تجزیه تحلیل کمی اسانس ها در عطر با استفاده از وضوح منحنی چند متغیره همراه با کروماتوگرافی گازی دو بعدی جامع Quantitative analysis of essential oils in perfume using multivariate curve resolution combined with comprehensive two-dimensional gas chromatography

۱٫ Introduction Quantification has always been a field of intensive study in analytical chemistry. The conventional approach is the univariate calibration model for a single analyte. Essentially, it consists on the correlation of the instrumental responses with the concentrations of the target analyte. In gas chromatography (GC) the instrumental response is the peak area, which can be estimated through conventional integration [1]. However, obtaining quantitative results can be problematic when the target is not a single analyte, but a complex mixture, such as the case when evaluating the presence of essential oils in perfumes [2]. The first step in conventional approaches to perform this task is to identify specific chemical markers present only in the targeted essential oil or essence. Temperature-programmed retention indices ofthe unknown peaks determined with polar and non-polar columns, combined with electron-impact ionization mass spectra, are required to correctly identify each peak [2]. For essential oils originating from several countries, characteristic components of the essential oil and their common biosynthetic precursors that must be met to define essential oil quality can be chosen as markers [3]. Once identified, some markers, or their ratios required to characterize an essential oil, are quantified in the complex mixture, i.e., commercial perfumes and the result is then used to estimate the amount of the essential oil in the complex mixture. The reference values for each marker or their ratios include the average, minimum and maximum, taking into account seasonal or climatic variations in essential oil composition [3]. This classical procedure can be misleading if co-elution is present or if the amount of the chemical marker is under the limit of quantification. For example, when co-elution is present, quantification of the chemical markers will be hampered as the resulting peak integration may be erroneous. Therefore, techniques which improve separation capacity can be of special relevance to these problems. Introduced in 1991 by Phillips and coworker [4], comprehensive two-dimensional gas chromatography (GC × GC) has become the benchmark technique for unraveling complex samples. The GC × GC system consists of conventional gas chromatograph fitted with two capillary columns connected in series, such that all sample portions emerging from the first column enter the second and are analyzed sequentially. The key interface that allows the injection of small and narrow fractions from the first to the second column is the modulator. A GC × GC analysis provides higher sensitivity, detectability and separation power [5]. Because GC × GC provides a true orthogonal separation system (when the retention mechanisms from the first and second dimension are independent), it is possible to observe well ordered distributions of chemically similar compounds in the retention plane [6]. In this way, numerous small ingredients presentin some essential oils can be separated, because every peak is submitted to two different mechanisms of separation and then detected, due to the higher sensitivity [7–۹]. Even without positive identification of the compounds, as in a GC × GC-FID analysis, this technique can be very useful for the analysis of essential oils, for example to reveal some regional or seasonal variations and to detect adulterations that would be unnoticed by GC analysis [2]. Even though GC × GC may provide the peak capacity and the sensitivity needed, the use of the conventional approach (use of marked compounds) to quantify a complex sample as an essential oil in a more complex mixture such as perfume sample is still a tedious and time-consuming task. The quantification of the individual chemical marker is, usually, obtained by conventional integration [10], but in this approach the chromatographic signal of the marker has to be well resolved. de Godoy et al. [11] proposed an alternative for quantification of targeted-compounds by using an interval multi-way partial least squares calibration whereby coelution did not affect the results. Furthermore, Zeng et al. [12] proposed an alternative moving window factor analysis and twostep iterative constraint method to extract the pure profile (either mass or absorbance spectra) in order to quantify targeted-analytes, in cases where co-elution is present. When compared to conventional gas chromatography with mass spectrometric detection (GC–MS) the amount of information obtained from a GC × GCFID chromatogram is considerably larger. Thus, instead of using a chemical marker to quantify a complex mixture in a complex sample, the whole two-dimensional chromatogram can be used in these analyses. As the intrinsic information obtained from GC × GC chromatograms is considerably larger and more complex, their manual(or conventional)interpretation can be problematic or even impossible. Consequently, the use of a chemometric approach is recommended, because it provides a reliable and non-subjective result[13]. Pedroso et al. applied multivariate calibration strategies to identify of gasoline adulteration, using GC × GC-FID data [14]. An important multivariate technique that has not been widely used with GC × GC-FID data is the algorithm proposed by Tauler et al. in 1995 called multivariate curve resolution (MCR) [15]. This method has been employed in analysis of complex mixtures through different analytical techniques [16–۲۰], such as high performance liquid chromatography coupled to diode array detection [21]. The theory behind the MCR algorithm has been discussed in previously papers [22–۲۵]. The most important feature of the MCR is the second order advantage, in which the calibration step can be built withfew samples insteadof a large calibrationset and,furthermore,itispossible toquantify the compounds ofinterest eveninthe presence of interferences not present in the calibration sample set. In the MCR algorithm the data setis decomposed into two matrixes, one related to concentration profiles and another related to instrumental profiles. These two matrixes are iteratively adjusted to the data set through an alternating least squares (ALS) procedure, which starts with an initial estimate of pure analyte instrumental profiles. During the ALS optimization, several constraints, such as non-negativity, unimodality, closure and selectivity, can be applied to obtain chemically meaningful solutions. Fig. 1 exemplifies how the MCR algorithm can be used with GC × GC-FID data in the case of a two component mixture. The first step is the unfolding of the GC × GC-FID chromatograms from a matrix to a vector. Next, the vectors of all samples are placed in the lines of the bidimensional data matrix D, which is decomposed into the matrix of concentration profiles C and the matrix with the chromatograms of each pure component S. Finally, after ALS optimization, the vectors obtained in matrix S are reshaped into two-dimensional GC × GC-FID chromatograms and the calibration curve can be built using the data contained in with matrix C. Thus, the combination of GC × GCFID and MCR-ALS provides the analyst the instrumental profiles of each pure compound besides the quantitative information. Consequently,this may be the best combination to unravelthese complex issues regarding one of the ultimate goals in analytical chemistry. In this paper, we proposed the use of the MCR-ALS method to analyze data obtained by GC × GC-FID. To evaluate the feasibility of this method, quantification ofthe essential oil of rosemary was performed in samples containing interferences (pineapple essence or a commercial perfume) not presentin the calibration set, which evaluates the secondorder advantage ofthe algorithm.Additionally,the amount of essential oil oflemon grass was quantified in commercial perfumes to evaluate the accuracy of the proposed method.