Evaluation of Thin-Layer Chromatography â Laser Desorption

Subscriber access provided by Kaohsiung Medical University

Fossil Fuels

Evaluation of Thin-Layer Chromatography – Laser Desorption Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometric Imaging for Visualization of Crude Oil Interactions Ali Zahraei, Peter Arisz, Alexander P. van Bavel, and Ron M.A. Heeren Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.8b00639 • Publication Date (Web): 07 Jun 2018 Downloaded from http://pubs.acs.org on June 7, 2018

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 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 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.

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 28 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

Energy & Fuels

Evaluation of Thin-Layer Chromatography – Laser Desorption Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometric Imaging for Visualization of Crude Oil Interactions Ali Zahraei†, Peter W. F. Arisz†, Alexander P. van Bavel‡, Ron M. A. Heeren†* †

Maastricht MultiModal Molecular Imaging (M4I) institute, Division of Imaging Mass

Spectrometry, Maastricht University, Maastricht, The Netherlands ‡

Shell Global Solutions International B.V., Amsterdam, The Netherlands

Keywords: FT-ICR mass spectrometry imaging, LDI, Crude Oil, TLC, Iatroscan, Intermolecular interactions, Aromaticity, Principle component analysis ABSTRACT: A light oil was separated into four chromatographic fractions that serve as proxy for SARA fractions. The fractions were (semi)quantified on a rod by TLC-flame ionization detection and characterized on a plate with laser desorption ionization-mass spectrometric imaging (TLC-LDIMS). Comparisons of (semi)quantitative TLC-FID and qualitative TLC-LDI-MS results showed that LDI-MS was most sensitive for detection of molecules in the polar P1 fraction, and to some extend for the aromatics fraction, while no signal was observed for the most polar P2 and saturates fractions. Based on these results limits of the compositional space as observed by the laser ionization technique were evaluated. The molecular speciation between and within the spots of the aromatics and the P1 fractions were analyzed and interpreted in terms of oil–SiO2 versus oil–solvent interactions as a function of molecular characteristics such as DBE, aromaticity (H/C ratio), heteroatom content, degree of alkylation, and shielding of heteroatoms. In addition, the high oil loading resulted in an interesting bifurcation of the aromatics spot which implies that oil-oil interactions can be enforced and studied in the TLC model system.

1 ACS Paragon Plus Environment

Energy & Fuels 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

INTRODUCTION The interface between oil and rock is believed to be mainly governed by polar components within the oil. These interactions alter the properties of the interface and can make it difficult to extract oil from reservoir rocks

1-2

. Crude oil separation on silica coated thin layer

chromatography (TLC) plates with direct laser desorption ionization mass spectrometry (LDIMS) readout 3 may serve as a simple on-line model system to study oil interactions with silica 4. Ideally, fundamental knowledge about physical-chemical behavior of oil at interfaces can be employed to improve oil recovery

2

and processing technologies, but we have to make the

reservation beforehand that subsurface conditions are far more complex and involve additional interactions with brines and other minerals at elevated temperature and pressure at geological time scales. Crude oil is intrinsically very complex and contains a plethora of natural organic compounds 5-9. Structural complexity of hydrocarbon molecules includes their sizes and shapes; naphthenic and aromatic ring systems, alkyl-substitution patterns and N, O, and S containing polar functional groups such as pyrrolic, pyridinic, quinolinic; hydroxyl, carbonyl, carboxyl; and thiophenic, sulfide, and sulfoxide units, respectively

10-11

. The number of elemental compositions rises

exponentially with increasing molecular weight and increasing heteroatom content in hydrocarbon compounds

7, 12-13

. Tens of thousands of distinctive mass peaks with unique

elemental compositions can be identified by high resolution Fourier Transform Ion Cyclotron Resonance (FTICR) MS measurements of crude oils 14. Each of the measured elemental compositions is composed of numerous isomeric structures including different functional groups. Information about isomeric structures cannot be assessed by mass spectra of (pseudo)molecular ions alone, but combined with chromatographic separation patterns are expected based on the sort of functional group and their exposure or shielding by alkyl-substituents such as convincingly demonstrated for carbazole model compounds by Li and Larter et al. 15, for example. Fractionation of crude oils often provides the opportunity to identify components that were not possible to detect directly in the parent crude oil 16. One of the approaches for petroleum characterization is by SARA analysis that classifies oils in terms of their proportion of Saturates, Aromatics, Resins and Asphaltenes fractions (SARA). 2 ACS Paragon Plus Environment

Page 2 of 28

Page 3 of 28 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

Energy & Fuels

SARA fractions are separated based on their solubility [as described in IP-143] in different solvents combined with their interaction with silica or alumina particles

5, 17-18

. The saturates

fraction is composed of a mixture of linear, branched, and cyclic alkanes. The aromatics fraction contains molecules with one or more aromatic rings with attached saturated hydrocarbon side chains. The resins fraction consists of polar molecules that dissolve in alkane solvents such as heptane or pentane, while asphaltenes are polar molecules that by definition precipitate in alkane solvents such as pentane or heptane but dissolve in aromatic solvents such as benzene, pyridine or toluene 20-21

19

. TLC is applied frequently to as a rapid and convenient proxy for SARA fractions

.

Iatroscan TLC based analysis is typically coupled with flame ionization detection (FID) to quantify the fractions. TLC-FID can be employed without considerable sample preparation, but a drawback of TLC-FID is that it provides no information at the molecular level. TLC combined with LDI-FTICRMS imaging is demonstrated as a direct chemical readout of fractionated petroleum samples 3 but no quantitative response factors are known for this technique. Here, we present the application and further validation of the TLC-LDI-FTICRMS imaging method presented by Smith et al. 3 to study molecular speciation resulting from of oil–SiO2, oil– solvent and oil-oil interactions that occur in a 3-step chromatographic separation. A light crude oil sample was chosen with the expectation that light oil produces less complex data compared to heavier oil types. The response of TLC-LDI-MS imaging was compared with TLC-FID data and optical observations of the developed TLC-plate in order to (semi)quantify and evaluate the limits of the compositional space that is observed by LDI-MS. The TLC-plate was imaged with a 16-fold higher spatial resolution, namely with a raster size of 250 µm compared to 1 mm used in the previous study, that allowed chemical speciation both between and within chromatographic spots. Observed trends in the chemical speciation were tentatively correlated with chemical structures.

EXPERIMENTAL SECTION Sample and Reagents.


Energy & Fuels 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

Xylene, toluene, n-heptane and dichloromethane (DCM) and methanol (MeOH) were purchased from Sigma-Aldrich (Zwijndrecht, the Netherlands). A hydrophobic type petroleum sample (API: 41.6, density: 0.8179 g/cm3, viscosity: 3.2769 cP at 20 oC and %S: 0.231%) was provided by Shell Global Solutions International B.V., the Netherlands. Proxy for SARA analysis by TLC-FID TLC-FID was measured on an Iatroscan MK-6S (SES GmbH - Analysesysteme) equipped with a rack with ten coated silica rods (Chromarods). The hydrogen flow rate of the FID was 160 mL per minute; the air flow was 2 L per min. Just prior to use, the rods were burned clean by a series of three blanks runs. A dilution series of 14, 37, 121 and 500 µg crude oil per µL of toluene was prepared and 1 µL of the solutions was applied at the starting point of the rods with a semiautomatic sample spotter (Model 3202). The rods were eluted sequentially in development tanks (DT-150) first with n-heptane for 40 minutes, second with a mixture of toluene / n-heptane (80/20 vol/vol) for 12 minutes and finally with a mixture of (95/5) DCM / MeOH for 3 minutes. The developed rods were scanned with the Iatroscan with a speed of 30 seconds per scan. The data were collected with SES-i-Chromstar software. TLC Development for MS imaging analysis. Silica gel 60 with fluorescence indicator 254 nm on TLC aluminum foil (56524-25EA SigmaAldrich, Netherlands) was used with a pore volume of 0.75 mL/g. The sample was diluted in xylene (1/1) and 4 µL was aliquoted at line 1 in Figure 1 on the bottom of a TLC plate with a spot size of roughly 2 mm diameter. The first mobile phase consisting of n-heptane was eluted to line 4 followed by a mixture of toluene / n-heptane (80/20 vol/vol) to line 3. Subsequently, DCM / MeOH (95/5 vol/vol) mobile phase was eluted to line 2. The plate was dried at room temperature for at least 5 min after every step. For reference MS measurement a diluted oil sample was spotted at line 5 after the fluorescence picture was taken. Photographs were taken of the fully developed plate exposed by room light and by fluorescent excitation with 254 nm UV light. LDI-FTICR MS imaging Analysis.


Page 4 of 28

Page 5 of 28 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

Energy & Fuels

Double-sided conductive tape (3M Science, USA) was used to attach the back of the developed TLC plate to a stainless-steel MALDI target. The LDI-FTICR-MS image was recorded with a raster size 250 µm without matrix in LDI-MS positive (+) ion mode on a SolariX XR FTICR mass spectrometer (Bruker Daltonics GmbH, Bremen, Germany) equipped with a 9.4 T magnet. A frequency-tripled Nd:YAG laser (355 nm, 3.49 eV photons) was employed for laser desorption directly from the TLC foil with 250 laser shots per position at 1 kHz repetition rate. The laser spot size was about 30 µm. The laser power and number of laser shots were adjusted to obtain a signal with sufficient signal-to-noise. Only the ions in the mass range set between m/z 250 and 600 were transferred with the quadrupole ion-guide from the ion-source into the FTICR cell. Such selective ion accumulation in the FT-ICR cell made petroleum detection easier as the signal-to-noise ratio increased significantly during the acquisition 22. The selected mass window was taken similar to the mass ranges reported in recent LDI-MS publications of crude oils

3, 23

.

24

The signal of light oils with API values larger than 40 hardly exceeds m/z 600 . LDI may form very abundant fullerenes above this mass that disturb the ICR signal 25. Spectra were acquired with a 2 MW transient size and were zero-filled with an additional 2 MW. The transient length was 1.82 s and gave a resolving power of about 350,000 at m/z 400. The mass spectra were externally calibrated using the red phosphorous in LDI-MS positive mode immediately prior to the experiment. After the experiment was completed an internal recalibration was performed by using the crude oil N1 series spectrum in positive mode. Peaklists were obtained with the Compass Data Analysis software package (Bruker Daltonics) through the selection of peaks with a signal-to-noise larger than five. Spectral interpretation was performed with the PetroOrg (Corilo, Yuri. EnviroOrg. Florida State University, 2013, Tallahassee, FL, USA; http://software.petroorg.com). PetroOrg spectral analysis. Peaklists of recalibrated spectra were created with Bruker DataAnalysis based on exact theoretical masses with a ±0.5 ppm error range. PetroOrg assignments were made with a maximum error of 2.0, 1.5 and 2.5 ppm for the whole oil, aromatics- and P1 spot, respectively. Normal boundary conditions for petroleum data (CcHhNnOoSs, c: 0-100, h: 0-200, n: 0-5, o: 0-5, s: 0-4) were used for these assignments. The data were filtered for a minimum of two 5 ACS Paragon Plus Environment

Energy & Fuels 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 28

consecutive carbons per homologous series, which removed all noise from single scattered data points. Data of assigned mass peaks can be visualized by the software as relative abundance class plots and iso-abundance-contoured plots of double bond equivalents (DBE= c – h/2 + n/2 + 1 9) versus the carbon number (#C). Principal Component Analysis Principal component analysis (PCA) was performed with an in-house written toolbox (ChemomeTricks) for MATLAB (The MathWorks, Natick, MA, USA)

26

. Sqlite data were

generated by the Bruker software and were peak-picked with 0.0001 Da bin size, n2 of 16, smoothing 2 and a threshold of 5. Background pixels were removed manually from the whole image (Figure 1C) until a clear separation was achieved between the aromatics and the P1 spots in PC1. PCA of the selected region of the aromatics spot resulted in a separation of the bottom and top zone by PC2 as shown in Figure 6B, and the top and bottom zones of the P1 spot were separated by PC1 as shown in Figure 6C.

RESULTS AND DISCUSSION TLC-FID Analysis Previous studies show that TLC development with a single mobile phase of 5% isopropanol in 95% heptane

3, 27

yields a continuous separation with poorly defined borders of the four proxy

SARA fractions. In this study four distinctly separated chromatographic fractions were obtained 28-29

by a sequential rod or plate development with n-heptane, 80/20 toluene / n-heptane and 95/5

DCM) / MeOH as mobile phase. The solvents subsequently eluted the fractions of saturates, aromatics, and the polar P1 fraction that was a proxy for the resins from the origin spot

21, 30-31

,

respectively. The most polar P2 fraction was a proxy for the asphaltenes which are non-mobile and remain at the origin of the plate 21. The separation was performed both on a flat TLC-plate that is suitable for optical analysis and direct MS imaging readout, as well as on an Iatroscan rod for semi-quantitative TLC-FID analysis.


Page 7 of 28 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

Energy & Fuels

TLC-FID analysis was used to estimate the sensitivity of LDI-MS for the chromatographic fractions. By first approximation, when the FID carbon-response factors for all fractions would be taken to be unity, the quantification by TLC-FID indicated that the light crude oil is composed of 59.8% saturates, 26.1% aromatics, 12.9% P1 and 1.1% P2. The accuracy of these numbers is discussed next. For Iatroscan analysis it is often granted that the FID signal increases linearly with the sample quantity and all carbon-response factors are equal reduced by heteroatoms in the organic structures

35

32-34

. However, carbon-response factors are

. Although the heteroatom content of oil is

limited to a few mass percent, it increases steadily with increasing polarity of the fractions. Saturates are almost pure hydrocarbons with carbon-response factor of 1.0; aromatics fraction contains some heteroatoms, while the P1 and P2 fractions contain substantial amounts, asphaltenes may contain even more than ten weight percent

36

, respectively. Therefore it would

be expected that the carbon-response factor decreases in the order from saturates to P2. However, the work by Bisht et al. revealed an opposite trend where the response of saturates is significantly lower than that of asphaltenes 20. This reversed effect can be explained by the loss of volatile compounds with boiling points 20 below 165 °C. The evaporation of volatiles probably mainly takes place during the drying of the rods after the development with the mobile phases. The amount of volatiles is most abundant in the saturates fraction, followed by aromatics fraction. Considering the high polarity and relatively high molecular weight of P1 and P2 the amount of volatiles can be ignored for these fractions. Our light oil, which is rich in saturates, likely contains a relatively large amount of volatile compounds. Therefore the measured amounts of saturates and aromatics were probably underestimated by the TLC-FID analysis. Considering the many uncertainties in the correction of the FID signal, the carbon-response factors in this study were taken to be equal for all fractions and a relatively large uncertainty in their (semi)quantification was accepted. Therefore the data derived from TLC-FID were rounded to whole percentages in the following text. Nevertheless, it was clear from the TLC-FID measurement that saturates make up the bulk of the volume of this light crude oil. The large amount of saturates fits well with the small amount of asphaltenes as indicated by P2

37

. These

two fractions are inherently incompatible with each other and asphaltenes would precipitate in


Energy & Fuels 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

situ due to the high saturates content in such light oil type. The TLC-FID results confirm the description that this was a hydrophobic oil type with a polar fraction of about 14%. TLC-plate images A flat TLC-plate was sampled with a large amount of 1.6 mg (for comparison, Smith et al. applied 0.2 or 0.4 mg 3) of oil at the origin position ‘1’ and developed subsequently with the three mobile phases. Sample overloading and multilayer formation was not a major concern since enforcement of oil-silica and oil-oil interactions was the primary interest in this study rather than obtaining neat chromatographic separation of diluted oil compounds.

Figure 1. A. Image of developed TLC plate in visible light, B. fluorescence with 254 nm UV, and C. total ion count (heat map) from LDI FT-ICR-MS.


Page 8 of 28

Page 9 of 28 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

Energy & Fuels

Figure 1 shows three images of the same plate as observed in visible light (A), fluorescent emission with 254 nm excitation light (B) and by the heat map of MS imaging where the total ion count over the mass range m/z 250 – 600 was summed for each pixel (C). The dotted lines indicate the original sample application position (1) and the three solvent fronts of DCM / MeOH (2), toluene / n-heptane (3) and n-heptane (4) development. Isolated SARA-proxy spots with some tailing in between the spots were observed for the P2 at dotted line 1, P1 just below line 2 and aromatics between lines 2 and 3. None of the applied readout techniques detected the saturates fraction. The most striking feature in the images is the bifurcation of the aromatics spot which was split into two equal branches. Data in Figure S1 in the supplement showed that this phenomenon was completely absent for another oil type. Some forcing in the chromatographic process leads to the observed demixing of oil compounds from the saturates and the aromatics fractions. The capability to load large amount of oil and study its development in 2-dimensions as partly driven by oil-oil interactions is a unique feature of the TLC plate model. The bifurcation phenomenon weakened somewhat at lower sample loadings, and therefore could be interpret as an oil-oil interaction. Inspection of TLC separations in other publications showed that this phenomenon is frequent, but none of the papers discussed this effect

3, 27

. Overlay of the three images in Figure 1 allows the comparison of optical and mass

spectrometric characteristics of the fractions, where zooming in on the bifurcation region is of special interest. The colors of fractions in Figure 1A agree well with the observations described by Pantoja et al. 38

: colorless for saturates, yellow-orange for aromatics and brown for P1 and P2. The darkness

of spots, however, is a poor indicator for the sample concentrations due to large differences in light absorption properties of the oil fractions. First of all, the saturates that made up 60% of the oil sample were observed by neither of the techniques, i.e. visible light, absorbance of the UV 254 nm excited fluorescence light nor LDI-MS. The absence of any LDI-MS signal is explained by the absence of chromophores in saturates that can absorb the 355 nm laser light. Improvement of the LDI-MS signal by application of silver-ion matrices

39

might be possible but was out of

the scope of this study. The aromatics fraction, which made up 26% of the oil, showed only a weak yellow band in visible light (Figure 1A). Examples of non-alkyl-substituted core structures that absorb light of 9 ACS Paragon Plus Environment

Energy & Fuels 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

this wavelength are four-ring aromatic compounds (C18H12, DBE of 12) as chrysene and tetracene, which substances are colored golden-yellow and pale orange, respectively. Fluorescence observed in Figure 1B after UV excitation revealed more details. There was a zone that emitted bright blue fluorescent light at the top and the inside of the bifurcated spot and there was a dark band at the bottom and outside of the spot. The dark band coincided both with the yellow band in Figure 1A and with the LDI-MS signal in Figure 1C. The blue fluorescent light zone stretched significantly further inwards the bifurcation cavity in Figure 1B than indicated by the location of the yellow band in the visible light image and the band of the LDI-MS signal in Figures 1A and 1C. Apparently inside the bifurcation gap there is a family of molecules that does not absorb any visible light and does not ionize with 355 nm laser light, but that becomes fluorescent when excited with 254 nm UV light. Emission of bright blue fluorescent light after UV excitation is characteristic for molecules with small aromatic cores 40. It can be concluded from the optical properties that oil molecules with aromatic cores composed of less than four aromatic rings migrated towards the top of the aromatics spot. The dark band at the bottom and outside of the bifurcated aromatics spot in Figure 1B was caused either by absorption of incoming UV excitation light or by absorption of emitted fluorescent visible light of chromophores in the aromatic molecules. The dark band coincided with the yellow colored band in Figure 1A and the LDI-MS signal in Figure 1C. Apparently the molecules in this band also interact strongly with 355 nm laser light, resulting in desorption and ionization of those molecules. These observations were indicative for the presence of stronger chromophores in the dark zone but the exact mechanism for the dimmed light remains inconclusive. The light absorption might be caused by larger aromatic core structures, the absorption of which is known to scale with their size. But it also might be the specific (conjugated) functional groups that harboring lone pair electrons from heteroatoms in the oil molecules contributing to the light absorption. The P1 fraction made up to 13% of the oil as quantified by TLC-FID. The P1 spot in Figures 1A and 1B gave the darkest bands in the visible and fluorescent light images, indicating that many P1 molecules had substantial light extinction coefficients. The bright background fluorescence from the silica plate made it impossible to infer conclusions about fluorescent properties of the P1 and the P2 fractions. Although the P1 content (13%) in the oil is lower than the aromatics 10 ACS Paragon Plus Environment

Page 10 of 28

Page 11 of 28 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

Energy & Fuels

fraction (26%), Figure 1C displays that the strongest LDI-MS signal came from the P1 fraction. Apparently the oil molecules in P1 fraction were desorbed and ionized more efficiently by the LDI-MS process than the aromatics fraction. This result also implies that a large and probably a major portion of the aromatics fraction remained undetected by LDI-MS. The small amount of only 1% P2 fraction still provided dark spots in Figures 1A and 1B. The lack of signal from the P2 fraction can be attributed to its low concentration combined with the large binding energy of polar molecules with the silica surface

28, 41

that could diminish

desorption efficiency on LDI. LDI - Mass spectral imaging The previous section concluded that the largest fraction of our light crude oil was not detected by LDI-MS and some inferences were derived from optical properties. The next challenge is to establish the molecular space that was probed by LDI-MS and to link molecular characteristics with speciation on the TLC plate. Figure 2 shows the LDI-FTICR mass spectra of the whole crude oil and the spots from the aromatics and P1 fractions. All samples were measured under identical experimental conditions and the ratio of the total abundance of the aromatics to the P1 fraction was about 1:2, which also quantified the result in the heat map in Figure 1C.


Energy & Fuels 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

Intens. x107

Page 12 of 28

Crude Oil_1

Whole oil

273.1512

1.5

1.0 367.2296 0.5

0.0 250 Intens. x106

300

350

400

450

500

550

600

m/z

new_aromatic

Aromatics spot

371.2607 317.2139

2.0 1.5 1.0 0.5 0.0 250 Intens. x107

300

350

400

450

500

550

600

m/z

new_resin

P1 spot

273.1517

4

3

353.2146

2

1

0 250

300

350

400

450

500

550

600

m/z

Figure 2. The FTICR mass spectra of the whole oil, and the spots of the aromatics and the P1 fraction. The base peaks in spectra of the P1 and aromatics spots were identified as C20H19N1 (m/z 273.1512) and C27H33N1 (m/z 371.2607), respectively. Both peaks belonged to the same homologue series of the N1 class with DBE 12. Although no isomer information is obtained by LDI-MS, carbazoles derivatives are usually the most abundant N1 class compounds in crude oils 12 ACS Paragon Plus Environment

Page 13 of 28 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

Energy & Fuels

42

and therefore these mass peaks can tentatively be assigned to benzocarbazoles core structures

with four and 11 additional carbon atoms as methylene groups in alkyl-substituents, respectively. This simple example clearly illustrates that within one compound class and type, i.e. the same heteroatom content and DBE, the molecules with increasing alkyl-substitution were more mobilized under normal-phase chromatographic conditions. This trend fully agrees with results of other studies 15, 28 and will be used for further interpretation.

Figure 3. Distribution of N1 DBE12 compounds between the aromatic (top) and P1 (bottom) spots. The example in Figure 3 visualizes in more detail the capability of MS imaging to study how N1 DBE 12 compounds with C4 through C10 alkyl substitution were distributed over the P1 and the aromatics spot. Again, no isomer information was obtained by LDI-MS, but it can be expected that compounds with N-atoms that were shielded by alkyl-substituents at their vicinal positions move with the mobile phase to the aromatics spot, while isomers equal number of CH2-groups but with exposed N-atoms were more retained by the silica stationary phase through hydrogen bonding 15. Next to the major mass peaks of the N1 DBE 12 ion-series, the crude oil spectrum may contain thousands of other mass peaks. The mass spectra were processed by PetroOrg software which assigns elemental compositions to mass peaks and also groups the assigned mass peaks into heteroatom classes. PetroOrg software has the option to add up the signals of radical-cations (M+.) and protonated ions (M.H+) that both can be formed as follows by LDI of the same


Energy & Fuels

molecular species

43

. The possibility of at least two ionization pathways required some further

evaluation of LDI-MS spectra, which is done in the supplement information. Figure 4 shows the class distribution of the sum of radical-cations and protonated ions of the N1, hydrocarbon (HC), O1, N1O1, S1, N1S1 and N2 classes as measured by LDI-MS imaging for the whole oil, the aromatics and the P1 fractions. The class selection was based on two criteria. First, the relative abundance was at least 1 percent in either one of the three spectra. Second, only relative abundances of more than 0.5% were reported. 90%

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 14 of 28

80%

Whole Oil

70%

Aromatics Spot

60%

P1 Spot

50% 40% 30% 20% 10% 0% N1

HC

O1

N1 O1

S1

N1 S1

N2

Heteroatom Class

Figure 4. Class distribution plot from summed intensities of radical-cations and protonated ions. In all samples the N1 class was most prevalent, followed by the hydrocarbon class. It was striking that the six major classes of the whole oil were also the major classes in both the P1 and the aromatics fractions. The minor N2 class was not detected above 0.5% abundance in the aromatics fraction. As observed through the selective LDI-MS ionization filter the applied chromatography provided the resolution to separate heteroatom free saturates from polar heteroatom bearing molecules, but the partitioning of the latter over the aromatics and the P1 fractions was gradual and non-specific on the heteroatom class basis. Closer comparison of the relative abundances in Figure 4 shows that the N1, N1O1 and N2 classes were enriched in the P1 fraction, while the HC, O1 and the S1 classes were enriched in the aromatics fraction. The relative abundance of the N1S1 class was about the same in all three 14 ACS Paragon Plus Environment

Page 15 of 28 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

Energy & Fuels

samples. In terms of oil-SiO2 interactions these results can be interpreted that the group of Ncontaining molecules was more strongly bond by the silica and therefore ended up in the more retained P1 fraction, while O1 and S1 molecules behaved more like hydrophobic entities that express less interaction with silica and dissolved better in the 80/20 toluene / n-heptane mobile phase.


Energy & Fuels 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 16 of 28

Figure 5. DBE versus #C plots of summed intensities of radical-cations and protonated ions of the seven most abundant classes. Figure 5 shows the DBE versus carbon number plots for the summed intensities of protonated and radical-cations of the seven selected heteroatom classes for the whole oil and the spots of the aromatics and the P1 fractions. The contoured plots of the whole oil resemble the weighted sum of the plots for the P1 and aromatics fractions, where fraction weights are given in Figure 4. In the case of the HC, O1 and S1 classes the aromatics fraction contributed more to the whole oil signal, while in the case of the N1, N1O1, N1S1 and N2 classes the P1 fraction were more prevalent. Comparison of the P1 versus the aromatics fraction shows that the contoured plots overlap to a great extent, which is in agreement with the rather similar relative class distributions in Figure 4. Table 1 lists abundance weighted parameters for the molecular weight, number of carbon atoms, DBE and H/C ratio per heteroatom classes in the aromatics and the P1 fractions. In general, most of the parameter values per class differ by less than 10% between both fractions, but several systematic trends can be observed that differentiate them. Except for the O1 class, all H/C values are smaller in the P1 fraction, which indicates that the polar behavior of P1 molecules is at least partly related to their aromaticity. Table 1. Abundance weighted average data of molecular weight, carbon number, DBE and H/C as derived from PetroOrg for the aromatics and the P1 fractions. Class

Abundance w eighted average molecular w eight

Abundance w eighted average #C

Abundance w eighted average DBE

Abundance w eighted average H/C

Aromatic

P1

∆(Ar. - P1)

Aromatic

P1

∆(Ar. - P1)

Aromatic

P1

∆(Ar. - P1)

Aromatic

P1

∆(Ar. - P1)

N1

421

382

39

31

28

3

13.9

14.9

-0.9

1.21

1.06

0.15

HC

388

423

-35

30

32

-3

15.4

16.9

-1.5

1.06

1.06

0.00

O1

411

440

-28

30

32

-2

15.9

15.4

0.5

1.05

1.15

-0.10

N1 O1

448

433

15

32

31

1

16.9

15.5

1.4

1.15

1.10

0.05

S1

398

410

-12

28

29

-1

15.7

17.2

-1.5

1.02

0.93

0.10

N1 S1

437

400

37

30

28

3

16.8

17.2

-0.4

1.15

0.89

0.26

N2

489

35

19.5

1.07

All the N-containing classes in the aromatics fraction are of higher molecular weight combined with a lower degree of aromaticity (H/C). The higher molecular weight of 421, 448 and 437 Dalton versus 382, 433 and 400 Dalton for N1, N1O1 and N1S1, respectively, implied that higher degrees of alkylation of N-compounds reduced the silica interaction and improved the 16 ACS Paragon Plus Environment

Page 17 of 28 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

Energy & Fuels

solubility in the mobile phases

15, 28

as was illustrated by Figure 3. The H/C ratios for the

corresponding classes were 1.21, 1.15 and 1.15 versus 1.06, 1.10 and .89, respectively. The effect of aromaticity is well established by the normal-phase chromatographic retention of the N1 class alkylcarbazoles (DBE 9) and benzocarbazoles (DBE 12) model compounds on silica 15. Benzocarbazoles are much more retained compared to tetramethyl-carbazoles, i.e. both C16N1 compounds but different types with DBE 12 versus 9, which emphasizes a significant increase in adsorption strength by the addition of more aromatic carbon-atoms in N-compounds. The adsorption strength of asphaltenes to silica was also reported to increase with increasing aromaticity of the molecules 28. On the contrary, lower molecular weight (388 versus 425 Dalton) combined with lower DBE (15.4 versus 16.9) were observed for the HC class in the aromatics fraction while the H/C ratio of 1.06 was similar in both fractions. Apparently, the partitioning of non-polar aromatic hydrocarbon molecules between silica and mobile phase depends on molecular size and absolute number of DBEs. A similar effect was observed for the S1 class, although the different H/C ratio of 1.02 versus 0.89 implied that an aromaticity effect may also play a role. The O1 class displayed a higher aromaticity combined with a lower molecular weight for the aromatics fraction. This may point to a different functionality of the oxygen atom in both fractions. For example, the P1 fraction may contain more phenolic oxygen while the aromatics fraction may contain more aromatic fluorenone or furan derivatives. Strong interactions of NO-classes from asphaltenes with SiO2 have been reported

4, 44

. This was

the class with the highest molecular weight in the aromatics fraction, which supports the strong binding of N1O1 functionalities. It should be noted that the observed H/C values in the range 0.89 to 1.21 were very low, especially for a light oil type. All H/C ratios for classes from the P1 fraction were smaller than 1.15, which is a common value for the most aromatic oil fraction, namely the asphaltenes 36. This result emphasizes the strong bias of LDI-MS for compounds with condensed aromatic structures and that LDI-MS only detects the small subfraction of molecules that easily ionize with two 3.5 eV photons.


Energy & Fuels 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

Principal component analysis (PCA) is frequently employed to elucidate and visualize individual m/z-features which relate to the major differences in an MS image plot as Figure 1C. Figure 6A shows an RGB-score plot image wherein the P1 and aromatics fractions are separated along the first principal component (PC1) in the PC1+ and PC1- direction, respectively. This report explored mainly the visualization of score plot images, and therefore the PCA interpretation was limited to a high level analysis. The top-30 mass peaks with the highest weighted scaled loadings are listed with their structural details in Table 2. All reported DBE values were integer values which indicate that the observed masses were radical-cations. All top-30 PC1+ loadings for the P1 fraction belonged to the N1 class with an average H/C of 0.98±0.03 and a DBE of 14.0±0.5, where the error is the standard error of the mean, ±(tn-1 * σ / √n) at p 0.05. The top-30 PC1- loadings for the aromatics fraction consisted of 18 HC class mass peaks with an average H/C of 1.03±0.04 and a DBE of 12.8±0.5 and twelve N1 class masses with an average H/C of 1.21±0.04 and a DBE of 10.9±0.4. From these results it can be concluded that the presence of an N-atom in combination with a relatively high degree of aromaticity results in stronger retention. The PCA interpretation can be extended with analyses of higher PCA functions and mass peaks of other heteroatom classes with lower loadings. However, processing the large amount of information generated by such analysis of oil data was outside the scope of our study. The PCA not only confirmed the results in Figures 5 and 6, namely that the N1 class with high aromatic signature was specific for the P1 fraction and HC class elements were specific for the aromatics fraction, but the added value of PCA is the identification of those (top-30) specific features that contribute most to the differentiation between both fractions. The colored bands in the plots imply that PC-loadings reveal information about groups of oil constituents that express similar SiO2 interactions.


Page 18 of 28

Page 19 of 28 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

Energy & Fuels

A

B

C

Figure 6. RGB plots of three PCAs. A. PCA of whole the TLC LDI-MS image with PC1+ (green), PC1(red) and PC2+ (blue); the P1 and aromatics spots are differentiated by PC1. B. PCA of the aromatics spot with PC2+ (red), PC2- (green) and PC1+ (blue); PC2 differentiates the top and the bottom of the aromatics spot and C. PCA of the P1 spot with PC1- (red), PC1+ (green) and PC3+ (blue); PC1 differentiates the top and the bottom of the P1 spot.

MS imaging with high spatial resolution allows performing PCA to study chemical speciation within the individual aromatics and P1 spots as illustrated by Figures 6B and 6C. The first figure shows that the top and the bottom of the curved shaped aromatics spot are differentiated by PC2. The corresponding 30 highest loadings are listed in Table 3, which are all radical-cations. All loading for the most retained bottom-zone described by PC2- were N1 class components, while all loading for the top-zone described by PC2+ are HC class components. In conclusion, within the aromatics spot N1-containing compounds were slightly more retained by the silica and therefore they were concentrated in a rather narrow band at the bottom of the spot. Figure 6C shows that the bottom and top of the P1 spot are differentiated by PC1. Table 4 lists the corresponding 30 highest scaled loadings, which are all radical-cations again. The PC1loadings for the top-zone were all N1 class compounds with an average H/C of 1.07±0.03 and a DBE of 13.3±0.4. The highest PC1+ loadings corresponding to the bottom-zone encompass also 13 N1 class compounds, which were characterized by a higher degree of aromaticity that was reflected by an average H/C of 0.84±0.06 and a DBE of 16.2±0.08. Also this result confirms again that the degree of aromaticity is an important parameter that determines the retention on 19 ACS Paragon Plus Environment

Energy & Fuels 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

silica. The remaining top-30 PC1- loadings were 16 highly aromatic hydrocarbons with an average H/C of 0.99±0.05 and a DBE of 14.8±0.7 and one N1O1 class compound. PCA of LDI-MS images confirmed that the retention of N-class compounds was favored by a low degree of alkylation, large aromatic core structures and nitrogen containing heteroatom classes. Figures 6B and C show that the heteroatom content was not distributed uniformly over the proxy SARA spots but that heteroatom classes were concentrated in chromatographic bands. The identification of this phenomenon implies that accurate TLC-FID quantification should take chromatographic banding of heteroatom classes into account. The lack of such correction may explain some of the variance observed when TLC-FID results are correlated with other techniques. Finally, we want to speculate about the origin of the bifurcation of the aromatics spot in Figures 1, 3 and 6. Obviously, the high sample loading enforced some sort of oil-oil interactions. The effect might result from enhanced viscosity caused by locally high concentration of dissolved high molecular weight saturates as paraffins or waxes for example. Dissolution of such compounds would increase the viscosity of the otherwise low viscous n-heptane mobile phase and consequently locally reduced the mobility of the liquid phase. Experiments with lower sample loadings indeed showed higher mobility of the bifurcation spot. When the solvent front of the second elution with 80/20 toluene / n-heptane hits the tail of the saturates fraction the viscosity would locally increase and slowed down or even blocked further solvent flow. This simple description, however, does not explain the presence of the blue fluorescent molecules inside the bifurcation gap. Apparently more complex molecular interactions occur in concentrated systems. Such interactions can be studied by MS imaging while in regular liquid chromatography they would give rise to undesired viscous finger effects.

CONCLUSIONS: Comparison of (semi)quantitative TLC-FID and qualitative TLC-LDI mass spec imaging results demonstrated that for light oil only a minor fraction of the compositional space of all oil components is observed by LDI-MS. The two-photon ionization mechanism at 355 nm only provides information about the oil components with ionization energies less than 7.0 eV, which 20 ACS Paragon Plus Environment

Page 20 of 28

Page 21 of 28 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

Energy & Fuels

are compounds with alkyl-substituted condensed aromatic rings. LDI-MS with a 355-nm laser source efficiently ionized a subfraction of the P1 and aromatics, which was emphasized by the low H/C values of the observed classes in the aromatics (average 1.11) and especially the P1 (1.04) fractions. In our experiment no saturates or P2 were detected. Despite the molecular space that was probed by LDI-MS is limited to polar aromatic molecules, our results demonstrated that TLC-LDI-FTICR MS imaging was a suitable model to study oil– SiO2 and oil–solvent interactions of those compounds that were detected. With high spatial resolution chromatographic trends in retention behavior could be monitored and were shown to depend on molecular features as the degree of alkylation and the degree of aromaticity. Although the on-line readout provides no direct information about isomers, based on examples as the differentiation of one type of compounds over both the P1 and the aromatics spots tentative conclusions can be inferred about shielded versus exposed functional groups. In addition, high sample loading was possible on the silica plate to enforce oil-oil interactions between different oil groups as manifested by the bifurcation phenomenon. It was demonstrated that MS imaging data of oil could be analyzed much deeper by principal component analysis, which could reveal a wealth of information. ACKNOWLEDGMENTS

The work is part of the LINK program which is financially supported by the Dutch Province of Limburg. We thank Ton van Loef and Irina Cotiuga from Latexfalt B.V., Koudekerk aan den Rijn, the Netherlands for the Iatroscan measurements and Gert Eijkel for assistance with the principal component analyses. We thank Erik Tegelaar, Pim Mul and Frans Korndorffer from Shell Global Solutions International B.V for providing feedback to our work and the funding support from Shell Global Solutions International B.V. AUTHOR INFORMATION Corresponding Author * Telephone: Tel: +31-433881499. Fax: +31 43 388 415. E-mail: [email protected].


Energy & Fuels 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 22 of 28

Page 23 of 28 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

Energy & Fuels

Table 2. Top-30 scaled loadings of PC1 of the whole TLC LDI-MS image in Figure 6A. P1 fraction in PC1+ (green in Figure 6A) Rank m/z (exp) Scaled Loading Formula 1 273.1512 34.6 C20H19N1 2 287.1668 23.4 C21H21N1 3 301.1825 13.7 C22H23N1 4 339.1983 10.5 C25H25N1 5 353.2138 9.9 C26H27N1 6 323.1668 8.6 C24H21N1 7 309.1511 8.2 C23H19N1 8 325.1826 8.2 C24H23N1 9 327.1982 7.7 C24H25N1 10 313.1826 7.7 C23H23N1 11 367.2294 7.1 C27H29N1 12 315.1980 7.0 C23H25N1 13 337.1824 6.7 C25H23N1 14 299.1667 6.7 C22H21N1 15 274.1545 5.9 C19(13C)H19N1 16 351.1981 5.9 C26H25N1 17 365.2133 5.8 C27H27N1 18 341.2137 5.5 C25H27N1 19 379.2290 5.5 C28H29N1 20 288.1702 5.2 C20(13C)H21N1 21 311.1668 5.0 C23H21N1 22 295.1354 4.7 C22H17N1 23 381.2449 4.3 C28H31N1 24 395.2606 4.3 C29H33N1 25 393.2448 3.9 C29H31N1 26 285.1511 3.8 C21H19N1 27 375.1978 3.7 C28H25N1 28 377.2135 3.4 C28H27N1 29 363.1977 3.3 C27H25N1 30 349.1824 3.3 C26H23N1

Class N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1

#C 20 21 22 25 26 24 23 24 24 23 27 23 25 22 20 26 27 25 28 21 23 22 28 29 29 21 28 28 27 26

H/C 0.95 1.00 1.05 1.00 1.04 0.88 0.83 0.96 1.04 1.00 1.07 1.09 0.92 0.95 0.95 0.96 1.00 1.08 1.04 1.00 0.91 0.77 1.11 1.14 1.07 0.90 0.89 0.96 0.93 0.88

Aromatics fraction in PC1- (red in Figure 6A) DBE Rank m/z (exp) Scaled Loading Formula 12 1 272.1557 -6.1 C21H20 12 2 286.1713 -6.1 C22H22 12 3 291.1979 -5.6 C21H25N1 14 4 317.2134 -5.5 C23H27N1 14 5 331.2290 -5.3 C24H29N1 15 6 300.1868 -5.1 C23H24 15 7 345.2444 -5.1 C25H31N1 14 8 366.2333 -4.9 C28H30 13 9 303.1978 -4.8 C22H25N1 13 10 312.1868 -4.7 C24H24 14 11 326.2023 -4.6 C25H26 12 12 277.1821 -4.6 C20H23N1 15 13 340.2179 -4.5 C26H28 13 14 359.2600 -4.5 C26H33N1 12 15 284.1556 -4.4 C22H20 15 16 352.2177 -4.2 C27H28 15 17 314.2024 -4.2 C24H26 13 18 373.2755 -4.2 C27H35N1 15 19 288.1869 -4.1 C22H24 12 20 274.1713 -4.1 C21H22 14 21 270.1401 -4.0 C21H18 15 22 371.2600 -3.9 C27H33N1 14 23 354.2333 -3.9 C27H30 14 24 289.1823 -3.9 C21H23N1 15 25 368.2488 -3.9 C28H32 13 26 305.2134 -3.9 C22H27N1 17 27 298.1712 -3.8 C23H22 16 28 338.2023 -3.8 C26H26 16 29 399.2907 -3.6 C29H37N1 16 30 380.2487 -3.6 C29H32


Class HC HC N1 N1 N1 HC N1 HC N1 HC HC N1 HC N1 HC HC HC N1 HC HC HC N1 HC N1 HC N1 HC HC N1 HC

#C 21 22 21 23 24 23 25 28 22 24 25 20 26 26 22 27 24 27 22 21 21 27 27 21 28 22 23 26 29 29

H/C 0.95 1.00 1.19 1.17 1.21 1.04 1.24 1.07 1.14 1.00 1.04 1.15 1.08 1.27 0.91 1.04 1.08 1.30 1.09 1.05 0.86 1.22 1.11 1.10 1.14 1.23 0.96 1.00 1.28 1.10

DBE 12 12 10 11 11 12 11 14 11 13 13 10 13 11 13 14 12 11 11 11 13 12 13 11 13 10 13 14 12 14

Energy & Fuels 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

Page 24 of 28

Table 3. Top-30 scaled loadings of PC2 of the aromatics spot in Figure 6B. Top of aromatics spot in PC2+ (red in Figure 6B) Rank m/z (exp) Scaled Loading Formula 1 288.1869 11.1 C22H24 2 274.1713 8.9 C21H22 3 314.2024 7.9 C24H26 4 302.2024 7.5 C23H26 5 276.1869 7.1 C21H24 6 290.2025 7.1 C22H26 7 342.2334 6.8 C26H30 8 316.2179 6.8 C24H28 9 328.2180 6.7 C25H28 10 356.2489 6.6 C27H32 11 300.1868 6.4 C23H24 12 368.2488 6.4 C28H32 13 370.2644 6.1 C28H34 14 354.2333 5.7 C27H30 15 286.1713 5.6 C22H22 16 340.2179 5.5 C26H28 17 330.2335 5.2 C25H30 18 344.2490 5.2 C26H32 19 366.2333 5.0 C28H30 20 304.2181 4.8 C23H28 21 382.2643 4.5 C29H34 22 396.2796 4.5 C30H36 23 358.2645 4.3 C27H34 24 380.2487 4.2 C29H32 25 394.2642 4.1 C30H34 26 352.2177 4.1 C27H28 27 408.2796 3.9 C31H36 28 326.2023 3.8 C25H26 29 398.2953 3.6 C30H38 30 318.2336 3.6 C24H30

Class HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC HC

#C 22 21 24 23 21 22 26 24 25 27 23 28 28 27 22 26 25 26 28 23 29 30 27 29 30 27 31 25 30 24

H/C 1.09 1.05 1.08 1.13 1.14 1.18 1.15 1.17 1.12 1.19 1.04 1.14 1.21 1.11 1.00 1.08 1.20 1.23 1.07 1.22 1.17 1.20 1.26 1.10 1.13 1.04 1.16 1.04 1.27 1.25

Bottom of aromatics spot in PC2- (green in Figure 6B) DBE Rank m/z (exp) Mass error in ppm Formula Class 0.1770 11 1 367.2294 C27H29N1 N1 0.7124 11 2 289.1823 C21H23N1 N1 -0.1852 12 3 313.1826 C23H23N1 N1 -0.0245 11 4 327.1982 C24H25N1 N1 -0.0849 10 5 353.2138 C26H27N1 N1 0.9127 10 6 369.2448 C27H31N1 N1 0.1700 12 7 341.2137 C25H27N1 N1 0.6756 11 8 355.2292 C26H29N1 N1 1.0884 12 9 303.1978 C22H25N1 N1 0.4616 12 10 381.2449 C28H31N1 N1 1.2807 12 11 277.1821 C20H23N1 N1 0.9135 13 12 291.1979 C21H25N1 N1 0.3577 12 13 299.1667 C22H21N1 N1 1.2412 13 14 343.2290 C25H29N1 N1 1.7075 12 15 357.2445 C26H31N1 N1 0.1501 13 16 273.1512 C20H19N1 N1 -0.3243 11 17 339.1983 C25H25N1 N1 1.1633 11 18 317.2134 C23H27N1 N1 1.9151 14 19 371.2600 C27H33N1 N1 1.0830 10 20 275.1666 C20H21N1 N1 -0.1062 13 21 301.1825 C22H23N1 N1 0.3331 13 22 315.1980 C23H25N1 N1 0.7837 11 23 329.2135 C24H27N1 N1 0.1637 14 24 287.1668 C21H21N1 N1 3.0436 14 25 395.2595 C29H33N1 N1 1.9595 14 26 383.2600 C28H33N1 N1 3.0420 14 27 409.2752 C30H35N1 N1 3.2219 13 28 397.2751 C29H35N1 N1 1.2801 12 29 331.2290 C24H29N1 N1 2.4035 10 30 385.2755 C28H35N1 N1


#C 27 21 23 24 26 27 25 26 22 28 20 21 22 25 26 20 25 23 27 20 22 23 24 21 29 28 30 29 24 28

H/C 1.07 1.10 1.00 1.04 1.04 1.15 1.08 1.12 1.14 1.11 1.15 1.19 0.95 1.16 1.19 0.95 1.00 1.17 1.22 1.05 1.05 1.09 1.13 1.00 1.14 1.18 1.17 1.21 1.21 1.25

DBE 14 11 13 13 14 13 13 13 11 14 10 10 13 12 12 12 14 11 12 11 12 12 12 12 14 13 14 13 11 12

Page 25 of 28 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

Energy & Fuels

Table 4. Top-30 scaled loadings of PC1 of the P1 spot in Figure 6C. Bottom of the P1 spot in PC1+ (green in Figure 6C) Rank m/z (exp) Scaled Loading Formula Class 1 309.1511 5.2 C23H19N1 N1 2 323.1668 4.1 C24H21N1 N1 3 295.1354 3.2 C22H17N1 N1 4 359.1664 2.1 C27H21N1 N1 5 373.1819 1.8 C28H23N1 N1 6 347.1665 1.6 C26H21N1 N1 7 345.1509 1.5 C26H19N1 N1 8 364.2177 1.5 C28H28 HC 9 336.1868 1.4 C26H24 HC 10 361.1821 1.4 C27H23N1 N1 11 350.2023 1.3 C27H26 HC 12 378.2332 1.2 C29H30 HC 13 374.2021 1.2 C29H26 HC 14 337.1824 1.2 C25H23N1 N1 15 366.2333 1.2 C28H30 HC 16 324.1868 1.2 C25H24 HC 17 352.2177 1.1 C27H28 HC 18 338.2023 1.1 C26H26 HC 19 310.1545 1.1 C22(13C)H19N1 N1 20 333.1509 1.1 C25H19N1 N1 21 360.1867 1.1 C28H24 HC 22 322.1712 1.1 C25H22 HC 23 376.2177 1.0 C29H28 HC 24 310.1712 1.0 C24H22 HC 25 380.2487 1.0 C29H32 HC 26 371.2235 0.9 C26H29N1O1 N1O1 27 324.1700 0.9 C23(13C)H21N1 N1 28 447.2902 0.9 C33H37N1 N1 29 392.2486 0.9 C30H32 HC 30 370.2644 0.9 C28H34 HC

#C 23 24 22 27 28 26 26 28 26 27 27 29 29 25 28 25 27 26 23 25 28 25 29 24 29 26 24 33 30 28

H/C 0.83 0.88 0.77 0.78 0.82 0.81 0.73 1.00 0.92 0.85 0.96 1.03 0.90 0.92 1.07 0.96 1.04 1.00 0.83 0.76 0.86 0.88 0.97 0.92 1.10 1.12 0.88 1.12 1.07 1.21

Top of the P1 spot in PC1- (red in Figure 6C) DBE Rank m/z (exp) Scaled Loading Formula -11.7 15 1 301.1825 C22H23N1 -11.2 15 2 353.2138 C26H27N1 -10.7 15 3 367.2294 C27H29N1 -9.3 18 4 287.1668 C21H21N1 -9.3 18 5 339.1983 C25H25N1 -8.9 17 6 315.1980 C23H25N1 -8.5 18 7 327.1982 C24H25N1 -8.2 15 8 341.2137 C25H27N1 -7.7 15 9 381.2449 C28H31N1 -6.7 17 10 355.2292 C26H29N1 -6.2 15 11 313.1826 C23H23N1 -5.9 15 12 329.2135 C24H27N1 -5.3 17 13 395.2606 C29H33N1 -4.5 15 14 369.2448 C27H31N1 -4.1 14 15 393.2448 C29H31N1 -4.0 14 16 379.2290 C28H29N1 -4.0 14 17 325.1826 C24H23N1 -3.8 14 18 343.2290 C25H29N1 -3.5 15 19 383.2604 C28H33N1 -3.1 17 20 407.2603 C30H33N1 -2.8 17 21 299.1667 C22H21N1 -2.8 15 22 357.2445 C26H31N1 -2.6 16 23 421.2759 C31H35N1 -2.5 14 24 340.2017 C24(13C)H25N1 -2.5 14 25 365.2133 C27H27N1 -2.4 13 26 302.1857 C21(13C)H23N1 -2.4 15 27 382.2482 C27(13C)H31N1 -2.4 16 28 288.1702 C20(13C)H21N1 -2.3 15 29 423.2913 C31H37N1 -2.3 12 30 273.1512 C20H19N1


Class N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1 N1

#C 22 26 27 21 25 23 24 25 28 26 23 24 29 27 29 28 24 25 28 30 22 26 31 25 27 22 28 21 31 20

H/C 1.05 1.04 1.07 1.00 1.00 1.09 1.04 1.08 1.11 1.12 1.00 1.13 1.14 1.15 1.07 1.04 0.96 1.16 1.18 1.10 0.95 1.19 1.13 1.00 1.00 1.05 1.11 1.00 1.19 0.95

DBE 12 14 14 12 14 12 13 13 14 13 13 12 14 13 15 15 14 12 13 15 13 12 15 14 15 12 14 12 14 12

Energy & Fuels 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

REFERENCES

1. Adams, J. J., Asphaltene Adsorption, a Literature Review. Energ Fuel 2014, 28 (5), 2831-2856. 2. Strand, S.; Puntervold, T.; Austad, T., Water based EOR from clastic oil reservoirs by wettability alteration: A review of chemical aspects. J Petrol Sci Eng 2016, 146, 1079-1091. 3. Smith, D. F.; McKenna, A. M.; Corilo, Y. E.; Rodgers, R. P.; Marshall, A. G.; Heeren, R. M. A., Direct Analysis of Thin-Layer Chromatography Separations of Petroleum Samples by Laser Desorption Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Imaging. Energy & Fuels 2014, 28 (10), 6284-6288. 4. Parida, S. K.; Dash, S.; Patel, S.; Mishra, B. K., Adsorption of organic molecules on silica surface. Adv Colloid Interfac 2006, 121 (1-3), 77-110. 5. Terra, L. A.; Filgueiras, P. R.; Tose, L. V.; Romao, W.; de Castro, E. V. R.; de Oliveira, L. M. S. L.; Dias, J. C. M.; Vaz, B. G.; Poppi, R. J., Laser desorption ionization FT-ICR mass spectrometry and CARSPLS for predicting basic nitrogen and aromatics contents in crude oils. Fuel 2015, 160, 274-281. 6. Tissot, B. P.; Welte, D. H., Petroleum Formation and Occurrence. Springer-Verlag Berlin Heidelberg: 1984. 7. Mullins, O. C.; Sheu, E. Y.; Hammami, A.; Marshall, A. G., Asphaltenes, heavy oils, and petroleomics. Springer Science & Business Media: 2007. 8. Rodgers, R. P.; McKenna, A. M., Petroleum Analysis. Anal Chem 2011, 83 (12), 4665-4687. 9. Marshall, A. G.; Rodgers, R. P., Petroleomics: Chemistry of the underworld. Proceedings of the National Academy of Sciences 2008, 105 (47), 18090-18095. 10. Liu, P.; Shi, Q.; Chung, K. H.; Zhang, Y.; Pan, N.; Zhao, S.; Xu, C., Molecular characterization of sulfur compounds in Venezuela crude oil and its SARA fractions by electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Energy & Fuels 2010, 24 (9), 5089-5096. 11. Acevedo, S. c.; Guzmán, K.; Labrador, H.; Carrier, H.; Bouyssiere, B.; Lobinski, R., Trapping of metallic porphyrins by asphaltene aggregates: a size exclusion microchromatography with highresolution inductively coupled plasma mass spectrometric detection study. Energy & Fuels 2012, 26 (8), 4968-4977. 12. Blomberg, J.; Mes, E. P. C.; Schoenmakers, P. J.; van der Does, J. J. B., Characterization of complex hydrocarbon mixtures using on-line coupling of size-exclusion chromatography and normalphase liquid chromatography to high-resolution gas chromatography. Journal of High Resolution Chromatography 1997, 20 (3), 125-130. 13. Lobodin, V. V.; Rodgers, R. P.; Marshall, A. G., Petroleomics and the analysis of complex organic mixtures with Fourier transform ion cyclotron resonance. See Ref 2012, 2, 415-42. 14. Rodgers, R. P.; Marshall, A. G., Petroleomics: Advanced characterization of petroleum-derived materials by Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS). Asphaltenes, Heavy Oils, and Petroleomics 2007, 63-93. 15. Li, M. W.; Larter, S. R.; Frolov, Y. B.; Bjoroy, M., Adsorptive Interaction between NitrogenCompounds and Organic and/or Mineral Phases in Subsurface Rocks - Models for Compositional Fractionation of Pyrrolic Nitrogen-Compounds in Petroleum during Petroleum Migration. Hrc-J High Res Chrom 1994, 17 (4), 230-236. 16. Marshall, A. G.; Rodgers, R. P., Petroleomics: The next grand challenge for chemical analysis. Accounts Chem Res 2004, 37 (1), 53-59. 17. Terra, L. A.; Filgueiras, P. R.; Tose, L. V.; Romao, W.; de Souza, D. D.; de Castro, E. V. R.; de Oliveira, M. S. L.; Dias, J. C. M.; Poppi, R. J., Petroleomics by electrospray ionization FT-ICR mass spectrometry coupled to partial least squares with variable selection methods: prediction of the total acid number of crude oils. Analyst 2014, 139 (19), 4908-4916. 26 ACS Paragon Plus Environment

Page 26 of 28

Page 27 of 28 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

Energy & Fuels

18. Panda, S. K.; Andersson, J. T.; Schrader, W., Mass-spectrometric analysis of complex volatile and nonvolatile crude oil components: a challenge. Anal Bioanal Chem 2007, 389 (5), 1329-1339. 19. Mullins, O. C., The Asphaltenes. Annu Rev Anal Chem 2011, 4, 393-418. 20. Bisht, H.; Reddy, M.; Malvanker, M.; Patil, R. C.; Gupta, A.; Hazarika, B.; Das, A. K., Efficient and Quick Method for Saturates, Aromatics, Resins, and Asphaltenes Analysis of Whole Crude Oil by ThinLayer Chromatography-Flame Ionization Detector. Energ Fuel 2013, 27 (6), 3006-3013. 21. Khan, S. A.; Sarfraz, S.; Price, D., TLC-FID Calibration and Accurate Weight Determination of SARA Fractions in Heavy Crude Oil. Petrol Sci Technol 2012, 30 (23), 2401-2406. 22. Anderson, D. M.; Mills, D.; Spraggins, J.; Lambert, W. S.; Calkins, D. J.; Schey, K. L., Highresolution matrix-assisted laser desorption ionization–imaging mass spectrometry of lipids in rodent optic nerve tissue. 2013. 23. Cho, Y.; Witt, M.; Kim, Y. H.; Kim, S., Characterization of Crude Oils at the Molecular Level by Use of Laser Desorption Ionization Fourier-Transform Ion Cyclotron Resonance Mass Spectrometry. Analytical Chemistry 2012, 84 (20), 8587-8594. 24. Smaniotto, A.; Montanari, L.; Flego, C.; Rizzi, A.; Ragazzi, E.; Seraglia, R.; Traldi, P., Can crude oils be distinguished by different component distribution obtained by laser desorption ionization mass spectrometry and evaluated by chemometrics? Rapid Commun Mass Sp 2008, 22 (10), 1597-1606. 25. Santos, V. G.; Fasciotti, M.; Pudenzi, M. A.; Klitzke, C. F.; Nascimento, H. L.; Pereira, R. C. L.; Bastos, W. L.; Eberlin, M. N., Fullerenes in asphaltenes and other carbonaceous materials: natural constituents or laser artifacts. Analyst 2016, 141 (9), 2767-2773. 26. Eijkel, G. B.; Kaletas, B. K.; van der Wiel, I. M.; Kros, J. M.; Luider, T. M.; Heeren, R. M. A., Correlating MALDI and SIMS imaging mass spectrometric datasets of biological tissue surfaces. Surf Interface Anal 2009, 41 (8), 675-685. 27. Chirinos, J.; Oropeza, D.; Gonzalez, J.; Ranaudo, M.; Russo, R. E., Determination of Vanadium/Nickel Proportionality in the Asphaltene Fraction of Crude Oil Using Thin-Layer Chromatography with Femtosecond Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry. Energ Fuel 2013, 27 (5), 2431-2436. 28. Chacon-Patino, M. L.; Blanco-Tirado, C.; Orrego-Ruiz, J. A.; Gomez-Escudero, A.; Combariza, M. Y., High Resolution Mass Spectrometric View of Asphaltene-SiO2 Interactions. Energ Fuel 2015, 29 (3), 1323-1331. 29. Kim, E.; Cho, E.; Moon, S.; Park, J. I.; Kim, S., Characterization of Petroleum Heavy Oil Fractions Prepared by Preparatory Liquid Chromatography with Thin-Layer Chromatography, High-Resolution Mass Spectrometry, and Gas Chromatography with an Atomic Emission Detector. Energ Fuel 2016, 30 (4), 2932-2940. 30. Fuhr, B. J.; Hawrelechko, C.; Holloway, L. R.; Huang, H. B., Comparison of bitumen fractionation methods. Energ Fuel 2005, 19 (4), 1327-1329. 31. Carbognani, L.; Gonzalez, M. F.; Pereira-Almao, P., Characterization of athabasca vacuum residue and its visbroken products. Stability and fast hydrocarbon group-type distributions. Energ Fuel 2007, 21 (3), 1631-1639. 32. Bisht, H.; Reddy, M.; Malvanker, M.; Patil, R. C.; Gupta, A.; Hazarika, B.; Das, A. K., Efficient and Quick Method for Saturates, Aromatics, Resins, and Asphaltenes Analysis of Whole Crude Oil by ThinLayer Chromatography–Flame Ionization Detector. Energy & Fuels 2013, 27 (6), 3006-3013. 33. Bissada, K. K.; Tan, J.; Szymczyk, E.; Darnell, M.; Mei, M., Group-type characterization of crude oil and bitumen. Part I: Enhanced separation and quantification of saturates, aromatics, resins and asphaltenes (SARA). Organic Geochemistry 2016, 95, 21-28. 34. Karlsen, D. A.; Larter, S. R., Analysis of Petroleum Fractions by Tlc-Fid - Applications to Petroleum Reservoir Description. Org Geochem 1991, 17 (5), 603-617.


Energy & Fuels 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

35. Holm, T., Aspects of the mechanism of the flame ionization detector. J Chromatogr A 1999, 842 (1-2), 221-227. 36. Speight, J. G., Chapter 2 Chemical and Physical Studies of Petroleum Asphaltenes. Developments in Petroleum Science 1994, 40, 7-65. 37. Bharati, S.; Rostum, G. A.; Loberg, R., Calibration and Standardization of Iatroscan (Tlc-Fid) Using Standards Derived from Crude Oils. Org Geochem 1994, 22 (3-5), 835-862. 38. Pantoja, P. A.; Lopez-Gejo, J.; Le Roux, G. A. C.; Quina, F. H.; Nascimento, C. A. O., Prediction of Crude Oil Properties and Chemical Composition by Means of Steady-State and Time-Resolved Fluorescence. Energ Fuel 2011, 25 (8), 3598-3604. 39. Mennito, A. S.; Qian, K. N., Characterization of Heavy Petroleum Saturates by Laser Desorption Silver Cationization and Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energ Fuel 2013, 27 (12), 7348-7353. 40. Ralston, C. Y.; Wu, X.; Mullins, O. C., Quantum yields of crude oils. Appl Spectrosc 1996, 50 (12), 1563-1568. 41. Wang, S. S.; Liu, Q.; Tan, X. L.; Xu, C. M.; Gray, M. R., Study of Asphaltene Adsorption on Kaolinite by X-ray Photoelectron Spectroscopy and Time-of-Flight Secondary Ion Mass Spectroscopy. Energ Fuel 2013, 27 (5), 2465-2473. 42. Li, M.; Larter, S. R.; Stoddart, D.; Bjoroy, M., Fractionation of pyrrolic nitrogen compounds in petroleum during migration: derivation of migration-related geochemical parameters. The Geochemistry of Reservoirs, Geological Society Special Publication 1995, 86, 103-123. 43. Balasanmugam, K.; Viswanadham, S. K.; Hercules, D. M., Characterization of polycyclic aromatic hydrocarbons by laser mass spectrometry. Anal Chem 1986, 58 (6), 1102-1108. 44. Reed, M. G., Retention of Crude Oil Bases by Clay-Containing Sandstone. Clays and Clay Minerals 1968, 16, 173-178.


Page 28 of 28

Evaluation of Thin-Layer Chromatography â Laser Desorption

Recommend Documents

Evaluation of Thin-Layer Chromatography â Laser Desorption