On-Line Combination of High Performance Liquid Chromatography

Dec 23, 2014 - Comparative Analysis of Typical Low Rank Coal Pyrolysis Technology Based on a Nonlinear Programming Model ... The Odor of Death: An Ove...
21 downloads 10 Views 5MB Size
Download PDF
Article pubs.acs.org/ac

On-Line Combination of High Performance Liquid Chromatography with Comprehensive Two-Dimensional Gas Chromatography-Triple Quadrupole Mass Spectrometry: A Proof of Principle Study Mariosimone Zoccali,† Peter Quinto Tranchida,† and Luigi Mondello*,†,‡,§ †

Dipartimento di Scienze del Farmaco e Prodotti per la Salute, University of Messina, Viale Annunziata, 98168 Messina, Italy Centro Integrato di Ricerca, University Campus Bio-Medico of Rome, Via Á lvaro del Portillo, 21, 00128 Rome, Italy § Chromaleont s.r.l. A start-up of the University of Messina, c/o Dipartimento di Scienze del Farmaco e Prodotti per la Salute, University of Messina, Viale Annunziata, 98168 Messina, Italy ‡

S Supporting Information *

ABSTRACT: The present contribution is focused on the online combination of high performance liquid chromatography (HPLC), cryogenically modulated comprehensive two-dimensional gas chromatography (GC × GC), and triple quadrupole mass spectrometry (QqQ MS), generating a very powerful unified separation-science tool. The instrument can be used in seven different combinations ranging from one-dimensional HPLC with a photodiode array detector to on-line LC × GC × GC/QqQ MS. The main focus of the present research is directed to the LC-GC × GC/QqQ MS configuration, with its analytical potential shown in a proof-of-principle study involving a very complex sample, namely, coal tar. Specifically, a normal-phase LC process enabled the separation of three classes of coal tar compounds: (1) nonaromatic hydrocarbons; (2) unsaturated compounds (with and without S); (3) oxygenated constituents. The HPLC fractions were transferred to the GC × GC instrument via a syringe-based interface mounted on an autosampler. Each fraction was subjected to a specific programmed temperature vaporizer GC × GC/QqQ MS untargeted or targeted analysis. For example, the coal tar S-containing compounds were pinpointed through multiple-reactionmonitoring analysis, while full-scan information was attained for the oxygenated constituents.

T

of the GC separation step is less important compared to that of sample preparation and mass spectrometry. In targeted GC/MS experiments, the use of quadMS, LR ToF MS, high-resolution (HR) ToF MS, and triple quadrupoles (QqQ MS) is popular.2−5 The combination of GC × GC with rapid-scanning quadMS, LR ToF MS, HR ToF MS, or QqQ MS generates very powerful analytical tools that can be employed in both untargeted and targeted applications.1 As such, GC × GC/MS can be considered as a unified analytical tool, inasmuch as it can potentially meet several sample-specific requirements. For example, on one hand, GC × GC/LR ToF MS can be used for the untargeted qualitative analysis of 100s (if not 1000s) of food volatiles,6 while on the other it can be exploited for the quantification of trace-amount phytosanitary compounds.7 As aforementioned, sample complexity can also be related to the heterogeneous nature of real-world samples. In many instances, it is convenient to isolate more homogeneous fractions prior to a GC analysis. It is obvious that prefractionation will generate simplified subsamples, also reducing problems related to matrix

he detailed characterization, or untargeted analysis, of very complex mixtures of organic volatiles (and semivolatiles) represents one of the main challenges in the field of analytical chemistry. Sample complexity, as such, is related not only to a potentially high number of constituents (e.g., ≥1000) but also to the range of different chemical classes (from apolar to highly polar) and concentration levels (from trace to high concentration levels). Consequently, the satisfactory profiling of a very complex mixture would require: (1) enhanced separation power; (2) high selectivity; (3) increased sensitivity; (4) wide dynamic range. High-resolution gas chromatography (GC) methods, such as comprehensive 2D GC (GC × GC), are advisible for the untargeted analysis of a complex sample. In terms of mass spectrometry (MS), unit-mass devices such as single quadrupole (quadMS) or low-resolution time-of-flight (LR ToF MS) are sufficient for the scopes of an untargeted experiment.1 In a targeted experiment, the analyst has an a priori knowledge of the name and number of solutes requiring investigation. For example, in the field of pesticide analysis, the development of GC/MS methods for the qualitative and quantitative determination of hundreds of pesticides (multiresidue methods) is rather common. In such analytical work, the role © 2014 American Chemical Society

Received: November 7, 2014 Accepted: December 23, 2014 Published: December 23, 2014 1911

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918

Analytical Chemistry

■

interferences. Additionally, the matrix contains considerable amounts of nonvolatile material, that should be separated from the analytes of interest before GC injection. Both prefractionation and the treatment of nonvolatile material can be performed by using different sample preparation processes. High performance liquid chromatography (HPLC), coupled with high resolution GC, in an on-line mode (LC-GC), is a very powerful analytical technique because of its enhanced selectivity and sensitivity. On-line LC-GC methods are particularly suited to the separation of compounds with similar physicochemical features, in samples characterized by several chemical classes (e.g., essential oils);8 such LC-GC approaches are usually employed for untargeted analyses. LC-GC techniques have also been widely used to eliminate nonvolatile material (e.g., lipids) before the second-dimension separation, normally in targeted experiments.9 The on-line combination of HPLC with GC × GC has been rarely reported; for example, Adam et al. and Edam et al. used LC with GC × GC for the analysis of different classes of compounds (saturates, monoaromatics, diaromatics) in diesel and in off- and on-line modes, respectively.10,11 In an off-line study, Biedermann and Grob analyzed the mineral oil aromatic hydrocarbon contamination in vegetable oil by performing an LC preseparation process prior to GC × GC separation (LC// GC × GC).12 It is noteworthy that all such studies were focused on petrochemicals,10−12 viz., very complex mixtures. Samples of apparent limited complexity, specifically orange and bergamot essential oils, were subjected to an off-line HPLC preseparation by Tranchida et al.13 The hydrocarbon and oxygenated fractions of both oils were then subjected to GC × GC/MS analysis, with an unprecedented number of analytes identified. Recently, on-line SFC-GC × GC experiments have started to appear;14,15 for example, Potgieter et al. used SFC-GC × GC/ LR ToF MS for the analysis of vacuum gas oil (VGO).14 The authors performed an SFC preseparation of VGO in saturated, unsaturated, and polar groups of compounds on a combination of three packed columns with a different selectivity. The same sample type and chemical-class preseparation was performed by Dutriez et al., in this case using SFC-GC × GC with a flame ionization detector;15 even though the interfaces used were characterized by a different design, both were based on the use of switching valves, with a direct link to the GC injector via a restrictor. To the best of the present authors’ knowledge, no previous description of on-line LC-GC × GC combined with QqQ MS has been reported. Such a method is a five-dimensional one, inasmuch as three chromatographic and two MS dimensions are combined. A proof-of-principle study was performed on a sample of coal tar: the first dimension was exploited to separate the coal tar in three fractions, namely: (I) nonaromatic hydrocarbons, (II) unsaturated compounds (with and without S), and (III) oxygenated compounds. The full-scan mode was used for the first and third fraction, while the second fraction was analyzed by using the full-scan and multiple-reactionmonitoring (MRM) modes. The MRM method was directed to S-containing aromatic compounds, the latter being targeted because the combustion process of such constituents can cause a series of negative effects to the environment and human health.16

Article

EXPERIMENTAL SECTION

Samples and Sample Preparation. The coal tar sample was attained from the Cambui ́ coal field, Figueira, Paraná (Brazil). Prior to LC analyses, the sample was diluted 1:200 (v/ v) in CH2Cl2. Standards of a C7−30 alkane mixture, perylene, 2propylthiophene, benzothiophene, 2-methylbenzothiophene, dibenzothiophene, 4,6-dimethyldibenzothiophene, and 4,6diethyldibenzothiophene were purchased from Sigma−Aldrich/Supelco (Bellefonte, PA, USA). A standard solution containing the S compounds was prepared at the 100 ppb (w/ v) level in CH2Cl2. LC-GC × GC/QqQ MS Analyses. All applications were carried out on an LC-GC × GC/QqQ MS system consisting of (1) a Shimadzu LC system (Kyoto, Japan), equipped with a CBM-20A communication bus module, two LC-30AD dualplunger parallel-flow pumps, a DGU-20A on-line degasser, an SPD-M20A photodiode array detector (PDA), a CTO-20A column oven, and an SIL-30AC autosampler. Data were acquired by the LabSolution v.5.71 software (Shimadzu). LC conditions were as follows: a 100 × 3 mm ID × 5 μm dp silica column (SUPELCOSIL LC-Si, Sigma-Aldrich/Supelco) was operated under the following gradient conditions (flow: 0.35 mL/min): 0−6.5 min (100% hexane), from 6.5 to 7.5 min 100% CH2Cl2 (until the end of the analysis). Injection volume was 2 μL. LC fractions were as follows: (a) nonaromatic hydrocarbons were transferred from 1.35 to 1.85 min (175 μL); (b) aromatic compounds (with and without S) were transferred from 1.85 to 7.00 min (1802 μL); (c) oxygenated compounds were transferred from 8.95 to 14.10 min (1802 μL). The three fractions were analyzed under slightly different operational conditions; the details in common are reported in this section, while specific application information is reported in the Results and Discussion section. (2) An AOC-5000 auto injector, equipped with a dedicated dual side-port syringe, employed as transfer device. Chromatography band transfer was achieved, in the stop-flow mode, through a modified 25 μL syringe. The LC−GC transfer device has been described previously.17 Briefly, the lower part of the syringe is connected via two transfer lines, to the LC detector exit and to waste. A Teflon plug is located at the end of the syringe plunger; the latter is characterized by a lower outer diameter with respect to the barrel ID, thus enabling mobile phase flow inside the syringe. In the waste mode, the syringe plug is located below both lines and the effluent is directed to waste. In the transfer position, the syringe plug is located between the lines and the effluent flows to the GC. (3) A GC × GC/MS system consisting of two GC2010 gas chromatographs (GC1 and GC2) and a TQ8030 triple quadrupole mass spectrometer (Shimadzu); the first GC oven was equipped with an Optic 4 injector (GL Sciences, Eindhoven, The Netherlands). The Optic 4 injector [PTV (programmed-temperature vaporizor)] was temperature-programmed as follows: from 35 °C (100 s) to 300 °C at 5 °C/s. The primary column, an SLB-5ms [(silphenylene polymer, practically equivalent in polarity to poly(5% diphenyl/95% methylsiloxane)] 30 m × 0.25 mm ID × 0.25 μm df (SigmaAldrich/Supelco), was connected to an uncoated capillary segment (1.5 m × 0.18 mm ID, used to create a double-loop), by using an SGE SilTite mini-union (Ringwood, Victoria, Australia). The uncoated capillary was then connected to a segment of Supelcowax-10 (100% polyethylene glycol) 1.0 m × 0.10 mm ID × 0.10 μm df column (Sigma-Aldrich/Supelco), by using another union (SGE). Modulation was carried out every 1912

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918

Article

Analytical Chemistry

Table 1. Chemical Classes and Related MRM Transitions (Trans.), with Collision Energies (CE) Reported in Parentheses and Calculated LoD and LoQ Values, along with Intra- and Interday RSD Values (Intraday Values Are Reported First) for Peak Areas (at the 100 ppb Level) and Ion Ratiosa class C2-T C3-T C4-T BT C1-BT C2-BT C3-BT C4-BT C5-BT DBT C1-DBT C2-DBT C3-DBT C4-DBT BNT C1-BNT C2-BNT

Trans. 1 m/z (CE) 111 126 140 134 147 162 176 190 204 184 198 212 226 240 234 248 262

> > > > > > > > > > > > > > > > >

77 (20) 77 (30) 125 (20) 90 (20) 45 (30) 147 (30) 161 (30) 175 (10) 175 (20) 152 (20) 166 (20) 197 (20) 211 (20) 225 (20) 202 (30) 215 (30) 234 (30)

Trans. 2 m/z (CE) 111 111 125 134 147 162 176 175 175 183 197 212 211 225 234 248 262

> > > > > > > > > > > > > > > > >

67 (20) 67 (20) 91 (20) 108 (10) 77 (30) 128 (30) 128 (30) 115 (30) 128 (30) 139 (10) 171 (20) 178 (30) 178 (20) 209 (20) 189 (40) 221 (30) 247 (40)

Trans. 3 m/z (CE)

LoD (μg/L)

LoQ (μg/L)

RSD 100 ppb

RSD ion ratio

3

11

11/11

2/7

14 23

47 75

5/6 9/11

7/6 3/2

4

12

5/8

3/3

2

6

3/8

1/1

1

2

2/11

1/1

147 > 77 (30)

234 > 208 (30) 248 > 202 (30)

a Chemical classes for which a pure standard compound was available are in bold. Trans. 1 is the quantifier. Chemical class abbreviations: T = thiophene; BT = benzothiophene; DBT = dibenzothiophene; BNT = benzonapthothiophene. The abbreviations C1, C2, etc., refer to the number of carbons in the alkyl substituents.

5 s, by using a loop-type modulator (under license from Zoex Corporation, Houston, TX, USA). The duration of the hot pulse (300 °C) was 300 ms. The GC1 oven temperature program was as follows: 50 to 280 °C at 3 °C/min (total GC × GC analysis time for all three fractions: 76.7 min). Carrier gas, helium, was supplied at an initial pressure of 175.8 kPa (constant linear velocity). For the GC2 oven temperature programs, see the Results and Discussion. The QqQ MS ionization mode was electron ionization (70 eV). Interface and ion source temperatures were 280 and 250 °C. Collision gas and pressure were Ar and 200 kPa. Full-scan acquisition mode had a mass range of m/z 45− 360; spectral production frequency was 33 Hz. For the MRM acquisition mode, transitions and collision energies are reported in Table 1; loop time was 0.02 s, corresponding to a 50 Hz spectral production frequency. A dedicated “5D Solution” software (Shimadzu) enabled the control of each instrument through each native software, as well as the transfer process. Data were collected by the GCMS Solution v.4.20 software (Shimadzu); the ChromSquare v.2.0 software was used for GC × GC data processing (Shimadzu). MS spectral matching was performed automatically by using the ChromSquare v.2.0 software and the NIST11 database.

■

Figure 1. LC-GC × GC/QqQ MS chromatogram of the first fraction of coal tar (the QqQ MS system was used in the full scan mode).

equipped with such a device. The system can be used in any of the seven configurations simply through selection of the appropriate software settings. In the present research, the system was employed in the LCGC × GC/QqQ MS and GC × GC/QqQ MS configurations. In the former configuration, it is possible to set different LC, PTV, GC × GC, and QqQ MS conditions during the same run. The sample subjected to analysis was of extreme complexity, namely, coal tar. The latter is obtained as a high-viscosity liquid byproduct from the pyrolysis of coal, with coke being the main product. Coal tar is a sample type of high complexity, with polycyclic aromatic hydrocarbons being the most abundant constituents. Linear, branched, and cyclic hydrocarbons are also present, along with heterocyclics containing S, O, and N.18 The use of an LC prefractionation step, in such an application type, is also useful to avoid the introduction of high MW material into the GC system. The coal tar sample was LC-separated in

RESULTS AND DISCUSSION

The LC-GC × GC/QqQ MS system used is the result of the combination of three (commercially available) independent instruments: LC-GC, cryogenically modulated GC × GC (with a single GC oven), and GC/QqQ MS (only the MS was exploited). It is a unified instrument inasmuch as it can perform, if necessary, one-dimensional, as well as heart-cutting (indicated with a hyphen between the chromatography methods) and comprehensive (indicated with a multiplication symbol) two-dimensional chromatography analyses: LC/PDA, LC-GC/QqQ MS, LC × GC/QqQ MS, LC-GC × GC/QqQ MS, LC × GC × GC/QqQ MS, GC × GC/QqQ MS, and GC/ QqQ MS. If flame ionization detection is required, then GC2 is 1913

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918

Article

Analytical Chemistry

Figure 2. LC-GC × GC/QqQ MS chromatogram of the third fraction of coal tar (the QqQ MS system was used in the full scan mode).

for the remaining analysis time. With regards to the GC × GC temperature program, no GC2 offset was used. As expected, the saturated hydrocarbons were aligned along a diagonal in the full-scan GC × GC/QqQ MS chromatogram (Figure 1). Even though the 2D separation in itself is of limited interest, a few considerations can be made. First, the selectivity of the LC step was certainly satisfactory, because no other coal tar constituents were present in the 2D chromatogram (apart from a few interferences, probably derived from the GC system); second, discrimination occurred for the more volatile hydrocarbons. The linear, branched, and cyclic compounds ranged from C11 to C28. PTV discrimination was evaluated by LC-GC × GC/QqQ MS injection of a C7−30 alkane mixture. The alkanes up to C9 were entirely lost, while discrimination was greatly reduced from C13 onward. The degree of discrimination was considered to be acceptable for the analytical scopes of the research, inasmuch as it was minimized to the extent that hydrocarbons from n-alkane C11 onward (in this case, up to nalkane C28) were detected in a satisfactory manner. The third fraction, containing the oxygenated compounds, was also analyzed in the full-scan mode (MS solvent cut time: 11.0 min). The PTV was operated with a split ratio of 250:1 during the LC transfer period (8.95−14.10 min), plus 1 min; then, the splitless mode was applied for 1.5 min, followed by a split ratio of 150:1 for the remaining analysis time. With regard to the GC × GC temperature program, a positive GC2 offset of +60 °C was applied, to reduce the extensive band broadening of the O-containing compounds in the second dimension. As can be observed in Figure 2, the chromatogram is characterized by very-high complexity and by the presence of group-type patterns (i.e., alkylated phenols). As expected, the prefractio-

three fractions. Optimization of the LC heart-cutting process was performed as follows: the elution window of the first fraction was established by transferring narrow (15 s wide) and sequential LC fractions of a C7−30 alkane mixture (from the dead volume onward) to the GC; the elution window of the second fraction (1.85 to 7.00 min) was established by considering the end of the first one, up to the elution of a five-ring aromatic pure standard compound, namely, perylene. The third fraction eluted between 8.95 and 14.10 min and was defined empirically through the GC × GC analysis of 15 s LC fractions of coal tar starting from 7.00 min. Different PTV-GC × GC/QqQ MS conditions were applied to each fraction, during the same run. During each PTV-GC × GC/QqQ MS analysis, the HPLC system entered the stop-flow mode. It is noteworthy that the occurrence of potential band broadening during the stop-flow mode could be observed through continuous PDA monitoring. In this respect, the system was found to be very stable (band broadening was insignificant), as were also the HPLC elution times, which were characterized by negligible variations between sequential runs. It is worthy of note that, in a hypothetical LC × GC × GC application, the number of fractions transferred from the LC to the GC dimension would be much higher, greatly extending the overall analysis time. Nonaromatic Hydrocarbons and Oxygenated Compounds (First and Third Fraction). The first fraction containing the nonaromatic hydrocarbons was analyzed only in the full-scan mode (MS solvent cut time: 10.5 min). The PTV was operated with a split ratio of 200:1 during the LC transfer period (1.35−1.85 min), plus 1 min; then, the splitless mode was applied for 1.5 min, followed by a split ratio of 150:1 1914

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918

Article

Analytical Chemistry

Figure 3. LC-GC × GC/QqQ MS chromatogram of the second fraction of coal tar (the QqQ MS system was used in the full scan mode). Abbreviations: B, benzene; N, naphthalene; O, olefine; I, indene; BPh, biphenylene; F, fluorene; P, phenanthrene; Py, pyrene.

alkyl substituent(s). Identification of the chemical classes illustrated was performed through MS database matching and on the basis of the chromatogram locations. As for the oxygenated compounds, the identification of chemical class constituents is not reported, because in such cases the presence of a pure standard would be necessary. Again, spectral similarities for single compounds, as well as spectral similarity ranges and average values for the indicated chemical classes (benzene, naphthalene, indene, fluorene, phenanthrene), are reported as Supporting Information (Table S1). As can be observed in the table, the majority of spectral similarities is over 90%. If one imagines the combination of the full-scan chromatograms illustrated in Figures 1, 2, and 3, then it is possible to form an idea of the overall complexity of coal tar and on the potential difficulties that one could encounter for a specific analytical scope. To confirm such a concept, the coal tar sample was subjected to direct full-scan GC × GC/QqQ MS analysis; the outcome was a very complex chromatogram available as Supporting Information (Figure S1). Initially, the MRM mode was used for the determination of aromatic polycyclic S hydrocarbons for which a commercial standard was available: one class of thiophenes, two of benzothiophenes, and three of dibenzothiophenes. MRM transitions were optimized considering each pure standard as representative of a specific class of compounds (Table 1). For example, 2-propylthiophene was representative of the C3-T group (quantification transition: m/z 126 > 77). It must be emphasized that the members of the same PASH class have very similar mass spectra, with the same significant ions,18 and so, the same MRM transitions were applied to each group. However, it is obvious that two isomers

nation process enabled the attainment of generally satisfactory MS database search matches. For example, the following spectral similarities were observed for single compounds and specific chemical classes (the majority of spectral similarities were over 90%): phenol, 94%; 5-indanol, 94%; C2-phenols, 91−93%; naphthalenols, 90−94%. In relation to Figure 2, spectral similarities for single compounds, as well as spectral similarity ranges and average values for the indicated chemical classes, are reported as Supporting Information (Table S1). Identification of oxygenated constituents within a specific chemical group is not provided because many isomers with similar spectra exist. It is noteworthy that, after a direct GC × GC/QqQ MS analysis of coal tar, MS database spectral similarities were much lower (and sometimes not present at all) for the same constituents. Unsaturated Compounds (with and without S). The second fraction was formed mainly of unsaturated compounds, with and without S. The same PTV conditions as for the third fraction were applied; with regard to the GC × GC temperature program, a positive GC2 offset of 10 °C was applied. Two types of MS applications were performed, namely, full scan and MRM (MS solvent cut time: 11.0 min). The full-scan GC × GC/QqQ MS chromatogram of the second fraction is shown in Figure 3; as can be seen, the bidimensional space is welloccupied with wrap-around (corrected by using a software function) occurring only for the phenanthrene and pyrene classes; moreover, and as expected, the chromatogram was characterized by a high degree of structural order: chemical classes are grouped together on the basis of the number of aromatic rings (benzene, naphthalene, etc.) and MW of the 1915

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918

Article

Analytical Chemistry

Figure 4. MRM LC-GC × GC/QqQ MS chromatogram expansion relative to the C4-T class; peaks 1−7 are all C4-thiophenes. Inset: raw MRM LC-GC × GC/QqQ MS chromatogram expansion relative to the separation of peak 4 and 3-methylbenzenethiol.

Table 1 (the first transition reported is the quantification ion). Standard solutions for MRM optimization were prepared in dichloromethane, at an approximate concentration of 10 mg L−1, and were analyzed by direct GC × GC/QqQ MS. Apart from the support of the pure standard compounds (positions in the 2D chromatogram), the MRM time windows (C3-T, BT, C1-BT, DBT, C2-DBT, C4-DBT) were defined through the investigation of the full-scan coal tar data (LC-GC × GC/QqQ MS). In fact, in many cases, valuable MS database information was attained for several aromatic S compounds. Additionally, MRM transitions and time windows were defined for compound classes for which no pure standard was available (Table 1). Such data were derived for C2-T, C4-T, C2-BT, C3BT, C4-BT, C5-BT, C1-DBT, C3-DBT, BNT, C1-BNT, and C2-BNT, through the study of 2D chromatogram locations and full-scan spectra, derived from LC-GC × GC/QqQ MS coal tar analysis. Figure 4 shows an MRM LC-GC × GC/QqQ MS chromatogram expansion relative to the C4-T class (C8H12S), in which seven constituents (numbered 1 to 7) were detected. Altogether, 129 sulfur compounds were identified (all tentatively, apart for those corresponding to the pure standards) by combining and cross-checking full scan and MRM information. Apart from the chemical class, no identification is reported because the same “most probable” MS database match was attained for a series of peaks, when investigating the full-scan data. The high selectivity of the MRM mode is evident: only a single “interfering” compound was present and was tentatively identified as 3-methylbenzenethiol (C7H8S) through the full-scan information. As can be seen

Figure 5. MRM LC-GC × GC/QqQ MS chromatogram expansion relative to the DBT class; peaks 1−3 are all naphthothiophenes.

relative to the same class (e.g., 2,3,4-trimethylthiophene and 2propylthiophene) can have a slightly different response for a specific MRM transition, and so, the ion ratios reported herein (for each aromatic S class) give a good, though not entirely correct, indication. It is noteworthy that there are hundreds of isomers considering all the classes of aromatic S compounds, the majority of which are not available commercially, thus justifying such an approach (e.g., there are 20 isomers in the C2-DBT class). MRM information for each S class is listed in 1916

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918

Article

Analytical Chemistry

Additionally, dedicated 3D chromatography software (not available at the moment) would be necessary to process the resulting multidimensional data. It is the present authors’ opinion that the instrumental advantages fully justify the expenses needed for the construction of the instrumentation proposed. Future research will be directed to the exploitation of the system in a variety of applicable fields.

in the inset in Figure 4, 3-methylbenzenethiol responded only to the qualifier transition (125 > 91) of the C4-T class. In this specific case, it was the high selectivity of GC × GC that avoided a complete first-dimension coelution with peak 4 and a great deviation from the predetermined ion ratio. It is worthy of note that two-dimensional peak intensities are derived from the sum of peak intensities relative to each transition. Another chemical class, namely, DBT, is illustrated in Figure 5. In this case, four constituents were detected, with one of these positively identified (the DBT standard was available). Apart from dibenzothiophene, the naphthothiophene class was pinpointed with the help of the full-scan information. MS database similarity matches were characterized by an average value of 91% for the naphthothiophenes. Considering the direct GC × GC/QqQ MS coal-tar analysis, the MS database spectral similarities were much lower for the naphthothiophene class, on average 80%. Again, it is worthy of note that the selectivity of GC × GC avoided a first-dimension coelution between DBT and a naphthothiophene. Even though method validation was not performed, some analytical information was gathered through the MRM LC-GC × GC/QqQ MS analysis of standard solutions of S-containing compounds at the 100 ppb level (see Samples and Sample Preparation). In each application, the volumes transferred to the GC were the same as for the coal tar second fraction. Peak area (quantification transition) intraday precision (n = 6), expressed as relative standard deviation (RSD), was between 2% and 11%; peak area interday precison, which was measured by analyzing the standard solution three times on three consecutive days, was between 6% and 11%. Ion ratio intra- and interday precision were also measured (considering the first and second transition in Table 1), with RSD values both in the range of 1−7%. Limits of detection (LoD) and quantification (LoQ) were extrapolated from the responses attained at the 100 ppb level, considering a signal-to-noise ratio of 3 and 10, respectively (n = 6). Minimum and maximum LoD values were 1 (C4-DBT) and 23 (C1-BT) ppb, while LoQ values ranged between 2 and 75 ppb. Precision, LoDs, and LoQs were extrapolated from the quantifier ion intensities reported in the native GC/MS software used (see LC-GC × GC/QqQ MS Analyses). Integer numbers were used to express LoDs and LoQs because in most cases the values derived were 2 orders of magnitude lower than the 100 ppb level and are to be considered as approximations. Finally, it is noteworthy that the LoD/LoQ values were extrapolated by injecting only 2 μL of solution into the HPLC (the same volume as the coal tar solution). With the sample loop employed, it would be possible to inject up to a volume of 50 μL, meaning that LoQ values could be potentially reduced to the low ppt level.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org/.

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: +39-090-6766536. Fax: +39-090-358220. E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The Project was funded by the “Italian Ministry for the University and Research (MIUR)” within the National Operative Project “Hi-Life Health Products from the industry of foods” (Project ID, PON01_01499).

■

REFERENCES

(1) Tranchida, P. Q.; Franchina, F. A.; Dugo, P.; Mondello, L. Mass Spectrom. Rev. 2014, DOI: 10.1002/mas.21443. (2) Garrido Frenich, A.; Gonzáles-Rodríguez, M. J.; Arrebola, F. J.; Martínez Vidal, J. L. Anal. Chem. 2005, 77, 4640−4648. (3) Koesukwiwat, U.; Lehotay, S. J.; Miao, S.; Leepipatpiboon, N. J. Chromatogr., A 2010, 1217, 6692−6703. (4) Maštovská, K.; Lehotay, S. J.; Hajšlová, S. J. Chromatogr., A 2001, 926, 291−308. (5) Hernández, F.; Portolés, T.; Pitarch, E.; López, F. J. Anal. Chem. 2007, 79, 9494−9504. (6) Risticevic, S.; DeEll, J. R.; Pawliszyn, J. J. Chromatogr., A 2012, 1251, 208−218. (7) Schurek, J.; Portolés, T.; Hajslova, J.; Riddellova, K.; Hernández, F. Anal. Chim. Acta 2008, 611, 163−172. (8) Marriott, P. J.; Shellie, R.; Cornwell, C. J. Chromatogr., A 2001, 936, 1−22. (9) Purcaro, G.; Moret, S.; Conte, L. TrAC, Trends Anal. Chem. 2013, 43, 146−160. (10) Adam, F.; Bertoncini, F.; Thiébaut, D.; Esnault, S.; Espinat, S.; Hennion, M. C. J. Chromatogr. Sci. 2007, 43, 643−649. (11) Edam, R.; Blomberg, J.; Janssen, H. G.; Schoenmakers, P. J. J. Chromatogr., A 2005, 1086, 12−20. (12) Biedermann, M.; Grob, K. J. Sep. Sci. 2009, 32, 3726−3737. (13) Tranchida, P. Q.; Zoccali, M.; Bonaccorsi, I.; Dugo, P.; Mondello, L.; Dugo, G. J. Chromatogr., A 2013, 1305, 276−284. (14) Potgieter, H.; van der Westhuizen, R.; Rohwer, E.; Malan, D. J. Chromatogr., A 2013, 1294, 137−144. (15) Dutriez, T.; Thiébaut, D.; Courtiade, M.; Dulot, H.; Bertoncini, F.; Hennion, M.-C. J. Chromatogr., A 2012, 1255, 153−162. (16) Soleimani, M.; Bassi, A.; Margaritis, A. Biotechnol. Adv. 2007, 25, 570−596.

■

CONCLUSIONS A highly flexible, unified, 5D LC-GC × GC/QqQ MS instrument has been herein described. The on-line nature of the system, compared to off-line approaches (e.g., LC//GC × GC, solid-phase extraction//GC × GC), reduces the risks of sample contamination, improves run-to-run precision, and enables the setting of batch-type applications. The system can be used in a variety of configurations, depending on the specific analytical requirements. Practically all of the requirements of any GC-based application can be covered along the line of the possible operational modes. It is clear that it will be hard to find a sample (along with analytical objective) requiring the most powerful configuration, specifically LC × GC × GC/QqQ MS. 1917

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918

Article

Analytical Chemistry (17) Purcaro, G.; Zoccali, M.; Tranchida, P. Q.; Barp, L.; Moret, S.; Conte, L.; Dugo, P.; Mondello, L. Anal. Bioanal. Chem. 2013, 405, 1077−1084. (18) Machado, M. E.; Caramão, E. B.; Zini, C. A. J. Chromatogr., A 2011, 1218, 3200−3207.

1918

DOI: 10.1021/ac504162a Anal. Chem. 2015, 87, 1911−1918