Interplay between Ethanol Adsorption to High ... - ACS Publications

Subscriber access provided by NEW YORK MED COLL

Article

Interplay between Ethanol Adsorption to High-Energy Sites and Clustering on Graphene and Graphite Alters the Measured Isosteric Adsorption Enthalpies Frantisek Karlicky, Eva Otyepkova, Pavel Banáš, Petr Lazar, Mikulas Kocman, and Michal Otyepka J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.5b06755 • Publication Date (Web): 17 Aug 2015 Downloaded from http://pubs.acs.org on August 25, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Interplay between Ethanol Adsorption to HighEnergy Sites and Clustering on Graphene and Graphite Alters the Measured Isosteric Adsorption Enthalpies František Karlický, Eva Otyepková, Pavel Banáš, Petr Lazar, Mikuláš Kocman, and Michal Otyepka* Department of Physical Chemistry, Regional Centre of Advance Technologies and Materials, Faculty of Science, Palacký University Olomouc, tř. 17. Listopadu 12, 771 46, Olomouc Czech Republic ABSTRACT We present a combined experimental and theoretical study aimed at understanding the behavior of polar probe ethanol on graphene and graphite hydrophobic surfaces. We measured isosteric adsorption enthalpies and entropies by inverse gas chromatography for coverages ranging from 0.1 to 20%. The adsorption enthalpies were found to vary with surface coverage and differed considerably between the materials at low coverage. However, they approached the same adsorption enthalpy value of –12.0 ± 0.4 kcal/mol for T centered at 303-393 K and coverages above 5%. We explained the observed behavior using molecular dynamics simulations by employing empirical force-field and density functional theory calculations on two graphene models: circumcoronene and infinite graphene. The simulations showed that various hydrogen-bonded ethanol clusters formed

1 Environment ACS Paragon Plus


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

spontaneously from isolated ethanol molecules on graphene and provided a distribution of cluster sizes. Non-local density functional theory was used to calculate adsorption enthalpies for various sizes of ethanol clusters. A theoretical adsorption enthalpy of –11.6 kcal/mol at 340 K was obtained from the weighted average of the cluster size distribution, while the adsorption enthalpy of single ethanol molecule to graphene was –6.3 kcal/mol at 323 K.


Page 2 of 26

Page 3 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. INTRODUCTION Adsorption of polar molecules to nonpolar surfaces is of broad scientific interest because its complexity can generate various nontrivial surface phenomena, such as hydrophobic hydration,1,

2

formation of nanobubbles,3 molecular self-assembly4 and

clustering.5 Polar molecules are attached to nonpolar surfaces mostly by van der Waals forces, with London dispersion forces being the predominant contribution.6,

7

Intermolecular

interactions between polar molecules, especially those capable of forming hydrogen bonds, are usually of comparable magnitude to the latter interactions. Thus, a delicate energy balance exists between the surface adsorption of molecules and mutual interactions among the adsorbed molecules. In addition, hydrophobic surfaces are not perfect and uniform but contain structural (defects, irregularities, cavities, pores, etc.) and chemical (impurities, contaminants, polymorphs, degree of crystallinity, etc.) features that create surface heterogeneities. All these phenomena complicate the interpretation of adsorption/desorption experiments providing averaged information. For example, besides interactions with the hydrophobic surface, measured adsorption enthalpies inherently include contributions stemming from interactions with high energy sites and mutual interactions among adsorbates.8 Undoubtedly, adsorption phenomena are important in many practical applications and technological processes. However, adsorption energies of single molecules are also important from a theoretical viewpoint because they can be utilized to benchmark and evolve theoretical methods, e.g., exchange-correlation functionals or dispersion corrections used in density functional theory (DFT).7, 9-11 In this work, we analyzed the adsorption of ethanol to two hydrophobic carbon-based materials, i.e., few-layer graphene and graphite nanopowders, as a function of the surface coverage. We measured the adsorption enthalpies by inverse gas chromatography under low coverages, ranging from 0.5 to 20% of a monolayer. The adsorption enthalpies on both 3 Environment ACS Paragon Plus


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

samples were found to be strongly coverage dependent and showed strikingly different behavior at low coverage, differing by about 7 kcal/mol (ΔHad of ~ –9 kcal/mol and ~ –16 kcal/mol for graphene and graphite, respectively). In contrast, they saturated to almost the same values between –11 and –12 kcal/mol at 10 - 20% coverage (1 kcal/mol is equivalent to 4.184 kJ/mol). The measured adsorption enthalpies at higher surface coverage agreed well with previous temperature programmed desorption (TPD) experiments.6 However, the experimentally observed enthalpy could not be attributed to the adsorption of single ethanol molecules to the graphene/graphite surface as the enthalpy of the latter process was estimated to be –6.2 kcal/mol by theoretical calculations.7 Our molecular dynamics (MD) simulations demonstrated a clear tendency of ethanol to form molecular clusters on the surface and provided a sufficiently robust statistical ensemble to evaluate the probability of the appearance of each cluster as a function of temperature. Combining these data with ab initio DFT calculations of the adsorption properties of ethanol clusters up to hexamers, we were able to estimate the saturated apparent adsorption enthalpy of ethanol to graphene at 10-20% surface coverage, which agreed well with the experimental value. We suggest that the different behavior of the isosteric adsorption enthalpies at low coverages of the materials originated from a delicate balance between ethanol clustering and adsorption of ethanol monomers to high energy sites. 2. METHODS 2.1 Materials. The samples used were graphene powder (Graphene Supermarket, Graphene Nanopowder AO-1) with a surface area of 915 m2/g, lateral size of ~1 μm and thickness of 3 nm and ultrafine natural nanographite (Graphene Supermarket, Nanostructured Graphite-250) with a surface area of 250 m2/g, lateral size of 100-500 nm and thickness of 10300 nm. Detailed characterization of both samples can be found in our previous article.12


Page 4 of 26

Page 5 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2.2 Experimental setup. Isosteric adsorption enthalpies (ΔHad) were measured using an SMS iGC-SEA instrument (Surface Measurement Systems, UK). Silanized gas chromatography columns were packed with either 18.8 mg of graphene or 26.2 mg of graphite powders. Prior to measurements, both samples were stored for 1 h under low pressure (50 ~mbar) and then conditioned for 1 h under a flow (10 sccm) of helium gas. Ethanol (Merck, LiChrosolv for liquid chromatography) vapor was used as the adsorbate. Primary chromatograms were recorded at temperatures from 303 to 393 K in steps of 10 K. For graphene, additional measurements at 368, 378, 388, 398 and 403 K were performed and in the case of 6% coverage, measurements were also made at 318, 328, 338, 348 and 358 K. Partial pressures and surface coverage values were calculated from the primary chromatograms using Cirrus Plus advanced version 1.2.1.2 (Surface Measurement Systems Ltd., UK). 2.3 Experimental data processing. To estimate the isosteric adsorption enthalpies (ΔHad) and entropies (ΔSad), we fitted the dependence of the partial pressure p of ethanol in the column and the target surface coverage on the column temperature. The target surface coverage was used instead of the directly measured actual coverage owing to inaccuracies in the estimation of the actual surface coverage in some measurements. This was particularly the case for adsorption on graphite at low temperatures as extensive tailing of the chromatographic peaks caused by significant adsorption to high energy sites hampered accurate peak area integration, leading to underestimation of the actual coverage. Therefore the target surface coverage, from which the total amount of injected ethanol was calculated, represented a more reliable estimation for the surface coverage. As in our recent study,12 the low surface coverages used allowed application of the Langmuir adsorption model, which leads to the following relation between the surface coverage, partial pressure and temperature (Eq. 1):



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 − ø

= =

Page 6 of 26

∆ ∆

Eq. 1

where , p and T stand for the surface coverage, ethanol partial pressure and column temperature, respectively, ΔHad and ΔSad correspond to the enthalpy and entropy of adsorption, respectively, K is the respective equilibrium constant and pø is the standard pressure of 760 Torr. Note that we used Eq. 1 directly for fitting instead of its usual logarithmic form combining Langmuir and van’t Hoff isotherms (see, e.g., Ref.

12

for more

details) because the experimental uncertainty in the pressure estimation was more equally (and thus statistically correctly) treated for all measured points in the adopted approach. ΔHad and ΔSad were fitted as the parameters of Eq. 1 using the least square fitting method. 2.4 Theoretical calculations. Force field simulations were performed using the all atom optimal potentials for liquid simulation (OPLS-AA) by Jorgensen et al.13 The structures and topologies of ethanol and acetone were taken from the Gromacs molecule & liquid database.14,

15

The graphene model consisted of 3,936 atoms, which were kept in fixed

positions on a planar hexagonal lattice with a bond distance of 1.4 Å. The model was placed into a simulation box with size 100×100×130 Å and periodic boundary conditions were applied in all three dimensions. The intermolecular interaction of graphene with ethanol molecules was simulated using the Lennard–Jones (LJ) potential by Cheng & Steele16 with a cutoff radius of 10.0 Å. To simulate the creation of clusters on graphene, 30 molecules of ethanol or acetone were placed on its surface, corresponding to ~10% coverage. The Newtonian equations of motion were integrated using 2 fs time step. All MD simulations were performed with a constant volume and temperatures of 300 K, 320 K, 340 K and 360 K. Each run was equilibrated for 2 ns. The distribution of ethanol molecules into clusters was analyzed every 20 ps over 30 ns of simulation time using agglomerative hierarchical cluster analysis with single linkage criterion. The cutoff distance between two oxygen atoms used to define a 6 Environment ACS Paragon Plus

Page 7 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


cluster was set to 3.6 Å for ethanol and 5.5 Å for acetone. It should be noted, that the distributions seem to be convergent from 30-ns-long MD simulations, because an extension of MD simulation to 150 ns, provided the distribution deviating by less than 2.1% (mean unassigned deviation) at 300 K. DFT calculations on graphene were performed using the projector-augmented wave (PAW) method in the Vienna Ab initio Simulation Package (VASP) suite.17, 18 The optB86bvdW DFT functional was employed to include a contribution from non-local correlation effects.19 The reliability of the scheme used for the electronic structure calculations of graphene−molecule complexes has been extensively tested in our previous work.7,

12

The

graphene sheet was modeled using a 6×6 supercell (72 carbon atoms) with a calculated C-C bond length of 1.44 Å. Bilayer graphene and graphite were modeled by two and four graphene sheets, respectively, with Bernal type (AB) stacking and an interlayer separation of 3.4 Å.12 The periodically repeated single-/multi-layers were separated by at least 18 Å of vacuum. The energy cutoff for the plane-wave expansion was set to 400 eV and a 3x3x1 k-point grid was used. The adsorption energy, ΔEad, was calculated as the difference between the energy of the most favorable configuration of the complex (between graphene and an ethanol cluster) and the sum of the energies of the isolated species (graphene and ethanol molecules in vacuum). In contrast, the interaction energy, ΔEint, corresponded to fragments (isolated species) with the geometry of the complex. The difference between the adsorption and interaction energies was termed the deformation energy, Edef, of the fragments, i.e., ∆ = EtOH ∆int + def + ∑ def . The enthalpy of adsorption, ΔHad, was calculated by adding the zerogr.

point energy (ΔΔE0), thermal (ΔΔET) and enthalpy (ΔΔEH) corrections to the adsorption energy, i.e., ΔHad = ΔEad + ΔΔE0 + ΔΔET + ΔΔEH.7 The corrections ΔΔE0, ΔΔET and ΔΔEH



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

were evaluated for the respective ethanol cluster on a finite model of graphene, circumcoronene (54 carbon atoms), i.e., ΔHgraphene ≈ ΔEgraphene + (ΔHcircumcoronene – ΔEcircumcoronene). For this purpose, geometry optimizations and frequency calculations were performed and partition functions and thermochemistry (at 323.15 K and 101.325 kPa) were obtained. We used the B97D20 functional, cc-pVTZ basis sets and Gaussian09 package21 for all calculations on circumcoronene. 3. RESULTS AND DISCUSSION 3.1 Adsorption enthalpies of ethanol to graphene/graphite samples. The adsorption process, as described by Eq. 1, fitted well the experimentally measured data corresponding to the adsorption of ethanol on the surface of graphite powder (with a coefficient of determination r2 above 0.9896, see Fig. 1, top). The obtained ΔHad values increased with increasing surface coverage and saturated above a coverage of ~5% (Fig. 2). The lower enthalpies at coverages below ~5% might be attributed to the preferred adsorption of ethanol to high energy adsorption sites, i.e., along steps, edges and cavities (see text below and Ref. 12

), present on the surface of graphite powder. We have previously observed similar behavior

for the adsorption of acetone on the same material and established that the graphite sample contained ~2% of high energy sites.12 The apparent isosteric adsorption enthalpy of ethanol on the graphite powder surface (at coverages above ~5%) ranged between –12.5 to –11.6 kcal/mol. Earlier thermal programmed desorption (TPD) experiments suggested a desorption activation energy of ethanol on highly oriented pyrolytic graphite (HOPG) of ΔEA = 12.0 ± 0.7 kcal/mol for a monolayer.6 This gives the adsorption enthalpy of –12.3 ± 0.7 kcal/mol assuming22 ΔHad = –ΔEA – ½RT (T = 323.15 K was chosen for relevant comparison). Therefore, our saturated adsorption enthalpy from inverse gas chromatography was in a good agreement with the earlier TPD value.


Page 8 of 26

Page 9 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Measured dependence of the surface coverage and partial pressure p, i.e., /1 − /ø , on the column temperature T (top and middle left) according to Eq. (1) for both graphene and graphite at 6% target surface coverage together with the corresponding logarithmic form of the dependence (top and middle right). Measured data points are depicted as blue diamonds and the fitted curves are in dark blue. In the case of graphene (middle plots), the points in red boxes correspond to the retention peaks (shown in the bottom graph) with insignificant net-retention time (i.e., not significantly differing from the dead-time determined by methane). The retention peaks are colored according to the respective temperature. 9 Environment ACS Paragon Plus


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Isosteric adsorption enthalpies (ΔHad) and adsorption entropies (ΔSad) of ethanol to graphite and graphene powders as a function of surface coverage obtained from fitting the experimental data. The error bars correspond to confidence intervals for a 5% level of significance.

In the case of ethanol adsorption on the graphene surface, we observed that the experimental data deviated from the behavior predicted by Eq. 1, particularly at the highest temperatures (see the middle right logarithm plot in Fig. 1). Bilinear behavior of the van’t Hoff plots (ln p vs. inverted temperature 1/T, see middle right plot in Fig. 1) is often attributed to two distinct mechanisms involved in the studied process, in this particular case two different mechanisms of ethanol adsorption/desorption to/from the graphene surface. To test this possibility, we carried out additional measurements at 6% surface coverage to sample the


Page 10 of 26

Page 11 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


temperature range 303-403 K with 21 data points spread at equidistant intervals of 5 K. However, we found that even the two-process model could not explain the observed behavior. Instead, we found that the non-linearity in the van’t Hoff plot stemmed from saturation of the partial pressure at high temperatures. Detailed inspection of the chromatographic peaks revealed that in the case of the high temperature measurements, the ethanol eluted almost at the same time as methane, which was used for the dead-time determination. The net-retention time of ethanol was therefore negligible in those measurements (see bottom panel of the Figure 1), leading to high uncertainty in the measured quantities.23,

24

We thus decided to

discard the peaks with net-retention time lower than 3σ from further data fitting, where σ stands for the standard deviation of the methane Gaussian peak. In other words, we only kept the measurements where ethanol had a significant retention time. Such filtered data were indeed found to fit Eq. 1. Although the isosteric adsorption enthalpy of ethanol on graphene increased with increasing surface coverage at very low surface coverages (below 1%), the dominant trend was decreasing enthalpy at higher surface coverages. In contrast, the adsorption enthalpy systematically increased with coverage on graphite (Fig. 2). We hypothesize that this difference might be explained by competition between adsorption to high energy sites and another adsorption process characterized by lower adsorption enthalpy. Since it was possible for ethanol to form clusters under the experimental conditions, because the enthalpy of ethanol liquefaction ranges from 10 to 8 kcal/mol for temperatures from 303 to 403 K,25 respectively, we attributed the other adsorption process to ethanol clustering. It is likely that at very low surface coverages (below 1%), the adsorption to the high energy sites is preferred over the clustering, due to the low partial pressure of ethanol. In such a regime, the occupancy of the high energy sites (in our case, the graphene sample contained ~0.2% of high energy sites12) would rapidly saturate with increasing surface coverage, and thus the apparent



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

isosteric enthalpy increases with surface coverage as the less energetically favorable adsorption process begins to dominate. At even higher surface coverage (1-6% of coverage), the high energy sites would already be saturated, and therefore ethanol clustering on the surface dominates. The increasing propensity for clustering with increasing surface coverage would explain why the apparent adsorption enthalpy declined with coverages > 1%. We suspect that the effect of clustering was hindered in graphite due to its higher number of high energy sites. The saturated adsorption enthalpy on the graphene for coverages above ~5% ranged from –10.9 to –11.3 kcal/mol. We would like to stress that as both the abovementioned effects (adsorption to the high energy sites and ethanol clustering) contribute simultaneously at low coverage, one should not fit the measured adsorption enthalpies to the two-state model introduced in our recent study.12 In principle, it is possible to derive a thermodynamic model accounting for clustering phenomena. However, in such a case, any statistical analyzes would face problems of over-fitting due to the high number of fitted parameters (adsorption enthalpies and entropies of each assumed cluster) compared to the number of measured data points. The presence of ethanol clustering on the graphene and graphite surfaces was supported by the striking difference between the observed (saturated) adsorption enthalpies of –12.0 ± 0.4 kcal/mol estimated by averaging the adsorption enthalpies for > 5% coverage, i.e., after saturation (discussed above), and the adsorption enthalpy of –6.4 kcal/mol of the ethanol molecule to graphene estimated from earlier theoretical calculations.7 Note that the adsorption enthalpy of ethanol to graphene of –7.3 ± 0.7 kcal/mol published in our previous study7 was higher than the saturated value of –12.0 ± 0.4 kcal/mol (see above) owing to the choice of experimental conditions, i.e., measurement at a single surface coverage of ~2% and temperature range of 303-343 K. Based on this experience, we would recommend measuring isosteric adsorption enthalpies over a sufficiently wide range of the surface coverage values.


Page 12 of 26

Page 13 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Finally, the isosteric adsorption entropy (ΔSad) of ethanol on the graphite powder surface (at coverages above ~5%, Figure 2) ranged from –25 to –27 cal/mol/K (for T centered at 303-393 K). The standard entropies of adsorbed molecules (Sad) on the single crystal surface was found to linearly track the entropies of the gas-phase molecule (Sgas) at the same temperature, Sad(T) = 0.7Sgas (T) − 3.3R.26, 27 Therefore, the entropy of adsorption, ΔSad(T) = Sad(T) − Sgas(T) = −0.3Sgas (T) − 3.3R, was a function of Sgas and T only. This means that it is reasonable to compare adsorption entropies of ethanol molecules on different surfaces. It has been previously reported that the adsorption entropy of ethanol on Ti sites of a TiO2 surface was –28.4, –27.4 and –26.8 cal/mol/K for temperatures of 395, 337 and 310 K, respectively.27 Surprisingly, these ΔSad values are in excellent agreement with our ΔSad values of ethanol on graphite. This agreement supported the idea that ΔSad does not dependent on the particular surface used for adsorption. 3.2 Adsorption properties of ethanol from theoretical calculations. MD simulations of ethanol molecules on the ideal surface of graphene (using ~10% coverage) clearly demonstrated a tendency for ethanol to form clusters (see Supporting Information for movie). The clusters lay flat on the graphene surface and cyclic planar forms of clusters were preferred (Figure 3). The number of molecules in the cluster on the surface increased/decreased mainly as a consequence of collisions with individual molecules moving on the graphene surface. Cluster analysis identified pentamers (EtOH)5 followed by tetramers (EtOH)4 as the predominant clusters (for a temperature range of 300 – 360 K). With increasing temperature, the frequencies of clusters slightly decreased in favor of monomers (inset of Figure 3, Table S1).



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3. Snapshot taken from MD simulation. Only the most frequently occurring ethanol clusters, cyclic pentamers and tetramers, and monomeric ethanol are shown. Inset: cluster size distribution agglomerated during the simulation for temperatures of 300 K, 320 K, 340 K and 360 K.

We further performed DFT calculations to obtain quantitative information about adsorption energies/enthalpies of the ethanol monomers and ethanol clusters to graphene. We optimized configurations of adsorbed ethanol clusters to graphene by utilizing the geometries obtained from the MD simulations. Indeed, the clustering of ethanol molecules was found to be thermodynamically preferential even on circumcoronene/graphene. The final DFT geometries of ethanol clusters were similar to the initial geometries obtained from the MD simulations. However, the hydroxyl groups of the ethanol molecules frequently switched from “trans” rotameric states to “gauche” states28, 29 (cf., Figure 3 and Figure 4). The shapes of the


Page 14 of 26

Page 15 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


adsorbed clusters differed significantly from their free forms (Figure 4). Whereas cyclic ethanol clusters on the surface were very flat, free clusters were more structured in space. The changes in the cluster and surface geometries upon adsorption increased with the size of the cluster (see deformation energy per ethanol molecule, Edef, Table 1). In addition, the planar geometrical structure of the ethanol clusters on the surface of graphene maintained a similar amount (per ethanol molecule) of dispersive interaction with the substrate as the single molecule (∆ED in Table 1).

Table 1. Adsorption energies and other quantitiesa characterizing the adsorption of various number (n=1-5) of ethanol molecules to circumcoronene, (C54H18 + n EtOH)/n → (C54H18…(EtOH)n)/n. All values are in kcal/mol, except the entropy (in cal/mol/K). n ∆Ead

∆E0

∆U

∆Had

∆G

∆∆E0

∆∆ET

∆∆EH

∆∆EG

∆Had-

∆Sad

∆Eint

∆ED

Edef

∆Ead

1 -9.0

-8.3

-6.9

-7.6

2.8

0.8

1.3

-0.6

10.4

1.4

-32

-9.2

-14.6

0.2

2 -10.9

-9.8

-8.7

-9.4

1.9

1.1

1.1

-0.6

11.2

1.5

-35

-11.4

-14.4

0.5

3 -12.9

-11.4

-10.5

-11.1

0.7

1.6

0.9

-0.6

11.8

1.8

-36

-13.7

-13.1

0.8

4 -14.5

-12.9

-12.1

-12.7

-0.5 1.6

0.8

-0.6

12.3

1.7

-38

-15.6

-13.7

1.1

5 -15.0

-13.4

-12.8

-13.4

-0.5 1.6

0.6

-0.6

13.0

1.6

-40

-16.1

-13.9

1.0

a

Adsorption energies (∆Ead and ∆E0 without and with the zero-point energy, respectively), internal energies (∆U), enthalpies (∆Had), Gibbs energies (∆G) and entropies (∆S) per ethanol molecule (including contributions of the zero-point energy (∆∆E0), thermal (∆∆ET), enthalpy (∆∆EH) and Gibbs energy corrections (∆∆EG)). The interaction (∆Eint, ∆ED is a part of ∆Eint due to the empirical dispersion term D) and deformation (Edef) energies per ethanol molecule are also provided for comparison.

The first model, in which graphene was modeled by circumcoronene C54H18, allowed evaluation of the different contributions to the adsorption enthalpy of ethanol to graphene.7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The adsorption energies of ethanol molecules per molecule (∆Ead) on circumcoronene were found to be proportional to the size n of cluster up to n=5 (Table 1). Whereas the zero-point energy corrections per ethanol molecule (∆∆E0) increased with n (from 0.8 to 1.6 kcal/mol), the thermal corrections (∆∆ET) decreased with n (from 1.3 to 0.6 kcal/mol) because of increased rigidity of the clusters. The pure enthalpy correction was exactly the same for various n (∆∆EH = –0.6 kcal/mol). The resulting difference between the energy and enthalpy (∆Had–∆Ead = ∆∆E0 + ∆∆ET + ∆∆EH) therefore did not depend on the cluster size and stayed within a relatively narrow range of 1.4–1.8 kcal/mol. The calculated adsorption entropies (ΔSad = (∆Had – ∆G)/T) increased with the cluster size and ranged from –32 to –40 cal/mol/K (Table 1). The averaged theoretical values of ∆Sad (weighted by the cluster distribution shown in the inset of Figure 3) of between –37 and –39 cal/mol/K (for temperatures 360 – 300 K) were in qualitative agreement with the experimental adsorption entropy, which ranged from – 25 to –27 cal/mol/K (for T centered at 303-393 K at coverage above ~5%, Figure 2). Slight overestimation of the calculated values w.r.t. experiment is to be expected considering the approximations used; the partition functions were evaluated from a finite model, the rigid rotor approximation was applied for rotations and the harmonic approximation was used for vibrations.


Page 16 of 26

Page 17 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. Ethanol clusters of various size (n = 2 - 6) in isolated form (left column), adsorbed on circumcoronene (middle column) and adsorbed on graphene (right column).

For the periodic model, we first calculated the adsorption energy of a single ethanol molecule adsorbed on a 6x6 graphene supercell. Its value of –7.7 kcal/mol obtained using the optB86b-vdW functional agreed well with the value of the adsorption energy of –7.9 kcal/mol published in our previous study7 (calculated using the optB88-vdW functional and 4x4 graphene). Next, we evaluated the adsorption energies of the clusters on graphene and



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

corrected them to the enthalpies derived from the circumcoronene model (see Methods section). The calculated adsorption enthalpies for different ethanol clusters are listed in Table 2. The lowest adsorption enthalpy was obtained for the ethanol pentamer, ∆Had = –14.8 kcal/mol. This finding supports the dominant occurrence of these clusters in the MD simulations (inset of Figure 3). To facilitate comparison with experiment, we calculated the weighted average of the adsorption enthalpy for different sized clusters, where the weights were obtained from the cluster size distribution (inset of Figure 3). The average theoretical values of ∆Had lay between –11.2 kcal/mol at 360 K and –13.0 kcal/mol at 300 K (Table 2), in close agreement with the experimental adsorption enthalpy of –12.0 ± 0.4 kcal/mol.

Table 2. Adsorption energies and enthalpies (kcal/mol) of ethanol clusters (EtOH)n on graphene per ethanol molecule at 10% surface coverage. The enthalpy corrections were calculated using the circumcoronene model of graphene (see Methods). Adsorbed system

∆Ead

∆Had

(EtOH)1 at 323.15 K

-7.7

-6.3

(EtOH)2 at 323.15 K

-9.8 (-10.3)a

-8.3 (-8.8)a

(EtOH)3 at 323.15 K

-12.7

-10.9

(EtOH)4 at 323.15 K

-15.6 (-16.1)a

-13.9 (-14.4)a

(EtOH)5 at 323.15 K

-16.4

-14.8

(EtOH)6 at 323.15 K

-15.7

-14.1

Weighted average at 300 K

-14.7

-13.0 (-13.5)b


-14.3

-12.7 (-13.2)b


-13.2

-11.6 (-12.1)b


-12.8

-11.2 (-11.7)b


Page 18 of 26

Page 19 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Experiment at 303-393 K a

-12.0 ± 0.4

calculated for adsorption to Bernal stacked bilayer graphene, b extrapolated values for

bilayer graphene

We also investigated adsorption of an ethanol dimer and tetramer on bilayer graphene because the effect of additional graphene layers on the adsorption energy was shown to be not negligible in our previous study of acetone adsorption.12 In the case of ethanol, this effect was smaller as the adsorption energy of both the ethanol dimer and tetramer on Bernal stacked bilayer graphene was lowered only by 0.5 kcal/mol per ethanol molecule (from –9.8 to –10.3 kcal/mol and from –15.6 to –16.1 kcal/mol per molecule, respectively, Table 2). Therefore, the averaged calculated value of ∆Had for the ethanol clusters on bilayer graphene was shifted by –0.5 kcal/mol with respect to ethanol clusters on the single sheet, resulting in adsorption enthalpies in the range from –11.7 to –13.5 kcal/mol in good agreement with our experimental value of ∆Had = –12.0 ± 0.4 kcal/mol and with the value from the TDP experiment of Ref.

6

(∆Had = –12.6 ± 0.7 kcal/mol). To elucidate the apparent difference between graphene and graphite at low coverages observed in experiment (~–9 kcal/mol and ~–16 kcal/mol for graphene and graphite, respectively, Figure 1), we calculated the adsorption of ethanol monomers on steps and edges, which are expected to be present on graphitic surfaces as high energy sites.12 The step was created by removing half of the upper layer in bilayer graphene (Figure 5). The adsorption energy on the step was –15.5 kcal/mol. The edge was modeled by using the four layer supercell of AB-stacked graphite. The molecule aligned itself so that the OH bond of the hydroxyl group pointed toward one of the carbon atoms on the edge (Figure 5). This geometry had an adsorption energy of –10.7 kcal/mol (cf., the value for adsorption on the flat surface of



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

–7.7 kcal/mol). The adsorption energy to Stone-Wales defect was –9.4 kcal/mol and the adsorption energies to vacancy and double vacancy defects do not significantly differ from the ethanol adsorption energy to surface (Figure 5). The difference between the calculated adsorption energies for the high energy sites and surface sites (–7.8 - –7.6 kcal/mol) corresponded quite well to the difference between low coverage and high coverage enthalpy values of ethanol adsorption to graphite powder obtained experimentally (~–5 kcal/mol, Figure 2).

Figure 5. Schematic showing the orientation and adsorption energy of an ethanol molecule adsorbed on a step, edge, surface of graphite/graphene and surface with Stone-Wales, vacancy and double vacancy defects.


Page 20 of 26

Page 21 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The difference in behavior of the ethanol molecules and other adsorbates examined in our recent studies7,

12

motivated us to perform additional calculations with acetone. In

analogical MD simulations, the acetone molecules adsorbed on graphene did not form clusters (the enthalpy of acetone liquefaction ranges from 7.3 to 6.7 kcal/mol at temperatures 303 to 343 K, respectively25) and preferred monomeric form (see cluster size distribution agglomerated during simulation in Table S2). The main reason for the different behavior of ethanol vs. acetone molecules lies in the capacity of ethanol to form intermolecular hydrogen bonds, which stabilize ethanol clusters. We also performed periodic calculations on a large 8x8 graphene supercell. We evaluated the overall energy balance for the creation and adsorption of selected ethanol and acetone clusters. The thermodynamic cycle shown in the Scheme 1 verifies that ethanol and acetone should have behaved differently according to the calculated adsorption energies, ∆Ead: it is energetically favorable for the ethanol molecules to both adsorb and form clusters (see also Table S3 and S4). On the other hand, clustering of the acetone molecule on graphene is almost energetically prohibitive because ∆Ead is significantly shifted to less negative values (Scheme 1).

Scheme 1. Thermodynamic cycle for the creation of an adsorbed ethanol tetramer (left) and



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

acetone dimer (right) on a graphene/graphite surface evaluated using a periodic model. All energies (in kcal/mol) are normalized to one ethanol (acetone) molecule.

4. CONCLUSIONS We measured the adsorption enthalpies of ethanol on graphene/graphite by inverse gas chromatography under a wide range of surface coverages from 0.1 to 20% and temperature of 303-393 K. The adsorption enthalpies varied significantly with the coverage and large differences between the samples were detected at low coverage. On the other hand, at higher coverages, the adsorption enthalpy approached the same limit of ∆Had = –12.0 ± 0.4 kcal/mol on both samples. We attributed this behavior to the surface heterogeneity and clustering of ethanol molecules on the surface, which was confirmed by theoretical calculations. Molecular dynamics simulations enabled the distribution of cluster sizes to be calculated for various temperatures. Adsorption enthalpies obtained from non-local DFT calculations for different sizes of ethanol clusters weighted by the size distribution provided a ∆Had value of between – 11.2 kcal/mol at 360 K and –13.0 kcal/mol at 300 K in agreement with the experimental value. Adsorption to high energy sites and clustering of ethanol molecules operated simultaneously at low coverages, explaining the different adsorption enthalpy trends observed for graphene and graphite. Clustering was more extensive at higher surface coverage, decreasing the apparent adsorption enthalpy. In contrast, monomer adsorption to high energy sites saturated with increasing surface coverage, leading to increasing apparent adsorption enthalpy. Finally, it should be noted that mutual interactions between adsorbates may significantly affect adsorption processes even at low coverage values and should be taken into account, particularly in studies addressing hydrophobic surface heterogeneity sampled by polar probes.


Page 22 of 26

Page 23 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ASSOCIATED CONTENT Supporting Information. Cluster size distribution agglomerated during the MD simulation of ethanol (Table S1) and acetone (Table S2) molecules on graphene, calculated adsorption properties of the ethanol clusters on circumcoronene (Table S3), calculated properties of ethanol molecule clustering (Table S4) and example MD trajectory demonstrating tendency of ethanol to form clusters (movie). This material is available free of charge via the Internet at http://pubs.acs.org/.

AUTHOR INFORMATION Corresponding Author * Tel. +420 585 634 756, E-mail: [email protected].

ACKNOWLEDGEMENT This work was supported by the Ministry of Education, Youth and Sports of the Czech Republic (project LO1305), Czech Science Foundation (P208/12/G016) and a student project of Palacký University (GA_PrF_2015_027). MO acknowledges support from Neuron fund for support of science. We thank Alessandre Tkatchenko for pointing out some discrepancies in experimental values, which contributed to our motivation to deeply understand ethanol adsorption on graphene/graphite.

REFERENCES



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1. Poynor, A.; Hong, L.; Robinson, I. K.; Granick, S.; Zhang, Z.; Fenter, P. A., How Water Meets a Hydrophobic Surface. Phys. Rev. Lett. 2006, 97, 266101. 2. Davis, J. G.; Gierszal, K. P.; Wang, P.; Ben-Amotz, D., Water Structural Transformation at Molecular Hydrophobic Interfaces. Nature 2012, 491, 582-585. 3. Attard, P., Nanobubbles and the Hydrophobic Attraction. Adv. Colloid. Interfac. 2003, 104, 75-91. 4. Lacher, S.; Matsuo, Y.; Nakamura, E., Molecular and Supramolecular Control of the Work Function of an Inorganic Electrode with Self-Assembled Monolayer of Umbrella-Shaped Fullerene Derivatives. J. Am. Chem. Soc. 2011, 133, 16997-17004. 5. Rouhana, L. L.; Moussallem, M. D.; Schlenoff, J. B., Adsorption of Short-Chain Thiols and Disulfides onto Gold under Defined Mass Transport Conditions: Coverage, Kinetics, and Mechanism. J. Am. Chem. Soc. 2011, 133, 16080-16091. 6. Ulbricht, H.; Zacharia, R.; Cindir, N.; Hertel, T., Thermal Desorption of Gases and Solvents from Graphite and Carbon Nanotube Surfaces. Carbon 2006, 44, 2931-2942. 7. Lazar, P.; Karlický, F.; Jurečka, P.; Kocman, M.; Otyepková, E.; Šafářová, K.; Otyepka, M., Adsorption of Small Organic Molecules on Graphene. J. Am. Chem. Soc. 2013, 135, 6372-6377. 8. Zettlemoyer, A. C., Hydrophobic Surfaces. J. Colloid Interface Sci. 1968, 28, 343-369. 9. Chakarova-Kack, S. D.; Schroder, E.; Lundqvist, B. I.; Langreth, D. C., Application of van der Waals Density Functional to an Extended System: Adsorption of Benzene and Naphthalene on Graphite. Phys. Rev. Lett. 2006, 96, 146107. 10. Mercurio, G.; McNellis, E. R.; Martin, I.; Hagen, S.; Leyssner, F.; Soubatch, S.; Meyer, J.; Wolf, M.; Tegeder, P.; Tautz, F. S.; Reuter, K., Structure and Energetics of Azobenzene on Ag(111): Benchmarking Semiempirical Dispersion Correction Approaches. Phys. Rev. Lett. 2010, 104, 036102. 11. Karlický, F.; Lazar, P.; Dubecký, M.; Otyepka, M., Random Phase Approximation in Surface Chemistry: Water Splitting on Iron. J. Chem. Theory Comput. 2013, 9, 3670-3676. 12. Lazar, P.; Otyepková, E.; Banáš, P.; Fargašová, A.; Šafářová, K.; Lapčík, L.; Pechoušek, J.; Zbořil, R.; Otyepka, M., The Nature of High Surface Energy Sites in Graphene and Graphite. Carbon 2014, 73, 448-453. 13. Jorgensen, W. L.; Tirado-Rives, J., Potential Energy Functions for Atomic-Level Simulations of Water and Organic and Biomolecular Systems. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 66656670. 14. Caleman, C.; van Maaren, P. J.; Hong, M. Y.; Hub, J. S.; Costa, L. T.; van der Spoel, D., Force Field Benchmark of Organic Liquids: Density, Enthalpy of Vaporization, Heat Capacities, Surface Tension, Isothermal Compressibility, Volumetric Expansion Coefficient, and Dielectric Constant. J. Chem. Theory Comput. 2012, 8, 61-74. 15. van der Spoel, D.; van Maaren, P. J.; Caleman, C., GROMACS Molecule & Liquid Database. Bioinformatics 2012, 28, 752-753. 16. Cheng, A.; Steele, W. A., Computer Simulation of Ammonia on Graphite. I. Low Temperature Structure of Monolayer and Bilayer Films J. Chem. Phys. 1990, 92, 3858-3866. 17. Blochl, P. E., Projector Augmented-Wave Method. Phys. Rev. B 1994, 50, 17953-17979. 18. Kresse, G.; Joubert, D., From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method. Phys. Rev. B 1999, 59, 1758-1775. 19. Klimeš, J.; Bowler, D. R.; Michaelides, A., Van der Waals Density Functionals Applied to Solids. Phys. Rev. B 2011, 83, 195313. 20. Grimme, S., Semiempirical GGA-Type Density Functional Constructed with a Long-Range Dispersion Correction. J. Comput. Chem. 2006, 27, 1787-1799. 21. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A. et al.. Gaussian 09, Revision A.02, Gaussian 09, Revision A.02, Wallingford CT, 2009. 22. Brown, W. A.; Kose, R.; King, D. A., Femtomole Adsorption Calorimetry on Single-Crystal Surfaces. Chem. Rev. 1998, 98, 797-831. 23. Ho, R.; Heng, J. Y. Y., A Review of Inverse Gas Chromatography and its Development as a Tool to Characterize Anisotropic Surface Properties of Pharmaceutical Solids. Kona Powder Part. J. 2013, 164-180.


Page 24 of 26

Page 25 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


24. Williams, D. R., Particle Engineering in Pharmaceutical Solids Processing: Surface Energy Considerations. Curr. Pharm. Design 2015, 21, 2677-2694. 25. Svoboda, V.; Majer, V., Enthalpies of Vaporization of Organic Compounds : A Critical Review and Data Compilation. Blackwell Scientific Publications: Oxford, 1985; Vol. 32. 26. Campbell, C. T.; Sellers, J. R. V., The Entropies of Adsorbed Molecules. J. Am. Chem. Soc. 2012, 134, 18109-18115. 27. Campbell, C. T.; Sellers, J. R. V., Enthalpies and Entropies of Adsorption on Well-Defined Oxide Surfaces: Experimental Measurements. Chem. Rev. 2013, 113, 4106-4135. 28. Dyczmons, V., Dimers of Ethanol. J. Phys. Chem. A 2004, 108, 2080-2086. 29. Soltesova, M.; Benda, L.; Peksa, M.; Czernek, J.; Lang, J., Determination of Size of Molecular Clusters of Ethanol by Means of NMR Diffusometry and Hydrodynamic Calculations. J. Phys. Chem. B 2014, 118, 6864-6874.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

TOC:


Page 26 of 26

Interplay between Ethanol Adsorption to High ... - ACS Publications

Recommend Documents