Letter pubs.acs.org/ac
Establishing Atmospheric Pressure Chemical Ionization Efficiency Scale Riin Rebane,* Anneli Kruve, Piia Liigand, Jaanus Liigand, Koit Herodes, and Ivo Leito Institute of Chemistry, University of Tartu, Ravila 14a, 50411, Tartu, Estonia S Supporting Information *
ABSTRACT: Recent evidence has shown that the atmospheric pressure chemical ionization (APCI) mechanism can be more complex than generally assumed. In order to better understand the processes in the APCI source, for the first time, an ionization efficiency scale for an APCI source has been created. The scale spans over 5 logIE (were IE is ionization efficiency) units and includes 40 compounds with a wide range of chemical and physical properties. The results of the experiments show that for most of the compounds the ionization efficiency order in the APCI source is surprisingly similar to that in the ESI source. Most of the compounds that are best ionized in the APCI source are not small volatile molecules. Large tetraalkylammonium cations are a prominent example. At the same time, low-polarity hydrocarbons pyrene and anthracene are ionized in the APCI source but not in the ESI source. These results strongly imply that in APCI several ionization mechanisms operate in parallel and a mechanism not relying on evaporation of neutral molecules from droplets has significantly higher influence than commonly assumed.
A
pheric pressure chemical ionization efficiencies (IEs) of a diverse set of molecules measured under identical conditions would be of great help in learning more about the APCI mechanism and helping in deciding about the most suitable ionization method for a particular application. Because of less frequent usage, different practical aspects of APCI have been less investigated than for ESI. In particular, ESI efficiency scales have been constructed in the positive ion6,7 as well as negative ion mode.8 In addition, sodium adduct formation in ESI has been researched.9 Comparisons of ESI and APCI have been done10 for different analytes, but IE scales of similar nature are not available for APCI. Such scales would be very valuable for better understanding of the processes in the APCI source. In this paper, we present an ionization efficiency scale for the APCI source in APCI+ mode under controlled experimental conditions containing compounds from different families such as pyridines, aromatic, aliphatic, and heterocyclic amines, tetraalkylammonium salts, and others, including some compounds not ionized by ESI. Several instruments were used for cross-checking the results. The general ionization efficiency trends are compared to the ESI.
tmospheric pressure chemical ionization (APCI) is currently the second most used ionization mode (after electrospray ionization, ESI) in mass spectrometry (MS) for liquid chromatography (LC). In the APCI source reagent ion plasma is maintained by a corona discharge between the sharp tip of a needle and the spray chamber serving as the counter electrode. Differently from ESI, APCI does not need ionization of the analyte in solution1−3 and the general assumption about the main mechanism of APCI is that ionization takes place in the gas phase and the analyte molecules are vaporized before ionization.4 Thereby, APCI has an advantage over ESI in that it can actively generate ions from neutrals and therefore is more suitable for low- to medium-polarity analytes.1 In positive ion APCI (APCI+), the nitrogen molecules are usually ionized first; they ionize solvent molecules, with the eventual formation of protonated solvent molecules (or clusters),3 which in turn transfer the proton to gas-phase analyte molecules. On the basis of these considerations, it seems obvious that, in order to produce ions in APCI+ with high efficiency, the analyte should be volatile and have high gasphase basicity. This does not always hold, however, and it has been observed that compounds of very low volatility and mediocre basicity can be ionized very well in APCI; at the same time, low molecular weight compounds with high gas-phase basicity can have poor ionization efficiency. It has been suggested that besides the gas-phase proton transfer also solution-phase processes are important in APCI and mixed ionization mode can be envisaged.5 This implies that the APCI process can be complex and should be investigated further. Quantitative data on atmos© XXXX American Chemical Society
■
EXPERIMENTAL SECTION The structures of the analytes together with their origin are presented in Table S1. The solvent was prepared from Received: December 22, 2015 Accepted: March 4, 2016
A
DOI: 10.1021/acs.analchem.5b04852 Anal. Chem. XXXX, XXX, XXX−XXX
Letter
Analytical Chemistry Table 1. LogIE Values for ESI7 and APCI
ultrapure water (Millipore Milli-Q Advantage A10), methanol (HPLC grade, Sigma-Aldrich, Germany), and formic acid (reagent grade, Sigma-Aldrich, Germany). The MS responses were recorded in a flow injection mode as previously done by Kruve et al.8 For every compound, the response versus concentration slope was measured. Converting these slopes to absolute ionization efficiency (IE) is complicated. Thus, the ionization efficiencies were expressed in relative terms, using pyrene as the reference compound. Thus, the IE of any compound M1 was calculated relative to pyrene, according to the following equation: RIE(M1/pyrene) =
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 19 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
IE([M1 + H]+ ) slope([M1 + H]+ ) + = IE([pyrene + H] ) slope([pyrene + H]+ )
(1)
where the slope of the analyte signal versus concentration was estimated via linear regression in the linear range of the signalconcentration plot. The measurements were carried out during 3 months, making 3−7 replicate measurements per compound. For better presentation of the data, the logarithmic scale (logIE) is used. Information about the experimental conditions and COSMO-RS calculations can be found in Table S2.
■
RESULTS AND DISCUSSION Choice of Analytes for the APCI Scale. Altogether, 40 analytes were used to construct the APCI efficiency scale (Tables 1 and S-1). In comparison to the ESI scale of refs 7 and 20, new analytes were added to the APCI scale in order to encompass a wide range of chemical properties (pKa ranging from −18 to 14 and logP ranging from −1.3 to 7.4) and ionization behaviors. Some analytes ionize well in ESI, and some are not expected to ionize in ESI at all. The APCI IE scale is expected to be strongly beneficial for developing derivatization reagents for the APCI source. Therefore, compounds related to derivatization such as derivatization reagents themselves, as well as amino acid derivatives of diethyl ethoxymethylenemalonate (DEEMM) and 9-fluorenylmethyloxycarbonyl chloride (Fmoc-Cl), were used in the scale. APCI is not affected much by small variations in the source parameters, e.g., nebulizing gas pressure and needle current, as reported in the literature4 and also confirmed by our observations. Therefore, default ionization conditions of the instrument were used, as with the earlier ESI IE scale.7 The main change made compared to ref 7 was using methanol instead of acetonitrile, since it is more common for APCI applications.2,3,11−15 Formic acid was used as additive, since inclusion of formic acid increases the abundance of the protonated form of basic analytes in both ESI and APCI.3 Constructing the APCI Efficiency Scale. The constructed IE scale is presented in Table 1. For the construction of the scale, the signal of the most abundant ion was used. In most cases, this was the protonated form of the analyte [M + H]+ (see Table S-1 for details). In APCI, some molecules can ionize as molecular ions, M·+,2 but this was not observed for the compounds in this study. All logIE values are presented relative to pyrene, chosen as the reference compound with the assigned logIE value 3.41 (see below), and range from −0.62 up to 4.32, i.e., spanning almost 5 orders of magnitude. The day-to-day reproducibility standard deviation of the logIE values did not exceed in most cases 0.35 logIE units, and the pooled standard deviation over all assigned logIE measurements was 0.27 logIE units.
tetrabutylammonium [2,4,6-(MeO)3C6H2]3P tetrahexylammonium 4-CF3-PhP1(pyrr) centralite I triphenylguanidine triphenylamine dimethyl phthalate diphenylamine tributylamine triphenylphosphine acridine 2,2′-bipyridine pyrene tetraethylammonium sudan I PheFmoc 1-naphthylamine GlyFmoc tetramethylguanidine anthtracene diphenyl phthalate 3-aminopyridine triethylamine aniline 2-methylpyridine DEEMM DBEMM PheDEEMM arginine GlyDEEMM pyridine diethylamine EBEMM phenylbenzoate azelaic acid methylbenzoate sulphanylamide benzoic acid myristic acid
logIE ESI7
logIE APCI
5.13
4.32 4.30 4.26 4.22 4.02 3.96 3.94 3.71 3.69 3.68 3.51 3.50 3.48 3.41 3.40 3.39 3.28 3.27 3.12 3.01 2.93 2.92 2.82 2.81 2.75 2.61 2.57 2.51 2.47 2.47 2.42 2.38 2.02 1.99 1.79 1.24 0.96 0.84 −0.47 −0.62
5.65 5.55
3.67 3.54 4.18 4.83 4.42
3.95
4.04 3.89 4.1 3.53 3.04 3.02
2.94 2.66 2.44 0 1.22
Flow Rate Suitability. The previous ESI IE measurements were done at flow rates of 8.3 μL/min7,9 or 0.2 mL/min.8 Generally, APCI is considered more suitable for higher flow rates,16 but there is no clear correlation between flow rates and ionization efficiency.17,18 Three flow rates were compared in this work: 0.05, 0.2, and 0.6 mL/min (measurements were carried out on the same day). Results on 12 different analytes (pyrene, tetrahexylammonium, 4-CF3-PhP1(pyrr), triphenylamine, diphenyl phthalate, dimethyl phthalate, triethylamine, diphenylamine, aniline, pyridine, diethylamine, and acridine) showed that, even though the absolute values of the ionization efficiencies vary, the order of the compounds remains the same for all flow rates and correlations between data obtained at different flow rates are R2 > 0.8, which is good considering the standard deviation of logIE values. A flow rate of 0.2 mL/min was used. Measurements with Different APCI Sources and Mass Spectrometers. In addition to the measurements carried out on the Agilent XCT ion trap instrument, two triple quadrupole B
DOI: 10.1021/acs.analchem.5b04852 Anal. Chem. XXXX, XXX, XXX−XXX
Letter
Analytical Chemistry
triphenylamine, dimethylphtalate, and methylbenzoate. The between-day standard deviations of their individual APCI logIE values were 0.17, 0.11, and 0.22 log units, respectively. However, triphenylamine and dimethylphtalate were always better ionized than, e.g., 1-naphthylamine or acridine, and methyl benzoate was always better ionized than benzoic acid, meaning that the relative shift of these three compounds within the APCI scale is with high probability caused by different ionization properties of the compounds, not by statistical fluctuations. Besides the correlation, some similar trends in IE are seen within compound families. In both ion sources, increasing the length of alkyl chains increases the ionization efficiency of the compounds. Differently from the ESI scale7 where the IE increases in the order of aniline, triphenylamine, and diphenylamine, in APCI, the IE order is aniline, diphenylamine, and triphenylamine where triphenylamine beats diphenylamine by 0.25 units. This IE order is exactly the opposite of the traditional understanding of the APCI: aniline has the lowest logIE, while being the smallest and most basic of the three! The comparison of ESI and APCI IE scales shows that there are compounds that are ionized in ESI and APCI and similar trends are observed; additionally, there are compounds that can be only ionized with APCI. Correlation of logIE Values to Molecular Properties. It has been shown that there is a relation between the ESI IE and the pKa value, hydrophobicity, surface activity, etc.3,21 As presented in the previous section, where comparison is available, the APCI IE scale is quite similar to the ESI IE scale. Nevertheless, differently from ESI, we did not find predictively useful correlations between APCI IE and molecular parameters with neither a multi- nor single-parameter approach. PLS analysis has been previously used to find parameters influencing ESI/MS response. We used all the physicochemical parameters calculated for cations as input parameters in a PLS model (Table S2) and logIE APCI values (Table 1). It was observed that several parameters had a very similar impact in the model. The first group of parameters having a similar influence was formed of ionization degree (at the used pH) and pKa values in different solvents. The second group contained weighted average negative sigma (WANS)22 and logP values; the third group contained molecular area and volume, and the fourth group contained hydrogen bonding acceptor and donor properties. The polarity and polarizability parameter did not group with other parameters and had loadings with different sign and magnitude. To avoid overparametrization of the model, only one parameter (the one with the highest absolute loading in PLS) from each group was used in the next step. The first four components of the model explain 84% of the parameters variability and 56% of the ionization efficiency variability (Table S3). However, it was observed visually that the fit was poorer for the compounds with lower ionization efficiency. To reveal the mechanism of APCI, the loadings of the PLS model can be analyzed (Table S3). The first PLS component explains the largest variation in physicochemical parameters. The model coefficients revealed that ionization degree, WANS, logP, molar volume, and polarizability parameter had a positive correlation with the ionization efficiency. This means that large polarizable and hydrophobic species already ionized in solution to give ions with delocalized charge (WANS) have high ionization efficiency in APCI. On the other hand, compounds having strong hydrogen bonding acceptor capacity have lower
(QqQ) instruments with different APCI sources were tested on a subset of 20 compounds: Agilent 6490 and Varian J-320. The correlation between the logIE values obtained with the Agilent 6490 and XCT was good (R2 = 0.84). Both, scan and selected ion monitoring mode, were used and gave consistent results. Correlation reveals the poor performance of the QqQ instrument with molecules of small m/z (lower than 100), which strongly deviate from the correlation. This poor performance at low m/z values is also addressed in the manual of the Agilent 6490 instrument.19 Therefore, aniline (m/z 94), pyridine (m/z 80), 2-methylpyridine (m/z 94), and diethylamine (m/z 74) were left out of the analysis. The order of the compounds remained the same in most cases. Differently from ion trap, for triple quadrupole, the highest IE was observed for centralite I and tetramethylguanidine has lower IE than on the ion trap. The IE scale measured on the Varian instrument was compressed as compared to the one obtained on the XCT instrument (approximately by 1.5 times). Possible reasons could be low flow rate (20 μL/min) and the specifics of the Varian instrument, which have led to compressed results also in earlier ESI IE measurements.20 These problems are also evidenced by the correlation coefficient R2 = 0.63. Overall, the agreement between the data from different instruments was considered acceptable, given the quite high reproducibility of the standard deviation of the measurements. For all further data treatment, the XCT instrument data were used. Comparison of ESI and APCI. In order to elucidate the relation between the ionization efficiencies in APCI and ESI sources, the APCI IE scale and ESI IE scale7 have 20 common compounds. In order to eliminate the effect of possible bias caused by different measurement times (the ESI scale measurements were carried out more than 5 years ago), the ESI measurements with four compounds (tetrahexylammonium, diphenyl phthalate, diphenylamine, and pyridine) were repeated so that both ESI and APCI absolute ion signal intensity measurements for these four compounds were carried out on the same day (as described in the Supporting Information). Taking into account the concentrations and signal intensities in both ESI and APCI, the absolute logarithmic difference (of the signal/concentration ratios of the four compounds) between the ESI and APCI scales was obtained. On the basis of these four measured compounds, the optimum value for the whole difference between the scales was determined using a least-squares approach. The average for common compounds of different day measurements showed that the logIE values of the same compounds on the APCI scale are on an average lower than on the ESI scale by 3.41 log units. The resulting comparison of the two scales is seen in Table 1. Most analytes have statistically lower IE in APCI, as confirmed by t-test. New compounds at the top of the APCI scale are centralite I, [2,4,6-(MeO)3C6H2]3P, triphenylphosphine, and triphenylguanidine. All four are expected to have good IE in ESI as well. The lower part of the scale includes compounds such as benzoic, myristic, and azelaic acid. These are expected to have poor ionization in the positive ion mode ESI. In addition, the APCI scale has some compounds, which can not be ionized in the ESI source, first of all the hydrocarbons pyrene and anthracene, which have very low polarity and are devoid of any protonation centers. In comparison to the ESI scale, three analytes show distinctly different behavior: C
DOI: 10.1021/acs.analchem.5b04852 Anal. Chem. XXXX, XXX, XXX−XXX
Letter
Analytical Chemistry
APCI than commonly assumed and the mixed ionization mechanism can be envisaged, whereby a mechanism similar to ESI dominates for some compounds. This scale can be used as an aid to choose between the ESI and APCI sources. In absolute terms, somewhat better ionization efficiency is observed in ESI, meaning that, for compounds suitable for ESI and APCI, ESI might be a more suitable choice in terms of detection limits. Additionally, APCI can be used for compounds, which are not suitable for ESI.
ionization efficiencies. Similar trends have been previously observed for ESI efficiency scales as well. It was of interest if the used compounds can be grouped by their suitability for the different ion sources. By using the same variables (all physicochemical parameters calculated for cations as in Table S2) as in PLS, a PCA analysis was conducted. By using 3 principal components, 2 groups of analytes were observed (Figure 1). By the chemical nature of the analytes in
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b04852. Detailed materials and methods as well as tables with exact logIE values and other additional information as noted in text. (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*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.
■
Figure 1. PCA analysis of the physicochemical properties of the compound set. The compounds were divided into “ESI compounds” (black dots) and “APCI compounds” (red dots) based on their structures.
ACKNOWLEDGMENTS This work was supported by the Estonian Ministry of Education and Research through the institutional research funding IUT14-20 (TLOKT14014I) and personal research funding PUT34.
the groups, they could be entitled as “APCI compounds” and “ESI compounds”, respectively (Table S-1). Under “ESI compounds”, we mean compounds that are expected to become charged inside the ESI droplet and ejected from the droplet due to charge−charge repulsion. On the other hand, we term analytes “APCI compounds” that are not expected to become markedly ionized in the droplets and are rather expected to become ionized in the gas phase. On the basis of physicochemical parameters, PCA revealed a large difference between the compounds. It was of interest whether ESI compounds ionize better than APCI compounds, but this was not the case. This could also explain the poor variability of ionization efficiency with PLS.
■
REFERENCES
(1) Gross, J. H. Mass Spectrometry, 2nd ed.; Springer: Heidelberg, 2011. (2) Fredenhagen, A.; Kühnöl, J. J. Mass Spectrom. 2014, 49, 727−736. (3) Kostiainen, R.; Kauppila, T. J. J. Chromatogr. A 2009, 1216, 685− 699. (4) Terrier, P.; Desmazières, B.; Tortajada, J.; Buchmann, W. Mass Spectrom. Rev. 2011, 30, 854−874. (5) Niessen, W. M. A. J. Chromatogr. A 1999, 856, 179−197. (6) Leito, I.; Herodes, K.; Huopolainen, M.; Virro, K.; Künnapas, A.; Kruve, A.; Tanner, R. Rapid Commun. Mass Spectrom. 2008, 22, 379− 384. (7) Oss, M.; Kruve, A.; Herodes, K.; Leito, I. Anal. Chem. 2010, 82, 2865−2872. (8) Kruve, A.; Kaupmees, K.; Liigand, J.; Leito, I. Anal. Chem. 2014, 86, 4822−4830. (9) Kruve, A.; Kaupmees, K.; Liigand, J.; Oss, M.; Leito, I. J. Mass Spectrom. 2013, 48, 695−702. (10) Thurman, E. M.; Ferrer, I.; Barceló, D. Anal. Chem. 2001, 73, 5441−5449. (11) Giorgianni, F.; Cappiello, A.; Beranova-Giorgianni, S.; Palma, P.; Trufelli, H.; Desiderio, D. M. Anal. Chem. 2004, 76, 7028−7038. (12) de Rijke, E.; Zappey, H.; Ariese, F.; Gooijer, C.; Brinkman, U.A.Th. J. Chromatogr. A 2003, 984, 45−58. (13) Steen, R. J. C. A.; Hogenboom, A. C.; Leonards, P. E. G.; Peerboom, R. A. L.; Cofino, W. P.; Brinkman, U.A.Th. J. Chromatogr. A 1999, 857, 157−166. (14) Henriksen, T.; Juhler, R. K.; Svensmark, B.; Cech, N. B. J. Am. Soc. Mass Spectrom. 2005, 16, 446−455.
■
CONCLUSIONS The first self-consistent ionization efficiency scale for APCI has been created, spanning for 5 logIE units and including 40 compounds with a wide range of chemical and physical properties: with a wide range of molecular masses; compounds with strong basic properties as well as compounds with weak basic properties; compounds that are hydrophilic as well as hydrophobic; compounds with low polarity as well as high polarity. The ionization efficiencies of different compounds in APCI and ESI are compared. Results show that there is more similarity than differences in the ionization mechanisms in APCI and ESI. As a rule, compounds that have high ionization efficiency in ESI have high ionization efficiency also in APCI, regardless of their molecule size or volatility. This could mean that solution-phase processes are much more important in D
DOI: 10.1021/acs.analchem.5b04852 Anal. Chem. XXXX, XXX, XXX−XXX
Letter
Analytical Chemistry (15) Huffman, B. A.; Poltash, M. L.; Hughey, C. A. Anal. Chem. 2012, 84, 9942−9950. (16) Garcia-Ac, A.; Segura, P. A.; Viglino, L.; Gagnon, C.; Sauvé, S. J. Mass Spectrom. 2011, 46, 383−390. (17) Asperger, A.; Efer, J.; Koal, T.; Engewald, W. J. Chromatogr. A 2001, 937, 65−72. (18) Herrera, L. C.; Grossert, J. S.; Ramaley, L. J. Am. Soc. Mass Spectrom. 2008, 19, 1926−1941. (19) Agilent 6400 Series Triple Quadrupole LC/MS System, Concepts Guide, Revision A; Agilent Technologies Inc.: Santa Clara, 2012. (20) Liigand, J.; Kruve, A.; Liigand, P.; Laaniste, A.; Girod, M.; Antoine, R.; Leito, I. J. Am. Soc. Mass Spectrom. 2015, 26, 1923−1930. (21) Cech, N. B.; Enke, C. G. Mass Spectrom. Rev. 2001, 20, 362− 387. (22) Kaupmees, K.; Kaljurand, I.; Leito, I. J. Solution Chem. 2014, 43, 1270−1281.
E
DOI: 10.1021/acs.analchem.5b04852 Anal. Chem. XXXX, XXX, XXX−XXX
