Evaluation of molecular mechanics calculated binding energies for isolated and monolayer organic molecules on graphite

https://doi.org/10.1016/j.jcis.2007.05.083Get rights and content

Abstract

The calculated molecule–surface binding energy, Ecal, for physical adsorption was determined using molecular mechanics MM2 parameters for a model graphite surface and various organic molecules. The results for Ecal were compared to published experimental binding energy values, E, from gas chromatography (GC) or thermal desorption (TD). The binding energies from GC were for isolated molecules in the Henry's law region of adsorption, and the binding energies from TD were for molecules in monolayer coverage on a highly oriented pyrolytic graphite (HOPG). A simple desorption model was used to allow the calculation of monolayer coverage to include both molecule–surface and molecule–molecule interactions and then the results were compared to experimental values. For the 14 TD organic adsorbates (polyaromatic hydrocarbons, alcohols, benzene, substituted benzenes, methane, chloroalkanes, N,N-dimethylformamide, and C60 Buckyball), the experimental versus calculated binding energies were E=1.1193Ecal and r2=0.967. The GC E values were also well correlated by calculated Ecal values for a set of 11 benzene and methyl substituted benzenes and for another set of 10 alkanes and haloalkanes. The TD Ecal mechanics computation provides a useful comparison to the one for GC data since adsorbate–adsorbate interactions as well as adsorbate–surface must be considered.

Graphical abstract

Molecule–graphite binding energies calculated from molecular mechanics, based on a monolayer model, EcalML, predict experiment binding energies obtained from thermal desorption, Eavg.

  1. Download : Download full-size image

Introduction

Molecular mechanics force fields [1], [2], [3], based on classical mechanics and optimized parameters, have been used to determine structural and energetic properties of isolated molecules and can be used to examine molecule–molecule and molecule–surface interactions. More sophisticated methods based on density functional theory (DFT) [4] and quantum mechanics/molecular mechanics (QM/MM) have been used to calculate gas–solid interactions; however, standard molecular mechanics parameters have been found to be quite useful in calculating gas–solid interactions or binding energies on a variety of carbon surfaces [5], [6]. Dispersion or van der Waals forces dominate the adsorption of neutral molecules on a carbon surface, and in prior work [5], [6] we have shown that existing MM2 parameters could be used to calculate gas–solid binding energies for porous and rough carbon surfaces. In this work we extend our calculations to smooth graphitic surfaces with either isolated molecule adsorption or complete monolayer coverage.

The interactions of molecules with graphitic surfaces have been investigated in various studies [7], [8], [9], [10], [11]. Using molecular mechanics, C34H70 was found to have a binding energy of 61.43 kcal/mol on highly oriented pyrolitic graphite (HOPG) [7]. Nitric oxide (NO) adsorption on a single-layer graphene gave adsorption energies of 90.1 kcal/mol from the Hartree–Fock method and 44.1 kcal/mol using density functional theory [11]. Thermal desorption from the basal plane of graphite gave desorption energies of 7.8 and 14 kcal/mol for (CF3CF2)2O and CF3CH2OH, respectively [8]. The interaction potential of N2 on a flat graphite surface was found to be 1150 K or 2.29 kcal/mol [9]. Thermal desorption and molecular simulations were used to examine the adsorption and desorption of acetone on highly oriented pyrolitic graphite [10]. Interactions of molecules with graphitic surfaces are of interest for theoretical and applied reasons including applications such as hydrogen storage [12], [13]. Interactions among adsorbed molecules and adsorbate–surface interactions are important in self-assembly systems, environmental monitoring, air purification, chemical warfare agent protection, and the use of a graphene layer or carbon nanotube for an electronic-nose chemical sensor [14], [15], [16].

In prior work [5] molecular mechanics calculations were used to determine binding energies (energies of adsorption) for a series of 9 alkanes and haloalkanes using a nanoporous parallel plate model for the adsorbent Carbosieve S-III. Carbosieve S-III (Supelco), is a carbon powder with fairly uniform, predominately 0.55 nm, slit width pores and a N2 BET surface area of 995 m2/g. MM2 and MM3 molecular mechanics were used to determine Ecal, the calculated binding energy, for each of the nine molecules on flat and nanoporous model surfaces. The nanoporous model was based on two sets of three graphene layers (127 rings per layer) separated by an internuclei separation of D (where d is the pore opening with D=d+0.34 nm). The MM2 parameters gave better results than MM3 and the porous model was superior to a flat model [5].

The pore model was further refined by moving the two sets of graphene layers relative to each other. The best correlations of experimental, E, and computed, Ecal, binding energies were for slit width pore separations in the range of d=0.480.55 nm [5]. The best result judging by r2 values for linear correlations of E versus Ecal were obtained with a slit width pore opening of d=0.50 nm or D=0.84 nm. These results compared well with a differential pore volume plot that showed the majority of peaks below 0.60 nm with a sharp peak maximum at 0.55 nm; and also with an observed molecular sieving estimate of between 0.51 and 0.56 nm [5].

In another prior study [6], a rough surface model was created to calculate binding energies for Carbopack B (Supelco) that has a N2 BET area of 100 m2/g. Three parallel graphene layers were used with 127 rings per layer. Two nanostructures were placed on this surface, and they could be brought closer together or moved apart. As the nanostructures were brought together, it represented an increasing surface roughness. Although the actual surface may not physically be like the model surface, the model led to calculated energies [6] quite similar to the experimental binding energies [17]. MM2 parameters were used and the binding energy regression for the best rough surface model gave E=1.018Ecal and r2=0.964. These results were for 16 molecules (alkanes, haloalkanes, and ethers) on the Carbopack B carbon powder. The experimental E values were previously determined by the temperature variation of second gas–solid virial coefficients determined from gas chromatography [17].

Having developed porous and rough surface models, the objective of the current work is to examine how well this computational approach works for smooth graphitic surfaces. Our prior work examined the binding energy for isolated molecules in the Henry's law region of surface coverage. We now wish to develop a new model for the binding energy expressed in monolayer coverage on a surface. We wish to compare these models designed to calculate binding energies for isolated molecule and monolayer coverage on a graphitic surface. Experimental data from gas–solid chromatography (GC) were previously used to find graphite binding energies for a set of 11 benzene and methyl substituted benzenes [18], [19] and for a set of 10 alkanes and haloalkanes [20]. Excellent experimental data from thermal desorption (TD) has been reported and used to determine monolayer coverage binding energies for polyaromatic hydrocarbons (PAH) [21] and for a diverse set of 14 organic molecules including: polyaromatic hydrocarbons, alcohols, benzene, substituted benzenes, methane, chloroalkanes, N,N-dimethylformamide, and C60 Buckyball [22].

The TD data provides an interesting comparison to the GC data since it is necessary to include adsorbate–adsorbate interactions as well as adsorbate–surface interactions. It is also possible to explore the role of hydrogen bonding among surface molecules. With a surface model that includes adsorbate–adsorbate interactions (presented in Section 3) and our MM2 mechanics based calculations, an improved correlation between E and Ecal relative to that previously reported for the TD data was obtained [22].

Section snippets

Binding energy values

Three available sets of experimental binding energy values, E, are used in the present work and are found in Table 1, Table 2, Table 3. The studies reflected in Table 1, Table 2 used gas chromatography to obtain E for isolated molecules on a graphite surface [18], [19], [20] and the work for Table 3 used thermal desorption to obtain E for monolayer coverage of molecules on a graphite surface [21], [22]. The methods used are discussed briefly below.

Henry's law constants based on retention

Theory

Two cases of molecule–surface interactions are considered in this work. The first case is for isolated molecules such as would be observed by gas–solid chromatography where the adsorbate molecules are in the Henry's law region of low coverage. The second case is for molecules adsorbed on a surface at monolayer coverage such as would be observed by carefully controlled thermal desorption experiments. Both of these examples represent physical adsorption where no adsorbate–adsorbent chemical bonds

Isolated benzene and methyl substituted benzenes

A surface model was created by representing graphene (a single layer of graphite) with 127 interconnected benzene rings. One, two, or three parallel layers of graphene were used to represent different model graphite surfaces. The graphene layers were placed in a packing arrangement with carbon atoms in alternate layers directly underneath each other as is found in the Bernal graphite structure. The computational work was done using CAChe software (version 6.1.12 from Fujitsu Computer Systems

Discussion

In prior work [5], [6] the MM2 parameters were found to provide calculated binding energy values that correlated well with experimental ones. In the current work, for the Ecal values found in Table 1, Table 2 and plotted in Fig. 1, Fig. 2, the r2 values for E versus Ecal correlations were 0.98 and 0.93, respectively. These results indicate that the flat model surface consisting of three parallel graphene layers provides a reasonable representation of the actual graphite surfaces and

Conclusion

Van der Waals and ππ interactions continue to have interesting new applications such as the recent synthesis by Sygula et al. of a concave hydrocarbon molecule designed to hold a C60 Buckyball by wrapping, like molecular tweezers, around a portion the fullerene [39]. Relatively simple calculations continue to be of interest to correlate binding energies based on weaker dispersive interactions. With no adjustment, the standard MM2 parameters for atom-carbon interactions prove to be surprisingly

Acknowledgements

We acknowledge the support provided by the Grote Chemistry Fund and the Wheeler Odor Research Center at the University of Tennessee at Chattanooga.

References (40)

  • D.L. Irving et al.

    Chem. Phys. Lett.

    (2004)
  • T.R. Rybolt et al.

    J. Colloid Interface Sci.

    (2006)
  • T.R. Rybolt et al.

    J. Colloid Interface Sci.

    (2006)
  • C.K.W. Adu et al.

    Chem. Phys. Lett.

    (2001)
  • T.R. Rybolt et al.

    J. Colloid Interface Sci.

    (1995)
  • T.R. Rybolt et al.

    J. Colloid Interface Sci.

    (1999)
  • H. Ulbricht et al.

    Carbon

    (2006)
  • H. Rydberg et al.

    Surf. Sci.

    (2003)
  • K.S. Kim et al.

    Chem. Rev.

    (2000)
  • N.L. Allinger

    J. Am. Chem. Soc.

    (1977)
  • N.L. Allinger et al.

    J. Am. Chem. Soc.

    (1989)
  • J. Lii et al.

    J. Am. Chem. Soc.

    (1989)
  • Y. Shuxia et al.

    J. Phys. Chem. B

    (2002)
  • N. Shukla et al.

    Langmuir

    (2001)
  • K. Murata et al.

    Nano Lett.

    (2001)
  • S. Kwon et al.

    Langmuir

    (2002)
  • A. Montoya et al.

    J. Phys. Chem. A

    (2000)
  • X. Sha et al.

    J. Am. Chem. Soc.

    (2004)
  • A.C. Dillon et al.

    Appl. Phys. A

    (2001)
  • J. Kong et al.

    Science

    (2000)
  • Cited by (0)

    View full text