Comparison of Atmospheric Pressure Chemical Ionization and Field

Oct 4, 2016 - E-mail: [email protected]., *Phone: (765) 494-0882. ... pressure chemical ionization mass spectrometry (APCI-MS) was compared to...
0 downloads 0 Views 692KB Size
Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Article

Comparison of Atmospheric Pressure Chemical Ionization and Field Ionization Mass Spectrometry for the Analysis of Large Saturated Hydrocarbons Chunfen Jin, Jyrki Viidanoja, Mingzhe Li, Yuyang Zhang, Elias Ikonen, Andrew Root, Mark Romanczyk, Jeremy M Manheim, Eric T. Dziekonski, and Hilkka I. Kenttamaa Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b02789 • Publication Date (Web): 04 Oct 2016 Downloaded from http://pubs.acs.org on October 4, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 10

1 2 3 4 5 6 7 8 9 10 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 59 60

Analytical Chemistry

10/4/2016

Comparison of Atmospheric Pressure Chemical Ionization and Field Ionization Mass Spectrometry for the Analysis of Large Saturated Hydrocarbons Chunfen Jina,b, Jyrki Viidanoja*,b, Mingzhe Lia, Yuyang Zhanga, Elias Ikonenb, Andrew Rootc, Mark Romanczyka, Jeremy Manheima, Eric Dziekonskia, Hilkka I. Kenttämaa*,a a

Department of Chemistry, Purdue University, West Lafayette, Indiana 47907, United States

b c

Technology Centre, Neste Corporation, P.O. Box 310, FI-06101 Porvoo, Finland

MagSol, Tuhkanummenkuja 2, FI-00970 Helsinki, Finland

ABSTRACT: Direct infusion atmospheric pressure chemical ionization mass spectrometry (APCI-MS) was compared to field ionization mass spectrometry (FI-MS) for the determination of hydrocarbon class distributions in lubricant base oils. When positive ion mode APCI with oxygen as the ion source gas was employed to ionize saturated hydrocarbon model compounds (M) in hexane, only stable [M-H]+ ions were produced. Ion-molecule reaction studies performed in a linear quadrupole ion trap suggested that fragment ions of ionized hexane can ionize saturated hydrocarbons via hydride abstraction with minimal fragmentation. Hence, APCI-MS shows potential as an alternative of FI-MS in lubricant base oil analysis. Indeed, the APCI-MS method gave similar average molecular weights and hydrocarbon class distributions as FI-MS for three lubricant base oils. However, the reproducibility of APCI-MS method was found to be substantially better than for FI-MS. The paraffinic content determined using the APCI-MS and FI-MS methods for the base oils was similar. The average number of carbons in paraffinic chains followed the same increasing trend from low viscosity to high viscosity base oils for the two methods.

An important task for the petroleum industry is to be able to design hydrocarbon refining processes with maximal cost efficiency, which includes the ability to use all the components of crude oil with minimal production of waste. In order to reach this target, knowledge on the size, structure, and amount of each type of compound in feeds and intermediate and end products is required.

nents. For the characterization of complex hydrocarbon mixtures, such as crude oil and its conversion products, an ionization/evaporation method is desired that yields a single type of an ionized molecule for different types of hydrocarbons in the sample and that causes minimal fragmentation. However, sensitive, reproducible and robust ionization methods capable of doing this are currently not available for the analysis of heavy saturated hydrocarbons.

Until now, 13C NMR spectroscopy and field ionization (FI) and field desorption (FD) mass spectrometry (MS) have been the most common techniques employed for characterization of heavy saturated hydrocarbon mixtures in petroleum industry (light hydrocarbons can be analyzed using GC/MS).1–4 However, 13C NMR spectroscopy can only provide bulk (average) information on these mixtures. In contrast, MS provides molecular level information, such as molecular weight and structural as well as quantitative information, on mixture compo-

In FI-MS, an intense electric field (108 V/cm) is generated on emitter carbon whiskers.2 The sample is introduced into the ion source by using a direct insertion probe (DIP) near the whiskers and heating is applied to gradually evaporate the sample. Positively charged molecular ions (M+●) are believed to be formed upon extraction of electrons from the molecules by quantum electron tunneling caused by the high electric field.5 However, the technique suffers from several limitations,

1

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 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 59 60

Page 2 of 10

10/4/2016 ions with little fragmentation. Both M+● and [M-H]+ ions were formed for cyclic saturated hydrocarbons, but only [M-H]+ ions for linear and branched saturated hydrocarbons.12 The employment of corona discharge ionization in conjunction with atmospheric solids analysis probe (ASAP) or atmospheric pressure gas chromatography (APGC) appears to produce analyte ions via nitrogen fixation and charge transfer accompanied by extensive fragmentation.10

including poor reproducibility, variable sensitivity, carry-over problems and occasional analyte ion fragmentation. FD method suffers from extensive fragmentation of the ionized analytes.6 Atmospheric pressure chemical ionization (APCI) is a general term that refers to corona discharge ionization followed by ion-molecule reactions at atmospheric pressure. APCI is showing increasing potential in mass spectrometric analysis of large saturated hydrocarbons.7-10 Exothermicity of the chemical ionization process can be tuned by selecting a suitable reagent ion for ionization of a given analyte. If an optimal reagent ion is available that ionizes analytes without depositing much internal energy into them, intact ionized molecules can be produced without fragmentation.

In all above studies, formation of multiple product ions upon ionization of each mixture component complicates the interpretation of the analysis results. Fragmentation can hinder accurate identification of the hydrocarbon types and determination of molecular weight distributions. We report here results obtained using an optimized APCI method and oxygen as ion source gas, which demonstrate that a sole ionized molecule, [M-H]+, is formed for linear, branched and cyclic saturated hydrocarbon model compounds in hexane solvent. The optimized APCI method is compared with FI as an alternative method for soft evaporation/ionization of pure saturated hydrocarbons and their mixtures as well as lubricant base oils prior to MS analysis. Lubricant base oils are essential for the automobile and machine industry. Crude oil derived lubricant base oils are composed of large paraffins and alkylnaphthenes.15-17 Desired properties of these oils include high boiling point, low friction, and high viscosity index. These properties are needed to provide resistance for corrosion and wearing, thermo-oxidative stability, as well as environmental friendliness.4,18-20 The ability to characterize these oils at the molecular level would help ensure that each oil shows optimal performance.

Different approaches have been studied for introduction of heavy hydrocarbon samples into an APCI source. These include direct infusion using a heated nebulizer, atmospheric pressure gas chromatography10 and laser-induced acoustic desorption (LIAD).11,12 Investigators have also tested different ion source gases, such as helium, nitrogen, synthetic air, and oxygen.12-14 The challenge in all this research has been the inability to produce a single type of an ionized molecule with minimal fragmentation for all mixture components. Marotta and coworkers have studied positive ion mode APCI of volatile hydrocarbons in synthetic air by using a triple quadrupole mass spectrometer.14 Like other investigators after them, they found that alkanes produce [M-H]+ ions upon APCI under these conditions. They concluded that the reaction producing these ions is hydride abstraction. In this research, various types of adduct ions and fragment ions were also observed for alkanes. Hourani and coworkers examined direct infusion positive ion mode APCI with N2 as the ion source gas and nheptane as the reagent.8 The technique was found to produce multiple types of ions for n-alkanes. Gao et al. examined N2, He and CO2 as ion source gases and employed volatile hydrocarbons as APCI reagents for the ionization of large saturated hydrocarbon model compounds.13 Many types of ionized molecules as well as fragment ions were observed. Positive ion mode LIAD-APCI was examined by Gao et al. using N2 as ion source gas for ionization of large saturated hydrocarbons with and without introducing volatile reagents into the ion source.11 In this study, less fragmentation was observed when a reagent was used as opposed to having no reagent. However, fragmentation was still a problem. In a subsequent study by Nyadong et al., positive ion mode LIAD-APCI-MS was used for the analysis of polyethylene polymers by using O2 as the ion source gas but no reagent.12 The technique was reported to yield predominantly molecular ions and pseudo-molecular

EXPERIMENTAL SECTION

Reagents and Materials. Butylcyclohexane (≥99%), cholestane (≥97%), squalane (≥98.5%), pentatriacontane (≥98%), C20-C40 n-alkane mixture (each 50 mg/l, all containing an even number of carbons) and n-hexane were purchased from Sigma-Aldrich and used as received. Three lubricant base oils were provided by NESTE Corporation, Finland. Ultrapure O2 (99.993%) was purchased from Praxair, Inc, and N2 was obtained from boil-off of a liquid nitrogen tank. All solutions of individual saturated hydrocarbon model compounds were prepared at 10 mM concentration in hexane while the base oil samples were prepared at 100 mg/ml and 50 mg/ml concentrations in hexane for APCI-MS and FI-MS experiments, respectively. C20-C40 n-alkane mixture was concentrated to achieve approximately ten times greater concentration (0.1-0.2 mM concentration for each component). Atmospheric Pressure Chemical Ionization (APCI) Mass Spectrometry. Direct infusion experiments were carried out

2

ACS Paragon Plus Environment

Page 3 of 10

1 2 3 4 5 6 7 8 9 10 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 59 60

Analytical Chemistry

10/4/2016 San Jose, CA, USA) was used for acquisition and analysis of the MS data.

by using a linear quadrupole ion trap (LQIT) mass spectrometer at unit mass resolution (LTQ, Thermo Fisher Scientific, San Jose, CA, USA) equipped with an APCI source. A Thermo Scientific Exactive Orbitrap mass spectrometer was used for high-resolution mass measurements and operated at a resolution of 100,000 by using ionized tricosane (n-C23) as the lock mass for the internal calibration. Direct infusion positive ion mode APCI conditions were as follows: heated nebulizer vaporizer temperature, 150 °C; sheath gas flow rate, 60 arbitrary units; auxiliary gas flow rate, 30 arbitrary units; ion transfer capillary temperature, 50 °C; discharge current, 4.5 mA; direct infusion flow rate of samples, 10 µl/min; ion transfer capillary voltage, 10 V; and tube lens voltage, 20 V. Xcalibur 2.2 (Thermo Fisher Scientific, Inc., San Jose, CA, USA) was used for acquisition and analysis of the MS data. All ionmolecule reactions were studied in the LQIT equipped with a pulsed valve sample introduction interface, which allows volatile molecules to be introduced directly into the ion trap.21 The internal nominal pressure of the LQIT was maintained at 0.8×10-5 torr.

Nuclear Magnetic Resonance (NMR) Spectroscopy. The base oil samples were analyzed on a Bruker 600 MHz Avance III NMR spectrometer by using a TCI cryoprobe. The 13C NMR spectra were acquired at 25 oC with inverse gated decoupling (no NOE) using a 90° pulse, 5 s recycle delay and as 200-300 transients. The separation of the naphthenic carbons from the paraffinic carbons in the 13C NMR spectra was carried out using the baseline subtraction method described previously.1 The broad underlying hump in the aliphatic carbon area is assumed to be mainly due to these naphthenic carbons. For the remaining paraffinic carbons, the average alkyl chain length was calculated using the formula4 C* = 2(Cp/tMe) where Cp is the total paraffinic carbon area and tMe is the terminal methyl area from 10-15 ppm. For these experiments, the lubricant base oil samples were dissolved in CDCl3 to prepare 50% weight/weight solutions. Cr(acac)3 was added as a relaxation reagent to achieve a concentration of 0.02 M.

Field Ionization (FI) Mass Spectrometry. All FI mass spectra were obtained in centroid mode using a Thermo Fisher Scientific double focusing sector (DFS) mass spectrometer equipped with a liquid injection field desorption ionization (LIFDI, Linden ChroMasSpec GmbH) source that was operated in FI mode. DFS MS was operated in the magnetic scan mode at a resolution of 2,000 (± 50). Ion source parameters were as follows: acceleration voltage, +5 kV; counter electrode voltage, -5 kV; reference inlet temperature, 80 °C; ion source temperature, 50 °C; flash duration, 150 ms; and inter scan delay, 150 ms. Two types of FI emitters were used: Linden ChroMasSpec GmbH FI-emitter 10 µm 20 mA type at 50 mA and CarboTec 10 µm Allround emitter at 90 mA. New emitters were preconditioned before the sample runs by applying emitter heating current for 2 h. DFS MS was scanned from m/z 50 up to 1000 at the rate of 7.5 s/decay.

RESULTS AND DISCUSSION

The aim of this study was to compare FI-MS and APCI-MS as alternative techniques for the analysis of lubricant base oils that are composed of heavy saturated hydrocarbons. FI-MS has been used for decades for the analysis of large saturated hydrocarbon mixtures and is an established technique,2,22,23 while APCI-MS is still under evolution. The APCI conditions and ion chemistry were optimized and the influence of APCI ion source gas on the observed ion types and degree of fragmentation for hydrocarbon model compounds was investigated using a linear quadrupole ion trap (LQIT) mass spectrometer. The ionization mechanism was explored via ion-molecule reactions performed inside the ion trap of LQIT. Finally, average molecular weights and hydrocarbon class distributions determined with FI-MS (using a Thermo Fisher Scientific double focusing sector (DFS) mass spectrometer) and APCIMS techniques for three lubricant base oils (low, mid and high viscosity) and the strengths and limitations of the techniques are compared. The three lubricant base oils originate from different base oil stock groups and represent a large viscosity range.

The direct insertion probe (DIP) was heated during the experiment from 50 °C up to 360 °C at a ramp rate of 25 °C/min. 2 µl of sample solution were injected into a sample holder (crucible: Mascom GmbH 0568770S-0568780S for low viscosity base oils and Mascom GmbH 0568760S for other base oils and model compound mixtures) and the solvent was allowed to evaporate at room temperature prior to analysis. The sample holder was placed into a DIP and introduced into the ion source via a vacuum exchange lock. The sample run was started immediately after the sample was introduced into the ion source. Xcalibur 2.2 program (Thermo Fisher Scientific, Inc.,

Atmospheric Pressure Chemical Ionization. Nitrogen is the most commonly used ion source gas in positive ion mode APCI, although oxygen, synthetic air and helium have been used in a few studies.12-14 The ion source gas has multiple functions in APCI that employs heated nebulizer for sample evaporation: 1) to nebulize the liquid sample, 2) to assist in the

3

ACS Paragon Plus Environment

Analytical Chemistry

10/4/2016 mass spectrum). Only [M-H]+ ions of paraffinic (saturated linear or branched) and naphthenic (saturated cyclic) hydrocarbons were detected. This result is in line with the model compound results shown in Figure 1. Both model compound and base oil sample analysis results demonstrate that using oxygen as the ion source gas greatly simplifies the product ion distributions and is more suitable for the analysis of lubricant base oils: if oxygen is used as the ion source gas, only one type of ionized molecules is formed from all the three types of hydrocarbons (linear, branched and cyclic) while two types of ionized molecules are formed when nitrogen is used as the ion source gas.

evaporation of the analyte molecules from liquid droplets and transfer them into the ionization region, 3) to act as reagent gas and the source of primary ions in corona discharge ionization and 4) to provide an atmospheric pressure medium (inert collision partners) that lowers the internal energies of the ions. Of these, the most important role of the ion source gas is probably the formation of primary ions and their subsequent ionmolecule reactions. For this reason, the ionization of saturated hydrocarbons was examined in both nitrogen and oxygen atmospheres. Three different types of model compounds, i.e., cyclic, branched, and linear saturated hydrocarbons, as well as three lubricant base oils (all in hexane), were used to evaluate the types of ionized molecules generated by the two ion source gases. The cyclic saturated hydrocarbon, cholestane (M), produced both [M]+● and [M-H]+ ions when nitrogen was used as the ion source gas, whereas only the [M-H]+ ion was produced when nitrogen was replaced by oxygen as the ion source gas (Figure 1a).

50 0

+ [M-H]

100 O2

b)

Relative Abundance

100 N2

[M-H] [M]+● 371 372

371

50

100

N2

[M-H] 421

In addition to the different types of ionized molecules generated upon APCI, the level of fragmentation, adduct ion formation as well as relative responses for different analytes were probed and optimized by using individual cyclic, branched and linear saturated hydrocarbons (Figure 3) and the C20-C40 nalkane mixture (Figure 2b) in hexane with oxygen as the ion source gas.

+

Fragment ions

0

50

+ [M-H] O2

254 198 226

421

50 0

0 360

370

m/z

380

366

100

50

100

[M]+• ions

a) APCI-MS N2

Relative Abundance

+

a)

Relative Abundance

1 2 3 4 5 6 7 8 9 10 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 59 60

Page 4 of 10

410

420

430

m/z

Figure 1. Direct infusion positive ion mode APCI mass spectra of a) cholestane (M; MW 372) and b) squalane (M; MW 422) with N2 or O2 as the ion source gas.

282 310

422 450

0

b) APCI-MS O2 100

394

338

[M-H]+ ions 309 337

365

281

393 421

50

0

449 477 505

533

561

[M]+• ions

c) FI-MS 366

100

394

On the other hand, the branched saturated hydrocarbon, squalane (M), produced [M-H]+ ions under both nitrogen and oxygen conditions (Figure 1b). Linear saturated hydrocarbons (a mixture of n-C20−C40) produced [M]+● ions and extensive fragmentation when nitrogen was employed as the ion source gas (Figure 2a), whereas only [M-H]+ ions were produced when oxygen was used (Figure 2b). Both [M]+● and [M-H]+ ions were formed when nitrogen was used for the ionization of the low viscosity lubricant base oil sample (Figure S3, top), whereas only [M-H]+ ions were produced when oxygen was used (Figure S3, middle).

478 506 534 562

338

50

422 450

478 506

534 562

310

0 200

300

400

500

600

m/z

Figure 2. Mass spectra of an n-alkane mixture (C20-C40) in hexane measured using direct infusion positive ion mode APCI with a) nitrogen and b) oxygen as the ion source gas and c) using FI (the mass spectrum shown above is an example, run-to-run variation was large). Small peaks in spectrum b) are background impurities due to low sample concentration.

The exact position of the outlet of the heated nebulizer relative to the corona discharge needle and MS capillary inlet and the temperatures of the heated nebulizer and ion transfer capillary were observed to affect the level of fragmentation and adduct ion formation, and these effects were reproducible from day to

To confirm the ion identities, the elemental compositions of the ionized components of the low viscosity base oil in hexane were determined using high resolution Orbitrap mass spectrometry and oxygen as the ion source gas (Figure S3, bottom

4

ACS Paragon Plus Environment

Page 5 of 10

10/4/2016 day and on different LQIT instruments. The same geometrical optimization couldn’t be done in case of FI because the positions of the emitter, counter electrode and crucible are fixed within the FI source. The shortest distance between APCI heated nebulizer and capillary inlet tip, a heated nebulizer temperature of 150 °C and ion transfer capillary temperature of 50 °C were found to be optimal for minimizing fragmentation and for obtaining a similar response for all components of the equimolar alkane mixture. The abundances of the fragment ions relative to the most abundant ion (ionized analyte) was only a few percent at optimal conditions. Abundances of the fragment ions of the cyclic model compound, cholestane (C27H48), branched model compound, squalane (C30H62), and linear model compound, pentatriacontane (C35H72), were 24%, 1-4% and 1% maximum, respectively, relative to the quasi-molecular ion [M-H]+ (Figure 3). Only the branched model compound, squalane, showed notable adduct ion formation ([M-H+O]+; 1% relative to the quasi-molecular ion [C30H62H]+) at optimal conditions. This type of adduct ions ([MH+O]+) have been proposed to be formed in the presence of humidity.12 However, no obvious correlation was found here between humidity and the extent of adduct ion formation. a)

276 316

219

200

m/z

371

Max 1-4% fragment ions 239

100 50 0

Comparison of Atmospheric Pressure Chemical Ionization and Field Ionization. FI-MS has been used in the petroleum industry for decades for the determination of the relative amounts of different hydrocarbon classes (paraffins and mono, di-, tri-, tetra-, penta- and hexanaphthenes) in base oils.12,24,25 FI-MS is beneficial because it ionizes large saturated hydrocarbons with little to no fragmentation. However, FI-MS suffers from unsatisfactory robustness: poor reproducibility of ion signals (within a day and between days; see Figures S4-S6), limited dynamic range (often not more than two orders of magnitude), dependence of signal levels on emitter age, limited lifetime of FI emitters, variable fragmentation of ionized analytes (caused by new emitter or volatile sample, dust in the ion source, bad emitter part, etc), carry-over problems, and laborious maintenance of the vacuum ion source. Fragment ions are formed via H2 losses from molecular ions. Small alkyl fragment ions with m/z values below 100 are also detected. The former leads to distortion in measured hydrocarbon class distributions, which results in overestimation of the amount of naphthenes and underestimation of the amount of paraffins. Strengths of FI include the requirement of only a small amount of sample and a very low extent of fragmentation when the FI emitter works properly. On the other hand, the strengths of APCI are related to better robustness: good reproducibility of ion signals, good dynamic range of 3−4 orders of magnitude, stable signal levels, long lifetime of the APCI corona needle, easier maintenance of the APCI source and the possibility to automate the experiments. Due to the continuous nature of direct infusion APCI, a large number of mass spectra can be averaged, leading to a good signal-to-noise ratio and precision of the measurements. The drawbacks of APCI include some fragment ion formation and challenges in achieving and maintaining a clean background.

[cyclic C27H48-H]+

300

100 50 0

b)

ion-molecule reactions under controlled conditions within the ion trap of LQIT. The results of this study are described in Supporting Information. They demonstrate that large saturated hydrocarbons can be ionized by hydride abstraction without fragmentation in the ion trap if the reagent ion is one of the carbocations derived from hexane. However, if the reagent ion is [O2]+●, this is not the case. The same may be true for APCI.

Max 2-4% fragment ions 189

Relative Abundance

1 2 3 4 5 6 7 8 9 10 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 59 60

Analytical Chemistry

200

[branched C30H62-H]+

295

421

300

[C30H62-H+O]+ 437

m/z

Max 1% fragment ions

c)

199

255

200

300 400 [linear m/z + 85 [C6H14-H] ion from hexane solvent C4H9+ ion from hexane solvent

100 50 57 0 50 100

150

200

250 300 m/z

350

400

C35H72-H]+ 491 450

500

Relative response factors (RRF) for different analytes can be used to evaluate the similarities of the results obtained using different methods. When another method is evaluated as an alternative to an existing method, differences in the analytical results between the methods can be as important as the accuracy of the methods. The observation of similar signals (72%, 100% and 70%, respectively) upon APCI/oxygen for an equimolar mixture of the compounds shown in Figure 3, cho-

Figure 3. Direct infusion positive ion mode APCI mass spectra of a) cholestane (MW 372), b) squalane (MW 422) and c) pentatriacontane (MW 492) in hexane obtained using oxygen as the ion source gas. Zoomed views on the mass spectra on the left top corner show the level of fragmentation. Percent numbers refer to the relative abundances of the fragment and quasi-molecular ions.

Possible mechanisms for the formation of the [M-H]+ ions upon APCI were investigated in more detail by performing

5

ACS Paragon Plus Environment

Analytical Chemistry

10/4/2016 high viscosity lubricant base oils, respectively. APCI-MS results are averages of 300 mass spectra.

lestane, squalane and pentatriacontane, in hexane, was promising. Figure 4 shows a comparison between molar RRFs for positive ion mode APCI and FI for the C20-C40 paraffin mixture discussed above, and C24 paraffin was taken as a reference for both techniques. While APCI and FI give a similar responses up to C28, the response factors differ between C28 and C40. APCI shows less variation in RRFs in that region than FI. This difference between the results obtained for APCI and FI suggest that APCI provides more accurate results and that the results obtained using APCI and FI should match well if the lubricant base oil is either composed of hydrocarbons with ≤C28 or ≥C28 carbon atoms but not both.

1.4

The same was observed for the mass spectra measured for low, middle and high viscosity lubricant base oils (Figures 57). Mass spectra measured for low and high viscosity lubricant base oils agree better for the two techniques than the mass spectra measured for the middle viscosity lubricant base oil (excluding the fragment ions, see Figures 5-7). Due to significant run-to-run variability in the mass spectra (many runs totally failed for the low viscosity sample but several also failed for the middle and high viscosity samples), FI mass spectra were selected from more than 30 replicates that were measured over a period of several weeks. The fragmentation patterns observed for APCI and FI are different based on model compound studies (Figure 3): large parts of the ionized molecules are lost upon APCI while H2 losses take place upon FI when the emitter is not performing optimally (Figure S7). This will distort the hydrocarbon class distributions determined based on FI mass spectra. For APCI, this only occurs if fragment ions overlap with ionized analytes.

FI APCI

1.2

Molar RRF

1 2 3 4 5 6 7 8 9 10 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 59 60

Page 6 of 10

1.0 0.8 0.6 0.4 0.2 0.0 18

20

22

24

26

28

30

32

34

36

38

40

Overlap of fragment ions with ionized analytes cannot be large for APCI-MS because the measured average carbon numbers and molecular weights are similar for APCI and FI (Table 1). Comparison of the relative amounts of hydrocarbons in different hydrocarbon classes as a function of carbon number for low, middle and high viscosity lubricant base oils is shown in Figures S8-S10 for APCI and FI. Even though the relative amounts of hydrocarbon classes (Table 2) differ between APCI and FI, many features of FI distributions are reproduced by APCI in Figures S8-S10. Hence, fragmentation must not have a large effect on the hydrocarbon distributions measured using APCI.

42

Carbon number Figure 4. Molar relative response factors (RRF) for the mixture of C20-C40 paraffins: comparison between APCI and FI. FI results are averages of six experiments (error bars correspond to one standard deviation). APCI results are averages of three experiments (error bars correspond to one standard deviation). Corresponding mass spectra are shown in Figure 2.

This was the case for the data shown in Table 1. The average carbon numbers and average molecular weights are similar for both the low and high viscosity lubricant base oils for APCI and FI but the results obtained for the middle viscosity sample differ (APCI fragment ions were excluded from the calculations; about fragment ions, see discussion below). Table 1. Carbon number range, average carbon number (CAVG number) and average molecular weight (MWAVG) for low, middle and high viscosity lubricant base oils determined using APCI-MS and FI-MSa. Lubricant base C number CAVG number MW AVG oil viscosity range APCI-MS FI-MS APCI-MS FI-MS 16-28 22.1 22.3 308 313 LOW 26-43 32.6 29.7 452 416 MID 29-53 39.3 39.7 547 555 HIGH

Figure 5. A positive ion mode APCI mass spectrum measured for the low viscosity lubricant base oil sample (in hexane by using oxygen as the ion source gas) (top, average of 30 mass spectra) and the corresponding FI mass spectrum (bottom, a single run). The FI mass spectrum is an example of a successful experiment

a

An average of three and six most representative mass spectra were used to determine the FI-MS results for low, middle and

6

ACS Paragon Plus Environment

Page 7 of 10

1 2 3 4 5 6 7 8 9 10 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 59 60

Analytical Chemistry

10/4/2016 among many failed experiments. The ionized analytes generated in APCI-MS and FI-MS are [M-H]+ and [M]+●, respectively.

Hydrocarbon classes Paraffins Mononaphthenes Dinaphthenes Trinaphthenes Tetranaphthenes Pentanaphthenes Hexanaphthenes

Low viscosity Middle viscosity High viscosity APCI-MS FI-MS APCI-MS FI-MS APCI-MS FI-MS 27.3 % 40.3 % 31.7 % 33.6 % 10.8 % 11.9 % 26.0 % 29.1 % 28.0 % 34.7 % 17.2 % 24.9 % 22.7 % 18.0 % 20.5 % 20.2 % 25.8 % 26.1 % 13.0 % 7.7 % 11.3 % 8.3 % 23.2 % 20.2 % 6.9 % 3.5 % 5.6 % 2.7 % 14.7 % 11.8 % 4.2 % 1.4 % 2.7 % 0.4 % 8.1 % 5.0 % 0.0 % 0.0 % 0.0 % 0.0 % 0.0 % 0.0 %

#

Three and six of the most representative mass spectra were averaged to obtain the FI-MS results for low- and middle-high viscosity lubricant base oils, respectively. APCI-MS results are averages of 300 mass spectra.

To evaluate the accuracy of the hydrocarbon class distributions measured using APCI and FI, the APCI-MS and FI-MS results were compared to NMR data. 13C NMR can provide an estimate of the relative amounts of paraffinic and naphthenic carbons and an average number of carbons in paraffinic chains for molecules present in complex mixtures. Results of these measurements and comparison to MS results are shown in Table 3. The likely naphthenic structures,16,17 their carbon numbers and relative contributions used to determine the average amounts of naphthenic carbons are shown in Figure S11. The APCI-MS and FI-MS results for paraffin content differ somewhat and neither agrees exactly with the NMR results. However, the average number of carbons in paraffinic chains follows the same increasing trend from low viscosity to high viscosity base oil for all the three methods.

Figure 6. A positive ion mode APCI mass spectrum measured for the middle viscosity base oil sample (in hexane by using oxygen as the ion source gas) (top, average of 30 mass spectra) and the corresponding FI mass spectrum (bottom, a single run). The ionized analytes generated in APCI-MS and FI-MS are [M-H]+ and [M]+● ions, respectively.

Table 3. Relative amounts of paraffinic carbons and the average number of carbons in paraffinic chains in low, middle and high viscosity lubricant base oils determined using APCI-MS, FI-MS and NMR. Lubricant Average number of carbons Relative paraffinic carbon base oil in paraffinic chains viscosity APCI-MS FI-MS NMR APCI-MS FI-MS NMR LOW 65.0 % 74.9 % 72.8 % 14.4 16.8 19.8 MID 78.6 % 79.9 % 72.0 % 25.7 23.7 23.1 HIGH 71.6 % 74.5 % 68.0 % 28.1 29.5 27.1

CONCLUSIONS A direct infusion APCI/MS method was optimized and studied as an alternative to FI/MS for the determination of the relative amounts of different hydrocarbon classes (paraffins and mono, di-, tri-, tetra-, penta- and hexanaphthenes) in lubricant base oils. The APCI/O2/hexane method produced predominantly [M-H]+ pseudo-molecular ions and a minimal amount of fragment and adduct ions for the different types of saturated hydrocarbon model compounds studied. Ion-molecule reaction studies demonstrated that carbocations derived from hexane solvent upon APCI ionize saturated hydrocarbons via hydride abstraction with minimal fragmentation in the linear quadrupole ion trap (see Supporting Information).

Figure 7. A positive ion mode APCI mass spectrum measured for the high viscosity lubricant base oil sample (in hexane by using oxygen as the ion source gas) (top, average of 30 mass spectra) and the corresponding FI mass spectrum (bottom, a single run). The ionized analytes generated in APCI-MS and FI-MS are [MH]+ and [M]+● ions, respectively.

Table 2. Hydrocarbon class distributions (wt %) for low, middle and high viscosity lubricant base oils measured using APCI-MS and FI-MS.#

7

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 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 59 60

Page 8 of 10

10/4/2016 The strengths of the APCI/MS method compared to FI/MS are related to better robustness of the APCI method: better reproducibility of ion signals, better dynamic range, more stable signal levels, long lifetime of APCI corona discharge needle, easier maintenance of the ion source and the possibility to automate the experiments. In addition, only APCI is well suited for the analysis of low viscosity lubricant base oils. The drawbacks of APCI are fragment ion formation and challenges in achieving and maintaining a clean background. Fragment ions obscure the starting point of the lubricant base oil molecular weight distribution. This calls for further development of the method to decrease the extent of fragmentation. Despite fragmentation, the APCI-MS method utilized here provides similar results as FI-MS for lubricant base oils: both methods give similar average carbon numbers and molecular weight distributions and mostly similar hydrocarbon class distributions. Comparison of the paraffinic content of the lubricant base oils shows similarities for APCI-MS and FI-MS. The average number of carbons in paraffinic chains follows the same increasing trend from low viscosity to high viscosity lubricant base oils for all the three methods.

Present Addresses Elias Ikonen, Thermo Fisher Scientific, Ratastie 2, FI-01620, Vantaa, Finland

Author Contributions All authors contributed to the preparation of the manuscript. / All authors have given approval to the final version of the manuscript.

Notes The authors declare no competing financial interest.

ACKNOWLEDGMENT The authors gratefully acknowledge Neste Co. for financial support for this project. We also thank Mr. Kari Kulmala of Neste for valuable discussions on base oils, and Mrs. Jaana Keskiniva and Mr. Juho Heininen of Neste for help in the field ionization mass spectrometry measurements. We also would like to thank the Finnish Biological NMR Center at the Institute of Biotechnology in Viikki, Helsinki, Finland for the use of the NMR instrument.

REFERENCES (1)

Saetre, R.; Somogyvari, A.; Yamashita, G. Abstr. Pap. Am. Chem. Soc. 1989, 197, 40–PETR. (2) Qian, K.; Dechert, G. J. Anal. Chem. 2002, 74, 3977–3983. (3) Liang, Z.; Hsu, C. S. Energy Fuels 1998, 12, 637– 643. (4) Sarpal, A. S.; Kapur, G. S.; Mukherjee, S.; Jain, S. K. Fuel 1997, 76, 931–937. (5) Gross, J. H. Mass Spectrometry: A Textbook, 2nd ed.; Springer: Heidelberg, 2004. (6) Klesper, G.; Röllgen, F.W. J Mass Spectrom. 1996, 31, 383-388. (7) Hourani, N.; Kuhnert, N. Anal. Methods 2012, 4, 730–735. (8) Hourani, N.; Muller, H.; Adam, F. M.; Panda, S. K.; Witt, M.; Al-Hajji, A. A.; Sarathy, S. M. Energy Fuels 2015, 29, 2962–2970. (9) Tose, L. V.; Cardoso, F. M. R.; Fleming, F. P.; Vicente, M. A.; Silva, S. R. C.; Aquije, G. M. F. V.; Vaz, B. G.; Romão, W. Fuel 2015, 153, 346– 354. (10) Wu, C.; Qian, K.; Walters, C. C.; Mennito, A. Int. J. Mass Spectrom. 2015, 377, 728–735. (11) Gao, J.; Ii, D. J. B.; Owen, B. C.; Jin, Z.; Hurt, M.; Amundson, L. M.; Madden, J. T.; Qian, K.; Kenttämaa, H. I. J. Am. Soc. Mass Spectrom. 2011, 22, 531–538. (12) Nyadong, L.; Quinn, J. P.; Hsu, C. S.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G. Anal. Chem. 2012, 84, 7131–7137.

Finally, it should be noted here that base oils can contain small amounts of impurities, such as alkenes and alkylaromatics. The impact of these impurities on determination of the hydrocarbon class distributions will be the subject of a future study. ASSOCIATED CONTENT Supporting Information Available Supporting information provides the results obtained in ionmolecule reaction studies, low-resolution and high-resolution APCI mass spectra of low viscosity lubricant base oil, APCI and FI comparison mass spectra and hydrocarbon type distributions as a function of carbon number determined using APCI-MS and FI-MS for low, middle, and high viscosity lubricant base oils, and naphthenic structures, their carbon numbers and relative contributions used to calculate average amounts of naphthenic carbons. This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION Corresponding Author *Professor Hilkka I. Kenttämaa. Tel: (765) 494-0882. Fax: (765) 494-9421. E-mail: [email protected] *Dr Jyrki Viidanoja. Tel: +358-50-4587526. E-mail: [email protected]

8

ACS Paragon Plus Environment

Page 9 of 10

1 2 3 4 5 6 7 8 9 10 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 59 60

Analytical Chemistry

10/4/2016 (13) Gao, J.; Owen, B. C.; Borton, D. J.; Jin, Z.; Kenttämaa, H. I. J. Am. Soc. Mass Spectrom. 2012, 23, 816–822. (14) Marotta, E.; Paradisi, C. J. Am. Soc. Mass Spectrom. 2008, 20, 697–707. (15) Espada, J. J.; Coto, B.; Pena, J. L. Energy Fuels 2009, 23, 888–893. (16) Lynch, T. Process Chemistry of Lubricant Base Stocks; CRC Press: Mississauga, 2008. (17) Sequeira, A. Lubricant Base Oil and Wax Processing; Marcel Dekker: New York, 1994. (18) Sharma, B. K.; Adhvaryu, A.; Sahoo, S. K.; Stipanovic, A. J.; Erhan, S. Z. Energy Fuels 2004, 18, 952–959. (19) Sharma, B. K.; Stipanovic, A. J. Tribol. Lett. 2004, 16, 11–19. (20) Sharma, B. K.; Adhvaryu, A.; Perez, J. M.; Erhan, S. Z. Fuel Process. Technol. 2008, 89, 984–991. (21) Jarrell, T.; Riedeman, J.; Carlsen, M.; Replogle, R.; Selby, T.; Kenttämaa, H. Anal. Chem. 2014, 86, 6533–6539. (22) Kuras, M.; Ryska, M.; Mostecky, J. Anal. Chem. 1976, 48, 196–198. (23) Scheppele, S. E.; Hsu, C. S.; Marriott, T. D.; Benson, P. A.; Detwiler, K. N.; Perreira, N. B. Int. J. Mass Spectrom. Ion Phys. 1978, 28, 335–346. (24) Zhou, X.; Shi, Q.; Zhang, Y.; Zhao, S.; Zhang, R.; Chung, K. H.; Xu, C. Anal. Chem. 2012, 84, 3192–3199. (25) Zhou, X.; Zhang, Y.; Zhao, S.; Chung, K. H.; Xu, C.; Shi, Q. Energy Fuels 2014, 28 , 417–422.

9

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 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 59 60

10/4/2016

For TOC Only

10

ACS Paragon Plus Environment

Page 10 of 10