Understanding the atmospheric pressure ionization of petroleumcomponents: The effects of size, structure, and presence of heteroatoms

Anna Katarina Huba a, Kristina Huba a, Piero R. Gardinali a,b,⁎a Department of Chemistry & Biochemistry, Florida International University, 3000 NE 151 Street, Biscayne Bay Campus, North Miami, Florida 33181, USAb Southeast Environmental Research Center (SERC), Florida International University, 3000 NE 151 Street, Biscayne Bay Campus, North Miami, Florida 33181, USA

H I G H L I G H T S

• Crude oil model compounds were ana-lyzed using HRMS and ESI, APPI, andAPCI sources.

• Molar intensities were used to evaluateionization efficiencies.

• A lack of ionization of alkanes was ob-served in all three sources.

• Size, structure, methylation, and het-eroatom content affected efficiencies.

• Overall, APPI was shown to provide themost favorable results.

G R A P H I C A L A B S T R A C T

a b s t r a c ta r t i c l e i n f o

Article history:Received 2 March 2016Received in revised form 7 June 2016Accepted 7 June 2016Available online 28 June 2016

Editor: D. Barcelo

Understanding the composition of crude oil and its changes with weathering is essential when assessing its pro-venience, fate, and toxicity. High-resolutionmass spectrometry (HRMS) has provided the opportunity to addressthe complexity of crude oil by assigningmolecular formulae, and sorting compounds into “classes” based on het-eroatom content. However, factors such as suppression effects and discrimination towards certain componentsseverely limit a truly comprehensivemass spectrometric characterization, and, despite the availability of increas-ingly better mass spectrometers, a complete characterization of oil still represents a major challenge. In order tofully comprehend the significance of class abundances, aswell as the nature and identity of compounds detected,a good understanding of the ionization efficiency of the various compound classes is indispensable. The currentstudy, therefore, analyzedmodel compounds typically found in crude oils by high-resolution mass spectrometrywith atmospheric pressure photoionization (APPI), atmospheric pressure chemical ionization (APCI), andelectrospray ionization (ESI), in order to provide a better understanding of benefits and drawbacks of eachsource. The findings indicate that, overall, APPI provides the best results, being able to ionize the broadestrange of compounds, providing the best results with respect to ionization efficiencies, and exhibiting the leastsuppression effects. However, just like in the other two sources, in APPI several factors have shown to affectthe ionization efficiency of petroleummodel compounds. Themain such factor is the presence or absence of func-tional groups that can be easily protonated/deprotonated, in addition to other factors such as size, methylationlevel, presence of heteroatoms, and ring structure. Overall, this study evidences the intrinsic limitations and

Keywords:APPIESIAPCIPetroleumModel compoundsHigh-resolution mass spectrometry

Science of the Total Environment 568 (2016) 1018–1025

⁎ Corresponding author at: Florida International University, Chemistry & Biochemistry and SERC, 11200 SW 8th Street, 33199 Miami, Florida, USA

http://dx.doi.org/10.1016/j.scitotenv.2016.06.0440048-9697/© 2016 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

benefits of each of the three sources, and should provide the fundamental knowledge required to expand thepower of crude oil analysis by high-resolution mass spectrometry.

© 2016 Elsevier B.V. All rights reserved.

1. Introduction

Despite current advances in alternative resources, petroleum is stillvital for the production of energy and as precursor for variousmaterials,such as plastics and medicines. (Speight, 2014) Petroleum dischargesduring production and transport add to what is already released intothe environment through natural seepage; since the fate and potentialeffect of this released crude oil is highly dependent on its composition,(Wang et al., 2013) both the characterization of crude oil and its evolu-tion with weathering are therefore of uttermost importance. In recentyears, unmasking the complexity of crude oil has beenmostly addressedthanks to the availability and development of high-resolution massspectrometry (HRMS). The high-resolution achieved by ion cyclotronresonance (ICR) or Orbitrap mass spectrometers provides the powerof separation needed for such a complex matrix, and allows for the as-signment of unequivocal molecular formulae (e.g., CcHhNnOoSs), whichcan then be sorted into “classes” based on heteroatom content (e.g.,NnOoSs).(Bae et al., 2010; Koolen et al., 2015; Ray et al., 2014) Themost commonly used ionization sources for crude oil studies by HRMSinclude electrospray ionization (ESI), (Bae et al., 2010; McKenna et al.,2013; Ray et al., 2014; Ruddy et al., 2014) atmospheric pressure photo-ionization (APPI), (Bae et al., 2010; Koolen et al., 2015; McKenna et al.,2013; Purcell et al., 2006; Ruddy et al., 2014) and atmospheric pressurechemical ionization (APCI). (Hsu et al., 2000; Panda et al., 2009; Qian etal., 2001; Rudzinski and Rai, 2005) Other sources used include atmo-spheric pressure laser ionization (APLI)(Gaspar et al., 2012; Panda etal., 2011) as well as laser desorption ionization (LDI). (Cho et al.,2013; Cho et al., 2012) Each of these ionization sources works throughspecific mechanisms, which are explained in detail elsewhere (Bruins,1991; Covey et al., 2009; deHoffman and Stroobant, 2007; Kauppila etal., 2002; Kauppila et al., 2015; Kebarle and Tang, 1993; Kebarle andVerkerk, 2009; Konermann et al., 2013; Raffaelli and Saba, 2003;Wilm, 2011), has its advantages and disadvantages, and is thus goingto be most useful for particular compound species. Overall, it is clearthat a comprehensive oil characterization, thus, most likely requiresthe use of a combination of these complementary sources, orfinding op-erating conditions that will allow at least marginal ionization of mostcomponents in the oil. The coupling of liquid and gas chromatographyto atmospheric pressure ionization sources and HRMS has been previ-ously reported and has shown some advantages. (Barrow et al., 2014;Lababidi et al., 2013; Schwemer et al., 2015) Nonetheless, separatingthe thousands of compounds in a crude oil is limited by the chromato-graphic resolution and cannot be achieved efficiently, and orthogonalseparations are limited by boiling point (GC) and/or functionality (LC).Therefore, a comprehensive crude oil characterization frequently relieson infusion analysis, which provides no chromatographic separation,and relies on the power of the ultra-high mass spectrometric resolutionto separate and assign the compounds. (Aeppli et al., 2012; Hall et al.,2013; McKenna et al., 2013; Ruddy et al., 2014). Even with direct infu-sion analysis, however, several factors (such as ion suppression effectsand discrimination with respect to specific structural features) affectthe ionization of petroleum components, and thus severely limit atruly comprehensive mass spectrometric characterization regardless ofthe operation or resolution of the instrument. (Huba and Gardinali,2016; Panda et al., 2009; Teräväinen et al., 2007) Hence, a comprehen-sive, detailed, semi-quantitative oil characterization still represents amajor challenge.

When obtaining mass spectra through high-resolution mass spec-trometry coupled to one of the previously mentioned sources, thou-sands of distinct peaks can be detected. For example, Fourier

transform ion-cyclotron mass spectrometry (FT-ICR MS) coupled to anESI source has been shown to produce more than 30,000 distinctpeaks,(Bae et al., 2010; McKenna et al., 2013) and as many as 50,000peaks have been reported for an Arabian light crude oil.(Schaub et al.,2008) The main benefit of high-resolution analysis is the capability toassign a unique elemental composition (CcHhNnOoSs) to each one ofthe peaks that are generated. The product of such an analysis is, there-fore, an enormous amount of elemental composition data, whichneeds to be carefully visualized, or grouped by using restrictions andstatistical methods, in order to make the data analysis manageable.Plots showing the relative abundance of the different compound “clas-ses” (e.g., the “O1” class comprising all molecular formulae containingone oxygen atom, the “O2” class all molecular formulae containingtwo oxygen atoms, etc.) are often used to obtain a general idea of thetype of compounds present in the oil. However, the relative abundanceof the classes is inevitably dependent on the overall ionizability of eachof the compounds making up that specific class. This means that theability of ionizing singly oxygenated compounds of multiple functional-ities (such as alcohols, phenols, or ketones)will affect their relative con-tributionwithin the “O1” class, but alsowith respect to, for example, the“O2” class whose compounds (e.g., carboxylic acids and quinones) mayalso have different ionization efficiencies. Therefore, a comparison ofthe abundances of the “classes” is intrinsically biased due to ionizationsource limitations. Moreover, some compound types might not be ion-ized at all based on the specific source being used. It becomes veryclear that a thorough and comprehensive understanding of the ioniza-tion potential and efficiency of the key compound types expected tobe present in crude oil will enormously benefit the interpretation ofhigh-resolution mass spectrometric data, both by providing a weighingscale for specific functionalities, and also by identifying “silent” areas ofthe spectra. This will aid in evaluating how close the spectral represen-tation of the sample is to its true composition, and to fully comprehendthe significance of class abundances.

The present study, therefore, focused on analyzing a series of modelcompounds by direct infusion high-resolution mass spectrometry withAPPI, APCI and ESI sources. Ionization efficiencies were evaluated bycomparing molar intensities, and were used to evaluate the role ofsize, polarity, and heteroatom contribution towards ionization ease.The relative formation of radical and protonated ionic species (whereapplicable), as well as the extent of ion suppression were investigated.

2. Materials and methods

2.1. Preparation of standards

Seven separate mixtures, as well as several single-compound stan-dards, were prepared and tested in all ionization modes (APPI, APCI,and ESI). Detailed composition of the standard mixtures is provided inthe Supplementary information section. All initial standard solutionswere in dichloromethane (DCM), and were then reconstituted to50:50methanol/toluene and spiked with 1% formic acid or 1% ammoni-um hydroxide for positive and negative mode, respectively. The sol-vents used were all Optima LC/MS grade, purchased from FisherScientific (Fair Lawn, NJ, USA).

2.2. High-resolution mass spectrometric analysis

Analysiswas carried out on a Q Exactive Orbitrap (Thermo Scientific,NJ, USA) by direct infusion through a 500 μL syringe (Thermo Scientific,NJ, USA) at a typical flow rate of 30 μL/min. The APPI ionization source

1019A.K. Huba et al. / Science of the Total Environment 568 (2016) 1018–1025

(Thermo Scientific, NJ, USA) was equipped with a krypton UV gas dis-charge lamp (Syagen Technology, Inc., Tustin, CA) that produces 10.0and 10.6 eV photons (120 nm). For both positive and negative ioniza-tion mode, N2 sheath gas at 40 psi was used to facilitate ionization,while the auxiliary port remained closed. Also, for bothmodes, the heat-ed vaporizer region was held at 350 °C, while the capillary temperaturewas set to 300 °C. The APCI source (Thermo Scientific, NJ, USA) param-eters for positive mode were a sheath gas at 10 psi, an auxiliary gas at7 psi, a capillary temperature of 350 °C, and a heated vaporizer regionat 400 °C. In negative mode the parameters were a sheath gas at32 psi, an auxiliary gas at 5 psi, a capillary temperature of 250 °C, anda heated vaporizer region at 450 °C. For both positive and negative ion-ization modes the discharge current was set to 4.00 μA. Finally, for theESI analysis, a heated electrospray (HESI) source (Thermo Scientific,NJ, USA) was used, and conditions for positive mode were a spray volt-age of 5.20 V, a heated vaporizer region at 300 °C, capillary temperatureof 300 °C, and sheath and auxiliary gas at 40 and 5 psi, respectively. Fornegative mode, the typical conditions were a spray voltage of 4.50 V, aheated vaporizer region at 300 °C, capillary temperature of 200 °C, andsheath and auxiliary gas at 35 and 30 psi, respectively. Each samplewas run in quadruplicate, and average and standard deviation valueswere calculated.

3. Results and discussion

The threemain atmospheric pressure ionization sources, namely ESI,APPI, and APCI, were used to analyze all standard mixtures andindividual compounds (since previous results showed only marginal

improvements when exploring the combination of APPI and APCI, thisspecific ionization method was not further explored in this study).Table 1 shows the primary different compound classes and whether ornot they were ionized in the three sources; since concentrations werekept consistent across the range of compounds, the differences ob-served are mostly due to ionization efficiency. It is very noticeable thatAPPI and APCI are able to ionize a much larger range of compound clas-seswhen compared to ESI.Moreover, when comparing APPI to APCI it isclear that even though most of the results overlap, APPI provides someadvantages over APCI as it enables the efficient ionization and detectionof two-ring PAHs, oxygen containing PAHs (such as dibenzofuran), andanhydrides. Overall, thus, APCI or ESI do not seem to provide any signif-icant advantages over the APPI ionization source as there is no com-pound class that can only be ionized in those sources, and one couldclaim that APPI is the most versatile atmospheric pressure ionizationsource for crude oil analysis. The ability of APPI (and to a lesser extentof APCI) to significantly expand the range of compounds that can be ion-ized (especially with respect to ESI) by being able to ionize compoundsthat are nonpolar and cannot be easily protonated/deprotonated, can beattributed to their ionizationmechanisms and has beenwidely reportedand explained.(deHoffman and Stroobant, 2007; Kauppila et al., 2015;Short et al., 2007) On the other hand, ESI, as expected, is limited to ion-izing acidic and basic compounds that easily lose or gain a hydrogenatom, respectively. Another noticeable feature illustrated in Table 1 isthe complete inability of any of the sources (even APPI) to ionize purealkanes; this can be attributed to their absence of either an aromaticring structure (which allows for the detection of PAHs), or of heteroat-om containing functional groups (i.e. lactones, anhydrides). The only

Table 1Ionization of themain compound classes in the three ionization sources (APPI, APCI, and ESI), in positive and negative ionizationmode. Compounds present at ≥1% relative abundance in aparticular source and mode are depicted with a checkmark.

Positive Negative

APPI APCI ESI APPI APCI ESI

Polycyclic aromatichydrocarbons (PAHs)

2 Ring PAH Naphthalene ✓ – – – – –3 Ring PAH Anthracene ✓ ✓ – – – –4 Ring PAH Chrysene ✓ ✓ – – – –5 Ring PAH Dibenzo(a,h)anthracene ✓ ✓ – – – –7 Ring PAH Hexaphenylbenzene ✓ ✓ – – – –Sulfur PAH Dibenzothiophene ✓ ✓ – – – –Oxygen PAH Dibenzofuran ✓ – – – – –Nitrogen PAH Pyridinic nitrogen: Dibenzo(a,h)acridine ✓ ✓ ✓ – – –

Pyrrolic nitrogen:7H-Dibenzo(c,g)carbazole

✓ ✓ ✓ ✓ ✓ ✓

Alkanes Alkanes Straight chain:Octadecane

– – – – – –

Cyclic:Decalin

– – – – – –

Functional group types Alcohol Aliphatic:Tetracosanol

– – – – – –

Aromatic:1-Pyrenemethanol

✓ ✓ ✓ – – –

Aldehyde Aliphatic:1-Octadecanal

– – – – – –

Aromatic:1-Pyrenecarbaldehyde

✓ ✓ ✓ – – –

Ketone Aliphatic:2-Nonadecanone

✓ ✓ ✓ – – –

Aromatic:1-Acetylpyrene

✓ ✓ ✓ ✓ – –

Carboxylic Acid Aliphatic:Stearic acid

– – – ✓ ✓ ✓

Aromatic:1-Pyrenecarboxylic acid

✓ ✓ – ✓ ✓ ✓

Phenol 4-Isopropylphenol – – – ✓ – ✓Lactone γ-Octalactone ✓ ✓ ✓ – – –Anhydride Phthalic anhydride ✓ – ✓ – – –

Polyoxygenated compounds Carboxylic Acid 2,6-Naphthalenedicarboxylic acid – – – ✓ ✓ ✓Ketone Anthraquinone ✓ ✓ ✓ ✓ ✓ –Alcohol 1,5-Dihydroxynaphthalene ✓ ✓ – ✓ ✓ ✓

1020 A.K. Huba et al. / Science of the Total Environment 568 (2016) 1018–1025

instance in which alkanes are ionized is when they present easily ioniz-able groups, such as a carboxylic acid (APPI, APCI, and ESI) or a ketone(APPI only). Even though the ionization of pure hydrocarbons was pre-viously achieved under very specific conditions (Tose et al., 2015), thelack of ionization of non-functionalized alkanes under common crudeoil analysis conditions (such as the ones employed in this study), evi-dences a significant limitation of crude oil characterization by atmo-spheric pressure ionization HRMS. This is especially noteworthy sincealkanes usually represent one of the most abundant compound classesof a typical oil. (Fingas, 2015)While Table 1 provides useful informationregarding the potential ionization, or lack thereof, of the main com-pound classes that one may expect to find in crude oil samples, it doesnot give any information on the relative ionization efficiencies betweenthe different compounds and ionization techniques.

In order to investigate the influence of structure on ionization effi-ciencies, the molar intensities of the ions were calculated (by dividingthe intensity by the molarity) and compared. Moreover, to provide abetter way of comparison, the relative molar intensities were then cal-culated by normalizing all values to the largest peak in a specific dataset. Fig. 1 illustrates the total ionization efficiencies (the sum of the rad-ical and protonated species) of the major compound classes that wereshown to be ionized in at least one of the three sources. The resultsfrom each ionization type and mode are shown normalized to thehighest abundance compound class in that particular source and mode.

Overall, from Fig. 1 it is clear that compounds prone to protonation ordeprotonation (such as pyrrolic andpyridinic nitrogens, ketones, and car-boxylic acids) have the highest ionization efficiencies irrespectively tothe source type. Besides that, it is obvious that the ionization efficienciesin the three ionization sources vary greatly based on compound type.While APPI and APCI show similar results with respect to their mostabundant classes (aromatic ketones and pyrrolic nitrogens in positiveand negative mode, respectively), the abundances of the other classesare highly variable, especially in negative mode. In APPI, for example,phenols represent the second most abundant class in negative mode,while this class is completely absent in APCI. In ESI on the other hand,the two classes that are preferentially ionized in positive and negativemode, respectively, are pyridinic nitrogens and aromatic carboxylicacids. As previously seen in Table 1, Fig. 1 also illustrates the ability ofAPPI to enable the ionization of species that are not easily protonated/deprotonated, such as PAHs and heteroatom containing PAHs. It alsoshows that APPI and APCI positive mode enable the efficient detectionof both the nitrogen containing compound classes (pyridinic and

pyrrolic), while ESI only marginally detects the pyrrolic species in posi-tive mode and would likely require a combination of the positive andnegative mode analysis. This advantage is in accordance with previouslyreported results obtained with the APPI ionization source. (Purcell et al.,2006; Purcell et al., 2007).

As mentioned before, Fig. 1 depicts the total ionization efficiencies,i.e., for compounds producing both radical and protonated ions the sumis calculated and plotted. In ESI the formation of the radical ion is very un-likely, and in APPI and APCI the respective ease to form radical or proton-ated ions depends on several factors, namely the composition of thesolvent and mobile phase, and the ionization energies and proton/elec-tron affinities of the solvents, mobile phase, and analytes.(deHoffmanand Stroobant, 2007) For example, in the presence of a dopant such astoluene, solvents like methanol or acetonitrile have shown to initiatethe formation of protonated ions.(Kauppila et al., 2002; Raffaelli andSaba, 2003) On the other hand, solvents which have low proton affinity(e.g., chloroform, hexane, and water) preferentially form radical ions bycharge exchange reactions. (Kauppila et al., 2002) Fig. 2a and bwere cre-ated in order to obtain amore detailed interpretation of the relative ratioof the radical versus protonated ion formation, and showcompound clas-ses that are relevant in APPI and APCI positivemode. The ratios of radicalversus protonated ions inAPPI andAPCI are very similar, and it is very ob-vious that some compound classes preferentially form radical ions, whileothers get protonated more easily. More specifically, heteroatom con-taining functional groups such as pyrrolic and pyridinic nitrogens, alde-hydes, ketones, and lactones preferentially form protonated ions, whilePAH species form mainly radical ions. However, for PAHs, it can be no-ticed that as the size of the PAH increases (by increasing the number ofrings) the formation of the protonated ionic species increases as well(e.g., protonated ions are not formed for a 2- or 3-ring PAH, while for a5-ring PAH there is a radical to protonated ion ratio of about 1). This re-sult is in accordancewith previous findings reporting that the protonatedion is often dominant for larger PAHs (Anacleto et al., 1995; Mansoori,1998; Marvin et al., 1999). In addition, one can also notice that largerPAHs show a higher relative molar intensity, and are thus ionized moreefficiently.

This trend is more clearly illustrated in Fig. 3a, which shows an in-crease in total ionization efficiency with PAH size in both APPI andAPCI. The statistical significance of the results was evaluated by aStudent's t-test and showed significant increases (p b 0.01) in ionizationas the ring size increases for all transitions (2–3, 3–4, 4–5 rings). The dif-ference in ionization energy can explain the higher relative molar

Fig. 1. Graph illustrating the total relative ionization efficiencies of the major compound classes that were ionized in at least one ionization source. Each compound was run inquadruplicate, all values portrayed represent averages, with the error bars representing the relative percent error.

1021A.K. Huba et al. / Science of the Total Environment 568 (2016) 1018–1025

intensity of 3-ring PAHs over 2-ring PAHs (8.1444± 0.001 eV for naph-thalene, (Lias, 2016) 7.439 ± 0.06 eV for anthracene, (Lias, 2016) and7.891 ± 0.001 eV for phenanthrene (Lias, 2016)), while the increasedformation of protonated cations can explain the remainder of thetrend (3–4, and 4–5 rings). One can also notice that, when normalized

to the highestmolar intensity peak (the 5-ring PAH), the relative ioniza-tion efficiency of smaller PAHs is better in APPI compared to APCI. Theability of APPI to better ionize small PAHs is in accordance with resultsobtained by Robb et al. showing a much better sensitivity for naphtha-lene in APPI with respect to APCI. (Robb et al., 2000) A separate trendwas found when looking at howmethylation affects the ionization effi-ciencies, and is illustrated in Fig. 3b. As can be seen, the relative molarintensity significantly (p b 0.01) increases as the level of methylationincreases for all transitions, with the exception of naphthalene tomethyl-naphthalene in the APCI source (both only marginally ion-ized). This correlates well and can be explained with the decreasingionization energies of methylated naphthalenes (8.1444 ± 0.001 eVfor naphthalene,(Lias, under review) 7.96 ± 0.03 eV for 1-methylnaphthalene,(Lias, 2016) and 7.78 ± 0.03 eV for 1.4-dimethyl naph-thalene (Nounou, 1966)). Further results regarding the methylationtrend with larger PAHs, however, suggest that this trend weakenswith the increase of PAH size, and eventually disappears for large(4-ring) PAHs. These same trends also hold true for heteroatom con-taining PAHs, such as the sulfur containing benzothiophene, and canbe observed in Fig. 4a and b.

Another interestingfinding thatwas obtained from the size distribu-tion study, was the varying nature in ionization efficiencies of specificcompounds with the same number of rings but different ring structure(i.e., isomeric compounds such as chrysene and benzo(a)anthracene).This result is shown in Fig. 5 and clearly points out how the ionizationefficiency of benzo(a)anthracene is significantly higher than the oneof chrysene (p b 0.05) in both ionization sources, but most prevalentlyin APPI. This trend can be explained, similarly to the methylationtrend, by looking at the ionization energies of the two compounds(7.60 ± 0.03 eV for chrysene,(Shahbaz et al., 1981) and 7.46 ±0.03 eV for benzo(a)anthracene(Akiyama et al., 1981)). An additionalinteresting aspect of ionization differences for isomeric compounds il-lustrated in Fig. 5, is the clear dissimilarity in the extent of protonationof the two compounds. While for benzo(a)anthracene the amount ofthe protonated ionic species is almost 50% of the total ions formed, forchrysene protonation is almost completely absent (b1% in APPI, andb4% in APCI). Since isomeric compounds often times have considerablydifferent physical, chemical, and toxicological properties,(Dabestaniand Ivanov, 1999) it is extremely important to use all available informa-tion (such as differences in ionization behavior) in order to interpret amixture of unknown compounds. Fig. 5 shows that, in this particularcase, the relative amount of protonated ion formation could suggestthe presence of one of the two isomers over the other.

So far, these results have shown that factors such as size, methyla-tion, and ring structure affect the ionization efficiencies of PAHs inAPPI as well as in APCI. An additional factor is shown in Fig. 6, whichtakes a closer look at the effect that heteroatoms have on the ionization

Fig. 2. Comparison of the average relative molar intensities of the radical versus theprotonated ion in (a) APPI and (b) APCI positive mode.

Fig. 3.Relative totalmolar intensities in APPI and APCI positivemode, showing (a) the effect of size (2 ring PAH=naphthalene, 3 ring PAH=mixture of anthracene/phenanthrene, 4 ringPAH = mixture of chrysene/benz(a)anthracene, and 5 ring PAH = dibenzo(a,h)anthracene), and (b) the effect of methylation on the ionization efficiencies of PAHs. Average molarintensities are depicted, with error bars showing the relative percent error.

1022 A.K. Huba et al. / Science of the Total Environment 568 (2016) 1018–1025

efficiency of PAHs. In order to exclude the contribution of any of the pre-viously described factors, four PAHs of equal size, methylation level, andring type (3 rings, 2 six membered rings and 1 five membered ring)were chosen and compared (i.e., fluorene, carbazole, dibenzofuran,and dibenzothiophene). As expected, carbazole is the compound thatis most efficiently ionized, and this can be attributed mainly to its easeto be protonated. This is also the only compound that is detected inESI since this technique relies on protonation and deprotonation. Allthe other PAHs (containing no heteroatoms, sulfur, and oxygen) donot have easily protonable groups and thus only significantly ionize bythe radical ionization mechanismwhich is less efficient and only possi-ble in APPI or APCI. When comparing APPI to APCI it can be seen thatAPCI provides a better ionization of carbazole (p b 0.01), but for allother compounds the ionization efficiency is significantly (p b 0.01) bet-ter for APPI. It can therefore be concluded from Fig. 6 that, overall, APPIprovides the best ionization efficiencies over the range of heteroatomcontaining compounds depicted in this study.

Lastly, this study aimed at illustrating how and to what extent ionsuppression effects (in the three atmospheric pressure ionizationsources here used) affect molar relative intensities. In order to do so, aproof of concept study with the same four PAHs (which covered arange of possible heteroatoms) shown in Fig. 6 were analyzed in indi-vidual solutions, additionally to the mixture containing about 50 otherPAHs. Fig. 7 illustrates the results that were obtained, and clearly

shows that significant suppression (p b 0.05) is present in APCI (forfluorene, dibenzofuran, and dibenzothiophene). APPI also shows somesuppression for fluorene (p b 0.01), while the ionization of carbazole(p b 0.01) is enhanced in the mixture with respect to the individual so-lution. The other two compounds in APPI do not show statistically dif-ferent ionization efficiencies whether present in a complex mixture ornot. Surprisingly, ESI does not show ion suppression with the exceptionof dibenzothiophene. One has to take into account, however, that theoverall ionization efficiencies of all compounds except carbazole, wasnegligible (b1%) in ESI when compared to the other two ionizationsources (see Fig. 6). Moreover, the mixture is solely comprised ofPAHs, which are compounds not ionized in ESI. The here obtained ESIresults were included to give a comprehensive picture of all the sources,and to show that, as expected, even though a complexmatrix was pres-ent the lack of ionization of the majority of compounds in such matrixlead to almost no suppression. In order to better demonstrate suppres-sion effects in ESI, a future study including a mixture containing com-pounds well ionized in ESI (such as pyridinic nitrogen compounds)should be performed. When comparing solely APCI and APPI, the latterprovides the more consistent ionization efficiencies and thus less sup-pression from thematrix. These, results are in agreementwith previousreports showing that APPI exhibits reduced suppression effects whencompared to both ESI and APCI.(Hanold et al., 2004; Short et al., 2007)Overall, it is evident that suppression does influence the ionization effi-ciencies, even for the least affected ionization method (APPI). This isclearly shown by the relative molar intensities of fluorene and carba-zole: while for the mixture the ionization efficiency for carbazole was

Fig. 4. Relative total molar intensities in APPI and APCI positive mode, showing (a) the effect of size (2 ring PAH= benzothiophene, 3 ring PAH= dibenzothiophene, and 4 ring PAH=naphthodibenzothiophene) and (b) the effect of methylation on the ionization efficiencies of a sulfur containing PAH. Averagemolar intensities are depicted, with error bars showing therelative percent error.

Fig. 5. Comparison of two 4-ring isomeric PAHs (benzo(a)anthracene and chrysene),showing significant differences in ionization efficiency and formation of the protonatedionic species (depicted by the shaded area). The compounds were both run inquadruplicate with the relative molar intensity representing the average value and theerror bars representing the relative percent error.

Fig. 6. Total relative molar intensities depicting the ionization efficiencies of heteroatomcontaining PAHs in the positive ionization mode. Relative molar intensities are anaverage of quadruplicate runs, and error bars represent the relative percent error.

1023A.K. Huba et al. / Science of the Total Environment 568 (2016) 1018–1025

significantly higher than the one for fluorene, for the individual solu-tions the two values were not statistically different. Thus, when refer-ring and comparing relative ionization efficiencies in comprehensivepetroleum studies by HRMS, which heavily rely on infusion data (withno prior chromatographic separation and thus particularly affected bysuppression effects), it is imperial to take the matrix into consideration.

Overall, the results here presented provide a fairly comprehensivepicture of the ionization behavior of major compound classes presentin crude oil under the specific conditions used in this analysis. Eventhough the APPI source seems to provide the best results (with thedata being concentration independent on a range from 0.2–2 ppm), sig-nificant limitationswith respect to the analysis of complex environmen-tal mixtures were shown. As previously mentioned, elementalcomposition data and the comparison of so called compound “classes”are fundamental tools used to interpret and compare oil composition.However, since this data is intrinsically biased due to limitations in theionization of the numerous different compound types present in thecrude oil, an understanding of such biases is fundamental in order toevaluate the significance of the results obtained. Several researchareas, such as understanding the compositional changes of crude oildue to weathering, rely on the accuracy of HRMS data to portray the“true” composition of the crude oil. (Huba and Gardinali, 2016; Ray etal., 2014; Ruddy et al., 2014) This is a very crucial step in understand-ing the ultimate fate and toxicity of oil released into the environ-ment. It is thus evident that this study provides knowledge that iscritical when interpreting crude oil characterization results byHRMS, and that the results here presented will benefit future re-search in this field.

4. Conclusion

The present study provides a comprehensive overview of the bene-fits and drawbacks of the three main atmospheric pressure ionizationsources (ESI, APPI, and APCI) with respect to the ionization of the prin-cipal compound classes expected to be found in crude oil. Moreover,computation ofmolar intensities allowed for semi-quantitative compar-isons of the relative ionization efficiencies. The overall complete lack ofionization of non-functionalized alkanes was a clear and significant lim-itation pertinent to all three sources. Out of the three sources, the resultsshowed that, if a comprehensive oil characterization is targeted, theAPPI source seems to provide the best results, by being able to ionizethe broadest range of compounds, as well as providing the best overallionization efficiencies, and less ionization suppression with respect toAPCI. ESI, on the other hand, showed severe limitations, as the amountof different compound classes that are ionized is significantly lowercompared to both APPI and APCI. This study, moreover, showed thatthe ionization efficiency is influenced by several factors: the presenceof easily protonated or deprotonated functional groups (primary fac-tor), the size, the methylation level, the presence/absence/type of het-eroatoms, the isomeric structure, and the presence/absence of acomplex matrix.

These results are critical information needed in order to interpretHRMS oil characterization results, and additionally provide the knowl-edge needed to aid in the selection of a specific ionization source withrespect to a compound type of interest. Therefore, these findings,which can be further applied to other high-resolution mass analyzersbeyond the Orbitrap, in addition to future studies expanding the range

Fig. 7.Comparison of ionization efficiencies of four PAHs analyzed inquadruplicate in individual solutions and in a PAHmixture, illustrating the extent of ionization suppression in (a) APPI,(b) APCI, and (c) ESI. Relative molar intensities and error bars represent the average value and the relative percent error, respectively.

1024 A.K. Huba et al. / Science of the Total Environment 568 (2016) 1018–1025

of compound classes and dopants used, will provide the fundamentalunderstanding required to greatly expand the power of HRMS analysisof crude oils.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.scitotenv.2016.06.044.

Aknowledgments

The Environmental Analysis Research Laboratory acknowledges thesupport from the Thermo Scientific Corporation in the development ofthis work. AKH would like to thank Kathia Sandoval for helpful discus-sions. This is contribution number 795 from the Southeast Environmen-tal Research Center at Florida International University.

References

Aeppli, C., Carmichael, C.A., Nelson, R.K., Lemkau, K.L., Graham, W.M., Redmond, M.C., etal., 2012. Oil weathering after the deepwater horizon disaster led to the formationof oxygenated residues. Environ. Sci. Technol. 46, 8799–8807.

Akiyama, I., Harvey, R.G., LeBreton, P.R., 1981. Ultraviolet photoelectron studies ofmethyl-substituted benz(a)anthracenes. J. Am. Chem. Soc. 103, 6330.

Anacleto, J.F., Ramaley, L., Benoit, F.M., Boyd, R.K., Quilliam, M.A., 1995. Comparison of liq-uid chromatography/mass spectrometry interfaces for the analysis of polycyclic aro-matic compounds. Anal. Chem. 67, 4145–4154.

Bae, E., Na, J.-G., Chung, S.H., Kim, H.S., Kim, S., 2010. Identification of about 30,000 chem-ical components in shale oils by electrospray ionization (ESI) and atmospheric pres-sure photoionization (APPI) coupled with 15T Fourier transform ion cyclotronresonance mass spectrometry (FT-ICR MS) and a comparison to conventional oil. En-ergy Fuel 24, 2563–2569.

Barrow, M.P., Peru, K.M., Headley, J.V., 2014. An added dimension: GC atmospheric pres-sure chemical ionization FTICR MS and the Athabasca oil sands. Anal. Chem. 86,8281–8288.

Bruins, A.P., 1991. Mass spectrometrywith ion sources operating at atmospheric pressure.Mass Spectrom. Rev. 10, 53–77.

Cho, Y., Witt, M., Kim, Y.H., Kim, S., 2012. Characterization of crude oils at the molecularlevel by use of laser desorption ionization Fourier-transform ion cyclotron resonancemass spectrometry. Anal. Chem. 84, 8587–8594.

Cho, Y., Jin, J.M., Witt, M., Birdwell, J.E., Na, J.-G., Roh, N.-S., et al., 2013. Comparing laserdesorption ionization and atmospheric pressure photoionization coupled to Fouriertransform ion cyclotron resonance mass spectrometry to characterize shale oils atthe molecular level. Energy Fuel 27, 1830–1837.

Covey, T.R., Thomson, B.A., Schneider, B.B., 2009. Atmospheric pressure ion sources. MassSpectrom. Rev. 28, 870–897.

Dabestani, R., Ivanov, I.N., 1999. A compilation of physical, spectroscopic andPhotophysical properties of polycyclic aromatic hydrocarbons. Photochem. Photobiol.70, 10–34.

de Hoffman, E., Stroobant, V., 2007. Mass Spectrometry: Principles and Applications.Wiley, England.

Fingas, M.F., 2015. Handbook of Oil Spill Science and Technology. JohnWiley & Sons, Ho-boken, New Jersey.

Gaspar, A., Zellermann, E., Lababidi, S., Reece, J., Schrader, W., 2012. Impact of differentionization methods on the molecular assignments of asphaltenes by FT-ICR massspectrometry. Anal. Chem. 84, 5257–5267.

Hall, G.J., Frysinger, G.S., Aeppli, C., Carmichael, C.A., Gros, J., Lemkau, K.L., et al., 2013. Ox-ygenated weathering products of deepwater horizon oil come from surprising pre-cursors. Mar. Pollut. Bull. 75, 140–149.

Hanold, K.A., Fischer, S.M., Cormia, P.H., Miller, C.E., Syage, J.A., 2004. Atmospheric pres-sure photoionization. 1. General properties for LC/MS. Anal. Chem. 76, 2842–2851.

Hsu, C.S., Dechert, G.J., Robbins, W.K., Fukuda, E.K., 2000. Naphthenic acids in crude oilscharacterized by mass spectrometry. Energy Fuel 14, 217–223.

Huba, A.K., Gardinali, P.R., 2016. Characterization of a crude oil weathering series by ultra-high-resolution mass spectrometry using multiple ionization modes. Sci. Total Envi-ron. 563-564, 600–610.

Kauppila, T.J., Kuuranne, T., Meurer, E.C., Eberlin, M.N., Kotiaho, T., Kostiainen, R., 2002. At-mospheric pressure photoionization mass spectrometry. Ionization mechanism andthe effect of solvent on the ionization of naphthalenes. Anal. Chem. 74, 5470–5479.

Kauppila, T.J., Syage, J.A., Benter, T., 2015. Recent developments in atmospheric pressurephotoionization-mass spectrometry. Mass Spectrom. Rev. 1–27.

Kebarle, P., Tang, L., 1993. From ions in solution to ions in the gas phase - the mechanismof electrospray mass spectrometry. Anal. Chem. 65, 972A–986A.

Kebarle, P., Verkerk, U.H., 2009. Electrospray: from ions in solution to ions in the gasphase, what we know now. Mass Spectrom. Rev. 28, 898–917.

Konermann, L., Ahadi, E., Rodriguez, A.D., Vahidi, S., 2013. Unraveling the mechanism ofelectrospray ionization. Anal. Chem. 85, 2–9.

Koolen, H.H.F., Swarthout, R.F., Nelson, R.K., Chen, H., Krajewski, L.C., Aeppli, C., et al.,2015. Unprecedented insights into the chemical complexity of coal tar from

comprehensive two-dimensional gas chromatography mass spectrometry and directinfusion Fourier transform ion cyclotron resonance mass spectrometry. Energy Fuel29, 641–648.

Lababidi, S., Panda, S.K., Andersson, J.T., Schrader, W., 2013. Direct coupling of normal-phase high-performance liquid chromatography to atmospheric pressure laser ioni-zation Fourier transform ion cyclotron resonance mass spectrometry for the charac-terization of crude oil. Anal. Chem. 85, 9478–9485.

Lias, S.G., Ion energetics data. In: Linstrom, P.J., Mallard, W.G. (Eds.), NIST ChemistryWebBook, NIST Standard Reference Database Number 69, (retrieved 2016)

Mansoori, B.A., 1998. Isomeric identification and quantification of polycyclic aromatic hy-drocarbons in environmental samples by liquid chromatography tandemmass spec-trometry using a high pressure quadrupole collision cell. Rapid Commun. MassSpectrom. 12, 712–728.

Marvin, C.H., Smith, R.W., Bryant, D.W., McCarry, B.E., 1999. Analysis of high-molecular-mass polycyclic aromatic hydrocarbons in environmental samples using liquid chro-matography–atmospheric pressure chemical ionization mass spectrometry.J. Chromatogr. A 863, 13–24.

McKenna, A.M., Nelson, R.K., Reddy, C.M., Savory, J.J., Kaiser, N.K., Fitzsimmons, J.E., et al.,2013. Expansion of the analytical window for oil spill characterization by ultrahighresolution mass spectrometry: beyond gas chromatography. Environ. Sci. Technol.47, 7530–7539.

Nounou, P., 1966. Etude des composes aromatiques par spectrometrie demasse I. Mesuredes potentials d'ionisation et d'apparition par la methode du potential retardateur etinterpretation des courbes d'ionisation differentielle. J. Chim. Phys. 63, 994.

Panda, S.K., Andersson, J.T., Schrader, W., 2009. Characterization of supercomplex crudeoil mixtures: what is really in there? Angew. Chem. 121, 1820–1823.

Panda, S.K., Brockmann, K.-J., Benter, T., Schrader, W., 2011. Atmospheric pressure laserionization (APLI) coupled with Fourier transform ion cyclotron resonance mass spec-trometry applied to petroleum samples analysis comparison with electrospray ioni-zation and atmospheric pressure photoionization methods. Rapid Commun. MassSpectrom. 25 (16), 2317–2326.

Purcell, J.M., Hendrickson, C.L., Rodgers, R.P., Marshall, A.G., 2006. Atmospheric pressurephotoionization Fourier transform ion cyclotron resonance mass spectrometry forcomplex mixture analysis. Anal. Chem. 78, 5906–5912.

Purcell, J.M., Rodgers, R.P., Hendrickson, C.L., Marshall, A.G., 2007. Speciation of nitrogencontaining aromatics by atmospheric pressure photoionization or electrospray ioni-zation Fourier transform ion cyclotron resonance mass spectrometry. J. Am. Soc.Mass Spectrom. 18, 1265–1273.

Qian, K., Rodgers, R.P., Hendrickson, C.L., Emmett, M.R., Marshall, A.G., 2001. Readingchemical fine print: resolution and identification of 3000 nitrogen-containing aro-matic compounds from a single electrospray ionization Fourier transform ion cyclo-tron resonance mass spectrum of heavy petroleum crude oil. Energy Fuel 15,492–498.

Raffaelli, A., Saba, A., 2003. Atmospheric pressure photoionization mass spectrometry.Mass Spectrom. Rev. 22, 318–331.

Ray, P.Z., Chen, H., Podgorski, D.C., McKenna, A.M., Tarr, M.A., 2014. Sunlight creates oxy-genated species in water-soluble fractions of deepwater horizon oil. J. Hazard. Mater.280, 636–643.

Robb, D.B., Covey, T.R., Bruins, A.P., 2000. Atmospheric pressure photoionization: an ion-ization method for liquid chromatography−mass spectrometry. Anal. Chem. 72,3653–3659.

Ruddy, B.M., Huettel, M., Kostka, J.E., Lobodin, V.V., Bythell, B.J., McKenna, A.M., et al.,2014. Targeted petroleomics: analytical investigation of Macondo well oil oxidationproducts from Pensacola beach. Energy Fuel 28, 4043–4050.

Rudzinski, W.E., Rai, V., 2005. Detection of polyaromatic sulfur heterocycles in crude oilusing postcolumn addition of tropylium and tandem mass spectrometry. EnergyFuel 19, 1611–1618.

Schaub, T.M., Hendrickson, C.L., Horning, S., Quinn, J.P., Senko, M.W., Marshall, A.G., 2008.High-performance mass spectrometry: Fourier transform ion cyclotron resonance at14.5 tesla. Anal. Chem. 80, 3985–3990.

Schwemer, T., Rüger, C.P., Sklorz, M., Zimmermann, R., 2015. Gas chromatographycoupled to atmospheric pressure chemical ionization FT-ICR mass spectrometry forimprovement of data reliability. Anal. Chem. 87, 11957–11961.

Shahbaz, M., Akiyama, I., LeBreton, P., 1981. Ultraviolet photoelectron studies of methylsubstituted chrysenes. Biochem. Biophys. Res. Commun. 103, 25.

Short, L.C., Cai, S.-S., Syage, J.A., 2007. APPI-MS: effects of mobile phases and VUV lampson the detection of PAH compounds. J. Am. Soc. Mass Spectrom. 18, 589–599.

Speight, J.G., 2014. The Chemistry and Technology of Petroleum. CRC Press.Teräväinen, M.J., Pakarinen, J.M.H., Wickström, K., Vainiotalo, P., 2007. Comparison of the

composition of Russian and north sea crude oils and their eight distillation fractionsstudied by negative-ion electrospray ionization Fourier transform ion cyclotron reso-nance mass spectrometry: the effect of suppression. Energy Fuel 21, 266–273.

Tose, L.V., Cardoso, F.M.R., Fleming, F.P., Vicente, M.A., Silva, S.R.C., Aquije, G.M.F.V., et al.,2015. Analyzes of hydrocarbons by atmosphere pressure chemical ionization FT-ICRmass spectrometry using isooctane as ionizing reagent. Fuel 153, 346–354.

Wang, C.Y., Chen, B., Zhang, B.Y., He, S.J., Zhao, M.M., 2013. Fingerprint and weatheringcharacteristics of crude oils after Dalian oil spill, China. Mar. Pollut. Bull. 71, 64–68.

Wilm, M., 2011. Principles of electrospray ionization. Mol. Cell. Proteomics 10(M111.009407).

1025A.K. Huba et al. / Science of the Total Environment 568 (2016) 1018–1025