Characterization of Asphaltenes Molecular Structure - Energy & Fuels


Characterization of Asphaltenes Molecular Structure - Energy & Fuels...

2 downloads 128 Views 69KB Size

422

Energy & Fuels 1998, 12, 422-428

Characterization of Asphaltenes Molecular Structure V. Calemma,* R. Rausa, P. D’Antona, and L. Montanari ENIRICERCHE S.p.A., Via Maritano 26, 20097 San Donato Milanese, Italy Received September 22, 1997. Revised Manuscript Received December 12, 1997

The features of molecular structure of seven n-C7 asphaltenes isolated from different feedstocks (i.e., vacuum residue, atmospheric residue and crude oil) have been investigated by pyrolysisgas chromatography/mass spectrometry (Py-GC/MS), and their average structural parameters have been estimated using a mathematical optimization method. The results of structural analysis pointed out substantial differences among the asphaltenes investigated. The size of aromatic fused ring systems ranged from 5.1 to 14.9, while a substantial fraction of peripheral carbon in hydroaromatic/aromatic sheets (0.28-0.40) is substituted with aliphatic chains whose average length was between 2.8 and 6.3. A significant fraction of aliphatic moiety (0.15-0.48) was found to be present as naphthenic structures. In all cases, the main classes of compounds identified during pyrolysis tests were homologous series of alkanes (up to C25-30), 1-alkenes (up to C25-30), branched paraffins, and alkyl-substituted aromatic compounds. The formation of pyrolysis products was explained on the basis of literature data regarding thermolysis reactions of model compounds mimicking the structural features of asphaltenes. The results of this study are consistent with the view that asphaltenes are a complex polydisperse mixture of molecules made up of polyaromatic and hydroaromatic units joined by aliphatic bridges and substituted with aliphatic chains containing up to 25-30 carbon atoms. However, both NMR data and results of structural analysis indicate that the major part of these chains (more than 80%) is formed by short alkyl groups (C1-C4).

Introduction It is well-known that asphaltenes, the heptaneinsoluble/toluene-soluble fraction of oil, are the cause of an array of problems associated with either thermal and catalytic processing of petroleum residues or recovery of crude oil. With respect to catalytic hydroprocessing, asphaltenes adversely affect the overall rate of hydrodesulfurization,1,2 act as coke precursors which in turn lead to catalyst deactivation,3 and may limit the maximum level of conversion achievable in hydrocracking process due to sludge formation.4 Asphaltenes are also responsible for the high viscosity5 of residues and, in the case of visbreaking, both the process severity and the stability of the product strongly depend on the asphaltene content of the feedstock and its tendency to form coke. Finally, the tendency of asphaltenes to precipitate during crude oil recovery can cause a sharp decline in oil flow or even blockage of the well.6 In these * Author to whom correspondence should be addressed. E-mail: [email protected]. (1) Speight, J. G. Fuel Science and Technology Handbook; Marcel Dekker: New York, 1990; p 143. (2) Berti, V.; Iannibello, A. Idrodesolforazione di residui di petrolio; Stazione Sperimentale per i Combustibili: San Donato Milanese, Italy, 1975; pp 42-62. (3) Bartholomew, C. H. In Catalytic Hydroprocessing of petroleum and distillates; Oballa, M. C., Shih, S. S., Eds.; Marcel Dekker: New York, 1994; pp 1-32. (4) Miyauchi, Y.; de Wind, M. Proc. Akzo Nobel Catalysts Symposium 1994; Hydroprocessing 1994, 123-140. (5) Speight, J. G. The Desulfurization of Heavy Oils and Residua; Marcel Dekker: New York, 1981; pp 145-170. (6) Cimino, R.; Correra, S.; Del Bianco, A.; Lockhart, T. P. In Asphaltenes Fundamentals and Applications; Sheu, E. Y.; Mullins, O. C., Eds.; Plenum Press: New York, 1995; pp 97-130.

circumstances, structural characterization of petroleum asphaltenes is a subject of considerable interest since a deeper knowledge of their molecular structure can be a valuable aid in achieving a better comprehension of their behavior during conversion processes of the bottom of the barrel and crude oil recovery. Elucidation of structural characteristics of the asphaltenes has been a subject widely investigated by means of numerous analytical techniques, and different models have been proposed.7,8 Dickey and Yen,9 mainly on the basis of results of X-ray analysis, proposed that petroleum asphaltenes consist of large sheets containing 100-300 carbon atoms and joined by aliphatic chains. Subsequently, the use of 1H NMR spectroscopy in conjunction with other analytical techniques10-14 (i.e. IR, GPC, VPO) led to a picture of asphaltene molecules consisting of one or more fused rings systems with a lower degree of condensation and containing naphthenic structures associated with the aromatic core. Unit sheets of moderate size are seen as joined by aliphatic bridges and bearing alkyl side chains with an average length between 4 and 6 carbon atoms. The simultaneous use of 1H and 13C NMR allowed a more accurate determination of several average structural parameters such as the aromatic carbon fraction, the average number per alkyl side chain and the percent of substitution of (7) Speight, J. G. In Asphaltenes and Asphalts, 1; Yen, T. F., Chilingarian, G. V., Eds.; Elsevier: Amsterdam, 1994; pp 7-65 and references therein. (8) Bestougeff, M. A.; Byramjee, R. J. In ref 7, pp 67-94 and references therein. (9) Dickie, J. P.; Yen, T. F. Anal. Chem. 1967, 39, 1847-1852.

S0887-0624(97)00185-0 CCC: $15.00 © 1998 American Chemical Society Published on Web 02/12/1998

Molecular Structure of Asphaltenes

peripheral aromatic carbon.15,16 Lately, the application of more sophisticated NMR techniques such as DEPT,17 GASPE, and PCSE provided a more detailed description of carbon-type distribution.18,19 Although the application of the aforementioned techniques allows direct determination of important structural features, a full description of the skeletal structure of asphaltenes requires the estimation of several parameters which are not still accessible by measurements or direct calculation. To get more information from a set of experimental data, several mathematical methods which combine data from NMR, elemental analysis, density measurements, and molecular weight determinations have been proposed. In 1970 Hirsch and Altgelt20 described a method for obtaining average structural parameters by solving three independent simultaneous equations describing the relation between structural groups and peripheral carbon, the compactness of fused ring system and the molar volume. Subsequently, different mathematical techniques have been proposed by Haley,21,22 Kiet,23 and other investigators.15,16 However, even though the latter methods are simpler and easier to apply, they are less comprehensive and require a larger number of assumptions. It is to be stressed that while this approach offers a powerful tool to establish the relative abundance of different carbon types, it provides only average values of the basic structural elements of asphaltene molecules. However, these methods can be an effective means to assess overall structural differences among different asphaltenes or to monitor their evolution during reactions. In the past years, analytical pyrolysis techniques (PyGC, Py-GC/MS) have been applied to characterize and gather information on the molecular structure of asphaltenes. Detailed analysis of the volatile products resulting from pyrolysis of asphaltenes can provide valuable data for a more detailed characterization of both the aromatic and the aliphatic moiety. Investigations by this method24-28 have produced strong evidences that a considerable fraction of the aromatic carbon in asphaltene molecules is made up of small aromatic systems (1-5 rings) while the length of the (10) Yen, T. F.; Wu, W. H.; Chilingar, G. V. Energy Sources 1984, 7 (3), 203-235. (11) Yen, T. F.; Wu, W. H.; Chilingar, G. V. Energy Sources 1984, 7 (3), 275-304. (12) Speight, J. G. Fuel 1970, 49, 76-90. (13) Jacobs, F. S.; Filby, R. H. Fuel 1983, 62, 1186-1192. (14) Speight, J. G. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1986, 31 (4), 818-825. (15) Dickinson, E. M. Fuel 1980, 59, 290-294. (16) Dereppe, J.-M. Moreaux, C.; Castex, H. Fuel 1978, 57, 435441. (17) Dereppe, J. M.; Moreaux, C. Fuel 1985, 64, 1174-1176. (18) Calemma, V.; Iwanski, P.; Nali, M.; Scotti, R.; Montanari, L. Energy Fuels 1995, 9, 225-230. (19) Snape, C. E.; Marsh, M. K. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1985, 30 (2), 247-261. (20) Hirsch, E.; Altgelt, K. H. Anal. Chem. 1970, 42, 1330-1339. (21) Haley, G. A. Anal. Chem. 1972, 44, 580-585. (22) Haley, G. A. Anal. Chem. 1971, 43, 371-375. (23) Kiet, H. H.; Malhotra, S. L.; Blanchard, L.-P. Anal. Chem. 1978, 50, 1212-1218. (24) Ritchie, R. G. S.; Roche, R. S.; Steedman, W. Fuel 1979, 58, 523-530. (25) Speight, J. G.; Pancirov, R. J. Liq. Fuels Technol. 1984, 2 (3), 287-305. (26) Solli, H.; Leplat, P. Org. Geochem. 1986, 10, 313-329. (27) Payzant, J. D.; Rubinstein, I.; Hogg, A. M.; Strausz, O. P. Geochim. Cosmochim. Acta 1979, 43, 1187-1193. (28) Payzant, J. D.; Lown, E. M.; Strausz Energy Fuels 1991, 5, 445453.

Energy & Fuels, Vol. 12, No. 2, 1998 423

aliphatic chains covers a wide range up to 30-40 carbon atoms. Since the asphaltenes are an extremely complex mixture of molecules, the use of a multidisciplinary approach through structural analysis and analysis of thermal degradation products can be a useful means to investigate their chemical structure. This paper reports the results from structural characterization of seven C7 asphaltenes, using both PY/GC/MS and average structural parameters determined by the method developed by Hirsch and Altgelt.20 Experimental Section Details of NMR analyses, GPC experiments and procedure for isolating asphaltenes from the different substrates have been described elsewhere.18 Briefly, n-heptane asphaltenes were recovered from crude oils Villafortuna (VIL) and Gaggiano (GA); atmospheric residues of Brent (BRE), Gela (GE), and Safaniya (SAFA); and vacuum residues of Arabian Light (AL) and Belayim (BEL). 1H NMR spectra were obtained from CDCl3 solution with a pulse width of 3.5 µs (30° flip angle) a recycle delay of 2 s and co-adding at least 200 scans. 13C NMR spectra were obtained by applying an inverse-gated decoupling technique to suppress the NOE effect. Chromium acetyl acetonate was added to ensure complete relaxation between pulses. Spectra were acquired with a pulse width of 2.7 µs a recycle delay of 2 s, and co-adding nearly 20000 scans. The gated spin-echo (GASPE) 13C NMR technique used was that proposed by Cookson and Smith29 with the average scalar coupling constant JCH of 125 Hz for aliphatic carbon. Asphaltenes density at 20 °C was determined in accordance with the method ASTM D4052 by means of a vibrating capillary densitimiter (PAAR DM48). Since at the test temperature the samples were solid, measurements were carried out using asphaltene solutions in xylene with a concentration between 2 and 3%. The asphaltene density was calculated by difference between the measured solution density and the known solvent density. Measurements carried out using solutions of different concentration showed that for values lower than 3%, the measured asphaltene density was independent of its concentration. Pyrolysis experiments were carried out on the System for Thermal Diagnostic Studies (STDS), developed at the University of Dayton, equipped with a HP-5971A mass detector. The system mainly consists of a “thermal” section made up of a CDS Pyroprobe 120 pyrolyzer connected to a programmable Lindbergh furnace in which a quartz reactor is placed. The whole section is linked up to a chromatographic section where, after cryofocusing at -60 °C on the head of a capillary column, the products generated during pyrolysis are analyzed by MS. A more detailed description of the system can be found in the literature.30 In the present case, to avoid secondary pyrolysis of the products, the transfer time from the pyrolyzer was minimized (i.e., 0.2 s) while the temperature of the transfer line from the pyrolyzer to the top of the GC column was kept at 250 °C. Pyrolyses were carried out at a nominal temperature of 650 °C with a heating rate of 20 °C ms-1 and a holding time of 20 s. Preliminary temperature measurements carried out by placing a very thin thermocouple into the pyrolyzer showed that the temperature seen by the sample could be 100 °C lower than the nominal one. The column used for separation was a (J&W Scientific) DB-1, 30 m × 0.25 mm i.d. Temperature programming was 1 min at -60 °C and then up to 300 °C with a linear heating rate of 20 °C/min and a holding time at the final temperature of 10 min. The acquisition of spectra was carried out in full scan mode, over a calibrated (29) Cookson, D. J.; Smith, B. E. Anal. Chem. 1985, 57, 864-871. (30) Rubey, W. A.; Grant, R. A. Rev. Sci. Instrum. 1988, 59 (2), 265269.

424 Energy & Fuels, Vol. 12, No. 2, 1998

Calemma et al.

Table 1. Analytical Data of Asphaltenes asphaltene

%w C

%w H

%w N

%w S

%w O

fcar

Gela Belayim Safaniya Arabian L. Brent Gaggiano Villafortuna

79.7 83.9 82.4 84.1 86.9 82.8 90.3

6.9 8.1 7.7 7.0 7.4 6.8 6.1

1.1 1.9 1.0 1.0 1.1 1.4 1.0

10.8 5.3 7.7 7.1 2.1 2.9 1.9

1.5 0.8 1.2 0.7 2.6 6.2 0.8

0.48 0.49 0.51 0.53 0.58 0.60 0.69

mass range from 16 to 650 amu. Products were identified by comparison of their MS spectra with those from computer libraries and retention times or, when possible, comparing their retention times with those of pure standards. After integration of the total ion chromatogram (TIC), peak areas were normalized on a weight basis, dividing their raw values by the weight of the original sample.

Results and Discussion Samples. Data and methodology of analytical characterization of asphaltene samples used in this work have been previously reported elsewhere.18 Here, for clarity, Table 1 shows the data of elemental analysis together with the fraction of aromatic carbon. Inspection of Table 1 shows that composition of asphaltenes covers a wide range. Particularly, the aromatic carbon fraction varies between 0.69 and 0.48 while sulfur content is between 2.9 and 10.8%w. Structural Analysis. The data required by the method are (1) molar volume; (2) elemental analysis (C, H, N, S and O); (3) the distribution of hydrogen among different structural groups, that is aromatic (Ha), benzylic CH + CH2 (Hb), benzylic CH3 (Hb3), aliphatic CH3 (Hl3), and the remaining CH2 and CH hydrogens β+ (Hr) present in aliphatic and naphthenic structures; (4) the distribution of heteroatoms (S, N, and O) functionalities; and (5) the asphaltene density at 20 °C. Molar fraction of hydrogen as Hb3 and Hb was calculated by the following equations:

fHb3 )

(

(f

CH3

)

%C - Hl3%H 4 RCH3 %H

fHb ) HR - Hb3

)

(1) (2)

where fCH3 is the molar fraction of carbon in CH3 groups, determined from 13C NMR and (GASPE) 13C NMR data; RCH3 is the fraction of H in CH3 in R position of H in CH3 present in both R and β positions to an aromatic ring. In our case, this parameter was assumed to be 0.5. Hl3 is the molar fraction of hydrogen present as CH3 groups in γ+ position, and HR is the molar fraction of hydrogen (i.e., CH3, CH2, and CH) present in R position. The molar fractions of HR, Hr, and Hl3 were directly obtained from 1HNMR spectra. Due to the large overlap of bands, the molar fractions of HR, Hr, and Hl3 have been determined using a curve-fitting procedure with a software supplied by the Bruker (Glinfit). The standard deviation of measurements was found to be 5%. According to the literature on asphaltenes,31,32 the (31) Mullins, O. C. In Asphaltenes Fundamentals and Applications; Sheu, E. Y.; Mullins, O. C., Eds.; Plenum Press: New York, 1995; pp 53-96. (32) Kelemen, S. R.; George, G. N.; Gorbaty, M. L. Fuel 1990, 69, 939-944.

distribution of sulfur and nitrogen among different functional groups was assumed to be as follows: nitrogen, 30% pyridinic and 70% pyrrolic; sulfur, 65% thiophenic, 20% φ-S-R, 15% R-S-R. As regards the oxygen distribution, due to the scarcity of literature data, after inspection of FT-IR spectra we assumed that oxygen was equally distributed among Ph-OH, CdO, and quinone groups. We are aware that the heteroatoms distribution used is quite rough; however, their atom fraction being quite small, we think that the error involved should have little effect on the final result. Subsequently, according to the method, the total number of carbon and hydrogen atoms per molecule and the number of hydrogen in each of the five hydrogen groups were calculated and the “reduction” of the average molecule to pure hydrocarbon was affected. The input data for the calculation of average structural parameters of these representative hydrocarbon “molecules” of the seven asphaltenes used are listed in Table 2. The latter were calculated by the equations shown here below.

MVX ) (Mn/F) + ∆MV

(3)

ΨXC ) ΨC + ∆ΨC

(4)

ΨXH ) ΨH + ∆ΨH

(5)

HFXi ) HFi + ∆HFi

(6)

where MVX is the molar volume of the asphaltene “reduced” to pure hydrocarbon; Mn the number average molecular weight of the asphaltene; F density of the asphaltene; ∆MV change in molar volume due to heteroatoms; ΨXC , ΨHH the number of carbon and hydrogen atoms per average molecule “reduced” to pure hydrocarbon; ΨC , ΨH the number of carbons and hydrogens atoms per average molecule; ∆ΨC, ∆YH the contribution of heteroatoms; HFXi the number of hydrogen atoms per average molecule “reduced” to pure hydrocarbon, falling into each of the five distinguishable groups (i ) Ha, Hb, Hb3, Hl3, and Hr); HFi the number of hydrogen atoms per average molecule falling into each of the five distinguishable groups; and ∆Fi contribution of heteroatoms. The values of ∆ΨC, ∆ΨH, ∆MV, and ∆Fi were calculated using the data provided in ref 20 concerning the change in molar volume and carbon and hydrogen atoms per molecule due to the various heteroatom functional groups. The determination of the average structural parameters was accomplished by solving the three implicit equations, relating to the compactness of fused ring system, to the peripheral distribution and to the molar volume, with the constraint of nonnegative values for all structural parameters calculated, and choosing the floating parameters (φ, a, b, ξ, and ψ)33 in such a way as to minimize the difference between the experimental and calculated values of fCar, and fCal. (33) φ ) ring compactness factor which is indicative of the shape of fused ring system. Its value ranges between 1 and 0 for peri- and katacondensed system, respectively. a ) fraction of peripheral aromatic carbons bonded to benzonaphthenic carbons. b ) average number of peripheral benzonaphthenic carbons per unit sheet. ξ, ψ ) parameters defining the fraction of naphthenic carbons bonded to aliphatic chains.

Molecular Structure of Asphaltenes

Energy & Fuels, Vol. 12, No. 2, 1998 425

Table 2. Input Parameters To Calculate the Structural Characteristics of Asphaltenes parameter

VIL

GA

BRE

GE

SAFA

BEL

AL

density Mn molar volume carbon atoms hydrogen atoms aromatic hydrogens CH2 + CH benzylic hydrogens CH3 benzylic hydrogens primary paraffinic hydrogens other hydrogens

1.2581 1181 950.6 90.34 75.32 17.99 11.62 8.92 23.41 13.38

1.2397 1420 1164.6 100.65 107.29 19.44 7.45 10.34 28.58 41.49

1.1825 1767 1511.8 130.41 137.89 27.10 27.82 5.37 28.71 48.92

1.2022 2050 1742.6 144.59 159.68 14.34 10.85 13.16 35.93 85.23

1.1080 1720 1577.6 123.53 143.74 14.30 12.32 7.90 30.45 78.63

1.1562 1450 1291.5 105.75 125.19 15.17 14.41 6.51 27.65 61.44

1.1911 1497 1278.3 109.52 114.07 14.85 13.32 7.105 25.87 53.09

Table 3. Molecular Structural Parameters of Asphaltenes

Table 4. Number Distribution of Alkyl Side Chain Lengths

parametera

VIL

GA

BRE

GE

SAFA

BEL

AL

Cnpht CN Rnpht Raro L LCH3 Cars fn fa Carq Bac Rafrs φ n

6.4 0.22 2.2 20.4 4.1 2.8 0.33 0.17 0.29 0.59 0.43 14.5 0.85 1.41

15.1 0.34 4.6 15.2 8.7 4.5 0.28 0.17 0.21 0.47 1.2 7.8 0.60 1.94

13.8 0.32 5.6 19.6 4.4 3.8 0.37 0.20 0.29 0.42 0.27 5.1 0.85 3.87

18.3 0.35 10.3 22.5 8.1 5.5 0.40 0.26 0.33 0.63 0.49 14.9 0.95 1.50

7.8 0.15 3.9 19.2 8.5 6.3 0.39 0.73 0.32 0.60 0.37 13.8 0.95 1.38

12.6 0.24 4.6 14.6 7.1 5.7 0.38 0.25 0.31 0.50 0.49 7.1 0.95 2.05

24.0 0.48 8.9 17.2 5.2 3.8 0.38 0.15 0.31 0.56 0.45 9.5 0.90 1.81

a C npht, % of naphthenic carbon; CN, fraction of aliphatic carbon as naphthenic carbon; Rnpht, number of naphthenic rings per molecule; Raro, number of aromatic rings per molecule; L, average length of aliphatic chains, excluding benzylic CH3; LCH3, average length of aliphatic chains, including benzylic CH3; Cars, % of peripheral aromatic carbon substituted; fa, fraction of peripheral aromatic carbon substituted by aliphatic chains; fn, fraction of peripheral naphthenic carbon substituted by aliphatic chains; Carq, % of internal aromatic carbons; Bac, average branching degree; Rafrs, number of aromatic rings per fused ring system; φ, compactness factor; n, fused ring systems per molecule.

The average values of the main structural parameters derived for the seven asphaltenes are listed in Table 3. The data point out significant differences in several of their important molecular parameters. The average number of aromatic cycles per fused ring system (Rafrs) ranges from 15 to 5 and does not show any correlation with the aromatic carbon content of the sample. The fraction of all carbon atoms present in naphthenic structures (Cnpht) is in each case significant and ranges from a low 0.06 to a high 0.24 for VIL and SAFA, respectively. When normalized to aliphatic carbon fraction naphthenic carbon (CN) varies between 22 and 48%. The fraction of aromatic carbon as internal carbon (i.e., aromatic carbons bonded to three other aromatic carbons; Carq) is between 0.42 and 0.63. With the exception of asphaltene Gaggiano, the data regarding the compactness of fused ring systems (φ) show that the hydroaromatic unit sheets are preponderantly pericondensed. The fraction of peripheral aromatic (fa) and naphthenic carbon (fn) bearing alkyl chains ranges from 0.21 to 0.33 and 0.15-0.73, respectively. The degree of substitution of carbons on the edge of aromatic sheet varies from 0.28 to 0.40, whereas values generally found in the literature indicate a more extensively substituted structure (0.45-0.70). There may be various reasons for such a difference. A likely explanation lies in the fact that in most of the cases the methods used consider the CH3 benzylic hydrogens, which can make up a

paraffinic chain length asphaltenes

C1-C2 (%)

C3-C4 (%)

C5+ (%)

VIL GA BRE GE SAFA AL BEL

56 66 26 47 47 46 41

32 29 52 n.d. 38 37 43

12 5 22 n.d. 15 17 16

significant fraction of benzylic hydrogen, as benzylic CH2. As a matter of fact, an estimate of CH3 benzylic hydrogens in the asphaltenes studied, carried out combining data of elemental analysis and NMR spectroscopy, shows that these hydrogens account for 25-60% of all benzylic hydrogens. Data pertaining to the alkyl substituents (L and LCH3) show that the average chain lengths are considerably affected when benzylic CH3 groups are considered. Actually, results of structural analysis indicate that the percentage of all alkyl side chains as benzylic CH3 varies from 17 to 44%. The aliphatic region of 13C NMR spectra presents several sharp peaks that can give interesting information on the distribution of alkyl side chains. In each case the most intense signal lies at 29.7 ppm which can be assigned to internal methylene [CH3CH2CH2(CH2)n-] carbons of long paraffinic chains. The other less prominent peaks at 14.1, 22.6, and 31.6 ppm are in accord with literature values34,35 of δ+-methyl groups of aliphatic side chains, and of methylene in R [CH3CH2(CH2)nAr; n g 2] and β position [CH3CH2CH2(CH2)nAr; n g 2] to a CH3 terminal, respectively. The band centered at 19.9 ppm stems from a branching CH3 group [-CH2CH(CH3)CH2-] of a long chain and from benzylic CH3. A common characteristic of the spectra is that the intensity of the peak at 31.6 ppm is considerably lower than that of the peaks at 14.1 and 22.6 ppm. As an aside, the ratio of the peak intensities at 14.1 and 22.6 ppm is in all cases nearly 1. From the relative intensities of the above-mentioned signals, determined by a curve-fitting procedure of aliphatic region with a software supplied by the Bruker, and data of a previous study,18 a semiquantitative estimation of the distribution of paraffinic chains has been worked out. The data reported in Table 4 shows that short alkyl substituents (C1-C4) make up most of paraffinic chains while C5 or longer alkyl groups form the remaining 5-20%. The results obtained are comparable to those found by (34) Gupta, P. L.; Dogra, P. V.; Kuchhal, K.; Kumar, P. Fuel 1986, 65, 515-519. (35) Rafenomanantsoa, A.; Nicole, D.; Rubini, P.; Lauer, J.-C. Energy Fuels 1994, 8, 618-628.

426 Energy & Fuels, Vol. 12, No. 2, 1998

Calemma et al.

Figure 1. Py-GC/MS chromatogram of asphaltene Brent. Numbers refer to the chain length of 1-alkene/alkane doublets (2, CH2Cl2). Table 5. Percentage Hydrocarbon Distribution for Pyrolysis Products hydrocarbons

SAF

GA

BRE

GE

VIL

BEL

AL

alkanes cycloalkanes isoalkanes 1-alkenes S-compounds alkyl polyaromatics alkylbenzenes

32.8 3.7 11.5 31.1 3.3 4.1 13.5

31.0 5.3 13.3 26.8 n.d. 4.2 19.4

29.8 10.1 9.2 30.1 n.d. 2.5 18.3

29.6 5.5 8.6 30.9 7.0 3.3 15.1

41.2 6.5 12.4 29.6 n.d. 1.6 8.7

34. 7.7 11.7 32.1 n.d. 1.8 12.7

33.6 8.3 11.2 31.5 1.6 2.3 11.5

Strausz et al.36 subjecting Athabasca asphaltene to ruthenium ion catalyzed oxidation. Another interesting aspect pointed out by the data in Table 1 is the presence of branched aliphatic chains (Bac) in asphaltene molecules. Flash Pyrolysis Experiments. In all cases GC-MS analysis of pyrolysate yielded a complex trace containing more than 200 resolved peaks on top of an unresolved hump more or less pronounced depending on the asphaltene sample. As example of results obtained, Figure 1 shows the pyrogram of asphaltene Brent. The TIC traces are qualitatively similar for all seven asphaltenes investigated and, as also observed by other investigators,24,26 the main classes of compounds formed during pyrolysis, which in our case account for roughly 80% of the recognized fraction, are alkane (from C5 up to C26), and the corresponding 1-alkene and alkyl-substituted aromatic compounds. The lack of light paraffins and olefins in the pyrograms (C1-C4) is due to the fact that the analysis apparatus cannot identify compounds below C5. The remaining 20% is formed by isoparaffins, cycloalkanes, branched olefins, and sulfur heterocyclic compounds such as thiophene, benzothiophene, and dibenzothiophene alkyl substituted. A complete listing of the yields of identified products groups is presented in Table 5. The values are normalized on the recognized fraction of the total area of the chromatogram. Alkane/1-Alkene. 1-Alkene and alkane homologues are in each case the most abundant classes in pyrolysis products. Hence, elucidation of reaction pathways and structural units giving rise to the aforementioned compounds is highly meaningful for structural studies of asphaltenes and a better understanding of their (36) Strausz, O. P.; Mojelsky, T. W.; Lown, E. M. Fuel 1992, 71, 1355-1363.

thermal behavior. The literature concerning the characterization of asphaltenes and the results of the present study indicate that a substantial fraction of peripheral aromatic carbon is substituted by carbon atoms belonging to either aliphatic chains or naphthenic structures. Studies performed by Savage and co-workers 37-39 and Mushrush and Hazlet40 on the pyrolysis of alkyl-substituted aromatic hydrocarbons provide the key to explain the formation of 1-alkene and alkane homologues during pyrolysis. The results of these studies showed that pyrolysis of alkyl aromatic hydrocarbons leads to three main pairs of compounds (i.e., Ar-CH3 + Cn-1 1-alkene; ArCHdCH2 + Cn-2 alkane; Ar-H + Cn alkane), and numerous minor products arising from random break off of side alkyl chains. The first pair of products results from the β-scission of γ alkylaromatic radical which yields to resonance stabilized benzylic radical and 1-alkene. The second pair stems from the easiest hydrogen abstraction step on R-position of alkyl aromatic compound followed by β-scission, while the third pair arise from the cleavage of the aryl-alkyl bond through a selective hydrogenolysis mechanism. Selectivities in each of the three pairs strongly depend on the type of reacting compound and operating conditions. However, the initial overall selectivity (i.e., the sum of selectivities in each of the three major products pair) at the highest temperature tested ranged between 0.6 and 0.8. Another reaction route which can lead to the formation of alkane/1-alkene homologues is the thermal decomposition of naphthenic structures with peripheral aliphatic substituents.41 However, in the latter case thermal degradation should occur in a less selective way in comparison with aryl-alkyl moieties. The results of structural analysis of the asphaltenes indicate that the percentage of alkyl side chains bonded to naphthenic carbon ranges between 3 and 25%. Therefore, the alkane/1-alkene homologues should chiefly derive from pyrolysis of the aryl-alkyl structures. A facet that needs to be examined is how much the distribution of alkene/1-alkene homologues in the pyrogram reflects that of alkyl chains in asphaltene. In other words, to what extent does the secondary degradation, the random breakdown of side alkyl chains, account for the products formed. As regards the first issue, it is to be stressed that pyrolysis occurs under a helium flow with a very short residence time in the reaction zone (0.2 s). Pyrolysis tests performed, in the same conditions used for the asphaltenes, with paraffins C30 showed that secondary decomposition of alkane should be lower than 5%. Moreover, a common characteristic of the pyrograms is the lack in the pyrolysis products of terminal diolefins (CH2dCH(CH2)nCHdCH2) and Ar(CH2)nCHdCH2 homologues (with the exception of styrene) which would result from extensive random degradation of side alkyl chains. This result together with literature data presented above supports the view (37) Smith, C. M.; Savage, P. E. AIChE J. 1991, 37 (11), 1613-1624. (38) Savage, P. E.; Klein, M. T. Ind. Eng. Chem. Res. 1987, 26, 488494. (39) Smith, C. M.; Savage, P. E. Ind. Eng. Chem. Res. 1991, 30, 331339. (40) Mushrush, G. W.; Hazlett, R. N. Ind. Eng. Chem. Fundam. 1984, 23, 288-294. (41) Savage, P. E.; Klein, M. T. Ind. Eng. Chem. Res. 1988, 27, 1348-1356

Molecular Structure of Asphaltenes

Energy & Fuels, Vol. 12, No. 2, 1998 427

Table 6. Number Average Alkyl Side Chain Lengths Calculated from Py-GC/MS Data asphaltenes

L

Gaggiano Brent Gela Safaniya Villafortuna Arabian Light Belaiym

9.5 9.4 9.7 9.2 11.0 9.8 9.8

that the breakdown of C-C bonds in alkyl chains in positions other than R, β, and γ to aromatic ring should be rather low. On this basis then, it is reasonable to assume that the 1-alkene/alkane homologues found represent a fairly accurate picture of the distribution of side alkyl chains longer than C5-C7. The number average chain lengths calculated from flash pyrolysis data is listed in Table 6. It is to be stressed that, in the light of the above-mentioned reaction pathways operating during thermal decomposition of aryl-alkyl structures, the average length of alkyl side chains in asphaltene molecules should be longer than 1-2 carbon atoms. Besides the straight-chain compounds, volatile products contain a significant amount of branched chains. The major part of the identified branched compounds (71-93%) was formed by isoalkanes and 1-isoalkenes with chain length up to 13 carbon atoms and with 1 or 2 methyl substituents per chain. It is to be stressed that the presence of these compounds is in agreement with the above-mentioned 13C NMR data and with the results of structural analysis. An aspect that needs to be addressed at this point is the apparent discrepancy between Py-GC/MS and structural analysis results. Particularly, Py-GC/MS results suggest a significant presence in the asphaltene molecules of aromatic structures substituted with long paraffinic chains up to C25-C30. In contrast, results of structural analysis indicate that the average length of aliphatic chains ranges from 2.8 to 6.3. We think the basic reason for this discrepancy is that the structural analysis, based mainly on NMR data, provides a value which is the number average of all aliphatic side chains in the asphaltene molecule and therefore emphasizes the short chains. Average values do not provide any information about the actual distribution about the components of the aliphatic chains which in this case can cover a wide range. Our NMR data presented before would suggest a very skewed distribution with high concentrations of short aliphatic chains and relatively low numbers of aliphatic chains longer than C5. The reason our pyrograms do not show any aliphatic chains shorter than C5 is that the system configuration used in this case did not allow to detect compounds lighter than C5. Aromatic Compounds. The data presented in Table 5 show that the content of aromatic compounds in pyrolysis products ranges from to 10.3 to 23.4%. The most abundant class is that containing mononuclear aromatics and includes benzene, toluene, styrene, xylene, and mono- and multisubstituted alkylbenzenes with alkyl groups up to C4. Other compounds present in lower concentration were unsubstituted and methylsubstituted naphthalene and indenes. Literature data38,42 indicate that an important aspect of molecular

structure of asphaltene is the presence of alkyl bridges connecting fused ring systems. In this connection, the extensive literature43-48 available on pyrolysis of the diarylalkanes (Ar(CH2)nAr′; n ) 1-6; Ar, Ar′: 1-4 rings) provides the basic information to interpret the formation of aromatic compounds. Data concerning pyrolysis of diarylalkanes with aliphatic bridges gC3 show that analogous to what is observed for aryl-alkyl compounds, reaction occurs through two major pathways which lead to primary products through the breakdown of bonds in β and γ position to the aromatic ring. Pyrolysis of diarylethanes indicates that, differently from higher homologues, decomposition occurs through homolytic scission of dimethylene bridges to produce benzyl radicals followed by hydrogen abstraction to give toluene and other minor products. Data published by Strautz et al.39 indicate that methylene bridges make up a considerable portion of aliphatic chains connecting aromatic fragments in asphaltene. Diarylamethanes are rather stable toward thermal decomposition and the homolytic rupture of the strong arylalkyl C-C bond of diphenylmethane will proceed at 650 °C with a half-life well beyond a year. However, with compounds having fused ring systems and in the presence of hydrogen coming from a donor solvent or the substrate itself, the thermal decomposition is much faster. Results obtained by Murata et al.48 on benzylsubstituted polyaromatic compounds, in the absence of donor solvent, indicate that in our experimental conditions breaking down of methylene bridges can be a significant reaction pathway during asphaltene pyrolysis. In addition to the breakdown of aliphatic bridges, the presence of aromatic hydrocarbons in the volatile products can also in part be due to dehydrogenation of naphthenic and hydroaromatic structures. As with aliphatic side chains so also for aromatic ring structures our structural analysis and PyGC/MS results seem to be inconsistent. In particular, structural analysis data show an average size of aromatic fused ring systems ranging between 5.1 and 14.9 while PY-GC/ MS see mostly one ring and little above three rings. The most likely reason for that is the wide distribution of the size of the condensed aromatic structures. PyGC/ MS data suggest in asphaltene molecules the presence of structural units where small aromatic moieties consisting of 1-3 condensed aromatic rings are joined to bigger fragments by aliphatic chains. The thermal decomposition of this type of structure would lead to the formation of alkylaromatic compounds with a molecular weight sufficiently low to volatilize at the decomposition temperature and to be determined by MS. Sulfur Compounds. Besides the products discussed so far, pyrolysis of asphaltene yielded aromatic sulfur compounds which include thiophene, benzothiophene, and dibenzothiophene, mostly alkyl-substituted with chains of up to C4. Determination of sulfur forms (42) Farcasiu, M.; Forbus, T. R. La Pierre, R. B. Prepr.sAm. Chem. Soc., Pet. Chem. 1983, 28 (2), 279-284. (43) Smith, C. M.; Savage, P. E. Energy Fuels 1991, 5, 146-155. (44) Poutsma, M. L.; Dyer, C. W. J. Org. Chem. 1982, 47, 49034914. (45) Sweeting, J. W.; Wilshire, J. F. K. Aust. J. Chem. 1962, 15, 89-105. (46) Miller, R. E.; Stein, S. E. J. Phys. Chem. 1981, 85, 580-589. (47) Poutsma, M. L. Fuel 1980, 59, 335-338. (48) Murata, S.; Nakamura, M.; Miura, M.; Nomura, M. Energy Fuels 1995, 9, 849-854.

428 Energy & Fuels, Vol. 12, No. 2, 1998

present in petroleum asphaltenes, by means XANES and XPS31,32 analysis, indicates that the most prominent sulfur functionalities are of thiophenic (100-51%) and sulfidic (0-50%) type while sulfoxide and sulfone make up only a few percent. Therefore, the aromatic sulfur compounds identified in this work may be regarded as primary products whose formation route should be similar to the one of aromatic hydrocarbons discussed before. Conclusions The results of this study indicate that the use of a multidisciplinary approach can be a valuable tool for a better understanding of several facets of asphaltenes molecular structure and allow to explain some inconsistencies found in the literature between structural characteristics of asphaltenes derived from spectroscopic methods and thermal decomposition experiments. In this connection, the data presented in this paper, concerning the alkyl side chains, provide a clear evidence of how the wide distribution of chain length can lead to different results according to the method used. Present data and those previously reported18 indicate that when the average length of alkyl side chains, L, is calculated considering all the alkyl substituent groups on aromatic rings (i.e., including benzylic CH3), both Dickinson’s and Hirsch and Altgelt’s methods yield similar results. However, the latter method points out that a considerable fraction of aliphatic chains is made up of CH3 benzylic groups while 13C NMR and Py-GC/ MS data indicate a significant presence of C5+ aliphatic chains. Besides the aliphatic chains, another important structural characteristic related with the aliphatic

Calemma et al.

moiety is the presence of naphthenic structures which form between 15 and 48% of the aliphatic carbon. A fundamental aspect of asphaltene molecular structure is connected with the aromatic moiety (i.e., aromatic ring size distribution, ring compactness factor, fraction of internal aromatic carbon, etc.). In our case, the results of structural analysis show quite a spread situation with values of average number of aromatic rings per fused ring system (Rafrs) ranging from 14.9 for GE to 5.1 for BRE asphaltenes. With the exception of asphaltene GA, the fused ring systems are predominantly peri-condensed. Clearly, Rafrs being an average quantity, it does not provide information about the actual size distribution of the fused ring system. However, Py-GC/MS data suggest that small (1-3 rings) aromatic systems should form a significant fraction of the aromatic moiety. From a general point of view, asphaltenes are an extremely complex collection of molecules having a polydisperse distribution of structural units. On the basis of our results, the asphaltene molecules can be described as formed by polyaromatic/ hydroaromatic units, whose size should cover a wide range, from 1 to 10-20 rings, depending on the asphaltene considered, joined by aliphatic bridges. A large fraction of peripheral carbon atoms of these units is substituted by aliphatic chains most of which are no longer than four carbon atoms. Acknowledgment. The authors are particularly grateful to the ENI group for financial support of this work and to Mrs. M. Anelli and Mr. W. Stringo for their precious technical contributions. EF9701854