Ab initio studies of carbon dioxide affinity to carbon compounds and minerals
/ locate / procedia European Geosciences Union General Assembly 2017, EGUDivision Energy, Resources & Environment, ERE
Ab initio studies of carbon dioxide a ffi nity to carbon compounds andminerals Mateusz Wlazło a, ∗ , Alexandra Siklitskaya a,b , Jacek A. Majewski a a Faculty of Physics, University of Warsaw, Pasteura 5 02-093 Warsaw, Poland b Institute of Physical Chemistry, Polish Academy of Sciences, Kasprzaka 44 /
52, Warsaw, Poland
Abstract
We have performed quantum chemical computational studies to determine carbon dioxide a ffi nity to carbon compounds and min-erals, which could be present in shales. These studies shed light on the microscopic mechanisms of the possible carbon oxidesequestration processes. Our studies reveal that the carbon oxide can be adsorbed to various forms of carbon structures and alsominerals such as periclase or illite. We find out that the strongest a ffi nity of carbon oxide towards carbon structures occurs whenthe carbon structures exhibit sp bonds.c (cid:13) Keywords: carbon sequestration, graphene, periclase, illite, density functional theory, ab initio molecular dynamics
1. Introduction
Carbon dioxide is one of the primary greenhouse gases implicated in global warming process. Alternative energysources that make no contribution to CO emission are still in the development phase and not likely to replace currentcarbon-based energy sources during the few next decades. Also carbon capture methods are intensively studied atpresent, however, it is di ffi cult to foreseen their e ff ective employment in industry and everyday life. Carbon dioxidesequestration process seems to be an alternative, and therefore, is of critical importance to maintain or even reducethe CO level in the atmosphere. Additionally, CO sequestration process can be connected to the Enhanced Oilrecovery in shale rocks. Therefore, accurate description of physicochemical mechanisms governing the CO and CH adsorption to various minerals and organic forms of carbon matter is an important step to understand mechanismsand to improve technologies of sequestration. The particular role of carbon in these processes is evident, since total ∗ Corresponding author
E-mail address: [email protected] c (cid:13) a r X i v : . [ c ond - m a t . m t r l - s c i ] J un Wlazło et al. / Energy Procedia 00 (2017) 000–000 -0.3 0 0.3 0.6 0.9 2.5 3 3.5 4 4.5 5CO height over graphene [Å]BLYPBLYP+vdW Fig. 1. Adsorption energy vs. distance of CO over ideal graphene obtained in calculations with (BLYP + vdW) and without (BLYP) the DFT-D2correction. The presence of an energy minimum after including the correction indicates that CO can be physisorbed on graphene and that it canonly occur via van der Waals interaction. Reprinted from [1]. organic carbon (TOC) is an important parameter that characterizes the potency of a shale formation in hydrocarbonproduction, and a high TOC value ensures that a reservoir is useful in gas production.In the present study, we employ quantum chemical computational techniques to study carbon dioxide adsorptionprocesses on the atomistic level in organic and inorganic shale rock constituents and to gain understanding of thesee ff ects on the micro (atomic) scale. We have selected a few materials that mirror the heterogeneity of shales. Inparticular, we focus on the organic components, which are modeled by pure carbon allotropes such as graphene andspiral carbon nanoparticles (spiroids), however, we provide also some findings for magnesium oxide mineral and illiterocks. We investigate the structural, energetic, and the thermodynamical aspects of CO adsorption and desorptionemploying ab initio molecular dynamics (AIMD) within the canonical ensemble (NVT) and are able to determine thepossible CO capture reactions.The present paper is organized as follows. In Section 2 we give a short and comprehensive description of themethodology employed in this study. The results of the study describing the mechanisms of the CO adsorption to thecarbon structures and minerals are presented in Section 3. Finally, the paper is concluded in Section 4.
2. Methods
The essential component of ab initio methods for predicting properties of a collection of atoms is a quantummechanical scheme to calculate the ground state energy of the system studied. In our studies of the physicochemistry ofthe carbon dioxide adsorption to surfaces of various materials, we employ density functional theory (DFT) calculationsthat constitute nowadays the method of choice for the materials science. The Hohenberg-Kohn-Sham [2, 3] realizationof DFT allows us to find not only the total energy of the system but also every observable of the system as a functionalof the ground state electronic density. The total energy functional, often referred to as the Kohn-Sham functional( E KS ), is written in terms of the kinetic energy functional ( T s ), the external potential acting on the electronic system( v ext ), the electron-electron interaction (Coulomb or Hartree interaction, E H ) and exchange-correlation interaction( E XC ): E KS [ n ] = T s [ n ] + (cid:90) d r v ext ( r ) n ( r ) + U H [ n ] + E XC [ n ] (1)The exact form of the XC term is unfortunately unknown and it has to be approximated [4]. The choice of a XCfunctional is a crucial parameter of a DFT calculation. Many successful approaches have been based on the localdensity approximation (LDA), which is derived from the homogeneous electron gas model. However, the so-calledgradient-corrected functionals, or generalized gradient approximations (GGA), are superior to LDA. Therefore, weuse the GGA form of the exchange correlation functional, as implemented in the so-called BLYP (Becke-Lee-Yang-Parr) form [5, 6]. Unfortunately, the GGA approach does not include dispersion interaction, which is a necessity if lazło et al. / Energy Procedia 00 (2017) 000–000 one wishes to account for van der Waals interaction. For this reason, we add another term to the KS functional. It is asemi-empirical, pairwise D2 correction [7], which has the following form: E D = − s (cid:88) A < B C AB R AB f damp ( R AB , R AB ) , (2)where the sum runs over pairs of atoms A and B, R AB is the interatomic distance, C AB are parameters obtained fromtime-dependent DFT [7], R AB is the sum of van der Waals radii of atoms A and B, s is a fitted parameter. The fullfunctional therefore reads: E DFT − D = E KS [ n ] + E D . (3)Including the D2 correction is necessary to obtain correct physisorption energy profiles, as it is demonstrated forthe case of adsorption of CO to the graphene surface in Figure 1. As it is seen, the adsorption energy as a functionof the distance of the CO molecule to the graphene surface has no minimum when the BLYP functional without vander Waals corrections is employed for exchange correlation functional, indicating that the CO does not adsorb tothe surface. Only inclusion of van der Waals forces to the energy functional guarantees physisorption of CO to thegraphene layer at the distance corresponding to the minimum of the BLYP + vdW total energy curve.Further in the text we will refer to adsorption or interaction energy. These quantities are calculated using the totalenergy functionals. For instance, we calculate the adsorption energy of a species on a surface by calculating the totalenergy of a system containing both the species and the surface. Then we subtract the sum of energies of systemscontaining only the surface and only the species, i.e.: E ads = E sur f . + species − ( E sur f . + E species ) . (4)Adsorption is energetically favored whenever E ads < ab initio molecular dynamics (AIMD) simulations, interatomic forces are calculated according to this relationand the motion of nuclei is propagated with Newton’s equations of motion. There are several di ff erent schemes to treatthe electronic system in such a simulation. Here we use the Car-Parrinello approach [9]. This scheme is extended bycoupling to external Nose-Hoover thermostats [10, 11]. This allows for simulations in the canonical (NVT) ensemble.
3. Results adsorption mechanisms There are several chemical mechanisms of adsorption of and CO on surfaces, including direct chemisorption,dissociative chemisorption and physisorption. Depending on many factors, such as electronic structure, molecularand surface charge distribution or geometry, di ff erent mechanisms will dominate the adsorption processes. For exam-ple, the methane molecule CH has a closed shell electronic configuration. All of its valence electrons form bondswithin the molecule. Therefore, direct chemisorption of such species can be ruled out immediately. An intermediatedissociation step has to occur first, i.e.: CH ( gas ) → CH ( ads ) + H ( ads ) (5)Even if the dissociation reaction is endothermic, there is an energy barrier for it to occur. The magnitude of thebarrier depends on the adsorbent. In many cases physisorption of CO is favored. It is much weaker than chemisorptionbut it plays a significant role in gas reactions in interporous space. Wlazło et al. / Energy Procedia 00 (2017) 000–000
Fig. 2. Optimized geometry of CO chemisorbed on a Stone-Wales defect site in graphene. Another example is the direct chemisorption of CO . Under normal conditions, the molecule is linear with two dou-ble bonds between C and O atoms. Chemisorption requires breaking one of the C-O bonds which greatly distorts theshape of the molecule. Such a distortion is associated with a high energy barrier. Typical reservoir conditions, however,are well above the supercritical point for CO . First principles calculations show slight non-linearity of supercritical CO [12]. This means that the activation energy for direct CO chemisorption is decreased in the supercritical fluidphase. ff erent classes of materials In this section we will describe adsorption mechanisms on a variety of carbon allotropes. As the first one we con-sider graphene, which is an atomically-thin, planar structure with hexagonal arrangement of atoms in a honeycomb-like lattice. Layers of graphene stacked on top of each other form graphite. It can also serve as a model of dehydro-genated kerogen. We will also consider two spherical forms of carbon – multishell fullerene and a carbon spiroid.
Graphene lattice is made up entirely of sp carbon atoms. Atoms are bonded by in-plane σ bonds and delocalized π bonds formed by type p orbitals that are perpendicular to the surface. When another molecule or radical approachesthe surface, it is possible for the π bond to be disturbed. If the species has a free electron of its own, the dangling p z orbital in graphene can overlap with the unoccupied orbital of the species. Then another σ bond is made and the sp carbon undergoes rehybridization to sp . Because sp carbon geometry is tetrahedral (as in the CH molecule) ratherthan planar (as in graphene), the atom moves out of plane by about 0.35 Å.Graphene lattice can feature a plethora of structural defects. The simplest ones are so-called point defects that onlychange a single atom, e.g. a lattice vacancy or single atom substitution by another atom type. A di ff erent type of defectis the Stone-Wales (SW) defect which is created by rotating one C-C bond by 90 degrees [13]. In the neighborhoodof such a defect the hexagonal lattice is disturbed an instead of hexagonal rings we obtain two heptagons and twopentagons. This defect does not change the σ bonds – every carbon atom still has three neighbors. Local stress thatappears in this structure is released by buckling around the defect [14]. This in turn disturbs the π -bond system andmakes the site more reactive towards chemisorption [1] Table 1. Dominant chemisorption mechanisms and CO adsorption energies for analyzed structures. See Figures 2 to 4 for optimized geometries. Species SW-defectedgraphene C spiroid Calcite( CaCO ) Periclase(MgO) Illite( Al KS i O )Dominant chemisorption mechanism Direct Direct Dissociative Direct DirectAdsorption energy [kJ mol − ] -143 -15 -98 -90 -92 lazło et al. / Energy Procedia 00 (2017) 000–000 CO chemisorbed on the edge of a C spiroid.Fig. 4. Optimized geometry of CO chemisorbed on cubic MgO (left) and illite oxygen site (right). The chemistry of SW-defected lattice is slightly di ff erent. This has an important implication when it comes to CO adsorption. On ideal graphene lattice, there is no stable chemisorption of CO and there is only weak physisorption.From geometry optimization of CO on SW defect we find a stable configuration in which CO is chemisorbed bytwo covalent C-C and C-O bonds (see Figure 2). Two other carbon allotropes have been analyzed: we have compared the multi-shell fullerene C @ C ( C fullerene enclosed in C fullerene) with the carbon onion containing also 300 carbon atoms but consisting of twoclosed fullerene shells. Actually, both forms of C , spiroid and spheroid ones, are observed in interstellar medium[15, 16].Geometry optimization of the C spiroid shows that its end features an opening around 9.3 Å wide and 3.6 Åhigh, as illustrated in Figure 3. The opening is terminated with carbon atoms with only two saturated bonds. Ourcalculations show, that the CO can be just chemisorbed on the edge of spiroid’s orifice. There is also the possibilityof physisorption of CO inside the spiroid, which constitutes the second e ff ective mechanism of CO capture byspiroid C , which apparently is facilitated by the large opening in this structure. We would like to mention that thespheroid forms of C are rather ine ffi cient in CO capture. We have performed AIMD simulations at temperatures up to 1200 K to determine which processes occur onwhich mineral surface. We found calcite to be inert to CO adsorption up to very high temperatures. Dissociativechemisorption is preferred [17] according to the equation CO ( gas ) → CO ( ads ) + O ( gas ). However, due to largeenergy of C-O bond breaking, it starts to occur above 1200 K. For two other minerals – periclase and illite – direct Wlazło et al. / Energy Procedia 00 (2017) 000–000 chemisorption is possible at low temperatures. In both cases bonding occurs via the carbon atom in CO and oxygenatom at the surface of the mineral. The arrangements of atoms around the adsorbed CO molecule on the MgO andillite are depicted in Figure 4.
4. Conclusions
In this work, we have performed the DFT and AIMD calculations that shed light onto physicochemical adsorptionmechanisms of CO on several forms of organic and mineral materials. The highest a ffi nity (adsorption energy) of CO was found on the defected graphene surface. This is due to the covalent bonds formed with C and O atoms in CO . Mineral surfaces exhibited similar adsorption energies. Among them, CO had the strongest a ffi nity to the calcitesurface. However, this reaction only occurs at extremely high temperatures starting at 1200 K. The weakest a ffi nity isexhibited at the C spiroid opening edge. The observed considerable scatter of the magnitude of the CO a ffi nity tothe various carbon structures clearly demonstrates that the surface chemistry of carbon is very much dependent on thestructural details and that the heterogeneity of organic matter can play an important role in adsorption properties ofshale samples. References [1] M. Wlazło, J. A. Majewski, First principles study of gas adsorption dynamics on pristine and defected graphene, Acta Physica Polonica A 129(2016) A142–A145. arXiv:http://dx.doi.org/10.12693/APhysPolA.129.A-142 , doi:10.12693/APhysPolA.129.A-142 .[2] P. Hohenberg, W. Kohn, Inhomogeneous electron gas, Phys. Rev. 136 (1964) B864–B871. doi:10.1103/PhysRev.136.B864 .URL http://link.aps.org/doi/10.1103/PhysRev.136.B864 [3] W. Kohn, L. J. Sham, Self-consistent equations including exchange and correlation e ff ects, Phys. Rev. 140 (1965) A1133–A1138. doi:10.1103/PhysRev.140.A1133 .URL http://link.aps.org/doi/10.1103/PhysRev.140.A1133 [4] R. Martin, Electronic Structure: Basic Theory and Practical Methods, Cambridge University Press, 2004.[5] A. D. Becke, Density-functional exchange-energy approximation with correct asymptotic behavior, Phys. Rev. A 38 (1988) 3098–3100. doi:10.1103/PhysRevA.38.3098 .URL http://link.aps.org/doi/10.1103/PhysRevA.38.3098 [6] C. Lee, W. Yang, R. G. Parr, Development of the colle-salvetti correlation-energy formula into a functional of the electron density, Phys. Rev.B 37 (1988) 785–789. doi:10.1103/PhysRevB.37.785 .URL http://link.aps.org/doi/10.1103/PhysRevB.37.785 [7] S. Grimme, Semiempirical gga-type density functional constructed with a long-range dispersion correction, Journal of Computational Chem-istry 27 (15) (2006) 1787–1799. doi:10.1002/jcc.20495 .URL http://dx.doi.org/10.1002/jcc.20495 [8] R. P. Feynman, Forces in molecules, Phys. Rev. 56 (1939) 340–343. doi:10.1103/PhysRev.56.340 .URL https://link.aps.org/doi/10.1103/PhysRev.56.340 [9] R. Car, M. Parrinello, Unified approach for molecular dynamics and density-functional theory, Phys. Rev. Lett. 55 (1985) 2471–2474. doi:10.1103/PhysRevLett.55.2471 .URL http://link.aps.org/doi/10.1103/PhysRevLett.55.2471 [10] S. Nos´e, A molecular dynamics method for simulations in the canonical ensemble, Molecular Physics 52 (1984) 255–268. doi:10.1080/00268978400101201 .[11] W. G. Hoover, Canonical dynamics: Equilibrium phase-space distributions, Phys. Rev. A 31 (1985) 1695–1697. doi:10.1103/PhysRevA.31.1695 .URL http://link.aps.org/doi/10.1103/PhysRevA.31.1695 [12] M. Saharay, S. Balasubramanian, Ab initio molecular-dynamics study of supercritical carbon dioxide, The Journal of Chemical Physics 120 (20)(2004) 9694–9702. arXiv:http://dx.doi.org/10.1063/1.1701838 , doi:10.1063/1.1701838 .URL http://dx.doi.org/10.1063/1.1701838 [13] A. Stone, D. Wales, Theoretical studies of icosahedral { C60 } and some related species, Chemical Physics Letters 128 (5–6) (1986) 501 – 503. arXiv:http://dx.doi.org/10.1016/0009-2614(86)80661-3 , doi:10.1016/0009-2614(86)80661-3 .URL [14] J. Ma, D. Alf`e, A. Michaelides, E. Wang, Stone-wales defects in graphene and other planar sp -bonded materials, Phys. Rev. B 80 (2009)033407. arXiv:http://dx.doi.org/10.1103/PhysRevB.80.033407 , doi:10.1103/PhysRevB.80.033407 .URL http://link.aps.org/doi/10.1103/PhysRevB.80.033407 [15] L. Henrard, P. Lambin, A. A. Lucas, Carbon Onions as Possible Carriers of the 2175 A Interstellar Absorption Bump, The AstrophysicalJournal 487 (2) (1997) 719–727. doi:10.1086/304645 .URL http://stacks.iop.org/0004-637X/487/i=2/a=719 lazło et al. / Energy Procedia 00 (2017) 000–000 doi:10.1016/j.nimb.2016.09.008 .URL [17] A. Siklitskaya, J. Majewski, Ab Initio Study of CH A ffi nity to the (001) MgO Surface 129. doi:10.12693/APhysPolA.129.A-145 .URL.URL