.D.

N.D.

N.D. not detected

The results of these analyses were compared and confirmed by quantifying the same analytes in the same perfumes with the official EU method [12, 13]. The results with the two methods were fully comparable, showing that both 1460 and 1910 can successfully be used to quantify target analytes with Watercol™ columns directly, without dilution (Table 4).

Conclusions

The results show that water-compatible gas chromatographic stationary phases based on ILs open new perspectives in the analytical and bioanalytical field, thanks to the fact that they can be applied routinely to the direct analysis of samples with water as main solvent. The peculiar selectivity of water-compatible IL columns can be applied to aqueous or hydroalcoholic samples, making them very promising, in particular for the fragrance and essential oil fields. The use of water-compatible IL columns in routine analysis in quality control laboratories means that time-consuming sampling procedures to transfer fractions or analytes of interest to solvents compatible with conventional columns can be avoided or simplified.

However, the results with samples where water is the main solvents show that these IL columns still present a slightly lower efficiency and higher activity than inert IL columns, meaning that significant efforts must still be made to achieve their full application in quality control. Further progresses are expected in column manufacturing, to improve their range of operative temperatures (where possible), their inertness and reduce their activity, exactly as it has occurred for the first generation of conventional IL columns, which recently resulted in the introduction of the inert series [11]. Moreover, in routine analysis, it is very often necessary that gas chromatography adopts mass spectrometry as detector; further experiments are under way to evaluate the effects of the introduction of aqueous samples on MS performance.

Notes

Acknowledgements

The authors are indebted to Supelco (Bellefonte, PA, USA) for providing the ionic liquid columns, to Dr. Len Sidisky (Supelco, Bellefonte, PA, USA) for helpful discussion and advice, and to Robertet SA (Grasse, France) for financial support to the laboratory.

Compliance with ethical standards

Conflict of interest

The authors declared that they have no conflict of interest.

Supplementary material

216_2018_922_MOESM1_ESM.pdf (355 kb)
ESM 1 (PDF 354 kb)

References

  1. 1.
    Belhassen E, Bressanello D, Merle P, Raynaud E, Bicchi C, Chaintreau A, et al. Routine quantification of 54 allergens in fragrances using comprehensive two-dimensional gas chromatography-quadrupole mass spectrometry with dual parallel secondary columns. Part I: method development. Flavour Frag J. 2018;33(1):63–74.CrossRefGoogle Scholar
  2. 2.
    DIRECTIVE 2003/15/EC of the European Parliament and of the Council of 27 February 2003.Google Scholar
  3. 3.
    Jayawardhana DA, Woods RM, Zhang Y, Wang CL, Armstrong DW. Rapid, efficient quantification of water in solvents and solvents in water using an ionic liquid-based GC column. LC GC N Am. 2012;30(2):142-+.Google Scholar
  4. 4.
    Ho TD, Zhang C, Hantao LW, Anderson JL. Ionic liquids in analytical chemistry: fundamentals, advances, and perspectives. Anal Chem. 2014;86(1):262–85.CrossRefPubMedGoogle Scholar
  5. 5.
    Frink LA, Weatherly CA, Armstrong DW. Water determination in active pharmaceutical ingredients using ionic liquid headspace gas chromatography and two different detection protocols. J Pharmaceut Biomed. 2014;94:111–7.CrossRefGoogle Scholar
  6. 6.
    Frink LA, Armstrong DW. Water determination in solid pharmaceutical products utilizing ionic liquids and headspace gas chromatography. J Pharm Sci-Us. 2016;105(8):2288–92.CrossRefGoogle Scholar
  7. 7.
    Frink LA, Armstrong DW. The utilisation of two detectors for the determination of water in honey using headspace gas chromatography. Food Chem. 2016;205:23–7.CrossRefPubMedGoogle Scholar
  8. 8.
  9. 9.
    Grob K, Grob G, Grob KJ. Testing capillary gas chromatographic columns. J Chromatogr. 1981;219:13–20.CrossRefGoogle Scholar
  10. 10.
    European Pharmacopoeia Online 9.2, http://online.edqm.eu/EN/entry.htm. 9th ed.
  11. 11.
    Cagliero C, Bicchi C, Cordero C, Liberto E, Rubiolo P, Sgorbini B. Analysis of essential oils and fragrances with a new generation of highly inert gas chromatographic columns coated with ionic liquids. J Chromatogr A. 2017;1495:64–75.CrossRefPubMedGoogle Scholar
  12. 12.
    IFRA (International Fragrance Association). GC/MS quantitation of potential fragrance allergens in fragrance compounds. 2007.Google Scholar
  13. 13.
    Cosmetics Europe. Technical guidance for the determination of fragrance material in cosmetic products. 2006.Google Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Cecilia Cagliero
    • 1
  • Carlo Bicchi
    • 1
  • Chiara Cordero
    • 1
  • Erica Liberto
    • 1
  • Patrizia Rubiolo
    • 1
  • Barbara Sgorbini
    • 1
  1. 1.Dipartimento di Scienza e Tecnologia del FarmacoUniversità degli Studi di TorinoTorinoItaly

Personalised recommendations