Fuel 90 (2011) 3602–3607

Contents lists available at ScienceDirect

Fuel

journal homepage: www.elsevier .com/locate / fuel

Pyrolysis kinetics of atmospheric residue and its SARA fractions

Elena Alvarez a, Gustavo Marroquín b, Fernando Trejo a,⇑, Guillermo Centeno b, Jorge Ancheyta b,José A.I. Díaz a

a Centro de Investigación en Ciencia Aplicada y Tecnología Avanzada, Unidad Legaria, del Instituto Politécnico Nacional (CICATA-IPN), Legaria 694, Col. Irrigación,México DF 11500, Mexicob Instituto Mexicano del Petróleo, Eje Central Lázaro Cárdenas Norte 152, Col. San Bartolo Atepehuacan, México DF 07730, Mexico

a r t i c l e i n f o

Article history:Received 29 July 2010Received in revised form 24 November 2010Accepted 25 November 2010Available online 5 February 2011

Keywords:Thermogravimetric analysisPyrolysis kineticsThermal decomposition

0016-2361/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.fuel.2010.11.046

⇑ Corresponding author. Fax: +52 55 5395 4147.E-mail address: ftrejoz@ipn.mx (F. Trejo).

a b s t r a c t

Thermal analysis of atmospheric residue from heavy crude oil and its SARA fractions was carried out andthe tendency of each fraction toward coke formation was determined. The coke yield was 16.3 wt.% foratmospheric residue, 43.1 wt.% for asphaltenes, 4.6 wt.% for resins, 3.8 wt.% for aromatics, and 0.3 wt.%for saturates. Pyrolysis kinetics of residue and its fractions, i.e., asphaltenes, resins and aromatics was alsoinvestigated. The TG experiments were conducted at three different heating rates of 8, 12, and 16 �C/minfrom room temperature up to 800 �C under nitrogen atmosphere to verify the weight variation withreaction temperature. Isoconversional analysis to fit data assuming first order kinetics was employed.Asphaltenes was the fraction that produces coke in higher amount having a range of activation energyof 41.0–58.6 kcal mol�1 whereas activation energy for atmospheric residue ranged from 11.5 to30.0 kcal mol�1.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Thermal decomposition of petroleum asphaltenes has receivedattention primarily because of its tendency toward coke formationunder thermal conditions which deactivates those catalysts used inhydroprocessing. Since asphaltenes produce substantial amountsof catalyst-deactivating coke during hydrotreating, they make therefining processes less efficient and more expensive. For this rea-son a key parameter for understanding residue processing is tostudy the chemistry of coke formation at different temperatures[1,2]. Various reaction pathways have been proposed for asphal-tene thermal decomposition and it has been reported that the mainproducts are alkanes ranging from C1 to C40 and polynuclear aro-matics (from 1 to 4 aromatic rings) [3,4]. During pyrolysis ofasphaltenes, CAS bonds are broken in the range of 350–400 �Cwhereas the cleavage of CAC bonds becomes dominant at temper-atures higher than 400 �C [5]. The formation of methane and othernormal alkanes during pyrolysis at mild conditions indicates thatasphaltene contains thermally fragile alkyl groups on its outersurface [6]. Experimental tests by thermogravimetric analysis(TG) were carried out by Yoshida et al. [7] who studied thermaldecomposition of coal-derived asphaltenes and concluded thatweight loss is rapid from 300 to 500 �C where asphaltenes showedthe greatest weight loss and then it reduced above 500 �C. Whenair is used in TG, three different reaction regions can be identified

ll rights reserved.

and known as low-temperature oxidation, fuel deposition andhigh-temperature oxidation as reported by Kok [8]. Hauser et al.[9] studied the asphaltene pyrolysis at 412 �C and determined thatthe aromatic/aliphatic carbon ratio increased from �1 to �3whereas the atomic H/C ratio decreased from �1.1 to �0.8.

On the other hand, reports on thermal analysis of petroleumfractions are more limited. Earlier studies [10] stated that pyrolysisof paraffinic, aromatic, polar, and asphaltene fractions from heavyoils can be interpreted in terms of a low-temperature phase involv-ing the volatilization of paraffinic and aromatic fractions and ahigh-temperature phase in which polar and asphaltene fractionspyrolyzed producing a particulate carbon residue. Douda et al.[2] carried out the pyrolysis of asphaltenes from Maya crudeobtaining maltenes (saturates, aromatics, and polar compounds),coke and gases as products. Maltenes were further separated byHPLC and it was found that saturate fraction is composed by tetra-cyclo-alkanes (18.2%), alkanes (15.9%), and hexacyclo-alkanes(10.9%), whereas aromatic fraction was distributed as follows:mono-aromatics (22.6%), di-aromatics (26.5%), thiophenoaromat-ics (19.5%), and penta-aromatics (1.3%). Douda et al. [11] also sta-ted that maltenes from Maya crude oil are stable up to 215 �C butat higher temperatures a fast conversion is observed up to 550 �Cand finally they are entirely decomposed at �1000 �C. The cokeyield from asphaltenes and resins varies from almost 25 wt.% tomore than 60 wt.% but thermal studies have been focused mainlyon asphaltene constituents which produce thermal coke in therange of 35–65 wt.%. Asphaltenes produce high yields of volatileproducts that include condensable liquids and gases as well

Table 1Properties of the heavy crude oil and its atmospheric residue.

Property Heavy crude oil Atmospheric residue

API gravity 12 5.6Sulfur content (wt.%) 5.29 6.08Metals (wppm)Ni 89 97V 441 493

Elemental composition (wt.%)C 83.98 82.56H 10.28 10.56N 0.40 0.74Saturates (wt.%) 15.83 11.75Aromatics (wt.%) 36.74 23.66Resins (wt.%) 18.61 34.27Asphaltenes (wt.%) 28.82 30.32

E. Alvarez et al. / Fuel 90 (2011) 3602–3607 3603

[12,13]. However, among the SARA fractions, asphaltenes contrib-ute the most to coke formation. Douda et al. [2] observed thatasphaltenes from Maya crude analyzed by TG showed a consider-able decomposition in the range of 300–680 �C with the maximumtemperature at 478 �C.

It is also assumed that heterocyclic nitrogen plays an importantrole in thermolysis of resins and asphaltenes by which the firstreactions involve thermolysis of aromatic alkyl bonds. Secondaryreactions are aromatization of naphthenic species and condensa-tion of aromatic rings that activates the coke formation. Thus, theinitial step in the coke formation from heavier fractions (resinsand asphaltenes) is the formation of volatile hydrocarbons and non-volatile heteroatom-containing systems. Products obtained areinsoluble in the surrounding medium enhancing the carbonizationto finally form coke [14,15]. It has been demonstrated previously[3,16] that H/C atomic ratio of asphaltenes decreases rapidly asfunction of reaction time during thermolysis and then approachesan asymptotic limit at higher reaction times doing that asphaltenesundergo cracking with the core practically unaffected. Experimen-tal data suggest that thermal cracking of asphaltenes and resinstoward coke is a complex process in which the following reactionstake place: (a) cracking of alkyl chains from aromatic groups,(b) dehydrogenation of naphthenes to form aromatics, (c) conden-sation of aromatics to higher fused-ring aromatics, and (d)dimerization/oligomerization reactions. Loss of alkyl chains alwaysaccompanies thermal cracking and dehydrogenation/condensationreactions are a consequence of hydrogen deficient conditions.

When TG experiments are carried out at different heating ratesand by applying different methods for analysis it is possible to ob-tain kinetic parameters. Based on isoconversional methods, a seriesof activation energies for asphaltenes were obtained by Schucker[17] as function of the volatilized fraction. Activation energies werehigher at higher conversions. In contrast, Shih and Sohn [18] foundthat the activation energy was almost constant at different conver-sion of asphaltenes. By using the direct Arrhenius plot method,Park et al. [19] determined the activation energy to be in the rangeof 146–246 kJ/mol in the carbonaceous decomposition region inthe interval between 350 and 600 �C. The first region that includesthe devolatilization of light organics in the range of 50–350 �C hadlower activation energies (19–23 kJ/mol).

The aims of this study were (1) to determine the thermal behav-ior by TG of an atmospheric residue (5.6 API) from heavy crude(12 API) and its fractions (saturates, aromatics, resins, and asphalt-enes) and (2) to obtain the pyrolysis kinetics of different fractionsfrom atmospheric residue. Activation energies and pre-exponentialfactor were calculated by using the isoconversional Friedman’smethod.

2. Experimental

2.1. Separation of different fractions from atmospheric residue

An atmospheric residue from heavy crude oil was used in orderto obtain different fractions by means of chromatographic separa-tion. Atmospheric residue was analyzed by elemental analysisaccording to standardized ASTM D 5291 method to determinethe content of C, H, and N, while sulfur was quantified by ASTMD 5453. The ASTM D 5863 method was applied to determine themetals content, mainly Ni and V, which were detected by atomicabsorption in a Varian Spectra AA300 apparatus. Table 1 showsthe properties of the heavy crude and its atmospheric residue usedin this study. Asphaltenes studied in this paper were obtained byprecipitation with n-heptane in a pressurized system with nitrogenas inert at pressure of 25 kg/cm2, temperature of 60 �C and con-stant stirring. Solids were filtered under vacuum and dried.

SARA fractionation was carried out in two chromatographic col-umns packed with clay and alumina. In this procedure, asphalteneswere separated from the atmospheric residue by adding pentane.During the test, 2 g of sample are mixed with 25 mL of pentaneand the mixture is stirred for 24 h. Then, the mixture is filteredand asphaltenes are recovered. The filtrate (pentane + maltenes)is poured into the columns adapted in series and washed with25 mL of heptane. Pentane is added in enough amount to complete280 ± 10 mL. After washing, columns are decoupled and washedagain individually. The column packed with clay is washed witha toluene–acetone mixture (50%/50%) to recover the polar fraction(resins) whereas saturate fraction is recovered with pentane andaromatics with toluene in the column packed with alumina.

2.2. Thermogravimetric analysis

Once the saturate, aromatic, resin, and asphaltene fractionswere recovered, thermogravimetric analyzes were carried out oneach fraction. TG/DTG analysis was performed into a Perkin Elmeranalyzer model TGA-7HT by using �10 mg of sample in each runwith nitrogen (99.999%) as carrier at flow rate of 50 mL/min. Inaddition, atmospheric residue was also analyzed by TG/DTG. Thetemperature ranged from ambient up to 800 �C. Aromatics, resins,asphaltenes, and atmospheric residue were analyzed at three dif-ferent heating rates temperatures, i.e., 8, 12, and 16 �C/min to ob-tain kinetic parameters toward coke formation. In the case ofsaturates, only one heating rate (8 �C/min) was used due to its pro-pensity to ease thermal decomposition. For comparison purposes,thermograms for all fractions and atmospheric residue were re-corded at heating rate of 8 �C/min.

3. Non-isothermal kinetics of atmospheric residue and itsfractions toward coke

A non-isothermal technique using various heating rates hasbeen applied to the determination of kinetics of asphaltene pyroly-sis. TG data were analyzed by using the Friedman’s method [20]and kinetics was fitted to first reaction order. The method is basedon the comparison of experiments which were performed at differ-ent linear rates of heating. By this method it is possible to deter-mine the activation energy of certain processes without knowingthe form of the kinetic equation. According to Schucker [17], ther-mal decomposition of heavy components can be written asfollows:

A!k aC þ ð1� aÞV ð1Þ

where A is the reactant, C is coke, V the volatile fraction, and a is astoichiometric coefficient. It is possible to obtain the following ki-netic expression for volatile products [18]:

3604 E. Alvarez et al. / Fuel 90 (2011) 3602–3607

1Vo

dVdt¼ koe�EART 1� V

Vo

� �ð2Þ

where Vo is the total amount of volatilized material. By linearizationof Eq. (2) and further transformations the following expression isobtained:

lndxdt

� �¼ ln½koð1� xÞ� � EA

RTð3Þ

where x = V/Vo is the volatile fraction that corresponds to asphal-tene conversion in thermal decomposition. Last equation is easilyused by substituting different values of conversion (x) which rangedfrom 0.1 to 0.8. The values of dx/dt and T were determined for eachconversion at different heating rates (8, 12, 16 �C/min) by using thefirst derivative obtained from TG. By linear regression of Eq. (3) aseries of activation energies (EA) and frequency factors (ko) are ob-tained as function of asphaltene conversion (x). The application ofthis method for Mexican heavy crude has been reported elsewhere[21].

4. Results and discussion

4.1. Physical and chemical characterization of the feedstock

Table 1 shows the physical and chemical properties of the heavycrude and its atmospheric residue. The atmospheric residue ischaracterized by heaving high sulfur and metals content as com-pared with the crude oil. Almost one third of the total compositionof AR corresponds to asphaltenes as observed in Table 1. However,the resins content is also high, and both resins and asphaltenescontribute to the aromatic nature of the residue. The resins-to-asphaltenes ratio of the residue, which is considered as an indica-tor of stability, exceeds slightly the unity. By comparing the SARAcomposition of heavy crude and atmospheric residue, it is observedthat AR is highly concentrated in resins and asphaltenes which isexpected due to lighter components were separated during atmo-spheric distillation.

4.2. Thermal decomposition of atmospheric residue and its SARAfractions

A comparison between atmospheric residue and its SARA frac-tions is shown in Fig. 1 at 8 �C/min of heating rate; the weight lossfor atmospheric residue varies from 100 �C to 500 �C. Initial volatil-ization is mainly due to light alkanes distillation which continuesup to 350 �C. From 350 to 500 �C there is a rapid volatilization indi-cating that cracking of heavy fractions such as asphaltene and res-ins occurred. From 500 to 800 �C volatilization is very low and it is

0

10

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500 600 700 800

Mas

s, w

t%

Temperature, C

SaturatesAromaticsResinsAsphaltenesAtmospheric residue

Fig. 1. Thermogram of atmospheric residue and its fractions at 8 �C/min of heatingrate under nitrogen atmosphere.

almost constant. The coke yield for atmospheric residue was16.3 wt.%.

In the case of asphaltenes, which are considered the heaviestfraction present in atmospheric residue, volatilization is carriedout from room temperature to 350 �C and represents almost 10%of conversion. Changes detected at �350 �C could be attributableto the scission of alkyl groups located in peripheral sites of asphal-tene structure as reported before [22]. However, the most impor-tant weight loss is observed in the range from 430 to 550 �Cwhere it is to be expected that intermolecular associations andchemical bonds such as sulfur bridges and CAC are destroyed. Inthis stage, asphaltenes could be transformed into gases and valu-able components, i.e., oils. Continuation of the pyrolysis makesasphaltenes undergo condensation reactions to form coke as finalresidue from 550 to 800 �C. In Fig. 1 it is seen that in the range from600 to 800 �C the coke yield for asphaltene fraction is almost con-stant (43.1 wt.%).

Resins initially present a weight loss of 7 wt.% from 50 to 150 �Ccorresponding to the distillation stage of light alkanes that were re-leased. In the range from 150 to 350 �C there is still distillation andthe weight loss is smooth and much more severe from 350 to500 �C due to cracking reactions. In this interval almost 80 wt.%of resins are thermally decomposed. By comparing each fractionin Fig. 1, it is observed that resins possess the widest range of tem-peratures in which this fraction is pyrolyzed and a possible expla-nation for its behavior has been reported by Phillips et al. [23] whostated that resins are able to take part in condensation reactionsduring pyrolysis and form heavier molecules such as asphaltenes.The broad range of temperatures could be due to the formationof free radicals in alkyl labile points that tend to neutralize andcondense in bigger molecules. As many of lighter compounds pres-ent in resins have been volatilized in the initial stages of heatingthe remaining bigger molecules, i.e., resins and asphaltenes formedfrom condensation still continue volatilizing and this phenomenoncould be responsible for extending the interval in which pyrolysisof resins is carried out. The coke yield for resins was 4.6 wt.%.

Aromatics did not undergo any significant weight loss up to100 �C. In the range of 100–320 �C the most prominent changesin this fraction occur likely by volatilization of mono-, di-, andtri-aromatics and higher aromatic compounds. However, this frac-tion undergoes cracking reactions from 320 to 480 �C and the cokeyield is 3.8 wt.% which is very close to that produced by resins. Sat-urates produce very little amount of coke (0.3 wt.%) because allcompounds of this fraction are volatilized completely as can beseen in Fig. 1, where the profile obtained for saturates is similarto aromatics. Saturates practically did not undergo cracking reac-tions and all changes of this fraction occur in the distillation stage.The thermograms for asphaltenes and atmospheric residue showthe same behavior beyond 400 �C being almost parallel from 450to 800 �C, which could indicate that in this range the reactionsinvolving heavy fractions like asphaltenes dominate the perfor-mance of the whole residue.

Fig. 2 shows the derivative weight for each fraction and atmo-spheric residue. It is observed that asphaltenes and resins presentthe maximum rate of weight loss at 467 and 455 �C, respectively.In the case of asphaltenes the maximum rate of weight loss wasobserved at 467 �C, which is very similar to that obtained by otherauthors [22,24,25]. Thermal decomposition of asphaltenes at rela-tively low temperature (<350 �C) occurs by elimination of groupson peripheral sites. Above this temperature asphaltenes undergosevere degradation on its structure as reported earlier [26]. Duringthe initial stages of pyrolysis, distillation of low molecular weightcompounds takes place but as temperature increases crackingmay also occur by producing volatile fragments as a consequenceof the volatilization rate due to progressive cracking of larger mol-ecules [2]. Since asphaltenes are formed by polycyclic aromatic

Fig. 2. Derivative weight of atmospheric residue and its fractions at 8 �C/min ofheating rate under nitrogen atmosphere.

Fig. 3. Volatile fraction of atmospheric residue at different heating rates.

0.0

0.5

1.0

1.5

2.0

2.5

0.0010 0.0012 0.0014 0.0016 0.0018 0.0020 0.0022ln

(dx

/dt)

1/T, K-1

0.10.20.30.4

0.5

0.60.7

0.8

Fig. 4. Arrhenius plot of atmospheric residue pyrolysis at different conversions.

E. Alvarez et al. / Fuel 90 (2011) 3602–3607 3605

rings linked to naphthenes and alkyl chains along with heteroele-ments the main molecular forces acting on asphaltenes are inter-molecular associations and weaker chemical bonds, which aredestroyed or weakened as temperature is increased. Gases are re-leased as product of alkyl chains breaking. However, at higher tem-perature (>450 �C), the stronger chemical bonds are broken andmolecular skeletons are destroyed releasing a big amount of gasesdue to decomposition of asphaltenes.

Atmospheric residue has a maximum rate of weight loss at431 �C and small shoulders are observed through its thermogram.A possible explanation for this performance is the volatilization ofdifferent compounds having lower boiling temperature. Saturatesshow more evidently this trend with multiple shoulders indicatingvolatilization of lighter compounds by distillation. The maximumrate of weight loss is observed at 212 �C. Aromatics also exhibitmultiple shoulders due to volatilization of different class of com-pounds, i.e., mono-, di-, tri-, and higher aromatics. The maximumrate of weight loss for aromatics is located at 395 �C. Table 2 sum-marizes the coke yield for each fraction and temperature at whichthe weight loss is the highest.

4.3. Pyrolysis kinetics for atmospheric residue and its fractions

Pyrolysis is an excellent process to generate char, oil, and gasesfrom coal, biomass, and oil, therefore thermogravimetric methodsto investigate reaction kinetics of thermal decomposition fromoil, coal, and bitumen have been adopted by several researchers[19,27–29]. The mechanisms involved in the pyrolysis of coal oroil-derived products are exceedingly complex and the influenceof many variables is not fully understood, reason why thermal deg-radation is not well described by an individual reaction. However,TG provides general information about the overall reaction kineticsand is commonly used to estimate kinetic parameters. In this studya model-free kinetic approach based on the isoconversional Fried-man’s method was used to analyze data obtained by TG. The meth-od gives simultaneously the activation energy and pre-exponentialfactor by assuming a reaction order without knowing the ratedependence on conversion.

Table 2Coke yield for atmospheric residue and its SARA fractions at heating rate of 8 �C/min.

Sample Coke yield (wt.%) Tmax weight loss (�C)

Atmospheric residue 16.3 431Saturates 0.3 212Aromatics 3.8 395Resins 4.6 455Asphaltenes 43.1 467

Aromatics, resins, asphaltenes, and the whole atmospheric res-idue were analyzed to obtain their kinetic parameters toward cokeformation; however, only the behavior of the heavier fractions(resins and asphaltenes) and atmospheric residue is reported here.Fig. 3 shows the volatilization of atmospheric residue consideredas conversion (x) against temperature. It is observed that volatiliza-tion is shifted toward higher temperature with increasing heatingrates as reported by other authors [19,30,31]. At the highest heat-ing rates reactions take place very fast and gases are readily re-leased diminishing the weight of the sample more notoriouslythan at lower heating rates. The Y-axis represents the fraction ofthe original sample weight, which has been volatilized by increas-ing temperature and it is seen that each curve approaches anasymptotic value at higher temperatures and conversions. Fig. 4shows the Arrhenius plot of ln (dx/dt) versus 1/T for fractional con-versions ranging from 0.1 to 0.8. The values of 0.9 conversion andhigher are not shown since they are uncertain likely due to the dif-ficulty associated with obtaining accurate values of dx/dt near thecompletion of the reaction. In all cases, good correlation coeffi-cients (r) were obtained as reported in Table 3. Activation energy(EA) is expressed in kcal mol�1 and pre-exponential factor (ko) inmin�1. There are some changes in activation energy at conversionshigher than 0.4 where the value changes significantly. At lowerconversions (from 0.1 to 0.4) only evaporation of lighter compo-nents is carried out; however, at higher conversions heavy frac-tions, i.e., resins and asphaltenes take place in the crackingreactions which require higher activation energy. It has been sug-gested that an increase in the activation energy reflects a change inthe nature of the rate-controlling step [32].

Fig. 5 shows the conversion of asphaltenes. The interval inwhich asphaltenes undergo the most important transformationsis from 430 to 550 �C approximately and compared with that ofthe atmospheric residue or even other fractions it is the narrowestinterval with cracking as the dominant reaction. Fig. 6 shows theArrhenius plot to determine the kinetic parameters. The coke yield

Table 3Activation energy and pre-exponential factors as function of asphaltene conversion.

x Atmospheric residue Asphaltenes Resins Aromatics

EA ko r EA ko r EA ko r EA ko r

0.1 11.5 4.1 � 105 0.989 41.0 1.4 � 1014 0.978 10.8 4.4 � 104 0.992 10.8 6.7 � 107 0.9910.2 15.3 4.4 � 106 0.999 54.9 1.8 � 1018 0.998 14.8 5.9 � 105 0.998 14.5 7.8 � 109 0.9970.3 12.4 1.0 � 105 0.995 56.7 4.2 � 1018 0.996 21.1 3.9 � 107 0.999 14.2 1.7 � 109 0.9910.4 12.7 4.8 � 104 0.991 52.8 1.6 � 1017 0.993 35.8 1.3 � 1012 0.999 17.6 4.6 � 1010 0.9920.5 20.1 9.5 � 106 0.985 58.6 6.7 � 1018 0.995 45.9 1.2 � 1015 0.996 17.9 1.3 � 1010 0.9900.6 24.8 1.2 � 108 0.994 52.1 4.6 � 1016 0.998 49.8 1.3 � 1016 0.998 21.6 1.1 � 1011 0.9920.7 27.8 2.7 � 108 0.994 53.1 5.9 � 1016 0.999 47.5 1.7 � 1015 0.994 19.8 8.9 � 1010 0.9940.8 30.0 2.4 � 108 0.980 58.6 8.1 � 1017 0.987 48.9 2.6 � 1015 0.999 19.2 1.2 � 1010 0.994

Fig. 7. Volatile fraction of resins at different heating rates.

1.5

2.0

2.5

3.0

0.4

0.60.8

/dt)

3606 E. Alvarez et al. / Fuel 90 (2011) 3602–3607

was the same for all the heating rates (43.1 wt.%) and activationenergy increased continuously having maximum peaks at 0.5 and0.8 of conversion as observed in Table 3. Activation energy ofasphaltenes is higher compared with the atmospheric residueand other fractions that could be associated to the formation of freeradicals produced by cracking reactions. To obtain the atmosphericresidue it is necessary the separation of lighter components by dis-tillation that makes the heavy components to concentrate in hea-vier fractions and under this condition the only prominentreaction is cracking.

In Fig. 7 the volatilization of resins is observed and it endsapproximately at 500 �C. The yield of coke was the same at thethree different heating rates (4.6 wt.%). Fig. 8 shows the Arrheniusplot for calculating the activation energy with good correlationcoefficients. It is observed how this value is changing with increas-ing the conversion and higher activation energies are obtained inthe range from 0.4 to 0.8. This behavior may be a consequence ofthe reactions that occur during its cracking producing free radicalsthat condense as coke. Activation energies and pre-exponentialfactors are summarized in Table 3.

Fig. 5. Volatile fraction of asphaltenes at different heating rates.

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

0.00130 0.00135 0.00140 0.00145 0.00150 0.00155 0.00160 0.00165

ln (

dx/d

t)

0.8

0.1

0.2

0.3

0.40.50.60.7

1/T, K-1

Fig. 6. Arrhenius plot of asphaltenes pyrolysis at different conversions.

-1.0

-0.5

0.0

0.5

1.0

0.0013 0.0014 0.0015 0.0016 0.0017 0.0018 0.0019 0.002 0.0021 0.0022

0.2

0.10.3

0.7

ln (

dx

1/T, K-1

Fig. 8. Arrhenius plot of resins pyrolysis at different conversions.

Profile of volatilization and Arrhenius plot for aromatics is notpresented; only the values of activation energy, pre-exponentialfactor, and correlation coefficient are shown in Table 3. The activa-tion energy was the lowest for this fraction compared with those ofasphaltenes and resins and it is similar to that obtained foratmospheric residue at low conversions. Thermogravimetric datashowed that asphaltenes are the main contributor to coke produc-tion during pyrolysis followed by resins and aromatics. Atmo-spheric residue presented intermediate values of activationenergy between aromatics and resins. At lower conversions theactivation energy was similar to that of aromatics and resins butat conversions higher than 0.5 this value tends to increase foratmospheric residue indicating probably that asphaltenes and res-ins are reacting.

5. Conclusions

Pyrolysis of an atmospheric residue from heavy crude and itsSARA fractions has been studied using thermogravimetric analysis.Results indicated that asphaltenes are the main fraction that formscoke during pyrolysis. The coke yield for asphaltenes was 43.1 wt.%

E. Alvarez et al. / Fuel 90 (2011) 3602–3607 3607

and it is to be expected that asphaltenes were transformed intogases, oils, and resins during cracking reactions. Resins and aro-matics exhibited similar amount of coke but the activation energyfor resins was higher than that of aromatics, whereas saturates vol-atilized almost completely. Atmospheric residue showed an in-crease in activation energy as conversion increased indicatingthat evaporation of lighter components occurred at the initialstages of heating followed by cracking of heavier fractions suchas resins and asphaltenes.

Kinetic data were adjusted by using an isoconversional methodto provide further insights about the pyrolysis of atmospheric res-idue, asphaltenes, resins, and aromatics. In all cases, good correla-tion coefficients were obtained and reaction was well fitted to firstorder. Activation energy for atmospheric residue ranged from 11.5to 30 kcal mol�1, for asphaltenes from 41.0 to 58.6 kcal mol�1, forresins from 10.8 to 49.8 kcal mol�1, and for aromatics from 10.8to 21.6 kcal mol�1. Differences in activation energies could meanchanges in the strength of bonds as pyrolysis is carried out.

Acknowledgements

We thank to CONACYT, Secretaría de Investigación y Posgradodel Instituto Politécnico Nacional (SIP-IPN) for financial supportand Instituto Mexicano del Petróleo for technical support. ElenaAlvarez also thanks to CONACYT for the Grant NO. 227051 andPrograma Institucional de Formación de Investigadores (PIFI).

References

[1] Gonçalves MLA, Teixeira MAG, Pereira RCL, Mercury RLP, Matos JR.Contribution of thermal analysis for characterization of asphaltenes fromBrazilian crude oil. J Therm Anal Calorim 2001;64:697–706.

[2] Douda J, Llanos ME, Alvarez R, López C, Montoya JA. Pyrolysis applied to thestudy of a Maya asphaltene. J Anal Appl Pyrol 2004;71:601–12.

[3] Schucker RC, Keweshan CF. Reactivity of cold lake asphaltenes. Prepr Pap – AmChem Soc, Div Fuel Chem 1980;25:155–65.

[4] Yasar M, Trauth DM, Klein MT. Asphaltene and resid pyrolysis. 2. The effect ofreaction environment on pathways and selectivities. Energy Fuels2001;15:504–9.

[5] Zhao Y, Gray MR, Chung KH. Molar kinetics and selectivity in cracking ofAthabasca asphaltenes. Energy Fuels 2001;15:751–5.

[6] Ali MF, Saleem M. Thermal decomposition of Saudi crude oil asphaltenes. FuelSci Technol Int 1991;9:461–84.

[7] Yoshida R, Takeda S, Teramoto S, Matsushita T, Takeya G. Thermal behavior ofcoal-derived asphaltenes. Fuel Process Technol 1984;9:307–13.

[8] Kok MV. Use of thermal equipment to evaluate crude oils. Thermochim Acta1993;214:315–24.

[9] Hauser A, Bahzad D, Stanislaus A, Behbahani M. Thermogravimetric analysisstudies on the thermal stability of asphaltenes: pyrolysis behavior of heavy oilasphaltenes. Energy Fuels 2008;22:449–54.

[10] Ciajolo A, Barbella R. Pyrolysis and oxidation of heavy fuel oils and theirfractions in a thermogravimetric apparatus. Fuel 1984;63:657–61.

[11] Douda J, Alvarez R, Navarrete J. Characterization of Maya asphaltene andmaltene by means of pyrolysis application. Energy Fuels 2008;22:2619–28.

[12] Speight JG. Petroleum asphaltenes. Part 2. The effect of asphaltenes and resinconstituents on recovery and refining processes. Oil Gas Sci Technol2004;59:479–88.

[13] Shiroto Y, Nakata S, Fukui Y, Takeuchi C. Asphaltene cracking in catalytichydrotreating of heavy oils. 3. Characterization of products from catalytichydroprocessing of Khafji vacuum residue. Ind Eng Chem Proc Des Dev1983;22:248–57.

[14] Dias JR. Handbook of polycyclic hydrocarbons. Part B. Polycyclic isomers andheteroatoms analogs of benzenoid hydrocarbons. New York: Elsevier; 1988.

[15] Magaril RZ, Aksenova EI. Study of the mechanism of coke formation in thecracking of petroleum. Int Chem Eng 1968;8:727–9.

[16] Savage PE, Klein MT, Kukes SG. Asphaltene reaction pathways. III: Effect ofreaction environment. Energy Fuels 1988;2:619–28.

[17] Schucker RC. Thermogravimetric determination of the coking kinetics of Arabheavy vacuum residuum. Ind Eng Chem Proc Des Dev 1983;22:615–9.

[18] Shih SM, Sohn HY. Nonisothermal determination of the intrinsic kinetics of oilgeneration from oil shale. Ind Eng. Chem Proc Des Dev 1980;19:420–6.

[19] Park YC, Paek JY, Bae DH, Shun D. Study of pyrolysis kinetics of Alberta oil sandby thermogravimetric analysis. Kor J Chem Eng 2009;26:1608–12.

[20] Friedman HL. Kinetics of thermal degradation of char-forming plastics fromthermogravimetry. Application to a phenolic plastic. J Polym Sci, Part C: PolymLett 1964;6:183–95.

[21] Trejo F, Rana MS, Ancheyta J. Thermogravimetric determination of coke fromasphaltenes, resins and sediments and coking kinetics of heavy crudeasphaltenes. Catal Today 2010;150:272–8.

[22] Karacan O, Kok MV. Pyrolysis analysis of crude oils and their fractions. EnergyFuels 1997;11:385–91.

[23] Phillips CR, Haidar YC, Poo C. Kinetic models for the thermal cracking ofAthabasca bitumen. Fuel 1985;64:678–91.

[24] Khulbe KC, Sachdev AK, Mann RS, Davis S. TGA studies of asphaltenes derivedfrom cold lake (Canada) bitumen. Fuel Process Technol 1984;8:259–66.

[25] Ritchie RGS, Roche RS, Steedman W. Pyrolysis of Athabasca tar sands: analysisof the condensable products from asphaltene. Fuel 1979;58:523–30.

[26] Moschopedis SE, Parkash S, Speight JG. Thermal decomposition of asphaltenes.Fuel 1978;57:431–4.

[27] Yagmur S, Durusoy T. Kinetics of the pyrolysis and combustion of Göynük oilshale. J Therm Anal Calorim 2006;86:479–82.

[28] Li S, Yue C. Study of different kinetic models for oil shale pyrolysis. Fuel ProcessTechnol 2004;85:51–61.

[29] Torrente MC, Galán MA. Kinetics of the thermal decomposition of oil shalefrom Puertollano (Spain). Fuel 2001;80:327–34.

[30] Sonibare OO, Egashira R, Adedosu TA. Thermo-oxidative reactions of Nigerianoil sand bitumen. Thermochim Acta 2003;405:195–205.

[31] Meng M, Hu H, Zhang Q, Li X, Wu B. Pyrolysis behaviors of Tumuji oil sand bythermogravimetry (TG) and in a fixed bed reactor. Energy Fuels2007;21:2245–9.

[32] Collett GW, Rand B. Thermogravimetric investigation of the pyrolysis of pitchmaterials. A compensation effect and variation in kinetic parameters withheating rate. Themochim Acta 1980;41:153–65.