Direct Infusion Resonance-Enhanced Multiphoton Ionization Mass

Sep 29, 2017 - Direct infusion resonance-enhanced multiphoton ionization (DI-REMPI) was performed on liquid samples, which were introduced to the ion ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/ac

Direct Infusion Resonance-Enhanced Multiphoton Ionization Mass Spectrometry of Liquid Samples under Vacuum Conditions Claudia Kruth,†,⊥ Hendryk Czech,†,⊥ Martin Sklorz,†,‡ Johannes Passig,†,‡ Sven Ehlert,†,§ Achille Cappiello,∥ and Ralf Zimmermann*,†,‡ †

Joint Mass Spectrometry Centre, Chair of Analytical Chemistry, Institute of Chemistry, University of Rostock, 18059 Rostock, Germany ‡ Joint Mass Spectrometry Centre, Cooperation Group “Comprehensive Molecular Analytics” (CMA), Helmholtz Zentrum München−German Research Centre for Environmental Health, 85764 Neuherberg, Germany § Photonion GmbH, Hagenower Strasse 73, 19061 Schwerin, Germany ∥ DiSPA, LC-MS Laboratory, University of Urbino, Piazza Rinascimento 6, 61029 Urbino, Italy S Supporting Information *

ABSTRACT: Direct infusion resonance-enhanced multiphoton ionization (DI-REMPI) was performed on liquid samples, which were introduced to the ion source via a direct liquid interface, to enable the investigation of dissolved aromatic compounds. Desolvation and nebulization of the samples were supported by a heated repeller using flow rates in the upper nL min−1 range. The obtained mass spectra of five pure polycyclic aromatic hydrocarbons as well as complex petroleum samples revealed predominantly molecular ions without evidence of solvent or dopant effects as observed in atmospheric pressure photoionization (APPI) and laser ionization (APLI) with limits of detection in the lower pmol range. Furthermore, it is demonstrated by the analysis of different complex oil samples that DIREMPI covers a larger m/z range than external volatilization of the sample prior to introduction to the ion source by using thermogravimetry (TG) hyphenated to REMPI time-of-flight mass spectrometry (TOFMS). Analogous to reported setups with direct liquid interface and electron ionization, direct-REMPI may be an option for soft ionization in liquid chromatography.

I

Basically, either VUV-photons (APPI) or UV-photons (APLI) ionize both matrix components, solvent and analytes, which fullfil the ionization criteria. Additionally, charges can be transferred from compounds of higher ionization energies (IEs) to lower IEs or low proton affinity (PA) to higher PAs, generating further molecular ions or adducts. Specific compounds are often added to the sample to enhance the ion yield of the analyte, which is known as dopant-assisted APPI and APLI. All of the mentioned ionization techniques share the disadvantage that not only molecular ions M+ but also substantial amounts of proton addition or abstraction ([M + H]+ or [M − H]−) are coincidentally formed. Furthermore, the composition of the matrix affects the ion yield,13 which may complicate the interpretation of the spectra. Photoionization (PI) under vacuum conditions, singlephoton and resonance-enhanced multiphoton ionization (SPI, REMPI), is known as a soft ionization technique mainly leading to molecular ions without adduct formation or substantial

onization techniques with ion sources under atmospheric pressure (API) denote a substantial development for the liquid chromatography mass spectrometry (LC/MS) instruments in several analytical fields. In particular, electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) are common API techniques, also due to their suitability for analysis of biologically relevant compounds over a wide range of volatility. However, both ESI and APCI favor the ionization of polar to medium-polar species.1 Closely related to APCI, atmospheric pressure photoionization (APPI)2,3 and atmospheric pressure laser ionization4 gained importance for nonpolar analytes with high sensitivity. Therefore, ESI, APCI, and APPI or chip-ESI and APLI were even combined in a multimode source for low-flow LC/MS or direct infusion mass spectrometry to cover a wider range of polarity.5−7 Apart from biological applications, APPI and APLI were applied to the analysis of crude oil, vitamin E congeners in soybean oil, molecular weight distribution of polyethylene, and gaseous toxic industrial compounds.8−11 A discussion on asphaltene mass spectra obtained from five different API techniques illustrates their complementary characters, which can be useful for the analysis of complex samples.12 © XXXX American Chemical Society

Received: July 6, 2017 Accepted: September 18, 2017

A

DOI: 10.1021/acs.analchem.7b02633 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

ionization yields. A turbomolecular pump (170 L s−1, Pfeiffer Vacuum, Germany) enables the fast removal of the solvent vapor and is directly attached below the ion source as well. The readout of the pressure gauge, located about 30 cm away from the center of the source, detects 10−4 mbar. For photoionization, an Nd:YAG laser (Minilite I, Continuum, California, USA) provides 266 nm radiation with 1 mJ per pulse and a repetition rate up to 15 Hz for REMPI. According to the wavelength and applied energy range, REMPI refers to a two-photon process ([1 + 1]-REMPI), which selectively ionizes aromatic compounds with low fragmentation.15 The originally installed electron ionization (EI) is kept operationally, e.g., for mass calibration of the quadrupole. Figure 1 illustrates the entire instrumental setup consisting of a direct liquid interface, ion source, Nd:YAG laser, and quadrupole mass spectrometer.

matrix effects.14,15 REMPI also refers to a part of the APLI mechanism and shows high selectivity and sensitivity for the ionization of aromatic compounds at a frequently applied wavelength of 266 nm, corresponding to the fourth harmonic of an Nd:YAG laser. Within a typical applied laser intensity range of 106−108 W cm−2 at 266 nm, only [1 + 1]-REMPI is expected, which means the absorption of one photon to excite a molecule and a second photon to exceed the ionization barrier.15 In a previous work, SPI by VUV-photons was demonstrated to be suitable for direct ionization of pharmaceutically relevant compounds from the liquid phase and as an option for LC coupling.16 As multiphoton ionization TOFMS with laser pulses from the nano- to the femtosecond range were shown to be a powerful detector for gas chromatography17,18 and also direct infusion (DI),19 this study aimed to implement REMPI for the analysis of aromatic compounds in liquid samples via a direct interface similar to SPI in the setup of Schepler et al.16,20 In the present fundamental study, we demonstrate the suitability of REMPI in connection with a direct liquid interface (in the following referred to as DI-REMPI) for the analysis of complex liquid samples. The ionization was conducted under high vacuum conditions, so molecule−ion collisions were drastically reduced, and predominantly radical cations were observed, which is typical for REMPI, without matrix effects. In contrast to previous studies, the introduction of the liquid sample into the ion source is not supported by additional gas flow or a vaporization laser. The performance of the instrument was investigated with a standard solution of six polycyclic aromatic hydrocarbons (PAHs) as well as with complex samples including diesel, heavy foil oil, and four crude oils of different origin. DI-REMPI spectra are compared to spectra obtained from thermogravimetry (TG) hyphenated to REMPI time-of-flight mass spectrometry (TOFMS) setup for evolved gas analysis, which represents an alternative photoionization mass spectrometric technique for the investigation of complex petroleum samples. Finally, a principal component analysis (PCA) was performed on the spectra of oil samples to exhibit the potential of the obtained data for rapid classification and identification of responsible oil constituents.

Figure 1. Instrumental setup of the DI-REMPI-MS and sample injection based on Schepler et al.16 The modifications comprise the exchange of the VUV lamp to an Nd:YAG laser and an enlargement of the light entrance. The electron gun was not used in these experiments and is therefore shadowed in this figure.

Regarding data acquisition, the quadrupole settings (DC and AC voltage for the corresponding m/z value) were synchronized to the laser shots driven by an external pulsedelay generator. This setup was constructed to show the proof of principle, but of course with a time-of-flight system, the duty cycle of the system in scanning mode would be enhanced by several orders of magnitude. For comparison purposes, measurements with a thermobalance (STA 409, PG Luxx, Netzsch Gerätebau GmbH, Germany) for thermogravimetry (TG) and evolved gas analysis by REMPI-TOFMS at 266 nm21 were carried out (Figure S1). Samples were placed in an Al2O3 sample pan under an inert N2 atmosphere (summed gas flow through thermobalance = 80 mL min−1, comprising 20 mL min−1 protective gas and 60 mL min−1 purge gas) and heated up from 30 to 1000 °C with a heating rate of 10 K min−1. PAH Standard, Complex Samples, and Sample Treatment. For flow injection experiments, solutions of six PAHs with different concentrations (0.1, 0.3, 1.0, 3.0, and 10.0 ng μL−1) were prepared in acetonitrile and dichloromethane (1:1, both LC-MS grade) to verify linearity of responses and to estimate limit of detections. To demonstrate analysis of complex matrices, four crude oils from different origins (North Sea, California, Greece, and Turkey) and one heavy fuel oil (HFO-180), representing marine fuels for ships, were dissolved in dichloromethane with a volumetric dilution of 1:20



EXPERIMENTAL SECTION Instrumental Setup. The instrumental setup is based on the work of Schepler et al.16 and enables the analysis of liquid samples by DI-REMPI-MS. Samples were introduced into a modified ion source of a quadrupole mass spectrometer (HP5970B MSD, Hewlett-Packard, California, USA) by flow injection in a constant solvent flow using a manual injection valve (VICI AG International, Switzerland) with an external 20 μL loop. The required low flow rate was provided by a nanoflow liquid chromatography pump (nanoAcquity, Waters Corporation, Massachusetts, USA). To enable a proper sample transfer from the capillary (fused-silica/PEEK, inner diameter of 25 μm) into the ion source, a direct interface was implemented as described in detail by Schepler et al.16 Briefly, the capillary ends about 1 mm away from the repeller, which is directly attached to the ion source and equipped with a heater to support the coincident desolvation and nebulization of the eluent. Compared to the previous setup, the temperature of the repeller was slightly increased to 280 °C, the flow rate was elevated to 400 nL min−1, and the openings in the ion volume for light entrance were enlarged to overall generate higher B

DOI: 10.1021/acs.analchem.7b02633 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry and filtrated using a 13 mm syringe filter (PTFE (hydrophobic), 0.22 μm). Furthermore, four diesel fuels with different biofuel contents (B0, B7, and B30) in the form of fatty acid methyl esters as well as one light subfraction of diesel (Haltermann solutions, Texas, USA) were also dissolved in dichloromethane with a dilution of 1:20 for analysis. Data Analysis. Quadrupole raw data were processed with DataAnalysis (BrukerDaltonik GmbH, Germany). The conversion of TOFMS raw data was performed by custom functions in Matlab 2014b (The Mathworks Inc., Massachusetts, USA), which was also used for further data evaluation and statistical analysis with commands from its Statistic Toolbox.



RESULTS AND DISCUSSION DI-REMPI of PAHs in a Solution. Mixtures of six PAHs (m/z 128 naphthalene, m/z 154 acenaphthene, m/z 178 phenanthrene, m/z 202 fluoranthene, m/z 228 chrysene, and m/z 252 benzo[e]pyrene) in different concentrations were analyzed by the setup as described in the previous section. Figure 2 depicts a REMPI spectrum of the PAH solution with a

Figure 3. Concentration series (five repetitions for each concentration) with instrumental response (black squares with error bars showing standard deviation from n = 5, left y-axis) coefficient of variation (gray triangles, right y-axis).

Figure 2. Mass spectrum of a PAH mixture (0.3 ng μL−1 in acetonitrile/dichloromethane 1:1) ionized by DI-REMPI@266 nm. The m/z 120 was not a constituent of the PAH standard solution and was identified as a contamination of propylbenzene from previous measurements.

corresponding to that level, ranging from 1.7 to 15 μg L−1 for all analytes, was defined as LOD at a signal-to-noise ratio (SNR) of 3, which is apparently higher than that of other competing analytical instruments for PAH analysis, such as GCAPLI-TOFMS with LODs in the ng L−1 range,22 but lower than that of recently reported laser vaporization multiphoton ionization for liquid samples.23 However, we want to emphasize that a benefit of our system is the low sample consumption. Thus, the results are given on an absolute scale per sample injection (Table 1), similar to on-column common for

concentration of 0.3 ng μL−1, which is dominated by the corresponding molecular ions of the six PAHs. The additional signal at m/z 120 is caused by a cross-contamination with propylbenzene. Some minor fragments of [M − 2]+, which are known as PAH fragments in EI, occur in the spectrum, but double-charged ions, which are for typical EI, were not detected. Furthermore, abundances of [M + 1]+ appear in accordance with isotopic distribution caused by 13C and are not generated by a proton transfer reaction. Finally, the ratio of the peaks to each other is similar to REMPI spectra under vacuum conditions, which supports the statement of negligible solvent effects and gives evidence for pressure inside the ion source below the threshold for substantial ion−molecule reactions. Regarding the different concentrations (Figure 3), a moderate linear response was found between concentration and intensity. However, for the highest concentration of 10 ng μL−1 the response left linear behavior, most probably due to high space charge occurring in the ion volume due to efficient ionization during the 10 ns laser pulse. Except naphthalene which shows the largest variance, all coefficients of variation appear below 30% for five repetitions of each concentration. The limits of detection (LODs) were estimated calculating the mean intensity with addition of the 3-fold standard deviation of an individual ion trace from a blank run. The concentration

Table 1. Estimated Limits of Detection (LOD, at SNR = 3) for Six PAHs in 1:1 Acetonitrile/Dichloromethane PAH

m/z

LOD [pmol injection−1]

naphthalene acenaphthene phenanthrene fluoranthene chrysene benzo[e]pyrene

128 154 178 202 228 252

2.4 0.87 0.19 1.1 0.2 0.17

chromatographic systems, appearing in the lower pmol-range. A quadrupole based on the state of technology or a TOFMS may further improve the LOD. However, the latter would require additional technical developments for more sophisticated pumping or ion guiding to meet the pressure requirements for this mass analyzer. Furthermore, relatively high C

DOI: 10.1021/acs.analchem.7b02633 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

despite the high availability of known dopants, which affect the ion yield and distribution between molecular and protonated molecular ions in photoionization techniques under atmospheric pressure. However, for other crude oil samples, such as the Turkish crude oil (Figure 5), odd m/z in a sequence of 14 Th denotes the most abundant peaks. However, the known homologue series of alkylated naphthalenes are visible, so inaccuracies in mass calibration can be excluded. Moreover, previous analysis by photoionization TOFMS with additional chromatographic resolution gave no evidence for nitrogencontaining (poly-)aromatic compounds in fairly high concentrations.26 Hence, these peaks of odd m/z either belong to protonated molecular ions from the formation of adducts or result from fragmentation. The working hypothesis was that they were caused by fragmentation of larger alkylated polycyclic aromatic structures during ionization. In contrast to the thermal analysis method, those large PAHs are introduced into the ion source but cover relatively low ionization energies. Thus, the excess energy δE during the ionization, which means the difference between the summed energy of the two 266 nm photons and IE of the molecule, may become sufficiently high to cause fragmentation even in REMPI. Each 266 nm photon has an energy of 4.66 eV, so a total energy of 9.32 eV is available for ionization. As an example, δE during the ionization of naphthalene (IE = 8.1 eV, NIST chemistry webbook) accounts for 1.2 eV, whereas for the larger PAHs pyrene (IE = 7.4 eV) and benzo[a]pyrene (IE = 7.1 eV) δE is almost doubled to 1.9 and 2.2 eV, respectively. Additionally, higher degrees of alkylation lower the ionization energy as well. The working hypothesis of fragmentation of large alkylated PAHs was tested by two experiments. First, three diesel samples with different amounts of biodiesel (0, 7, and 30% of fatty acid methyl esters (FAME)) and a light diesel were analyzed by the same procedure (Figure 6). It was assumed that diesel, which refers to a distillation cut of the crude oil, contains only minor amounts of those large PAH structures due to the boiling point range of its components from approximately 200 to 350 °C and carbon numbers from 11 to 20.27 All spectra from DI-REMPI analysis contain a homologue series of 1- to 5-ring (poly)aromatic hydrocarbons, while the spectrum of the light Haltermann diesel sample is clearly shifted toward lower m/z with a base peak at m/z 128, which belongs to naphthalene. The vast majority of the diesel constituents comprised C, H, and O, which lead to peaks of even-numbered m/z. Oddnumbered m/z values clearly show lower abundances than even-numbered m/z, in particular for the light Haltermann diesel, so fragmentation plays a minor role in the ionization process. For testing the hypothesis of proton transfer reaction and ion−molecule interaction, the effect of dopant addition, known from APLI and APPI, was examined by toluene addition. Toluene was added in four different concentrations of 1%, 3%, 10%, and 30% v/v to the diluted heavy fuel oil sample. However, neither a change in the mass spectrometric pattern nor an influence on the peak intensities was observed. Solely the peak intensity of toluene at m/z 92 (IE = 8.83 eV) increased with a linear and direct proportional response (Figure 7). Fragmentation rules and pathways of alkylated PAHs from electron ionization are known from the literature and can be adapted to suppose structures of the observed odd-numbered ion series. Zhao et al.29 included fragments of alkylated PAHs in GC/MS analysis for quantification and proposed fragmentation

LODs result from the fact that only during 10 ns of the 67 ms between consecutive laser shots ions are generated. Laser of higher repetition rate would likely improve the LODs. Analysis of Complex Samples. The performance of the instrumental setup was tested with complex matrices by means of four crude oils and one heavy fuel oil. Additionally, the samples were analyzed by TG-REMPI-TOFMS as reference analysis with a similar ionization process but at higher thermal stress for the sample. A widely used temperature ramp of 10 K min−1 was applied to volatilize approximately 8−9 mg of each crude oil in the TG, while evolving gases were detected by REMPI-TOFMS. The combination of TG with photoionization TOFMS was successfully applied to the analysis of crude oils and other complex samples.24,25 The sum of detected compounds in TG analysis from room temperature to 280 °C appears in the m/z range from 100 to 300 with highest intensities for alkylation series of 2- and 3-ring PAHs, such as a series of alkylated naphthalenes (m/z 128, m/z 142, m/z 156, and m/z 170) and alkylated phenanthrenes (m/z 178, m/z 192, m/z 206, and m/z 220) (Figure 4).

Figure 4. Mass spectra of North Sea crude oil from DI-REMPI (top) and TG-REMPI-TOFMS analysis (bottom). The mass traces from TG-REMPI-TOFMS were summed up to 280 °C for comparison with the DI-REMPI experiments where the repeller was heated to the same temperature. Most abundant homologue alkylation series belong to benzene, naphthalene, phenanthrene, and chrysene/benz[a]anthracene. Red lines denote the cumulative distribution function of the m/z intensities, indicating that in DI-REMPI compounds of higher molecular weight are accessible compared to those in TG-REMPITOFMS.

In the contour plot (Figure S2), a continuous increase in detected m/z with increasing temperature can be observed, indicating volatilization of the crude oil constituents without decomposition. Decomposition through pyrolysis starts at approximately 250 °C, pointed out by the occurrence of thermal fragments with lower m/z. On the contrary, the DIREMPI spectrum shows an enlarged range of accessible m/z at equal temperatures and a higher average m/z as well (DIREMPI, 254; TG-REMPI, 191). Although naphthalene and phenanthrene alkylation series also denote the most prominent ions in DI-REMPI, higher m/z can be detected, such as highly alkylated benzenes (m/z 246, m/z 260, m/z 274, and m/z 302; cyan squares), which may thermally decompose in the TG before they are introduced into the TG-REMPI-TOFMS. As typical for REMPI, molecular ions dominate the spectrum D

DOI: 10.1021/acs.analchem.7b02633 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 5. DI-REMPI mass spectra of the five investigated oils. Apart from the North Sea oil (C), odd-numbered ion series (red down-pointing triangles) dominate the spectra. The vast majority of the crude oil constituents, accessible by REMPI, are composed of C, H, S, and O. Therefore, only even-numbered molecular ions can be formed, which excludes the assignment to molecular ions (as radical cations).

Figure 7. Addition of toluene in concentration of 1, 3, 10, and 30% (v/ v) to heavy fuel oil with linear response as depicted in the inset. Toluene is a frequently used dopant in APLI28 but neither affects the total ion yield nor the mass spectrometric pattern apart from its own increase at m/z 92 (black triangles).

Figure 6. DI-REMPI mass spectra of four diesel samples, (a) B0 diesel, (b) B7 diesel, (c) B30 diesel, and (d) light reference diesel from Haltermann, which are strongly dominated by the ion series of alkylated naphthalenes (m/z 128, 142, 156, ...) and phenanthrenes (m/ z 178, 192, 206, 220, ...) but contain only minor amounts of oddnumbered fragment ions. Moreover, the different contents of biodiesel by means of FAME do not affect the DI-REMPI mass spectra.

from m/z 141 and m/z 241 (Figure 8), which are more abundant in the other oil samples and assigned to them in the following section. Moreover, steric interactions in the tropylium-derived ions, such as in angled aromatic structures (Figure 8, center and right), may further eliminate two protons by bridging two aromatics leading to an ion series of m/z 189, 203, 217, 231, 245, 259, ... Therefore, information about the structure of isobaric alkyl PAHs can be obtained based on the of ratio these two fragment ion series. Additional chromatographic separation of the alkylated PAHs may reveal which fragmentation pathways require low excess energy so that fragment ions even appear in otherwise soft REMPI. Altogether, these two experiments in combination with fragmentation pathways on EI support the working hypothesis

pathways for 9-butylphenanthrene as a model compound. The molecular ion can rearrange after the loss of C3H7+ to form a tropylium-derived ion with two condensed aromatic six-rings or from direct benzylic bond cleavage, which gives the ion of m/z 191 and explains the following ion series of m/z 205, 219, 233, ... through increasing alkylation as observed for Turkish crude oil. This concept can be further extended to ions of more condensed six-rings explaining the additional ion series starting E

DOI: 10.1021/acs.analchem.7b02633 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 9. Biplot of PCA scores referring to single measurements of oil samples (colored circles) and loadings referring to variables (gray squares). For reasons of clarity, only the most relevant m/z values are labeled.

Figure 8. Supposed structures of parent PAH-derived ions which cause odd-numbered ion series in oil samples by their alkylated derivatives. Angled tropylium-derived ions can eliminate two hydrogen atoms followed by bridging two noncontiguous aromatic rings with a C−C-bond. Although each single step is known from the ionization of PAHs and their alkylated derivatives, theoretical calculations should be performed to verify the hypothesis of the proposed ion structures.

crude of high quality with relatively high amount of light hydrocarbons,31 which confirms its result from PCA. The association of heavy fuel oil (green) with lighter aromatic hydrocarbons might not be intuitive but can be elucidated by its production. After the crude oil has been distilled under vacuum, the residue is redissolved in middle distillates, such as diesel.27 The kinematic viscosity in this blend fuel must not exceed 180 mm2 s−1, which is necessary for its use as marine fuel. Thus, the similarity between the most important m/z of PC1 in the negative direction and the most abundant peaks of the diesel REMPI spectra (Figure 6) does not surprise. Moreover, it seems that the positive values of PC1 represent the amount of larger alkylated aromatic structures in the oil samples through photoionization fragments of odd-numbered m/z as discussed in a previous section. In particular, the Turkish crude oil was described as the apparently heaviest and most viscous one of the crude oil samples,32 so it does not appear in the same space as m/z from smaller PAHs. Furthermore, the Californian crude oil was characterized as medium-heavy25 and in fact is less explained by photoionization fragments in the positive direction of PC1 but also shows a low contribution of positive values of PC2 containing alkylated naphthalenes. Greek crude oil seems to have a higher molecular similarity to the Turkish crude oil but with a lower contribution of larger alkylated aromatic structures, which agrees with earlier findings by Wohlfahrt et al.26 In total, the PCA results back up the assignments for heaviness to the crude oils from thermal analysis (Figure S3).

that peaks of odd-numbered m/z rather belong to fragments of larger alkylated aromatic structure than protonated molecular ions. Classification of the Oil Samples by Principal Component Analysis (PCA). A principal component analysis (PCA)30 was carried out to demonstrate the potential of the generated data for rapid classification of complex oil samples and for proper visualization of the entire data set but also to show that REMPI mass spectrometry provides data beyond fingerprinting and enables chemical interpretation of the results. Briefly, PCA computes the eigenvectors of the m-bym covariance matrix of the centered m-by-n REMPI data. Here, it contains six measurements for each of the five crude oils (ntotal = 30) in the m/z range from 78 to 586 (m = 509). Finally, PCA generates m linearly uncorrelated latent variables called principal components (PCs). The first PC explains as much of the total variance as possible which also holds for the subsequent PCs under the constraint that they have to be orthogonal to the first PC. In this context, the n-by-m score matrix denotes the matrix needed to convert the PCs to the original data and illustrates the relationship between the samples. Prior to PCA, each mass spectrum was normalized to its base peak as is often done with EI spectra to discover similarities. Figure 9 displays the two first principal components explaining 44.5% and 23.9% of the total variance, respectively, and the scores (colored dots) of individual oil samples and measurements. The different oil samples clearly show lower inner-class variance than inter-class dispersion and are well separated from each other. From the PC loadings, the responsible variables, i.e., m/z, can be identified from the appearance of the samples in the score space. In the negative direction of the x-axis and positive direction of the y-axis, even-numbered m/z values have the highest influence on segregation of the samples and belong to the alkylation series of phenanthrene (m/z 178, 192, 206, 220, 234, and 248), naphthalene (m/z 142, 156, and 170), and chrysene/benzo[a]anthracene (m/z 228, 242, 256, 270, and 284). North Sea crude oil (yellow) is generally regarded as a



CONCLUSIONS In this study, REMPI was made accessible for ionization after direct infusion of the liquid phase into the high vacuum region similar to previous work with SPI and EI.16 Mass spectra of a PAH standard solution and complex petroleum samples by means of diesel, crude, and heavy oil show that the advantages of negligible solvent effects, the absence of protonated molecular ions, and low fragmentation are predominantly maintained for the investigated m/z range from 78 up to 586. The addition of toluene as a possible candidate for charge transfers affects neither the overall ion yield nor the mass spectrometric pattern. Compared to TG-REMPI-TOFMS, which thermally desorbs sample constituents for evolved gas analysis, the m/z range was substantially increased. Low-volatile F

DOI: 10.1021/acs.analchem.7b02633 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(6) Schmitt-Kopplin, P.; Englmann, M.; Rossello-Mora, R.; Schiewek, R.; Brockmann, K. J.; Benter, T.; Schmitz, O. J. Anal. Bioanal. Chem. 2008, 391, 2803−2809. (7) Syage, J. A.; Hanold, K. A.; Lynn, T. C.; Horner, J. A.; Thakur, R. A. J. Chromatogr., A 2004, 1050, 137−149. (8) Schrader, W.; Panda, S. K.; Brockmann, K. J.; Benter, T. Analyst 2008, 133, 867−869. (9) Cotte-Rodriguez, I.; Justes, D. R.; Nanita, S. C.; Noll, R. J.; Mulligan, C. C.; Sanders, N. L.; Cooks, R. G. Analyst 2006, 131, 579− 589. (10) Méjean, M.; Brunelle, A.; Touboul, D. Anal. Bioanal. Chem. 2015, 407, 5133−5142. (11) Kéki, S.; Nagy, L.; Kuki, Á .; Zsuga, M. Macromolecules 2008, 41, 3772−3774. (12) Gaspar, A.; Zellermann, E.; Lababidi, S.; Reece, J.; Schrader, W. Anal. Chem. 2012, 84, 5257−5267. (13) Kauppila, T. J.; Kersten, H.; Benter, T. J. Am. Soc. Mass Spectrom. 2014, 25, 1870−1881. (14) Hanley, L.; Zimmermann, R. Anal. Chem. 2009, 81, 4174−4182. (15) Boesl, U. J. Mass Spectrom. 2000, 35, 289−304. (16) Schepler, C.; Sklorz, M.; Passig, J.; Famiglini, G.; Cappiello, A.; Zimmermann, R. Anal. Bioanal. Chem. 2013, 405, 6953−6957. (17) Li, A.; Uchimura, T.; Watanabe-Ezoe, Y.; Imasaka, T. Anal. Chem. 2011, 83, 60−66. (18) Zimmermann, R.; Welthagen, W.; Gröger, T. J. Chromatogr. A 2008, 1184, 296−308. (19) Ishigami, H.; Tsuda, Y.; Uchimura, T. Anal. Methods 2014, 6, 5615−5619. (20) Cappiello, A.; Famiglini, G.; Pierini, E.; Palma, P.; Trufelli, H. Anal. Chem. 2007, 79, 5364−5372. (21) Streibel, T.; Mitschke, S.; Adam, T.; Weh, J.; Zimmermann, R. J. Anal. Appl. Pyrolysis 2007, 79, 24−32. (22) Schiewek, R.; Schellenträger, M.; Mönnikes, R.; Lorenz, M.; Giese, R.; Brockmann, K. J.; Gäb, S.; Benter, T.; Schmitz, O. J. Anal. Chem. 2007, 79, 4135−4140. (23) Uchimura, T.; Tokumoto, G.; Batnyam, O.; Chou, C. W.; Fujita, S. Anal. Sci. 2016, 32, 255−257. (24) Streibel, T.; Geißler, R.; Saraji-Bozorgzad, M.; Sklorz, M.; Kaisersberger, E.; Denner, T.; Zimmermann, R. J. Therm. Anal. Calorim. 2009, 96, 795−804. (25) Geissler, R.; Saraji-Bozorgzad, M. R.; Gröger, T.; Fendt, A.; Streibel, T.; Sklorz, M.; Krooss, B. M.; Fuhrer, K.; Gonin, M.; Kaisersberger, E.; Denner, T.; Zimmermann, R. Anal. Chem. 2009, 81, 6038−6048. (26) Wohlfahrt, S.; Fischer, M.; Saraji-Bozorgzad, M.; Matuschek, G.; Streibel, T.; Post, E.; Denner, T.; Zimmermann, R. Anal. Bioanal. Chem. 2013, 405, 7107−7116. (27) Fahim, M. A.; Al-Sahhaf, T. A.; Elkilani, A. Fundamentals of Petroleum Refining; Elsevier B. V.: Oxford, 2010. (28) Kauppila, T. J.; Kersten, H.; Benter, T. J. Am. Soc. Mass Spectrom. 2015, 26, 1036−1045. (29) Zhao, Y.; Hong, B.; Fan, Y.; Wen, M.; Han, X. Ecotoxicol. Environ. Saf. 2014, 100, 242−250. (30) Jolliff, I. T. Principal Component Analysis, 2nd ed.; Springer Verlag: New York, 2002. (31) Johnson, P. The Structure of British Industry; Routledge, Taylor & Francis, 2002. (32) Geissler, R.; Saraji-Bozorgzad, M.; Streibel, T.; Kaisersberger, E.; Denner, T.; Zimmermann, R. J. Therm. Anal. Calorim. 2009, 96, 813− 820. (33) Termopoli, V.; Famiglini, G.; Palma, P.; Piergiovanni, M.; Cappiello, A. Anal. Chem. 2017, 89, 2049−2056.

aromatic compounds were made accessible, which do not reach the TOFMS ion source because of thermal decomposition. Nevertheless, odd-numbered m/z in a selection of the oil samples occurred in the spectra, which were assigned to tropylium ion-derived structures and could be linked to large aromatic constituents in heavy crude oils. From the respective fragmentation ion series, even structural information can be gained from the spectra. Altogether, for liquid samples DI-REMPI-MS provides rapid characterization of aromatic analytes in complex mixtures, such as crude oil, and may be applied to online analysis of educts in reaction solutions of highly parallelized combinatorial chemistry studies. However, the performance of this rather low-cost quadrupole setup can be substantially improved by switching the mass analyzer to a state of technology quadrupole or a timeof-flight system. Additionally, a laser of higher repetition rate would enhance the duty cycle and improve the LODs further. In doing so, sensitivity, time, and possibly mass resolution may be substantially enhanced as demanded by liquid chromatography, which has been successfully applied with the combination of EI-MS and direct liquid interface.20,33



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.7b02633. Instrumental setup of the TG-REMPI-TOFMS, contour plot of the total TG-REMPI-TOFMS analysis of North Sea crude oil, and curves of mass loss from TG analysis (PDF)



AUTHOR INFORMATION

Corresponding Author

*Phone: +49 0 381 498 66460. Fax: +49 0 381 498 118 6527. E-mail: [email protected]. ORCID

Hendryk Czech: 0000-0001-8377-4252 Achille Cappiello: 0000-0002-6416-304X Ralf Zimmermann: 0000-0002-6280-3218 Author Contributions ⊥

C.K. and H.C. contributed equally to this work.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS C.K. and H.C. thank the European Social Fund for Germany (ESF) for scholarships. Furthermore, A. Ulbrich and C. Grimmer are acknowledged for providing the TG-REMPITOFMS data.



REFERENCES

(1) Covey, T. R.; Thomson, B. A.; Schneider, B. B. Mass Spectrom. Rev. 2009, 28, 870−897. (2) Robb, D. B.; Covey, T. R.; Bruins, A. P. Anal. Chem. 2000, 72, 3653−3659. (3) Syage, J. A.; Evans, M. D.; Hanold, K. A. Am. Lab. 2000, 32, 24− 29. (4) Schmidt, S.; Appel, M. F.; Garnica, R. M.; Schindler, R. N.; Benter, T. Anal. Chem. 1999, 71, 3721−3729. (5) Short, L. C.; Hanold, K. A.; Cai, S. S.; Syage, J. A. Rapid Commun. Mass Spectrom. 2007, 21, 1561−1566. G

DOI: 10.1021/acs.analchem.7b02633 Anal. Chem. XXXX, XXX, XXX−XXX