Atmospheric Pressure Photoionization and Laser Desorption

Subscriber access provided by NEW YORK UNIV

Article

Atmospheric Pressure Photoionization and Laser Desorption Ionization Coupled to Fourier Transform Ion Cyclotron Resonance Mass Spectrometry to Characterize Asphaltene Solubility Fractions: Studying the Link between Molecular Composition and Physical Behavior Estrella Rogel, Michael Moir, and Matthias Witt Energy Fuels, Just Accepted Manuscript • Publication Date (Web): 02 Jun 2015 Downloaded from http://pubs.acs.org on June 4, 2015

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.

Energy & Fuels 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 39

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

Atmospheric Pressure Photoionization and Laser Desorption Ionization Coupled to Fourier Transform Ion Cyclotron Resonance Mass Spectrometry to Characterize Asphaltene Solubility Fractions: Studying the Link between Molecular Composition and Physical Behavior. Estrella Rogel,1,* Michael Moir,1 Matthias Witt,2 1

Chevron Energy Technology Company, Richmond, CA 94801, USA. 2

Bruker Daltonik GmbH, 28359 Bremen, Germany.

* To whom correspondence should be addressed

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

ABSTRACT

In the present work, a series of asphaltene solubility fractions extracted from a heavy crude oil have been analyzed using laser desorption ionization (LDI) and atmospheric pressure photoionization (APPI)

coupled to Fourier transform ion cyclotron resonance mass

spectrometry. The objective is to find how molecular compositional data compares and correlates with information found using other techniques. In order to accomplish this comparison, a methodology to link the compositional information and the macroscopic behavior is presented and successfully used. In general, we found that average H/C ratios and molecular weights were lower, average densities and average solubility parameters larger than those determined by other techniques. Values obtained from APPI data were closer than those obtained by LDI data to the reference values. This is indicative that in order to estimate bulk properties, APPI might be better suited than LDI. Solubility parameter distributions obtained using APPI data for the classes were shift to larger values when compared with similar distributions obtained by the solubility profile test. Interestingly, the less soluble the fraction, the closer are the APPI distributions to the solubility profile distributions. All these results seem to be related to the preferential ionization of the most aromatic molecules in the sample. The preferential detection of high aromaticity molecules suggests that APPI might be particularly suitable to look into the molecular information of the less soluble and therefore, most troublesome molecules in petroleum. These are the molecules likely to form deposits. Further confirmation of these findings is now underway.


Page 2 of 39

Page 3 of 39

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

INTRODUCTION Understanding the relationship between composition of asphaltenes and their behavior is of central importance in the development of predictive, preventive and remediation tools for asphaltene deposition. However, asphaltene analysis is a difficult task: Asphaltenes are defined as a solubility class, and therefore, they are comprised of an incredibly large number of molecules that might have different chemical characteristics even though they share similar solubilities. This makes the task of understanding the nature of this fraction, based on its detailed chemical characterization, close to impossible. In an ideal world, all these components could be separated and analyzed individually. In the real world, however, this is a daunting task in the case of asphaltenes: earlier conservative estimates indicated that asphaltenes contain at least 105 different types of molecules.1 Currently, the most straightforward approach to look into the molecular structures of asphaltenes is the use of ultrahigh resolution mass spectrometry. Additionally, this technique proved to be successful in providing a glimpse into the extraordinary complexity of crude oils,2,3 shale oils,4 asphaltenes5,6 and asphaltene deposits7,8 among other materials. However, since these materials contain a wide range of components of diverse nature, no single ionization technique can ionize all of the compounds present in the material.4 In fact, detection is going to depend on the applied ionization technique and therefore may provide misleading information6. In practical terms, this means that different aspects can be studied depending on the ionization technique. On the other hand, there are some techniques that can ionize a larger range of different molecules in comparison to others. For instance, APPI can efficiently ionize polycyclic aromatic compounds whether they show basic, acidic or neutral characteristics7 while ESI (Electrospray Ionization) and APCI (Atmospheric pressure chemical ionization) preferentially ionize polar compounds.6


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

Asphaltene studies using mass spectrometry have shed light about different aspects of this petroleum fraction. For instance, it has allowed the identification of paraffins and other compounds trapped by asphaltenes that could be used as geochemical markers.8 In particular, the use of APPI Fourier transform ion cyclotron resonance mass spectrometry has been used to confirm that asphaltenes are more aromatic than maltenes containing the same carbon number,9 and to validate the compositional continuum of petroleum components model3 first developed by Boduszynski.10 Fourier transform ion cyclotron resonance mass spectrometry had great success in describing the compositional differences between asphaltenes obtained in the lab by pressuredrop and solvent-drop techniques11 as well as between field asphaltene deposits and those obtained at lab conditions.12 Comparisons of asphaltene bulk properties obtained using standard techniques and mass spectroscopy measurements have been published. Aromaticity,6,13 and molecular weight14,15 have been the preferred explored properties. Another less explored and sometimes overlooked aspect of a detailed composition characterization is linking molecular knowledge to the macroscopic behavior observed for the samples, i.e., tendency towards precipitation, reactivity, viscosity, etc. In the present work, laser desorption ionization (LDI) and atmospheric pressure photoionization (APPI) were used to study the compositional characteristics of asphaltenes. In order to simplify the study, a residue was separated in solubility fractions and the fractions were subsequently analyzed by Fourier Transform ion cyclotron resonance mass spectrometry (FTICR MS). Our main goal is to examine the link between the distribution and characteristics of the detectable molecules in asphaltenes, with experimental bulk data obtained using more conventional approaches and link the FT-ICR MS compositional data to the macroscopic behavior. In particular, we are interested in the evaluation of changes in solubility as function of


Page 4 of 39

Page 5 of 39

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

the compositional shift product of the fractionation. In this work, we compare solubility parameter distributions obtained by solubility profile analysis16 and distributions obtained from the FT-ICR MS compositional data.


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

EXPERIMENTAL SECTION Samples. A residue was separated by solubility according to a procedure described in detail before.17 The fractionation was performed using an Accelerated Solvent Extractor Dionex 300. A sample of the material is weighed (mass around 5.0 g) and dissolved in 50 mL of dichloromethane. 50 g of PTFE are added to the solution and stirred for 1 h at room temperature. The solvent was removed by heating at 60°C under nitrogen. The PTFE supported sample is placed into a 100 mL stainless steel cell and extracted with a series of solvents/solvent blends. This procedure yields 6 fractions: heptane solubles (Fraction1), 15/85 CH2Cl2/n-heptane solubles (Fraction 2), 30/70 CH2Cl2/n-heptane solubles (Fraction 3), CH2Cl2 solubles (Fraction 4) and 90/10 CH2Cl2/CH3OH solubles (Fraction 5) and Fraction 6 which is composed of the remaining material and it is extracted using a 90/10 CH2Cl2/CH3OH blend at high temperature (120o C). Fraction distribution and standard data have been published for the fractions elsewhere.18 A summary is shown in Table 1. Molecular weights were determined by size exclusion chromatography (SEC) using a 30 cm x 7.5 mm PLgel mixed E column (Agilent Technologies) recommended for oligomers and polymers up to 25,000 g/mol. Solutions of the fractions were prepared in dichloromethane using concentrations of 100 ppm. The solutions were eluted with a 90/10 dichloromethane/methanol blend at a flow rate of 1.0 mL/min. The temperature was kept constant at 25 oC. A HPLC Agilent model 1100 liquid chromatograph equipped with an evaporative light scattering detector Alltech 2000 was used. The molecular weights were calculated based on a calibration that uses porphyrins, dyes and polyaromatics as standards. Molecular weight values are reported in Table 1.


Page 6 of 39

Page 7 of 39

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

FT-ICR MS Analysis. The samples were analyzed using a solariX XR FT-ICR mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany) equipped with a 12 T refrigerated actively shielded superconducting magnet (Bruker Biospin, Wissembourg, France) and the ParacellTM analyzer cell. Apollo II Dual ESI/MALDI ion source was used. Samples were analyzed using positive ion mode using APPI and LDI. The transient length of the mass spectrometric measurements was 3.3 seconds for APPI as well as LDI measurements. Sine apodization was applied and spectra were processed in absorption mode resulting in a resolving power of 1,300,000 at m/z 400. The spectra were externally calibrated with arginine clusters in electrospray ionization in positive ion mode. During the acquisition each scan was single point calibrated with a known mass (lock mass calibration). The final spectrum was internally calibrated in DataAnalysis 4.2 (Bruker Daltonics) with a known homologous series using quadratic calibration. All RMS mass errors of the internal calibration were below 110 ppb. The RMS mass errors of the internal calibration of the LDI and APPI measurements were on average 55 ppb and 92 ppb, respectively. Samples were prepared by diluting them 1:100 in toluene as stock solution. For APPI measurements, the stock solution was diluted 1:200 in 50/50 CH3OH/toluene for final spray solution (50 ppm). This solution was directly injected to the APPI source with a syringe pump at a flow rate of 10 µL/min. The ion accumulation time was 30 ms and 200 single scans were added for final mass spectrum. For LDI measurements, 0.5 µL of the stock solutions were spotted on a stainless steel target. Between 80 and 300 laser shots were used for a single scan with Nd:YAG laser at 355 nm wavelength using a laser repetition rate of 400 Hz. The laser power was set to only 13% to minimize fragmentation. 300 single scans were added for final mass spectrum.


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 8 of 39

Internal mass calibration, spectral interpretation and export of mass lists were performed using DataAnalysis 4.2 (Bruker Daltonics). The analysis of the data including calculation of molecular formulas and relative abundances of compound classes was performed using Composer 1.0.6 (Sierra Analytics). Elemental composition assignment was based on Kendrick mass defect sorting. Maximum mass error of 0.5 ppm and maximum number of heteroatoms of N=3, O=3 and S=3 were allowed for molecular formula calculation. Double-bond- equivalence (DBE) values representing the number of rings plus the number of double bonds in a given molecular formula were calculated using the following equation:19 DBE=c-h/2+n/2+1

(1)

For the elemental formula CcHhNnOoSs.


Page 9 of 39

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

RESULTS AND DISCUSSION Mass Spectra and Class Distributions of the Fractions. Figure 1 shows the LDI and APPI spectrum of asphaltene fraction 1. The comparison of the mass spectra profiles of both ionization techniques shows that APPI provides a broader distribution that goes to higher molecular weights while LDI produces narrower molecular weight distributions with lower values. Class distribution (radical cation classes) plots are shown in Figure 2 for both APPI and LDI. According to this figure, APPI results indicate that for the more abundant fractions (Fraction 2 to 4), the predominant classes are S and N containing classes. For fraction 1 (or “maltenes”), S and HC are the most abundant classes. In contrast, LDI shows a large relative abundance of N class followed by N2-and NS-classes in all the fractions. S-and S2-classes are clearly diminished in comparison with APPI. This is expected since LDI is very sensitive toward aromatic nitrogen-containing species.4 Another interesting aspect of the comparison between LDI and APPI for all the fractions is the evaluation of aromaticity and size of the detected molecules. Figure 3 shows a comparison between average aromaticities determined from APPI and LDI for the different classes. The aromaticity for each class is calculated as:20 Fa=DBE/(C+N)

(2)

C and N are average carbon and nitrogen. DBE represents the average double bond equivalent. These values were determined for each class as weighted average values based on intensity. The results shown in Figure 3 indicate that average aromaticities obtained by LDI tend to be larger than those obtained by APPI. Interestingly, the results for nitrogen-containing classes


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 10 of 39

are very close, while the larger differences found correspond to classes S-, S2- and HC- that, obviously, do not contain nitrogen. A similar comparison (not shown) obtained for the total carbon number C shows that average carbon numbers are lower for LDI than for APPI. APPI shows that the aromaticity of the classes increases for fraction 1 to fractions 3-4 and then, decreases for fractions 5 and 6, reaching a maximum value for fraction 3 or 4. In contrast, LDI aromaticity values do not decrease significantly or even increase for the last two fractions. Additonally, for fractions 5 and 6, the difference between APPI and LDI average aromaticities is more significant for S-, S2- and HC- classes. Comparison of Average and Bulk Properties. Figures 4 and 5 show the comparison of average properties determined by standard characterization, i.e. elemental analysis and molecular weight with results obtained by APPI and LDI, respectively. Based on the relative abundance of the classes it is possible to roughly estimate the nitrogen and sulfur abundance in the fractions. For APPI measurements, Figure 4a shows the comparison between relative abundance of classes containing N and S and the content of these elements in the fractions obtained by standard techniques. These results indicate that the relative abundance of N and S containing classes correlates qualitatively with the real content of these elements, showing maximum abundances for the middle fractions 3 and 4. In Figure 5a, a similar comparison is presented for LDI measurements. In this case, protonated and radical cations are taken into account. The relative abundance of N does not follow the same tendency as the elemental analysis indicated. This result seems to be a consequence of the large presence of nitrogen protonated species in the last two fractions, which is probably related to the use of dichloromethane/methanol blend in the separation for these two fractions. This blend has a large polar character as indicated by its acceptor number which leads


Page 11 of 39

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

to the extraction of the more polar species. This leads to larger average aromaticities calculated by LDI than by APPI in fractions 5 and 6 as shown in Figure 3. Average H/C molar ratios for the predominant classes are compared in Figures 4b and 5b. These values were calculated as weighted average values by intensity. In figure 4b, APPI H/C ratios are lower than those corresponding to the fractions (determined by standard elemental analysis) with a couple of exceptions corresponding to the class HC. In contrast, in a rather recent work using APPI,6 a larger H/C value is reported in comparison to the bulk value. However, in this case a non-weighted distribution was used in the calculations. Lower H/C values than those found by standard elemental analysis indicate the preferential ionization of aromatic species. The larger APPI H/C values for HC classes in fractions 5 and 6 are consistent with the previous finding that these two last fractions contain trapped maltenes as reported before21 and therefore, lower aromaticities (or larger H/C) than the rest of the components of these fractions. The presence of trapped maltenes has been previously reported by other authors.8 In fact, it can be seen that the increase of the APPI H/C ratio for fraction 6 corresponds also with a noticeable increase in the relative abundance of HC class in this fraction using APPI (see Figure 2). This agrees with the presence of maltenes in these last two fractions. On the other hand, LDI H/C molar ratios of the classes are lower than those corresponding to the fractions and lower also than those obtained using APPI. As H/C is correlated to aromaticity, it is clear that LDI tends to ionize molecules with larger aromaticities than APPI. Similarly, Figure 4c and 5c shows the comparison between the average molecular weights of the classes and the molecular weight obtained by SEC for both ionization techniques.


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 12 of 39

As a rule of thumb, ionization efficiencies of APPI and LDI are rather low for molecules in asphaltenes larger than 1500 Da. In fact, it has been reported that APPI has poor sensitivity in the high molecular weight range.21 Also, it can be noticed that LDI produces lower molecular weight values than APPI. This difference in MW is similar to LDI analyses made by others, which report lower MW values using this technique.23-25 On the other hand, there is some degree of correlation between the tendencies observed in the bulk values (SEC) and the classes for both ionization techniques. Several of the classes detected by APPI showed an increase in molecular weight from fraction 1 to fractions 2-3 and then a decrease from fractions 4 to 6. A similar situation was found for LDI, molecular weights for some of the classes increase from fraction 1 to fraction 4 and then decrease for fractions 5 and 6. Both sets of data (APPI and LDI) point out to a behavior where molecular weight decreases for fractions 5 and 6 for most of the classes. SEC values followed similar tendencies, i.e. an initial increase in molecular weight for fractions 1 to 3 and a decrease after fraction 3. Increasing molecular weight as solubility decreases has been reported based on VPO measurements26 which are affected by asphaltene aggregation. In fact, it has been shown that low solubility fractions have larger tendencies toward aggregation than highly soluble fractions27 and this affects these measurements. In the results shown in Figure 5c and 6c, a significant increase in molecular weight was not observed. The increases are rather small. In the analysis of these results is worth to consider that different fractionation procedures can yield fractions with completely different properties28 and therefore, different property tendencies. Figures 4d and 5d show the comparison of density experimental values for the fractions and densities calculated for the most abundant classes. Notice that for the comparison only fractions 1 to 4 are used because experimental density could not be determined for fractions 5


Page 13 of 39

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

and 6 due to the scarcity of the samples.

In order to calculate the average densities for each

class, we calculate the density for each molecule according to the following equation:18 ρ=-0.064*H+1.6793

(3)

Where H represents the hydrogen content (wt%). This method is based on a correlation previously established for asphaltenes and their solubility fractions and can predict asphaltene densities with an error of 0.7 %. The density for each class was obtained as the weighted average value based on intensity. This equation was used for all the classes except HC, since it is not reliable when hydrogen contents are larger than 9 wt% and class HC contains a large proportion of these molecules. Figures 4d and 5d indicate that, on average, most of the molecules detected by mass spectrometry have higher densities than the bulk values for the fractions. In fact, they compare better with fractions obtained from a deposit (dotted line),29 indicating that these techniques ionize denser materials that are characterized by a large hydrogen deficiency. This might be an indication that these techniques are particularly suitable for the analysis of asphaltene deposits that are characterized by molecules with high aromaticities. In general the comparisons shown in Figures 4 and 5 indicated that APPI produces average properties closer to the bulk values determined by standard techniques than LDI. Even more, from the point of view of aromaticity for Nitrogen containing classes, APPI and LDI produce similar results as shown in Figure 3. These results are indicative that in order to estimate bulk properties, APPI is better suited than LDI. Another comparison was established between boiling points for the fractions evaluated by simulated distillation18 and values calculated for the classes based on average H/C ratios and molecular weights. To this end, the Altgelt and Boduszynski equation26 that correlates molecular


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 14 of 39

weights with boiling points and H/C ratios was used to calculate the boiling points of the classes obtained by APPI: MW = 170 + 2.67x10-7(AEBP3/(H/C)0.9), for AEBP > 500oF

(4)

where AEBP represents the mid-atmospheric boiling point. MW is the molecular weight and H/C is the molar hydrogen to carbon ratio. The results indicate an excellent agreement between the data obtained for fractions 1 and 2 and their classes as it can be seen in Figure 6. For fractions 3 and 4, the equation produces values that are too low in comparison with values obtained from simulated distillation experiments.18 However, if the molecular weights corresponding to fractions 3 and 4 are multiplied by 2 or 4, respectively, the boiling points are within the range of the experimental values. This tendency is the result of the influence from asphaltene aggregation on the correlation as the molecular weight values used for its development were obtained partially from VPO measurements. This was already cautiously indicated by the authors of the correlation.26 These calculations also seem to point out that aggregation can occur for both fractions 3 and 4, but not for fractions 1 and 2, or at least that it does to a significantly lesser extent. Solubility Profile and Other Distributions. Solubility Parameters for the molecules of the different classes obtained by APPI were calculated based on a procedure described elsewhere.18 This procedure is based on the use of density values calculated according to equation (3) and the third-rule that correlates density with solubility parameter for hydrocarbon molecules30 δ = 17.347ρ +2.904

(5)


Page 15 of 39

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

Where δ is the solubility parameter (MPa0.5) and ρ is density (g/cm3). Weighted average solubility parameters for the classes were calculated and the values between 20 MPa0.5 and 23 MPa0.5, as expected, based on the average hydrogen contents. These values are close to the solubility parameters reported for known polyaromatics as well as to values reported for asphaltenes using different methods.31-35 Figure 7 shows a typical solubility parameter distribution as a function of the size of the molecule for class NS of fraction 3. In this figure, it can be seen that the solubility parameter decreases with the carbon number of the molecule. This is a consequence of a decreasing aromaticity as the molecule becomes larger. Similar distributions and tendencies were found for other classes from different fractions. In Figure 7, two lines represented the lower and upper limit of the solubility parameter calculated according to equation (4). The hydrocarbons with the lowest solubility parameters are linear alkanes and the lower line reflects their solubility parameter values as a function of C. The upper boundary represents the same concept but for those molecules with the largest solubility parameters among the hydrocarbons present in petroleum related materials: pericondensed aromatic molecules. They represent the upper limit for the double-bond equivalent (DBE) to carbon. They follow the established rule that molecules with a double-bond equivalent (DBE) to carbon larger than 90% of the carbon number are not present in petroleum and similar materials.36 Also, molecules that break this rule are not planar as opposed to the pericondensed aromatics that are in the boundary. Because of their low hydrogen content, pericondensed molecules exhibit the largest densities and therefore, the largest solubility parameters according to equations(3) and (5). Figure 8 shows a comparison between solubility parameter distributions of the same class (class N) in different fractions. According to Figure 8, there is a significant overlapping among


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 39

classes coming from different fractions, but it can also be observed that solubility parameters shift to larger values from fractions 2 to 4. Higher solubility parameters indicate that molecules become less soluble in hydrocarbons. This is the expected behavior based on the fractionation procedure where the fractions become more and more difficult to solubilize (from fraction 1 to fraction 4) as reported in a previous work.18 Similar results are found for all the other classes when the same class is compared for the different fractions. A shift to larger solubility parameters is always observed. Also noticeable is the significant overlapping in the solubility/carbon number space. This significant overlapping in fractions obtained by solubility has been already pointed out by previous publications37,38 and corroborated for these fractions in an earlier work.18 The shift of the solubility parameter distribution for the same class can be better visualized in a two-dimensional distribution. For example, Figure 9 shows this shift to larger solubility parameters as the fraction becomes less soluble for class S2. The same tendencies were found using van Krevelen method39 to calculate solubility parameters instead of the procedure described in previous paragraphs. The comparison of the solubility parameter distributions for different classes belonging to the same fraction reveals that, in general, for all the studied asphaltene fractions, the average solubility parameter increases in the following class order: N

Atmospheric Pressure Photoionization and Laser Desorption

Recommend Documents