Identification of Ion Series Using Ion Mobility Mass Spectrometry

Article pubs.acs.org/ac

Identification of Ion Series Using Ion Mobility Mass Spectrometry: The Example of Alkyl-Benzothiophene and Alkyl-Dibenzothiophene Ions in Diesel Fuels Florian Maire,†,⊥ Kieran Neeson,‡ Richard Denny,‡ Michael McCullagh,‡ Catherine Lange,† Carlos Afonso,*,† and Pierre Giusti§ †

Normandie Université, COBRA, UMR 6014 et FR 3038; Université de Rouen; INSA Rouen; CNRS, IRCOF, 1 Rue Tesnière, 76821 Mont-Saint-Aignan Cedex, France ‡ Waters Corporation, Manchester, United Kingdom § TOTAL Refining & Chemicals, Total Research & Technology Gonfreville, BP 27, 76700 Harfleur, France S Supporting Information *

ABSTRACT: Ion mobility-mass spectrometry (IMMS) has been presented as a promising method for analysis of highly complex mixtures. This coupling adds an additional postionization separation dimension to MS. The IM separation of ions is obtained in the millisecond time scale and can be particularly helpful when chromatographic separation is not possible. For obtaining relevant information about the samples, data processing is usually the bottleneck because of the high amount of data generated with IMMS. In the current work, we present a new workflow using specific comparison software dedicated to IMMS data, which allows one to compare m/z−drift time plots to highlight differences between samples. Two diesel fuels have been compared, i.e., the feed and the product of hydrodesulfurization (HDS) process, and this approach allowed us to clearly highlight the variation of intensity of several ions distributed along the plots of both samples. Accurate mass measurements and post IM collision induced dissociation experiments allowed us to identify two series of polycyclic aromatic sulfur-containing heterocycle (PASH) compounds among the matrix ions. influence of an electric field, leading to a separation in the millisecond time scale based on their ion-neutral collision cross section. The IMMS data yields a drift time vs m/z 2D-plot in which the different species are separated and organized depending on their properties.4,5 The hyphenation of ASAP and IMMS has been presented as an efficient approach for detection and identification of impurities or additives, especially if the structure of the targeted ions is clearly different from those of the matrix.6−8 The major difficulty with the petroleum products analysis is that the thousands of components which constitute the sample cover a wide range of structures (e.g., from linear alkanes to the highly compact asphaltenes) but also have very similar molecular structures which differ from each other only by alkyl-chain length or presence of heteroatoms. The result of this complexity is a continuous, feather-like distribution of ions along the m/z vs drift time plot which leads to a time-consuming data processing.3 Interest of the petroleum industry in detection and quantification of polycyclic aromatic sulfur-containing heterocycles (PASHs) is significant due to regulations in many countries concerning the sulfur content in gasoline and diesel fuel.9−12 Sulfur levels in gasoline and diesel fuel affect direct

T

he comparison of samples from complex mixtures and in particular of petroleum compounds remains a difficult task. With mass spectrometry, sample introduction, ionization, ion separation, and data processing are the main difficulties. The atmospheric solid analysis probe (ASAP) developed by McEwen has been shown to be an efficient technique for ionization of low polarity species in a solvent free approach.1,2 Crude oil can be analyzed using the ASAP ionization source showing that compounds with relatively high boiling point can be easily desorbed and ionized.2,3 No sample preparation is needed, so solubility of the compounds has not been considered; the single-use glass capillary tube is simply dipped in the sample. The sample is not continuously sprayed through the ion source like with the classical atmospheric pressure sources electrospray ionization (ESI), atmospheric pressure chemical ionization (APCI), and atmospheric pressure photoionization (APPI). Signal suppression due to clogging in the spray capillary or tubing is avoided. Furthermore, memory effect due to carry-over from previous sample is also strongly limited. However, one of the limitations of the ASAP source is that a chromatographic separation is difficult to implement. Ion mobility-mass spectrometry (IMMS) has been presented as a promising tool for characterization of complex mixtures such as oil samples.3 This coupling adds a postionization separation dimension to MS based on mass, shape, and charge state. In the ion mobility cell, ions drift in a buffer gas under the © XXXX American Chemical Society

Received: March 10, 2013 Accepted: May 2, 2013

A

dx.doi.org/10.1021/ac400731d | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

1000 ppm. IMMS experiments were performed using a SYNAPT G2 HDMS instrument (Waters Corp., Manchester, UK) fitted with an ASAP source. This instrument is a hybrid quadrupole/time-of-flight mass spectrometer, which incorporates a traveling wave (T-wave)-based mobility separation device. The instrument, the T-Wave device, and the ASAP probe have been described in detail elsewhere.1,27 A glass capillary tube, fixed to the ASAP probe holder, was dipped in the sample and introduced in the ionization source. A nitrogen flow of 1200 L h−1 heated at 650 °C was used to perform the thermal desorption of the sample. The ASAP was conditioned to experimental temperature for 1 h before experiments, and a blank was recorded for 1 min before each sample introduction. For all experiments, ASAP mass spectra were acquired in positive ion mode over the m/z 50−2000 range. The corona discharge voltage was 4 kV, and sampling cone voltage was 40 V. Trap and transfer collision gas flows were set at 2.0 mL min−1 (0.02 mbar for Ar cell pressure). Helium cell gas flow was set at 180 mL min−1, and IMS gas flow (N2) was set at 90 mL min−1 of (3.0 mbar for IM cell pressure). For ASAP-IMMS experiments, T-wave height and velocity were set, respectively, at 40 V and 652 m s−1. The T-wave velocity was tuned during the IMMS/MS experiment to partially resolve isobaric ions. Argon was used as target gas for MS/MS experiments. Data acquisition and mass spectra processing were carried out with MassLynx (version 4.1). The DriftScope (version 2.1) software was used for the processing of the IMMS data. HDMS Compare build 01 (Waters Corp., Manchester, UK) was used for the m/z vs drift time plots of unprocessed and processed diesel fuels. The purpose of the algorithm is (i) to place two images derived from mass spectrometric data on a common intensity scale and drift time axis so that a difference image may be constructed, (ii) to identify regions in the images where the intensity differences are significant (see a detailed description in the Supporting Information). Gas chromatography (fitted with a sulfur chemiluminescence detector, GC-SCD) and liquid chromatography (fitted with an ultraviolet detector, LC-UV) experiments were performed for comparison.

emissions of sulfur dioxide. Moreover, sulfur reduces the effectiveness of vehicle emission control technologies such as catalytic converters and diesel particle traps. In refineries, hydrodesulfurization (HDS) is the chemical process used to desulfurize sulfur-containing compounds, which are converted into hydrogenated derivatives (Scheme 1).13−16 Scheme 1. Reaction Pathways for Hydrodesulfurization of Dibenzothiophene9,13

Typical methods for the analysis of the volatile samples involve the use of gas chromatography (GC) with electron ionization−mass spectrometry17,18 or with different kinds of specific detectors such as flame photometric detection (FPD),19 sulfur chemiluminescence detection (SCD),20 and atomic emission detection (AED).21 On the other hand, liquid chromatography can be used for nonvolatile samples.22 It was shown that ultrahigh resolution mass spectrometry can be used to provide evidence of sulfur containing species in petroleum products. Under such conditions, chromatographic separation can be avoided as isobaric species can be separated by the mass spectrometer.23,24 The very high amount of data obtained with Fourier transform instruments (FT/MS) and the difficulties and time needed to process it are somehow similar as the issues encountered with ASAP-IMMS data. The high mass accuracy of FT/MS instruments allows a unique elemental composition for each ion to be obtained. Several strategies have been used to display the data in bidimensional plots such as Kendrick25 or van Krevelen diagrams.26 In order to provide evidence of differences between samples from data obtained by IMMS, a specific approach should also be developed owing to the specificities of the obtained data. In this work, we present a new workflow for a qualitative comparison of complex mixtures using ASAP-IMMS analysis. Hydrodesulfurization of gas oil is the process investigated to demonstrate the proof of principle. This system was chosen because it was fully investigated by other conventional methods. A feedback about the catalytic process is provided quickly due to the fast analysis of samples and an efficient data processing based on a comprehensive comparison of m/z vs drift time plots.

■

RESULTS AND DISCUSSION Standard PASH Compounds. ASAP ionization is known to produce ions by both proton and charge transfer.2 To determine which mechanism occurs for the ionization of PASH compounds under our experimental conditions, we analyzed BT, DBT, 4,6-diMeDBT, and BDBT spiked in the processed diesel fuel. The major ion observed for each standard PASH compound was the M+• molecular ion obtained through the charge transfer process as shown in Figure 1a. Tandem mass spectrometry experiments were carried out on each M+• ion. Collisional activation was performed after the ion mobility separation, i.e., in the transfer cell. The traveling wave velocity influences significantly the separation of the isobaric ions. By increasing the T-wave velocity from 650 to 1700 m/s, Figure 1b, the 4,6-diMeDBT ion (m/z 212.0655, drift time 5.11 ms) has been separated by ion mobility from the matrix ion. As activation was performed after the ion mobility separation, the drift time alignment of the precursor ion and fragment ions occurs.28 By extracting the mass spectrum at the drift time of the 4,6diMeDBT ion (i.e., 5.11 ms), we can observe that loss of H• (m/z 211.0589) is the most abundant loss observed from this ion. Informative fragmentations could be detected which are the loss of methyl radical (m/z 197.0415) from the precursor

■

EXPERIMENTAL SECTION Benzothiophene (BT), dibenzothiophene (DBT), 4,6-dimethyldibenzothiophene (4,6-diMeDBT), and benzo[a]dibenzothiophene (BDBT) were purchased from Sigma Aldrich (St. Louis, MO). The unprocessed diesel fuel (i.e., HDS unit feed) and processed diesel fuel (i.e., desulfurized diesel fuel) were provided by the European Research and Technical Center of TOTAL Refining and Chemistry (Rogerville, France) and have been analyzed without any sample preparation. BT, DBT, 4,6-diMeDBT, and BDBT have been analyzed by being dissolved in the processed diesel fuel at a concentration of B

dx.doi.org/10.1021/ac400731d | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

To identify BT and DBT ion series in the m/z−drift time plot of unprocessed diesel fuel, experimental m/z values of extracted peaks were processed as a Kendrick plot.25 Using standard PASH compounds, we previously showed that M+• ions of BT and DBT were formed using the ASAP probe. Theoretical Kendrick Mass Defect (KMD) of BT ion (C8H6S+•) and DBT ion (C12H8S+•) as well as their alkyl substituent should be, respectively, 131 and 171. By filtering even m/z ions and appropriate KMD ranges, ion series with KMD close to those expected were highlighted in unprocessed diesel fuel, Figure 2a and Table S2, Supporting Information.

Figure 1. ASAP-IMMS(/MS) experiment of 4,6-diMeDBT (m/z 212.0655) spiked in processed diesel fuel and displaying an isobaric signal at m/z 212.1534. (a) Partial mass spectrum with matrix ions shown as dashed lines. (b) Evolution of the ion mobility separation depending on the T-wave velocity, overlay of mobilograms for a Twave velocity set at 652 m/s (red) and 1700 m/s (blue). (c) CID spectrum of 4,6-diMeDBT M+• (precursor ion is highlighted by ∗).

ion, the neutral loss of sulfur atom (m/z 179.0778), and the neutral loss of CS (m/z 167.0864) both from m/z 211.0589 (Figure 1c). This observation is consistent with results obtained previously using APCI.29 For each standard PASH compound, the specific neutral losses of CS or S were observed (Figure S1, Table S1, Supporting Information). Unprocessed and Processed Diesel Fuel ASAP-IMMS Experiments. ASAP-IMMS mass spectra as well as m/z−drift time plots of unprocessed and processed diesel fuels exhibit approximately the same ion distribution (Figure S2, Supporting Information). The use of a peak detection algorithm on the m/ z−drift time plots highlighted 688 peaks (320 even m/z peaks) for unprocessed diesel fuel and 553 peaks (260 even m/z peaks) for processed diesel fuel. This indicates that the chemical modification of the diesel fuel during the HDS process could be, at least partially, detected by ASAP-IMMS experiments. Since the main goal of the HDS process is to convert sulfur compounds present in the feed, we focused on the detection of BT and DBT derivative compounds, which are usually the most abundant PASHs in raw diesel fuel.12 It should be noted that BT, DBT, and their alkyl substituent present significant reactivity differences during the HDS process. BT and their alkyl substituent present a faster reaction rate compared to DBTs. In addition, it is known that, among the different isomers of alkyl-DBTs, those with an alkyl group in the vicinity of the sulfur (such as 4,6-diMeDBT) are the most difficult to desulfurize.12

Figure 2. Detection of BT and DBT ions series in the unprocessed diesel fuel using data processing: (a) Partial Kendrick plot of even m/z ions. (b) Partial m/z−drift time plot of even m/z ions.

Superimposition of each series on m/z−drift time plot showed a regular distribution of BT and DBT ions along the plot, Figure 2b. Such behavior is expected, as ions of the same series tend to follow a trend line in the m/z−drift time plot.4 IMMS analysis allowed an effective separation of BT and DBT ion series based on aromatic ring number. DBT ions have a lower mobility (i.e., lower collision cross-section, more compact structure) than BT ions with the same carbon number (see extracted ion IMS spectra presented in Figure S3 and Table S2, Supporting Information). Unprocessed Diesel Fuel ASAP-IMMS/MS Experiments. To confirm identification of BT and DBT ion series, tandem mass spectrometry experiments were performed. Precursor ions were selected in the quadrupole, separated in the IM cell, and finally collisionally activated in the transfer cell. As in Figure 1, this sequence leads to a “mobility aligned product-ion spectra” (Figure 3a,c).30 The loss of H• was the major fragmentation pathway observed for low mass alkylated PASHs (Figure 3b,d). CID spectra showed also the specific loss of a sulfur atom, which clearly confirms identification of sulfur C

dx.doi.org/10.1021/ac400731d | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Comparison of m/z−Drift Time Plots of Unprocessed and Processed Diesel Fuel. Comparison of unprocessed and processed diesel fuel m/z−drift time plots was carried out using HDMS Compare. This software was developed specifically for the comparison of 2D plots obtained by IMMS. The main differences between each 2D plot are highlighted in different colors, Figure 4. The species present mainly in the unprocessed

Figure 3. ASAP-IMMS/MS experiments on PASH ions detected in unprocessed diesel fuel. (a) m/z−drift time plot of isobaric ions at m/z 212 for a T-wave velocity set at 1700 m/s. (b) CID spectrum of C14H12S+• ion (precursor ion is highlighted by ∗). (c) m/z−drift time plot of isobaric ions m/z 162 for a T-wave velocity set at 1700 m/s. (d) CID spectrum of C10H10S+• ion (precursor ion is highlighted by ∗).

containing ion series (Figures S4 and S5, Supporting Information). This specific fragmentation pathway was more difficult to observe for high mass alkylated PASHs. However, because members of an ion series differ only by alkyl groups, we assume that assignment of the first members of such a series is sufficient to identify all members. The presence of identified BT, DBT, and their alkyl derivatives in unprocessed diesel fuel was confirmed by a GC-SCD experiment, Table S4, Supporting Information. It should be noted that we did not find experimental conditions that allowed ion mobility separation of the isomers of alkyl-BT and alkyl-DBT compounds. Such isomers are partially resolved using GC-SCD.31 However, ASAP-IMMS methodology allowed us to overcome the coelution of high mass alkyl-BT and low mass alkyl-DBT compounds generally observed by GC-SCD.32 When comparing results obtained by GC-SCD and IMMS (Table S4, Supporting Information), it can be noticed that the relative abundance are very similar for the DBTs although some differences can be noticed for the low the mass Cx-BT (x ≤ 3). The relative abundance of low mass alkyl-BT detected by ASAP-IMMS is indeed significantly lower than in GC-SCD. This is probably due to ionization and/or transmission discrimination.

Figure 4. Global difference image using HDMS Compare for processed diesel fuel versus unprocessed diesel fuel. Pink indicates ions of higher intensity in processed diesel fuel; green indicates ions with higher intensity in unprocessed diesel fuel.

sample are displayed in green whereas the species present mainly in the processed diesel fuel are displayed in pink. The areas in white indicate no significant difference, and the color intensity is related to the differences between the two plots according to the color chart as shown in Figure 4. Extraction of mass spectra from these highlighted areas allows one to identify these specific ions. The green areas correspond to alkyl BT and DBT ion series present in the unprocessed diesel fuel. The mass spectra corresponding to the m/z 145−201 and 1.86− 2.83 ms range display clearly ion series corresponding to alkylBTs and ion series corresponding to alkyl-DBTs (Figure S5, Supporting Information). The pink areas in Figure 4 correspond to intense series of ions mainly present in the processed diesel fuel. Mobility of ions highlighted in the processed diesel fuel disclosed more compact structures than alkyl-BT or alkyl-DBT ions. Accurate mass measurements yielded elemental compositions that were consistent with a [M − H]+ ions series of hydrogenated derivatives compounds such D

dx.doi.org/10.1021/ac400731d | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

network, CNRS, and the European Regional Development Fund (ERDF) for financial support.

as cyclohexylbenzene. These species are expected to be produced by partial hydrogenation of alkyl-DBTs during the HDS process33 as shown in Scheme 1 (Figure S6 and Table S5, Supporting Information). This attribution was supported by results obtained by the LC-UV experiment (Table S6, Supporting Information), which have shown the increasing of monoaromatic compounds mass percentage in the sample after the HDS process whereas the mass percentage of diaromatic compounds decreased. The [M − H]+ ions are most likely related to M+• molecular ions that lost a hydrogen atom as shown previously.34 Experiments were repeated several times showing that (i) the ions distribution is very reproducible for sample analyzed in the same experimental conditions and (ii) the increase of intensity of ions highlighted after the HDS process is very specific to the processed diesel fuel (Figure S7, Supporting Information).

■

■

CONCLUSIONS Alkyl-BT and alkyl-DBT ions were detected in a complex mixture thanks to the high peak capacity of the ion mobility/ high resolution mass spectrometry combination. Separation of isobaric ions in the IM cell allowed one to identify alkyl-BT and alkyl-DBT ions that were characterized using MS/MS experiments. The m/z−drift time plots comparison algorithm highlighted clearly alkyl-BT and alkyl-DBT presence in the unprocessed diesel fuel as well as significant intensity differences in the processed diesel fuel which could be due to ions of hydrogenated derivative compounds produced by the catalytic hydrodesulfurization process. We demonstrated that ASAP-IMMS is a fast analytical tool for identification or control of specific compounds in complex mixtures. We believe that this workflow can be applied for comparison of other systems involving small molecules in metabolomics or lipidomics approaches.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional mass spectra, tables, and details concerning IMMS data processing. This material is available free of charge via the Internet at http://pubs.acs.org.

■

REFERENCES

(1) McEwen, C. N.; McKay, R. G.; Larsen, B. S. Anal. Chem. 2005, 77, 7826−7831. (2) McEwen, C. N. In Encyclopedia of Analytical Chemistry; Meyers, R. A., Ed.; Wiley: New York, 2010. (3) Ahmed, A.; Cho, Y. J.; No, M.-h.; Koh, J.; Tomczyk, N.; Giles, K.; Yoo, J. S.; Kim, S. Anal. Chem. 2011, 83, 77−83. (4) Woods, A. S.; Ugarov, M.; Egan, T.; Koomen, J.; Gillig, K. J.; Fuhrer, K.; Gonin, M.; Schultz, J. A. Anal. Chem. 2004, 76, 2187− 2195. (5) Zhong, Y.; Hyung, S.-J.; Ruotolo, B. T. Analyst 2011, 136, 3534− 3541. (6) Pan, H.; Lundin, G. Eur. J. Mass Spectrom. 2011, 17, 217−225. (7) Barrère, C.; Maire, F.; Afonso, C.; Giusti, P. Anal. Chem. 2012, 84, 9349−9354. (8) Maire, F.; Coadou, G.; Cravello, L.; Lange, C. M. J. Am. Soc. Mass Spectrom. 2013, 24, 238−248. (9) Knudsen, K. G.; Cooper, B. H.; Topsoe, H. Appl. Catal., A 1999, 189, 205−215. (10) Choudhary, T. V.; Parrott, S.; Johnson, B. Environ. Sci. Technol. 2008, 42, 1944−1947. (11) Pawelec, B.; Navarro, R. M.; Campos-Martin, J. M.; Fierro, J. L. G. Catal. Sci. Technol. 2011, 1, 23−42. (12) Stanislaus, A.; Marafi, A.; Rana, M. S. Catal. Today 2010, 153, 1−68. (13) Girgis, M. J.; Gates, B. C. Ind. Eng. Chem. Res. 1991, 30, 2021− 2058. (14) Daage, M.; Chianelli, R. R. J. Catal. 1994, 149, 414−427. (15) Gates, B. C.; Topsoe, H. Polyhedron 1997, 16, 3213−3217. (16) Mochida, I.; Choi, K.-H. J. Jpn. Pet. Inst. 2004, 47, 145−163. (17) Nishioka, M. Energy Fuels 1988, 2, 214−219. (18) Becker, G.; Nilsson, U.; Colmsjo, A.; Ostman, C. J. Chromatogr., A 1998, 826, 57−66. (19) Song, C.; Ma, X. Appl. Catal., B: Environ. 2003, 41, 207−238. (20) Andari, M. K.; Abu-Seedo, F.; Stanislaus, A.; Qabazard, H. M. Fuel 1996, 75, 1664−1670. (21) Landau, M. V. Catal. Today 1997, 36, 393−429. (22) Rudzinski, W. E.; Rai, V. Energy Fuels 2005, 19, 1611−1618. (23) Hughey, C. A.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G. Energy Fuels 2001, 15, 1186−1193. (24) Muller, H.; Adam, F. M.; Panda, S. K.; Al-Jawad, H. H.; Al-Hajji, A. A. J. Am. Soc. Mass Spectrom. 2012, 23, 806−815. (25) Hughey, C. A.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G.; Qian, K. Anal. Chem. 2001, 73, 4676−4681. (26) Stanford, L. A.; Kim, S.; Rodgers, R. P.; Marshall, A. G. Energy Fuels 2006, 20, 1664−1673. (27) Giles, K.; Williams, J. P.; Campuzano, I. Rapid Commun. Mass Spectrom. 2011, 25, 1559−1566. (28) Revesz, A.; Schroeder, D.; Rokob, T. A.; Havlik, M.; Dolensky, B. Angew. Chem., Int. Ed. 2011, 50, 2401−2404, S2401/2401−S2401/ 2415. (29) Herrera, L. C.; Ramaley, L.; Grossert, J. S. Rapid Commun. Mass Spectrom. 2009, 23, 571−579. (30) Damen, C. W. N.; Chen, W.; Chakraborty, A. B.; van Oosterhout, M.; Mazzeo, J. R.; Gebler, J. C.; Schellens, J. H. M.; Rosing, H.; Beijnen, J. H. J. Am. Soc. Mass Spectrom. 2009, 20, 2021− 2033. (31) Lopez, G. C.; Becchi, M.; Grenier-Loustalot, M. F.; Paiesse, O.; Szymanski, R. Anal. Chem. 2002, 74, 3849−3857. (32) Wang, F. C.-Y.; Robbins, W. K.; Di, S. F. P.; McElroy, F. C. J. Chromatogr. Sci. 2003, 41, 519−523. (33) Yang, H.; Fairbridge, C.; Ring, Z. Energy Fuels 2003, 17, 387− 398. (34) Yang, Z.; Attygalle, A. B. J. Am. Soc. Mass Spectrom. 2011, 22, 1395−1402.

AUTHOR INFORMATION

Corresponding Author

*Address: Université de Rouen/IRCOF/UMR 6014 CNRS, 1 Rue Tesnière, 76821 Mont-Saint-Aignan, France. Phone: +33 2 35 52 29 40. E-mail: [email protected]. Present Address ⊥

Leiden Academic Centre for Drug Research (LACDR), Leiden University, Einsteinweg 55, Leiden, The Netherlands.

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 This work was supported by the European Regional Development Fund (ERDF) N°31708, Région Haute Normandie, Total Raffinage Marketing: RUC342. The authors thank TOTAL, the Région Haute-Normandie, the Crunch’ research E

dx.doi.org/10.1021/ac400731d | Anal. Chem. XXXX, XXX, XXX−XXX