Quantum-Chemistry Study of Asphalt Oxidative Aging: An XPS-Aided

Article pubs.acs.org/IECR

Quantum-Chemistry Study of Asphalt Oxidative Aging: An XPS-Aided Analysis Tongyan Pan,*,† Yang Lu,‡ and Stephen Lloyd† †

The Catholic University, 620 Michigan Avenue, N.E., Washington, DC 20064, United States Institute for Research in Electronics and Applied Physics, Energy Research Building, No. 223, University of Maryland, College Park, Maryland 20742-3511, United States

‡

ABSTRACT: Asphalt derived from crude oil (or petroleum) is an important base organic material for many industrial purposes. Oxidative hardening occurs throughout the service life of asphalt materials, which could significantly change the desired physicochemical properties. The study of asphalt oxidative hardening has thus far been focused on the changes in the physical properties, mainly the viscosity and ductility of bulk asphalt. Such phenomenological approaches meet the direct engineering needs, however do not help understand the fundamental physicochemical mechanisms of asphalt hardening. From this standpoint, this paper aims at exploring the chemical basis of asphalt oxidative hardening by establishing an ab initio quantum chemistry (QC) based physicochemical environment, in which the possible chemical reactions between asphalt ingredients and oxygen, as well as the incurred changes in their physical behavior, can be readily studied. X-ray photoelectron spectroscopy (XPS) was used to validate the bulk asphalt model, of which the results showed high agreement to the model predictions, and thereby prove the validity of the QC-based models developed for studying the oxidative behavior of general asphalt materials.

■

INTRODUCTION Asphalt derived from crude oil (petroleum) is the primary binding and/or waterproofing component in road pavements and shingle roofs.1 Today 95% of the road pavement we use is made of asphalt mixtures. Oxidized grades of asphalt, offering good dimensional stability which is essential to prevent tiles buckling in high or low temperatures, are often used as the main constituent of the backing material on carpet tiles.2 In addition to their important role in the construction industry, asphalt materials today are commonly used as a base organic material, such as a key component of the vibration-lessening material in computers, dampening components in sound deadening panels of cars and household appliances, or an additive to increase weather resistance and permanence of paints, inks, and rubbers.2,3 When added with oil, the asphalt ingredient improves the wetting of fillers and pigments and can make the color much darker. Being an organic product from the remains of ancient organisms, asphalt is subject to chemical oxidation by reactions with atmospheric oxygen. Oxidation can cause the hardening of asphalt, leading to the sacrifice of its desirable physical properties. For example, in asphalt pavements, oxidative hardening is responsible for mixture embrittlement that could speed up pavement cracking; as a roofing material (in shingles), asphalt embrittlement from oxidative hardening can promote the loss of protective granules, substrate shrinkage, and cracking.4−6 In general, oxidative hardening of asphalt is believed to be caused by the generation of oxygen-containing polar chemical functionalities on asphalt molecules, which in turn can cause agglomeration among molecules due to increased physicochemical associations such as hydrogen bonding, van der Waals force, and Coulomb force.7,8 In addition to oxygen-containing functionalities, oxidation also can cause aromatization of certain © 2012 American Chemical Society

asphalt molecules that facilitates further agglomeration of asphalt components in ambient conditions.9 However, the systematic study of asphalt oxidation from the chemical perspective has not been seriously attempted due to one underlying challenge in the state of knowledge: the complexity of chemical composition in asphalt, which has naturally led to the lack of an effective tool for studying the manifold chemical reactions involved in asphalt oxidation.7−9 Asphalt is composed of molecules with different chemical bonds, sizes, and polarities, ranging from nonpolar hydrocarbons such as waxes to highly polar or polarizable hydrocarbon molecules containing aromatic ring systems that incorporate heteroatoms such as oxygen, nitrogen, and sulfur.7 Molecular association of polar components in asphalts can form agglomerates at the nanoscale level, of which the stabilities are highly dependent on temperature. As temperature increases, the intermolecule bonds within agglomerates can break and the sizes of agglomerates will be reduced.8 It is the reversible nature of such interactions that gives asphalts their unique hightemperature viscosity susceptibility at the macro level. Also, the diversity in compositions makes asphalts from different sources differ much in their oxidation behavior, as clearly demonstrated by a previous experimental road test which included different types of asphalts.10 With the complexity in asphalt composition, it is experimentally impracticable to track the sheer number of chemical reactions possibly involved in asphalt oxidation. Instead, the state of knowledge has concentrated on identifying the polar functional groups in asphalt and characterizing the Received: Revised: Accepted: Published: 7957

March 17, 2012 May 11, 2012 May 16, 2012 May 16, 2012 dx.doi.org/10.1021/ie3007215 | Ind. Eng. Chem. Res. 2012, 51, 7957−7966

Industrial & Engineering Chemistry Research

Article

Table 1. ReaxFF Parameters for Determining Bond Order and Ebond of Asphalt (A C−H−O−S System) bond

De (kcal/mol)

pbe,1

pbe,2

pbo,1

pbo,2

pbo,3

pbo,4

pbo,5

pbo,6

C−C C−H H−H C−O O−O H−O C−S H−S O−S S−S

145.4070 167.1752 188.1606 171.0470 90.2465 216.6018 128.9942 151.5159 100.0000 96.1871

0.2176 −0.4421 −0.314 0.36 0.995 −0.4201 0.1035 −0.4721 0.5563 0.0955

−0.1940 1.0000 1.0000 −0.2660 −0.1850 1.0000 −0.2398 1.0000 −0.4577 −0.2373

5.9724 8.5445 5.7082 5.0637 6.2396 5.9451 5.6731 7.0050 7.1145 6.4757

1.0000 0.0000 0.0000 0.0000 1.0000 0.0000 1.0000 1.0000 1.0000 1.0000

8.6733 0.0000 0.0000 7.4396 7.5281 0.0000 8.1175 0.0000 12.7569 9.7875

1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

−0.7816 0.0000 0.0000 −0.1696 −0.2435 0.0000 −0.5211 0.0000 −0.4038 −0.4781

0.3217 0.5969 0.6816 0.3796 0.9704 0.9143 0.6000 0.6000 0.6000 0.6000

powerful computers and parallel computing methods, advances in the statistical mechanics, and new experimental data. The QC-based methods today are readily implemented using different computer-based programs, of which the dominant one is the Density Functional Theory (DFT).19 Even though recent computing capacity has significantly improved the speed of QC computation that enables accurately predicting the geometries and vibrational energies, extensive use of QC computation today is still limited to small atomic/ molecular systems and therefore is not practical for studying asphalt oxidation. From this standpoint, it is necessary to have an accurate force-field based method that enables quick evaluation of interatomic bonding and forces. Such a first principle based force-field method has the advantage of being able to simulate chemical reactions while obtaining fast computation speed as traditional force field methods do, and therefore is highly desired for simulating asphalt oxidation that involves thousands of atoms for a realistic simulation.

associating mechanisms or forces among the groups that essentially lead to molecular agglomerations.9 Considerable progress has been made in this direction via component separation and characterization based on the reactivity and/or polarity of the various molecular types present in asphalt.11,12 With the classical separation methods, molecules in asphalt can be grouped into molecular types or fractions with a narrower range of properties based on their chemical functionality. Research efforts made in characterizing polar functional groups of asphalt is significant as asphalt materials are produced from natural organisms over millions of centuries, which has limited the types of polarity groups present in spite of its extreme molecular complexity. However, restrained by the available technologies for determining the polarity of asphalt ingredients, particularly the technologies for discerning the subtle chemical differences among asphalt molecules, the existing studies do not offer much molecular information regarding asphalt oxidation except providing some general groups/categories of distinct dissolubilities.11,12 Nevertheless, such separation and classification of molecular types are useful in providing chemically definitive component fractions of asphalt for more advanced characterization, such as molecularbased studies, and verifying the results of such studies. Within this context, a research study was recently launched to explore the molecular-scale processes of asphalt oxidation, aiming at elucidating the chemical mechanisms of asphalt oxidative hardening and the physicochemical characteristics of oxidized asphalts under a condition similar to that of a typical service environment. This manuscript presents the results from the study, focusing on reporting the development of an atomistic model, validation of the model, and the utilization of the model in simulating asphalt oxidation.

■

FUNDAMENTALS OF REACTIVE FORCE FIELD FOR ASPHALT OXIDATION One first principle based force-field method is the Reactive Force Field (ReaxFF) developed by van Duin et al.16 that, since its advent, has been parametrized and implemented to a variety of materials and processes, including hydrocarbon reactions, transition-metal-catalyzed nanotube formation, and high-energy materials. Recently ReaxFF has been extended to more materials including various metals, ceramics, silicon, and polymers, and is now used as a general tool for chemical simulations. A reactive force field of a C−H−O−S system as originally developed by van Duin et al.16,20 is re-evaluated and used in this study to simulate the oxidation of asphalt in exposure to oxygen. The overall energy of the C−H−O−S ReaxFF contains a series of energy contributions per eq 1, all determined using QC. The actual number of energy contributions of a ReaxFF system depends on the type of chemical species and processes to be modeled. The ReaxFF for simulating asphalt oxidation in an oxygen environment involves complex chemical reactions among the five element species of asphalt and external oxygen.

■

QUANTUM CHEMISTRY PERSPECTIVE OF ASPHALT OXIDATION Although a fundamental understanding of asphalt oxidation is highly desired when dealing with the problems related to asphalt aging, this task cannot be readily accomplished within the frame of traditional organic or polymer chemistry that has centered on experimenting bulk asphalt using different devices. The past few decades has seen significant advances made in using spectroscopic techniques for characterizing molecular and atomic phenomena that involve essentially the transfer of electrons.13−18 Because the oxidation of asphalt is mainly a chemical process, an approach capable of simulating electrontransfer processes is desired. In this study, the quantum chemistry (QC) is resorted to as such an approach. In recent years, the efforts made in atomic-level description of phenomena have been accelerated by the availability of

Esystem = Ebond + Eval + +Etors + EH − bond + EvdWaals + ECoulomb

(1)

In eq 1, the bond energy Ebond describes the chemical energy between each pair of bonded atoms. Valence angle energy Eval accounts for the energy contribution from valence angle; torsion rotation energy Etors ensures proper dependence of the energy of torsion angle for bond order approaching trivial and 7958

dx.doi.org/10.1021/ie3007215 | Ind. Eng. Chem. Res. 2012, 51, 7957−7966


Article

that the valence angle energy contribution disappears smoothly during bond dissociation. Equation 4b deals with the effects of over/under coordination in central atom j on the valence angle energy. Atom undercoordination/overcoordination parameter Δj is defined for atoms as the difference between the total bond order around the atom and the number of its bonding electrons valence. The equilibrium angle Θ0 for Θijk depends on the sum of π-bond orders around the central atom j and changes from around 109.47° for sp3 hybridization (π-bond = 0) to 120° for sp2 (π-bond = 1) to 180° for sp (π-bond = 2) based on the geometry of the central atom j and its neighbors. Table 2 lists

bond order greater than 1; van der Waals interactions energy EvdWaals accounts for the van der Waals interactions; and Coulomb interactions energy ECoulomb between all atom pairs adjust for orbital overlap between atoms at close distances. The fundamental ReaxFF assumption is that the bond order BO′ij between a pair of atoms is dependent on the interatomic distance rij according to eq 2, in which the parameter ro is the bond radius and the series of parameters ps describe the bond order. In calculating bond orders, ReaxFF distinguishes between contributions from σ bonds, π bonds, and ππ bonds. The bond orders BO′ij are updated in each time step. The energy of the system is finally determined by summing up all the energy contributions per eq 1. The three pairs of parameters: pbo,1 and pbo,2, pbo,3 and pbo,4, and pbo,5 and pbo,6 in eq 2 correspond to the orders of the σ bond, the first π bond, and the second ππ bond, respectively, of which the values are given in Table 1. The values of the exponential terms is unity below a particular interatomic distance ro and negligible at a longer distance. The bond energy is calculated from the bond order BO′ij.

Table 2. Parameters for Determining Valence Angle Energy

BOij′ = BOijσ + BOijπ + BOijππ ⎡ ⎡ ⎛ rij ⎞ pbo2 ⎤ ⎛ rij ⎞ pbo4 ⎤ = exp⎢pbo1 ·⎜ σ ⎟ ⎥ + exp⎢pbo3 ·⎜ π ⎟ ⎥ ⎢⎣ ⎢⎣ ⎝ ro ⎠ ⎥⎦ ⎝ ro ⎠ ⎥⎦ ⎡ ⎛ rij ⎞ pbo6 ⎤ + exp⎢pbo5 ·⎜ ππ ⎟ ⎥ ⎢⎣ ⎝ ro ⎠ ⎥⎦

(2)

The connectivity related terms in eq 1 such as the bond energy, valence angle, and torsion angle energy terms are also bond order dependent and will disappear upon bond dissociation. This feature of ReaxFF ensures a smooth transition of the energy and force from a bonded system to a nonbonded system. In addition to the valence interactions which depend on overlap, there are repulsive interactions at short interatomic distances due to Pauli principle orthogonalization and attraction energies at long distances due to dispersion. These interactions, comprising van der Waals and Coulomb forces, are included for all atom pairs, thus avoiding awkward alterations in the energy description during bond dissociation. In this respect, ReaxFF is similar in spirit to the central valence force fields used earlier in vibrational spectoscropy. The following sections introduce these energy contribution terms. Bond Energy. Ebond of the C−H−O−S system was determined according to eq 3. De, pbe,1, and pbe,2 are bond parameters. Upon the dissociation of a bond, the bond order BO′ij approaches zero making the bond energy term Ebond disappear (eq 3). To simulate oxidation of asphalt, the energy contributions developed by van Duin et al.16,20 are used to determine the overall system energy―hence the ReaxFF potential function―for the atomic-level modeling of asphalt oxidation, with the parameters given in Table 1.

( ( ) )⎤⎦⎥

⎡ Ebond = −De ·BOij′·exp⎣⎢pbe1 1 − BOij′

angle atoms

Θ0 (degree)

pval,1 (kcal/mol)

pval,2(1/radian2)

pval,3

pval,4

C−C−C C−C−H H−C−H C−H−H C−H−C H−H−H C−C−O O−C−O H−C−O C−O−C C−O−O O−O−O C−O−H H−O−O H−O−H C−H−O C−H−S O−H−O H−H−O C−C−S C−S−C H−C−S C−S−H C−S−S H−S−H H−S−S H−H−S

70.0265 69.7786 74.6020 0.0000 0.0000 0.0000 72.9588 80.0708 66.6150 79.1091 83.7151 80.0108 78.1533 84.1057 79.2954 0.0000 0.0000 0.0000 0.0000 74.9397 86.9521 74.9397 86.1791 85.3644 93.1959 84.3331 0.0000

13.6338 10.3544 11.8629 0.0000 3.4110 27.9213 16.7105 45.0000 13.6403 45.0000 42.6867 38.3716 44.7226 9.6413 26.3838 0.0019 0.0019 0.0019 0.0019 25.0560 36.9951 25.0560 36.9951 36.9951 36.9951 36.9951 0.0019

2.1884 8.4326 2.9294 6.0000 7.7350 5.8635 3.5244 2.1487 3.8212 0.7067 0.9699 1.1572 1.3136 7.5000 2.2044 6.0000 6.0000 6.0000 6.0000 1.8787 2.0903 1.8787 2.0903 2.0903 2.0903 2.0903 6.0000

0.1676 0.1153 0.1367 0.0000 0.0000 0.0000 1.1127 1.1127 0.0755 0.6142 0.6142 0.6142 0.1218 0.1218 0.1218 0.0000 0.0000 0.0000 0.0000 0.0559 0.0559 0.0000 0.0000 0.0559 0.0000 0.0000 0.0000

1.0400 1.0400 1.0400 1.0400 1.0400 1.0400 1.1880 1.1880 1.0500 1.0783 1.0783 1.0783 1.0500 1.0500 1.0500 1.0400 1.0400 1.0400 1.0400 1.0400 1.0400 1.0400 1.0400 1.0400 1.0400 1.0400 1.0400

the parametric values used to determine the valence angle energy Eval.

( ) (

) ( )

Eval = f1 BOij ·f1 BOjk ·f2 Δj

2⎤ ⎡ · pval1 − pval1 exp⎣⎢ −pval 2 Θ0(BO) − Θijk ⎦⎥

{

( )

(

( )

f2 Δj =

(3)

Valence Angle Energy. Eval was determined according to eq 4. Just like bond energy, it is important that the energy contribution from valence angle term goes to zero as the bond orders in the valence angle go to zero. Equation 4 determines the energy associated with deviations in valence angle Θijk from its equilibrium value Θ0. The term f1(BO) per eq 4a ensures

}

(4a)

p

f1 BOij = 1 − exp( −BOij val3)

pbe2

)

( ) 1 + exp(Δj ) + exp(−pval 4 ·Δj )

(4b)

2 + exp Δj

(4c)

Torsion Rotation Energy. Etors ensures that dependence of the energy of torsion angle ωijk accounts properly as bond order approaches trivial and bond order is greater than 1. This is given in eqs 5, 5a, and 5b. Θijk and Θjkl are valence angles. Δ is the atom undercoordination/overcoordination parameter. 7959



Article

Parameters ptor3 and ptor4 in eq 5b equal zero for the specific types of elements in this study. Table 3 lists the values of

interaction (eq 7a), excessively high repulsions between bonded atoms (1−2 interactions) and atoms sharing a valence angle (1−3 interactions) are avoided. Dij is the basic energy term of an atomic pair; rvdW is the van der Waal radius. The parameter αij is set to be 1 for the elements of this study. The other parameters for EvdWaals are given in Table 5.

Table 3. Parameters for Determining Torsion Rotation Energy (X Represents Any of the Elements C, H, O, or S) torsion angle C−C−C−C C−C−C−H H−C−C−H X−C−H−X X−H−H−X X−C−O−X X−H−O−X X−O−O−X X−C−C−X X−C−S−X X−S−S−X X−H−S−X

V2 (kcal/mol) 23.2168 45.7984 44.6445 0.0000 0.0000 16.7344 0.1000 68.9706 0.9305 30.3435 −42.7738 0.0000

V3 (kcal/mol) 0.1811 0.3590 0.3486 0.0000 0.0000 0.5590 0.0200 0.8253 0.0000 0.0365 0.1515 0.0000

ptor −4.6220 −5.7106 −5.1725 0.0000 0.0000 −3.0181 −2.5415 −28.4693 −24.2568 −2.7171 −2.2056 0.0000

⎧ ⎡ ⎛ f (rij) ⎞⎤ ⎪ ⎟⎟⎥ EvdWaals = Dij ·⎨exp⎢αij·⎜⎜1 − 5 ⎪ rvdW ⎠⎥⎦ ⎩ ⎢⎣ ⎝ ⎡ ⎛ f (rij) ⎞⎤⎫ ⎪ 1 ⎟⎟⎥⎬ − 2·exp⎢ ·αij·⎜⎜1 − 5 rvdW ⎠⎥⎦⎪ ⎢⎣ 2 ⎝ ⎭ ⎡ p f5 (rij) = ⎢rij vdW 1 + ⎢⎣

(

)

· 1 − 2cos ωijkl +

(

Dij (kcal/mol)

rvdW (Å)

γw

αij

PvdW1

0.1818 0.0600 0.088 0.2099

1.8857 1.603 1.9741 2.0677

2.0784 4.4187 7.7719 4.9055

9.5928 9.3951 10.219 9.9575

1.5591 1.5591 1.5591 1.5591

2

)}

⎤ 1 V2 1 + cos 3ωijkl ⎥ ⎦ 2

(

)

Coulomb Interactions Energy. ECoulomb is considered between all atom pairs. To adjust for orbital overlap between atoms at close distances a shielded Coulomb potential ECoulomb is used (see eq 8). qi ·qj ECoulcomb = C· 3 ⎤1/3 ⎡ 3 + γ r 1/ ⎢⎣ ij ij ⎥ ⎦ (8)

(5)

(

) = ⎡⎣1 − exp(BOij)⎤⎦ ·⎡⎣1 − exp(BOjk)⎤⎦ ·[1 − exp(BOkl)]

f3 BOij , BOjk , BOkl

( )

(5a)

(

)

f4 Δj , Δk =

(7a)

C H O S

)

{

1/ pvdW 1

atom units

Etors = f3 BOij , BOjk , BOkl ·sin Θijk ⎡1 ·sin Θjkl⎢ V1·exp ptor · BOjk − 1 + f4 (Δi , Δk ) ⎣2

⎛ 1 ⎞ pvdW 1⎤ ⎜⎜ ⎟⎟ ⎥ ⎝ γw ⎠ ⎥⎦

Table 5. Parameters for Determining van der Waals Interaction Energy

parameters V1, V2, and ptor used to determine the torsion rotation energy Etors.

(

(7)

(

) 1 + exp(−Δj − Δk ) + exp(Δj + Δk ) 2 + exp −Δj − Δk

Atomic charges are calculated using the electron equilibration method.21,22 The initial values for the electron equilibration method parameters (η, χ, and γ) listed in Table 6 were determined by Njo et al.23 The optimization parameter γij in eq 8 is included for orbital overlap correction in the electron equilibration method.

(5b)

Hydrogen Bond Interactions. Equation 6 describes the bond-order dependent hydrogen bond term for an X−HZ system as incorporated in ReaxFF. Table 4 lists the values of

Table 6. Parameters for Determining Coulomb Interactions Energy

Table 4. Parameters for Hydrogen Bond Interactions Energy hydrogen bond

r°hb

phb1

phb2

phb3

atom type

χ (eV)

γ (Å)

2.0431 2.6644 2.1126 1.9461

−6.6813 −3.9547 −4.5790 −4.0000

η (eV)

O−H−O O−H−S S−H−O S−H−S

3.5000 3.5000 3.5000 3.5000

1.7295 1.7295 1.7295 1.7295

C H O S

6.9235 9.8832 7.8386 8.2545

5.7254 3.8196 8.500 6.500

0.8712 0.7625 1.0804 1.0336

■

parameters rohb, phb1, phb2, and phb3 used to determine the torsion rotation energy EHbond.

ATOMISTIC MODELING OF ASPHALT OXIDATION By chemical composition, asphalt in industrial use―primarily petroleum asphalt―is a molecular system of different chemical elements: carbon (∼85%), hydrogen (∼11%), and some trace species including sulfur (1−5%), nitrogen (0.3− 1.1%), oxygen (0.2−0.8%), vanadium (4−1400 ppm), and nickel (0.4−110 ppm). Bulk asphalt consists of molecules with different structures formed by atoms of these chemical species. As in general organic substances, structures of the asphalt

EHbond = phb1 ·⎡⎣1 − exp(phb2 ·BOXH )⎤⎦ ⎡ ⎛ ro ⎞⎤ 8⎛ ΘXHZ ⎞ r ⎥ ⎟ ⎜ ·exp⎢phb3 ⎜ hb + HZ o − 2⎟ · sin ⎝ ⎢⎣ ⎝ rHZ rhb 2 ⎠ ⎠⎥⎦

(6)

van der Waals Interactions Energy. EvdWaals accounts for the van der Waals interactions using a distance-corrected Morse-potential given in eq 7. By including a shielded 7960



Article

Table 7. Molecules Included in the Bulk Asphalt Model

Table 8. Thermodynamics Study of Representative Asphalt Molecules (130 °C, 1 ATM Oxygen Partial Pressure)

asphalts in their service conditions, a typical petroleum asphalt can be divided into three representative functional groups: asphaltenes (relatively large molecules insoluble in straightchain alkanes such as n-heptane or n-pentane), resins (naphthene aromaticsalkane-soluble and elute in aromatic solvents such as benzene, and polar aromaticsalkane-soluble and elute in more polar solvents, such as an aromatic/alcohol mixture), and oils or saturates (molecules that elute immediately in n-heptane).11,12 The resin and oil portions together also are known as maltenes or petrolenes. Asphaltenes, with higher molecular weight and polarity than resins and saturates, if not properly dispersed by the resinous components of maltenes can cause reduced asphalt compatibility and asphalt durability thereby.11,12 In this study, a bulk asphalt model is made by mixing a portion of each component functional group to study the oxidation of asphalts under general service conditions. The normal-docosane (n−C22H46), a saturated aliphatic hydrocarbon with the highest concentration in the speciation reported by Kowalewski et al.,25 was selected to represent the saturate/oil portion of asphalt. Its melting point Tm = 44 °C and boiling point Tb = 369 °C are consistent with the waxy component of asphalt. The normal-docosane was once used by

molecules are more important than the amount of each element present. Some heteroatoms, i.e., sulfur, nitrogen, oxygen, are attached to carbon atoms in different configurations, forming different polar molecules or functional groups due to imbalance of electrochemical forces which are weaker than the primary chemical bondthe covalent bond that holds atoms within asphalt molecules. Molecular interactions among the polar molecules of asphalt are the primary mechanism of agglomerations that strongly influence the physical properties and macroscopic performance of asphalt. Usually a small amount of polar molecules can have a great effect on most engineering behavior of asphalt, such as the resistance to binder-aggregate stripping that in general is believed to be controlled by the adsorption of polar molecules to the surface of mineral aggregates.8,12 A proper balance among the different functional groups hence is important for producing durable and resistant asphalts against detrimental physical property changes during oxidative aging.24 A classic method for separating asphalt into different functional groups is per their solubilities in a series of organic solvents, supplemented with chromatographic analyses (namely, the Corbett method).11,12 To study the oxidation of 7961



Article

Zhang and Greenfield in their asphalt model.26 An asphaltene component of asphalt taken from the NMR studies by Artok et al.,27 i.e., C64H52S2, was adopted to represent a portion of the asphaltene molecules. This asphaltene molecule contains a moderate-size aromatic core and a few branches that represent a statistically average asphaltene molecule in general bulk asphalt. For the resin group, the 1,7-dimethylnaphthalene with the number of aromatic rings and side chains intermediate between that of saturates and asphaltenes was selected, which resembles some asphalt molecules depicted earlier.28 The ball− stick schemes of the selected molecules are shown in Tables 7 and 8. With the selected representative molecules of asphalt, a series of single-molecule models, including a bulk asphalt model (per fractions in Table 7) and one model for each representative molecule (Table 8), were developed to study the oxidation behavior of each asphalt molecule and the bulk asphalt under the condition of temperature = 130 °C and one standard atmospheric pressure (1 ATM). In the simulated condition, in addition to the asphalt molecules only oxygen molecules are included in the models that are separated at an average distance of 1.2 nm (same as the intermolecule distance in air at 1 ATM). The chemical functionality developed in asphalt during the oxidation at 130 °C was found to be similar to that developed during normal pavement aging at ambient temperatures.29 Also, it was shown that oxidation of asphalts in exposure to 100% oxygen is equivalent to that exposed to air at 300 psi pressure (such as in a Pressure Aging Vessel (PAV) condition), which is believed to be in the range typically found for asphalts with five or more years of pavement service.30 The use of 1 ATM pressure for modeling is important for studying the escape of existing small-molecule-weight components of asphalt or newly generated ones during asphalt oxidation from bulk asphalt, which might be an important mechanism contributing to asphalt hardening. Thermodynamics and kinetics are two distinct aspects of a chemical reaction. While thermodynamics describes the possibility and direction of a reaction in terms of the free energy, ΔG, released or consumed during a chemical reaction; kinetics concerns how fast the chemical reaction can reach equilibrium, i.e., the rate of the reaction, as influenced by factors of reaction condition such as the temperature and concentrations of reactants. The thermodynamics aspect of asphalt oxidation in this study is evaluated by observing the oxidation behavior of the individual asphalt molecules in a domain of oxygen molecules all at the same concentration. The kinetics of asphalt oxidation depends on factors such as asphalt composition, products formed in oxidation, reaction temperature, oxygen partial pressure, and other physicochemical effects of the system. Accordingly, the kinetics of the different molecular species was evaluated by studying the bulk asphalt model oxidized under the same condition, i.e., at 1 ATM pressure (oxygen only) and 130 °C using the compositional fractions shown in Table 7. The overall mixture composition, i.e., the relative amounts of the three ingredient functional groups used for making the bulk asphalt, was determined based on the measurements by Storm et al.,31 who applied NMR to asphalt fractions separated by the common alkane precipitation method (heptane used) and identified the balance among aromatic and alkane carbons in the asphaltene, resin, and saturate mixtures through the different peak positions of different kinds of carbon atoms. An asphaltene fraction of

19.6 wt % was selected in this study, which is close to the 22 wt % reported by Storm et al.31 As such, each ReaxFF model contains an individual (for thermodynamic study) or a number (for kinetics study) of the saturate, resin, and asphaltene molecules, forming a domain embedded in oxygen molecules (see Table 8 for the singlemolecule models, and Figure 1a and Table 7 for the bulk asphalt model). The bulk asphalt model was packed to a density of 1000 kg/m3 prior to adding oxygen molecules into the modeled domain, a typical value for asphalts at ambient conditions. The ReaxFF models were run at a Velocity Verlet plus Berendsen ensemble32,33 at a time step of 0.25 fs using a parallel computing algorithm developed based on the MessagePassing Interface (MPI) Standard. The Virginia Tech’s suite of high performance computersSystem Xwas used for the computation. System X is a supercomputer comprising 1100 Apple PowerMac G5 cluster nodes and capable of running at 12.25 Teraflops (meaning 1012 FLoating point OPerations per Second). Each ReaxFF model was kept running until the equilibrium of oxidation was reached, with no more new species observed for one hour. The bulk asphalt model after oxidation simulation is shown in Figure 1b.

■

VALIDATION OF THE ATOMISTIC MODEL X-ray photoelectron spectroscopy (XPS), also known as the electron spectroscopy for chemical analysis (ESCA), is a powerful surface technique for accurately detecting the presence and relative quantities of chemical elements (except hydrogen and helium) in a sample surface (usually 1−10 nm deep and 10 μm wide depending on the input energy). XPS analysis can obtain information of the state and environment of atoms in the sample, which can be further analyzed for information of the surface structure. Such powerful ability of XPS is particularly useful for studying the oxidative aging of asphalt binders,34−36 for which a sufficiently thin sample (e.g., a few μms) can be easily prepared and oxidized with uniform oxidation throughout the entire depth of the sample in a relatively short period of time (e.g., 2 h at 130 °C).37 The XPS technique was used in this study to validate the bulk asphalt model by determining and comparing the amounts of generated functional groups before and after asphalt oxidation. XPS analysis is based on the photoelectric effect stemming from the ejection of electrons from the surface of a sample in exposure to electromagnetic radiation of sufficient energy. Electrons emitted have characteristic kinetic energies proportional to the energy of the radiation, according to eq 9, in which KE is the kinetic energy of the electron, h is Planck’s constant, ν is the frequency of the incident radiation, Eb is the ionization or binding energy, and φ is the work functiona constant dependent on the X-ray photoelectron spectrometer used. In an XPS analysis, a level of energy radiation is used to expel core electrons from a sample, and the kinetic energies of the resulting core electrons are measured. Using eq 9 with the kinetic energy KE and known frequency ν of radiation, the binding energy Eb of the ejected electron can be determined. By Koopman’s theorem,38 which states that binding or ionization energy is equivalent to the negative of the orbital energy, the energy of the orbital from which the electron originated can be determined. These orbital energies are characteristic of the element and its state. KE = hυ − Eb − φ 7962

(9)



Article

Figure 1. Bulk asphalt model to study kinetic of asphalt oxidation: (a) before oxidation; (b) after oxidation.

software associated with the Physical Electronics 5600 XPS device detects the transform of sigma carbon−carbon single bonds (before oxidation) to carbon−oxygen double-bonded ketones (after oxidation), and the transform of sulfides (before oxidation) to sulfoxides (after oxidation) at the carbon and sulfur peaks, as predicted by the ReaxFF-based atomistic models as shown in Table 8 and Figure 3. Furthermore, the generation of ketones and sulfoxides during oxidation, as quantified by XPS for the same asphalt materials as modeled atomistically, is shown in Figure 3, which further validates the ReaxFF-based bulk asphalt model by showing close results in the amounts of newly generated ketones and sulfoxides. The slight difference between the experimental results and numerical predictions might come from the

A Physical Electronics 5600 XPS setup was used in this study, which provides the relative frequencies of binding energies of electrons measured in 0.1 electron-volts (0.1 eV). The binding energies were then used to identify the elements to which each peak corresponded, as elements of the same kind in different states and environments usually have different characteristic binding energies. Furthermore, comparing the areas under the peaks gave relative percentages of the elements detected in a sample. An initial survey XPS scan was conducted to identify the elements present in the asphalt sample (Figure 2a), followed with a subsequent high-resolution scan of the peaks of interest, i.e., the carbon and sulfur in this study, to identify ketones and sulfoxides as observed in the atomistic simulation (Figure 2b). For the high-resolution scan, the computer 7963



Article

different products. The carbon atoms in chain alkane, i.e., saturateC22H46, are quite resistant to oxidation; instead the chain can break into shorter chains relatively easily. This could constitute an important mechanism for asphalt hardening as liquid portions of asphalt vaporize into air under general service conditions. The carbon atoms in aromatic (C12H12) show a partial level of oxidation, i.e., oxidized at the benzylic position, resulting in a ketone structure. The oxidation did not occur to the carbon atoms forming benzene rings. For the asphaltene molecules, i.e., the asphalteneC64H52S2, sulfoxidation occurs to one sulfur atom forming a sulfoxide that consists of a double bond with an oxygen atom (SO), and ketonization occurs to one benzylic carbon atom. In asphaltenes, sulfoxidation seems to be easier than ketonization as the sulfinyl appears earlier than the carbonyl in the same oxidized asphaltene molecule. However, no aromatization was observed for the individual asphalt molecules. The snapshot of the oxidation of bulk asphalt at 2 h, as shown in Figure 1b, displays two obvious phenomena: (1) generation of low-molecular-weight saturate molecules (with shorter chains than the two original saturates), which tend to leave the bulk asphalt via diffusion; and (2) agglomeration of asphaltenes and aromatics as attracted by the oxygen-bearing functional groups. Some unbroken long saturates were also entangled with such agglomeration due to larger molecular weight and/or electrostatic forces. These two observations are in agreement with the phenomena of chain breaking of saturates, sulfoxidation, and ketonization as observed in the preceding thermodynamics models (on individual molecules). Undoubtedly, the two mechanisms contribute a lot to the oxidative hardening of asphalt. The generated new species is summarized in Figure 3l the generation rate of sulfoxides seems to exceed that of ketones in the early stages, and gradually slows down and lags behind. The generation rate of shorter chains shows a similar trend as sulfoxides; the generation rate of ketones however continues at a higher rate than both sulfoxides and shorter-chain alkanes, which contribute more to the agglomeration of aged asphalt in the long run. Based on this QC-based atomistic study, ketones and sulfoxides are the major oxidation products formed in oxidative aging of asphalt, which is in agreement with results of existing laboratory experiments such as by Functional Group Analyses.11,12 Moreover, such oxidation products formed are consistently observed among field asphalts from different sources,29,30 and are in good agreement with the general oxidation chemistry of hydrocarbon and sulfur-containing molecules.34−36 Ketones are a group of organic compounds carrying a carbonyl group, CO, in which the carbon atoms are bonded to two adjacent carbon atoms. Ketones are weak acids due to the α-proton adjacent to the carbonyl group that are much more acidic compared to simple hydrocarbons, and can be removed by common bases such as HO− and RO−. Sulfoxides are another group of chemical compounds, containing a sulfinyl functional group, SO, attached to two carbon atoms. The generation of ketones and sulfoxides are thermodynamically and kinetically favorable in general asphalt under the typical open-to-air conditions. Although the bond between sulfur and oxygen atoms differs from conventional double bonds like that between carbon and oxygen in ketones, sulfoxides and ketones are both polar, containing an electronegative oxygen atom producing a dipole in interaction or

Figure 2. XPS spectra of modeled asphalt before and after oxidation: (a) survey scan; (b) high-resolution scan.

Figure 3. Experimental vs numerical results of the oxidation of bulk asphalt (2 h shown).

machine or operation errors related to XPS. In preparing samples for XPS analysis, the asphalt samples were vacuumed at the room temperature following the procedure established by Ruiz et al.37 The samples (in dry powders) were then pressed onto a piece of thin indium foil (0.1 mm thick) as the sample substrate. Graphite tape was not used as sample substrate for the carbon-based asphalt to avoid peaks from the graphite tape, which would otherwise add to the carbon peak, potentially skewing or overwhelming the data in XPS spectrum.

■

SIMULATION RESULTS AND DISCUSSIONS Per Table 8, which presents the thermodynamic aspect of asphalt oxidation, even under the same oxidation condition the different asphalt molecules oxidize at different rates and lead to 7964



Article

Schabron, J. F. Binder Characterization and Evaluation. Vol. 2: Chemistry; Report No. SHRP-A-368; SHRP, National Research Council: Washington, DC, 1993. (9) Petersen, J. C. Asphalt Aging: A Dual Oxidation Mechanism and Its Interrelationships with Asphalt Composition and Oxidative Age Hardening; Transportation Research Record 1638; TRB, National Research Council, Washington, DC, 1998; pp47−55. (10) Hveem, F. N.; Zube, E.; Skog, J. Progress Report on the ZacaWigmore Experimental Asphalt Test Project. Spec. Tech. Publ., Am. Soc. Test. Mater. 1959, 277, 1−45. (11) Corbett, L. W. Composition of Asphalt Based on Generic Fractionation Using Solvent Deasphalteneing, Elution-Adsorption Chromatography and Densiometric Characterization. Anal. Chem. 1969, 41, 576−579. (12) Petersen, J. C. Chemical Composition of Asphalt as Related to Asphalt Durability: State of the Art; Transportation Research Record 999; TRB, National Research Council: Washington, DC, 1984; pp 13−30. (13) Pan, T.; van Duin, A. C.T. Steel Surface Passivation at a Typical Ambient Condition: Atomistic Modeling and X-ray Diffraction/ Reflectivity Analyses. Electrocatalysis 2011, 2 (4), 307−316. (14) Pan, T.; van Duin, A. C.T. Passivation of Steel Surface: An Atomistic Modeling Approach Aided with X-ray Analyses. Mater. Lett. 2011, 65 (21−22), 3223−3226. (15) Pan, T. Quantum Chemistry-Based Study of Iron Oxidation at the Iron-Water Interface: An X-Ray Analysis Aided Study. Chem. Phys. Lett. 2011, 511 (4−6), 315−321. (16) van Duin, A. C. T.; Dasgupta, S.; Lorant, F.; Goddard, W. A. ReaxFF: A Reactive Force Field for Hydrocarbons. J. Phys. Chem. A 2011, 105. (17) Pan, T.; Xi, Y. Physicochemical Nature of Iron Oxidation in a Damp Atmospheric Condition. Acta Metall. Sin. 2011, 24 (6), 415− 422. (18) Pan, T.; Lu, Y. Quantum-Chemistry Based Studying of Rebar Passivation in Alkaline Concrete Environment. Int. J. Electrochem. Sci. 2011, 8. (19) Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and Molecules; Oxford University Press: New York; ISBN 0-19-504279-4; 1989. (20) Chenoweth, K.; van Duin, A. C. T.; Goddard, W. A. ReaxFF reactive force field for molecular dynamics simulations of hydrocarbon oxidation. J. Phys. Chem. A 2008, 112, 1040−1053. (21) Mortier, W. J.; Ghosh, S. K.; Shankar, S. ElectronegativityEqualization Method for the Calculation of Atomic Charges in Molecules. J. Am. Chem. Soc. 1986, 108, 4315. (22) Janssens, G. O. A.; Baekelandt, B. G.; Toufar, H.; Mortier, W. J.; Schoonheydt, R. A. Comparison of Cluster and Infinite Crystal Calculations on Zeolites with the Electronegativity Equalization Method (EEM). J. Phys. Chem. 1995, 99, 3251. (23) Njo, S. L.; Fan, J.; van de Graaf, B. Extending and Simplifying the Electronegativity Equalization Method. J. Mol. Catal. A 1998, 134, 79. (24) White, R. M.; Mitten, W. R.; Skog, J. B. Fractional Components of Asphalt: Compatibility and Interchangeability of Fractions Produced From Different Asphalts. Proc. Assoc. Asphalt Paving Technol. 1970, 39, 498−531. (25) Kowalewski, I.; Vandenbroucke, M.; Huc, A. Y.; Taylor, M. J.; Faulon, J. L. Preliminary Results on Molecular Modeling of Asphaltenes Using Structure Elucidation Programs in Conjunction With Molecular Simulation Programs. Energy Fuels 1996, 10, 97−107. (26) Zhang, L.; Greenfield, M. L. Relaxation Time, Diffusion, and Viscosity Analysis of Model Asphalt Systems Using Molecular Simulation. J. Chem. Phys. 2007, 127, 194502. (27) Artok, L.; Su, Y.; Hirose, Y.; Hosokawa, M.; Murata, S.; Nomura, M. Structure and Reactivity of Petroleum-Derived Asphaltene. Energy Fuels 1999, 13, 287−296. (28) Groenzin, H.; Mullins, O. C. Molecular Size and Structure of Asphaltenes from Various Sources. Energy Fuels 2000, 14, 677−684.

association with other dipoles or induced dipoles, thus both contributing to agglomeration of oxidized asphalt.

■

SUMMARY AND CONCLUSIONS Based on a QC-based reactive force field re-evaluated for the C−H−O−S system, the oxidation of asphalt is modeled and validated with XPS analyses. Important conclusions are made as follows. 1. Two distinct stages of asphalt aging can be identified based on the composition and structure of generated chemical species and their generation speeds. Asphalts initially exhibit a high chain-breaking trend and a high reactivity with oxygen, causing a rapid spurt in the formation of low-molecular-weight alkanes and sulfoxides. This spurt is followed by a slower rate of oxidation and hardening. 2. Different oxidation mechanisms appear operative during these two periods. During the initial spurt, sulfoxides are the major oxidation product that controls viscosity increase. Following the spurt, ketones become the major product. The ratio of ketones to sulfoxides formed therefore depends on sulfur content in asphalt. 3. The asphalt oxidation involves many stable radicals, and a number of intermediate chemicals that can only be observed at the atomic scale due to the high chemical instabilities of such intermediate products. Significant decomposition of saturates and evaporation of smallmolecular-weight hydrocarbon were observed in the simulation. Ketones and sulfoxides account for most of the oxidative reactions in the simulated asphalt. Therefore for the oxidative hardening of asphalt in typical service conditions, chain breaking of saturates, sulfoxidation, and ketonization at benzylic carbons are the major mechanisms.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: (202) 319-5165; fax: (202) 319-6677; e-mail: pan@cua. edu. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) McNichol, D. Paving the Way: Asphalt in America, 1st ed.; National Asphalt Pavement Association, 2005. (2) Usmani, A. M. Asphalt Science and Technology; Marcel Dekker, New York, 1997. (3) Wess, J.; Olsen, L. D.; Sweeney, M. H. Asphalt (Bitumen), Concise International Chemical Assessment Document 59; World Health Organization: Geneva, 2004. (4) Petersen, J. C.; Branthaver, J. F.; Robertson, R. E.; Harnsberger, P. M.; Duvall, J. J.; Ensley, E. K. Effects of Physicochemical Factors on Asphalt Oxidation Kinetics; Transportation Research Record 1391; TRB, National Research Council: Washington, DC, 1993; pp 1−10. (5) Davis, T. C.; Petersen, J. C. An Inverse GLC Study of Asphalts Used in the Zaca-Wigmore Experimental Test Road. Proc., Assoc. Asphalt Paving Technol. 1967, 36, 1−15. (6) Plancher, H.; Green, E. L.; Petersen, J. C. Reduction of Oxidative Hardening of Asphalts by Treatment with Hydrated Lime: A Mechanistic Study. Proc. Assoc. Asphalt Paving Technol. 1976, 45, 1−24. (7) Petersen, J. C. A Dual Sequential Mechanism for the Oxidation of Asphalts. Pet. Sci. Technol. 1998, 16 (9−10), 1023−1059. (8) Branthaver, J. F.; Petersen, J. C.; Robertson, R. E.; Duvall, J. J., Kim; S. S., Harnsberger, P. M.; Mill, T.; Ensley, E. K.; Barbour, F. A.; 7965



Article

(29) Petersen, J. C.; Plancher, H.; Miyake, G. Chemical Reactivity and Flow Properties of Asphalts Modified by Metal Complex-Induced Reaction with Atmospheric Oxygen. Proc. Assoc. Asphalt Paving Technol. 1983, 32, 32−60. (30) Griffen, R. L.; Simpson, W. C.; Miles, T. K. Influence of Composition of Paving Asphalts on Viscosity, Viscosity-Temperature Susceptibility, and Durability. J. Chem. Eng. Data 1959, 4, 349−354. (31) Storm, D. A.; Edwards, J. C.; DeCanio, S. J.; Sheu, E. Y. Molecular Representations of Ratawi and Alaska North Slope Asphaltenes Based on Liquid- And Solid-State NMR. Energy Fuels 1994, 8, 561−566. (32) Verlet, L. Computer ‘Experiments’ on Classical Fluids: I. Thermodynamical Properties of Lennard-Jones Molecules. Phys. Rev. 1967, 159 (1), 98−103. (33) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W.; DiNola, A.; Haak, J. R. Molecular Dynamics with Coupling to an External Bath. J. Chem. Phys. 1984, 81. (34) Gelius, U.; Heden, R. F.; Hedman, J.; Lindberg, B. J.; Manne, R.; Nordberg, R.; Siegbahn, K. Molecular Spectroscopy by Means of ESCA III: Carbon Compounds. Phys. Scr. 1970, 2 (1−2), 70−80. (35) Kelemen, S. R.; George, G. N.; Gorbaty, M. L. Direct Determination and Quantification of Sulfur Forms in Heavy Petroleum and Coals: 1. The X-ray Photoelectron Spectroscopy (XPS) Approach. Fuel 1990, 69, 939−944. (36) Lindberg, B. J.; Hamrin, K.; Johansson, G.; Gellius, U.; Fahlman, A.; Nordllng, C.; Slegbahn, K. Molecular Spectroscopy by Means of ESCA II: Sulfur Compounds, Correlation of Electron Binding Energy with Structure. Phys. Scr. 1970, 1 (5−6), 288−298. (37) Ruiz, J. M.; Carden, B. M.; Lena, L. J.; Vincent, E. J.; Escaller, J. C. Determination of Sulfur in Asphalts by Selective Oxidation and Photometric Spectroscopy for Chemical Analysis. Anal. Chem. 1980, 54, 688−691. (38) Koopmans, T. Ü ber die Zuordnung von Wellenfunktionen und Eigenwerten zu den Einzelnen Elektronen Eines Atoms. Physica 1934, 1 (1−6), 104−113.

7966


Quantum-Chemistry Study of Asphalt Oxidative Aging: An XPS-Aided

Recommend Documents