Introduction

Chemical ionization using proton-transfer-reactions with hydronium ions (H3O+) is widely used for trace gas detection in air. Both proton-transfer-reaction mass spectrometry (PTR-MS) [1] and selected-ion flow-tube mass spectrometry (SIFT-MS) [2] instruments use H3O+ ions for ionization and allow online measurements of a wide range of trace gases in air with high sensitivity, fast time response, and without the need for pre-concentration and sample treatment. These techniques have proven useful in a variety of research applications including atmospheric chemistry, plant studies, food science, and medical applications [3,4,5,6].

In PTR-MS, a discharge provides a bright source of H3O+ ions, and the proton-transfer reactions take place at enhanced collision energies to avoid cluster ion formation. Recent instrument developments have improved the sensitivity of PTR-MS and the resolving power of the mass analyzers used [5,6,7,8], vastly increasing the number of VOCs that can be detected in different air samples.

In SIFT-MS, H3O+ ions produced in the source are mass selected by a quadrupole mass filter and react in a flow tube at lower collision energies. Cluster ion formation is avoided by diluting the sample gas in helium. Due to ion losses caused by the additional mass filter in between the source and the flow reactor, the reagent ion signal and thus the sensitivity of SIFT-MS is less than that of PTR-MS. In contrast, the ion source produces no unwanted reagent ions, which makes the mass spectra easier to interpret, and the lower collision energies reduce product-ion fragmentation compared with PTR-MS. The reagent ion employed in SIFT-MS can also be switched rapidly, allowing for analysis of a single sample by multiple reagent ions.

The large number of compounds that can be detected by PTR-MS and SIFT-MS also provides a challenge to the quantification of trace gases. Calibrations using diluted standard mixtures or other methods have typically been used to determine instrument sensitivity. However, accurate calibrations for hundreds to thousands of trace gases are very time consuming and alternative methods need to be considered. In principle, instrument sensitivity can be calculated from the reaction rate coefficients and the collision conditions in the reactor [4]. Rate coefficients have not been measured for all trace gases; however, and in PTR-MS, they need to be known at the elevated collision energies used in the reactor. Rate coefficients can be estimated from the polarizability and dipole moment of the trace gas [9]. The latter are also not known for all trace gases and can be further estimated from the molecular mass and the presence of specific functional groups empirically [10]. In addition, product-ion fragmentation is a large uncertainty in calculating sensitivity and has not been systematically studied.

There is an extensive and growing body of literature on the detection of different trace gases by PTR-MS and SIFT-MS using proton-transfer reactions with H3O+ ions. In this work, we have collected the existing information in an online, publicly available library. The library allows users of PTR-MS and SIFT-MS to look up at which m/z’s trace gas of interest is detected and, vice versa, which trace gas may have been responsible for a detected product ion. The library developed in this work can also serve as a dataset to study instrument sensitivity and product-ion fragmentation using, for example, linear regression and machine learning techniques. To illustrate this potential, a few analyses of product-ion fragmentation are presented here. Further growth of the library is anticipated, and users are encouraged to submit their peer-reviewed data for this purpose.

Methods

An online, publicly viewable Google Sheet (www.tinyurl.com/PTRLibrary) was created using data from 49 publications [1, 4, 11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58] and currently contains nearly 1000 entries of different trace gases detected by PTR-MS and SIFT-MS. The library as it exists at the time of this publication has been included as electronic supplementary material of this paper. The information contained in the sheet consists of the following:

  1. 1.

    Compound identifiers: The protonated ion mass [M+H]+ is used as a leading identifier of trace gases even though in some cases it is not detected in PTR-MS. It is the intended detection m/z in PTR-MS, and the difference between protonated ion mass and detection m/z allows fragmentation to be studied systematically. Compound name, elemental composition, and CAS number are also included as compound identifiers.

  2. 2.

    Product ion m/zs: Detection m/zs and signal fractions on each of these m/zs were obtained from calibration and/or measurements in which a gas chromatographic pre-separation method was used [51]. In some cases, only the detection m/z is included in a publication, in which case, the signal fraction is omitted from the library. All detection m/z in the library are calculated as exact masses and account for the charge state of the ion to facilitate use of the library with high-resolution mass spectrometers. Because isotopic abundances of the elements are well documented, entries in the library are limited to the most abundant monoisotopic mass of each ion. For example, bromoacetone is detected by PTR-MS with no product-ion fragmentation [51], and the library entry lists m/z 137 (C3H5BrOH+) as the only detected m/z with an intensity of 100%, corresponding to a monoisotopic composition of 1H, 12C, 79Br, and 16O. For the library entries where [M+H3O]+ is a reported detection m/z, we make no attempt to discern the mechanism by which the association with H2O occurred.

  3. 3.

    Instrument details and reference: The electric field strength divided by the number density of the gas inside the reactor (E/N) determines the collision energy of ions in the reactor and therefore the degree of fragmentation and cluster formation. We include E/N, instrument type, and reactor pressure for library entries. Instrument calibration factors are presented in units of normalized counts per second per ppb (ncps ppbv−1), where the signal of the protonated trace gas has been normalized per million counts of H3O+ reagent ion. While the settings of electrostatic lenses can affect product-ion fragmentation and the flow of water vapor can impact water cluster formation, these parameters are not well documented and are not included in the library. The composition of drift gas in SIFT-MS measurements is included under the “Comments” column. The publication used to construct each entry is listed under the “Reference” column, with links to the publication included in the “doi” column. We envision the library as a tool for researchers to find the original publications, and our hope is that researchers will continue to cite the publications found through the library.

  4. 4.

    Physical molecular properties: Polarizabilities and dipole moments were taken from the CRC Handbook of Chemistry and Physics where available [59], and otherwise from the reference listed in the library. Proton affinities are taken from a series of reviews [60]. When reported, measured proton-transfer reaction rate coefficients are included in the library, and we have also followed the procedure of Sekimoto et al., which is based on that of Su, to estimate proton-transfer reaction rate coefficients from each compound’s polarizability and dipole moment [10, 61].

  5. 5.

    Chemical molecular properties: SMILES strings and number of functional groups are included in anticipation of studies by linear regression or machine learning to predict sensitivity and product-ion fragmentation patterns from the dataset.

No attempt was made to synthesize information about the same VOC from different studies, with each study contributing a separate library entry for a given compound. Similarly, entries with incomplete information were also included. Our philosophy is that the combined information can serve as a dataset for further analysis, and any attempts to combine multiple studies of a single compound into a single library entry would only reduce the amount of information available.

Product-Ion Fragmentation

A summary of product ion fragmentation is presented in Fig. 1. While the majority of the detected m/zs are at the parent m/z [M+H]+ expected from proton transfer reactions, there are many detected fragments. As illustrated by the lines in Fig. 1, the library demonstrates that there are both common fragmentation patterns, such as neutral loss of water and common detected ions, such as C2H3O+. The library also facilitates systematic exploration of fragmentation patterns of compounds with different functional groups, as demonstrated for oxygenated organic compounds in Fig. 2. The ketones included in the library show virtually no fragmentation, alcohols and aldehydes exhibit neutral loss of water and alcohols, acids lose water and formic acid, and esters lose both alkenes and alcohols.

Figure 1
figure 1

Detected ion m/zs for each parent compound in the PTR library, shaded by relative contribution to total signal. The most abundant ions are shaded black. Patterns emerge for common neutral losses as lines with slope of 1 and for common detected fragments as horizontal lines. A selection of these fragmentation patterns are labeled

Full size image
Figure 2
figure 2

Neutral losses for oxygenates included in the PTR library for different functional groups, presented as the fraction of total signal detected at each neutral loss for all compounds in the library with the given functional group

Full size image

To summarize the effect product-ion fragmentation can have on bulk properties of a PTR-MS spectrum, we have created histograms of the carbon number, O:C ratio, H:C ratio, and number of ring plus double bond equivalents (DBE) for the compounds in the PTR library and for the product ions at which those compounds are detected. These results are presented in Fig. 3. If no product-ion fragmentation occurred in PTR-MS, the bulk properties of the parent compounds and the product ions would be identical. For entries in the library, product-ion fragmentation decreases the average carbon number from 6.27 to 5.62, decreases the O:C ratio from 0.181 to 0.162, decreases the H:C ratio from 1.86 to 1.76, and increases the average DBE from 1.85 to 2.03. One proton from each product ion was assumed to be the result of the proton transfer reaction, and that proton is excluded from the calculations of H:C and DBE. In general, product-ion fragmentation involves carbon loss, oxygen loss, and an increase in the extent of unsaturation of an ion. This is consistent with the many observations in the library of oxygenates undergoing neutral loss of water or saturated alcohols, and researchers should account for these losses when relating the bulk properties of a PTR-MS spectrum to a VOC sample. While the results presented in Figs. 1–3 are broad, general observations about product-ion fragmentation, our hope, is that this library will enable more systematic studies of this process and other aspects of proton transfer reactions and that libraries of additional reagent ion chemistries will be developed in the future.

Figure 3
figure 3

Histograms comparing the bulk properties of parent compounds in the PTR library to the product ions of those compounds. Product-ion fragmentation documented in the PTR library leads to a decrease in average carbon number, O:C ratio, and H:C ratio and an increase in average ring plus double bond equivalents (DBE)

Full size image