Petr
Lazar
,
Eva
Otyepková
,
Martin
Pykal
,
Klára
Čépe
and
Michal
Otyepka
*
Regional Centre of Advanced Technologies and Materials, Department of Physical Chemistry, Faculty of Science, Palacký University Olomouc, tř. 17. listopadu 12, 77 146 Olomouc, Czech Republic. E-mail: Michal.Otyepka@upol.cz
First published on 25th April 2018
Nanomaterials have a high surface-to-mass ratio and their surface properties significantly affect their features and application potential. Phosphorene, a single layer of black phosphorus (BP), was the first homoatomic two-dimensional material to be prepared after the discovery of graphene. The structure of phosphorene resembles the honeycomb arrangement of graphene, but its layers are buckled and highly anisotropic. We studied how this difference affects the surface properties of BP, namely the free surface energy and adsorption affinity of various organic molecules. Using inverse gas chromatography, we measured the total surface free energy of BP powder to be 90 mJ m−2 and showed that it was dominantly determined by dispersion forces, but, unlike on graphene, with a notable contribution from specific acid–base interactions. We further measured adsorption enthalpies of volatile organic compounds on BP and rationalized them using density functional theory calculations. Polar molecules showed an increased affinity due to a significant contribution of dipole–dipole interactions to the molecule–surface bonding, because the buckled surface of BP causes higher diffusion barriers than those on graphene, hinders molecular in-plane motion and supports mutual orientation of molecular dipoles over longer distances, in contrast to graphene.
Phosphorene is a single layer derivative of bulk orthorhombic black phosphorus (BP).9 BP was first prepared by Bridgman10 in 1914 and remained a curiosity in materials science until 2014, when phosphorene became the first homoatomic 2D material to be isolated since the discovery of graphene.6,11 BP has a layered structure with individual layers held together by weak van der Waals forces, which facilitates exfoliation into phosphorene.12 The structure of phosphorene resembles the honeycomb structure of graphene in terms of its connectivity, but it is not planar. Instead, BP layers are buckled, giving rise to anisotropic bond angles and highly anisotropic properties within the basal plane. Unlike graphene, which is a zero band-gap semiconductor, phosphorene is a direct semiconductor exhibiting intrinsic p-type conductivity and a band-gap energy dependent on the number of layers.6,13 The band-gaps of BP and phosphorene are about 0.3–0.5 eV and 1.0–2.0 eV, respectively.14,15 Moreover, the electronic properties of phosphorene can be tuned by its covalent functionalization using nucleophilic reagents.16 The presence of a band-gap, its tunability and p-type conductivity with hole ballistic transport offer several advantages for electronic device construction in comparison with graphene.17,18 In addition, BP devices display carrier mobilities that vastly exceed those typical of a MoS2 transistor.6,19 Therefore, BP bridges the current gap between graphene and 2D semiconductors based on transition metal dichalcogenides, prompting enormous scientific interest in this material.20
BP is sensitive to the surrounding environment. This sensitivity is generally considered a disadvantage because the surfaces of 2D materials are prone to adsorption of various contaminants, which affects the real performance in applications. In contrast to the inert van der Waals nature of many layer-type materials (e.g. graphene and MoS2), which renders them hydrophobic, the reactivity of a BP surface in air is strongly influenced by its interaction with water. Pristine BP is hydrophobic, whereas oxidation by O2 turns the surface progressively hydrophilic.21 On the other hand, this sensitivity opens up applications in analytical chemistry. For example, a vapour sensor based on layered BP has been suggested to selectively detect methanol vapour22 and a field-effect transistor sensor device fabricated using 2D phosphorene nanosheets has been shown to exhibit ultrahigh sensitivity to NO2 in dry air that is dependent on its thickness.23 Therefore, it is desirable to assess the affinity of phosphorene to various molecules and airborne contaminants, such as volatile organic compounds, which are present in chemical labs and also in the daily environment. Thus, surface adsorption on phosphorene is of fundamental interest for its applications.
In this study, we assessed the surface energy and affinity of small molecules to the surface of BP by using inverse gas chromatography (iGC) and used density functional theory (DFT) calculations and molecular dynamics (MD) simulations to interpret the experimentally measured affinities. iGC is a surface characterization technique that can provide useful information about the surface nature, heterogeneity, etc.24–26 One advantage of this method is that the coverage of probe molecules on the surface can be readily controlled. iGC provides averaged information about the complete surface because it is based on the interaction of gas probes, which flow through the bed containing the studied material. Adsorption/desorption events result from the probe–surface interaction and affect the retention of the probe by the sample. Thus, we measured the surface free energy of BP and its dispersive and specific acid–base components from the retention times of n-alkanes and acid–base probes (see Materials and methods for details). We found that the surface energy of BP arises mainly from dispersive interactions but has also a considerable acid–base component, in contrast to other 2D materials, such as graphene and fluorographene, in which acid–base component is negligible. The adsorption enthalpies of non-polar molecules (benzene, dioxane and cyclohexane) were very similar to those measured on graphene, whereas polar molecules (acetonitrile, nitromethane and acetone) exhibited significantly higher affinity to BP than to graphene. We explained this discrepancy by showing that the barrier against surface diffusion of molecules on BP is almost four times higher than the barrier on graphene. Thus the surface potential of BP traps molecules and hinders their in-plane motion, which allows mutual orientation of molecular dipoles and induces a significant contribution of dipole–dipole interactions to the molecule–surface bonding. We modelled the surface behaviour of molecules directly using MD simulations. Molecules on a BP surface formed pronounced cluster chains across individual valleys of BP and were highly aligned with respect to each other. In an analogical simulation on graphene, the molecules formed one large cluster, but there was no preferential orientation of molecules within the cluster, i.e. the molecules interacted as randomly oriented dipoles.
The surface of single-layer BP (aka phosphorene) was modelled by a 4 × 4 supercell (64 phosphorus atoms). The periodically repeated sheets were separated by at least 16 Å of vacuum, and a 3 × 3 × 1 k-point grid was used to sample the Brillouin zone. The forces between the surface and the molecule were fully relaxed. We tested the influence of dipole–dipole interaction between periodic images of molecules. We found that, first, the dipole corrections in VASP did not change the total energies (the total energies changed less than 1 meV) and, second, that the adsorption energies did not change on changing the size of a supercell (we tested 3 × 3, 4 × 4, 5 × 5 and 6 × 6 supercells). Thermal corrections for the enthalpy were taken from our previous work.30
MD calculations were carried out using the GROMACS 5.0 software33 using the OPLS-AA force field.34 Topologies of individual small molecules were taken from the GROMACS Molecule & Liquid database.35 Intermolecular interactions of phosphorene and graphene were treated using the potentials of Sresht et al.36 and Cheng and Steele,37 respectively, and truncated after 10 Å. Both surfaces were modelled as uncharged and periodic within the x–y plane (with a simulation box size of ∼120 × 100 × 200 Å) and were kept fixed in the centre of the box during the simulation. Periodic boundary conditions were applied in all three dimensions. To simulate the collective behaviour of molecules on the surfaces, 60 molecules were placed randomly into the upper half of the simulation box. The simulations were run with a 2 fs time step in the NVT ensemble at 100 and 300 K. Coordinates were stored for analysis every 20 ps. The total simulation time for cluster creation was 30 ns (the first 2 ns were used for equilibration). Clusters were defined using agglomerative hierarchical cluster analysis with a single-linkage criterion using a maximum cluster distance of 7.9 and 6.5 Å for acetonitrile and nitromethane, respectively.
![]() | (1) |
The total surface energy of a material, γt, consists of two components, the dispersive (γd) and acid–base surface energies (γab), i.e. γt = γd + γab. The dispersive surface energy of a material originates from the London (dispersion) interactions. We employed the temperature corrected Dorris and Gray method for calculating γd, which uses a series of n-alkanes as probes to measure the free energy of adsorption. The dispersive free energy of one methylene group (ΔGCH2) can be calculated from the slope of the alkane line by plotting the probe adsorption free energies versus carbon number n of the alkane probe:
![]() | (2) |
ΔGCH2= −NAaCH2WCH2, | (3) |
![]() | (4) |
![]() | (5) |
Hence, the retention volume was measured for a set of alkane probes and the isosteric surface energy calculated at a given surface coverageν using Eqn (5). This was repeated for a range of coverage values to generate a surface energy profile as a function of coverage. The acid–base component of the surface energy, γab, is associated with specific interactions between a probe and the surface, e.g. hydrogen bonding, and was determined using the van Oss–Good–Chaudhury (vOGC) approach39 with the Della Volpe scale.40 Dichloromethane was used as a monopolar acid probe and ethyl acetate was used as a monopolar basic probe to characterize the basic and acidic properties of the solid surface, respectively.41
![]() | (6) |
Molecule | Exp. ΔHad on BP [kcal mol−1] | Temp. range [°C] | Calc. ΔHad on BP [kcal mol−1] | Exp. ΔHad on Ga [kcal mol−1] |
---|---|---|---|---|
a iGC-determined values on graphene taken from ref. 30 and 47. | ||||
1,4-Dioxane | −11.7 ± 0.6 | 30–90 | −12.1 | −10.8 ± 0.1 |
Benzene | −11.9 ± 0.1 | 30–90 | −12.2 | −11.9 ± 0.3 |
Cyclohexane | −11.2 ± 0.6 | 30–90 | −10.5 | −11.4 ± 0.3 |
Acetone | −12.1 ± 0.2 | 40–90 | −9.3 | −8.2 ± 0.3 |
Acetonitrile | −10.9 ± 0.1 | 40–90 | −7.2 | −7.6 ± 0.3 |
Nitromethane | −10.3 ± 0.3 | 30–90 | −7.3 | −6.3 ± 0.1 |
Tetrachloromethane | −8.2 ± 0.8 | 30–90 | −8.6 |
The shape of the measured γ vs. coverage curve indicates that (1) the number of high-energy sites on BP is rather low because they influence γ only up to 5% coverage, and (2) their relative energy is rather high because the energy of 146 mJ m−2 at low coverage is higher than the respective energy measured by iGC on graphite.46,49 However, there was a distinct difference when comparing previous iGC experiments on layered materials. On BP, the acid–base component made a notable contribution to the surface free energy, amounting to ∼13% (Fig. 1), although the dispersive component was still dominant. In the case of graphite and graphite fluoride, the acid–base component was essentially zero and the surface free energy originate from dispersive interaction alone. The acid–base component arises from polar interactions beyond dispersion.24 The significant acid–base component on BP can be related to a recent theoretical study by Tománek et al., which demonstrated that the nature of the interlayer binding in BP is richer than simple van der Waals interaction.50 Instead, additional interlayer interaction is associated with significant charge redistribution between the in-layer and interlayer regions caused by the nonlocal correlation of electrons in adjacent layers. It should be noted that Tománek et al. studied the nature of the interlayer interaction in layered BP using quantum Monte Carlo calculations,50 which describe the correlation of electrons explicitly and treat covalent and dispersive interactions on the same footing.
![]() | ||
Fig. 2 Isosteric adsorption enthalpies of volatile organic molecules as a function of surface coverage (in % of a monolayer). |
The decrease was reasoned to be due to the adsorption of molecules into high-energy adsorption sites (e.g. steps and cavities46), which were present in the graphite powder in much higher amounts than in graphene nanopowder. In our case, the adsorption profiles (Fig. 2) suggest that the amount of high-energy sites in BP is lower than in graphite.
We further performed DFT calculations to obtain quantitative information about the adsorption energies of molecular monomers to the surface of BP. It should be noted that the calculations represented adsorption to a pristine and clean single layer surface, i.e. a surface without any oxidation. It should be noted that our test calculation for bilayer BP showed only a minor change of the adsorption energy. The magnitude of the adsorption energy was slightly increased for benzene (−15.2 kcal mol−1 compared −14.0 kcal mol−1 for single layer BP) but remained essentially the same for acetonitrile (−8.2 kcal mol−1vs. −8.4 kcal mol−1).
The results are summarized in Table 1 and show that the molecules could be divided into two distinct groups. The first group showed good agreement between theory and experiment and comprised 1,4-dioxane, benzene, tetrachloromethane and cyclohexane. For this group, the calculated and measured adsorption enthalpies agreed within one kcal per mol, which was roughly the expected accuracy of both experiment and theory. Notably, the adsorption enthalpies of these molecules were very similar to those previously measured for adsorption on graphene,47 indicating that they showed purely van der Waals character of bonding. In the second group of molecules (acetone, acetonitrile and nitromethane), there was a larger deviation between theory and experiment. In particular, the magnitudes of the measured adsorption enthalpies of acetone and acetonitrile were significantly higher than the magnitudes of the respective theoretical values and also the magnitudes of the values previously determined for graphene.
We considered the possibility of hydrogen bonding for acetone and acetonitrile, i.e. the molecules for which the measured adsorption enthalpy most exceeded its calculated value. For acetone, the formation of a hydrogen bond was thermodynamically unfavourable to adsorption onto a clean surface, whereas for acetonitrile, a hydrogen bond was present but the bond was too weak to influence the adsorption enthalpy. Acetonitrile bonded to the lone pair oxygen and the magnitude of the adsorption enthalpy of the resulting bond (−7.8 kcal mol−1) was marginally higher than that of the enthalpy of van der Waals bonding onto a clean surface (−7.2 kcal mol−1). The hydrogen-bonded acetonitrile molecule was oriented along the y direction with its C–N group slightly tilted towards the surface. The length of the C–H⋯O bond was 3.12 Å (Fig. 3). Nevertheless, the magnitude of the theoretical adsorption enthalpy was still much lower than that measured in the iGC experiment (−10.9 ± 0.1 kcal mol−1).
This type of hydrogen bonding is relevant in other contexts. Pumera et al. have recently reported22 that methanol molecules have a high affinity to a BP surface, which could be utilized in a selective device for methanol sensing based on an electrode modified with BP. We calculated the adsorption enthalpy of methanol on a clean surface of BP, but its value of −5.1 kcal mol−1 did not indicate any particular affinity of methanol towards BP. However, assuming that a lone pair oxygen was present at the surface, the methanol molecule was able to bind to the oxygen via a hydrogen bond and the affinity was significantly increased (−8.3 kcal mol−1). The C–H⋯O bond length was 2.85 Å, in which the H⋯O distance was 1.88 Å. This result suggests that surface oxidation of BP may affect the affinity of some molecules and may also complicate interpretation of the surface selectivity of BP-based sensing devices.
Hence, the surface of BP shows significantly different properties from those of graphene. In contrast to superdiffusive motion on graphene, the molecules adsorbed on BP are trapped in ‘valleys’ created by its puckered structure and their translational as well as rotational degrees of freedom are hindered. Due to the hindered motion, molecules may appear ordered at a local level. Such ordering is particularly important for molecules possessing a dipole moment. Dipole–dipole electrostatic interactions can contribute to the net molecule surface bonding and in so doing increase the surface affinity of polar molecules.
The contribution of dipole–dipole interactions to the bonding on BP explains why the magnitudes of calculated adsorption enthalpies of acetonitrile and nitromethane were lower than the respective experimental values (Table 1). Among the molecules studied, the calculations seemed to underestimate the affinity of the polar molecules (acetonitrile, nitromethane, and acetone) towards BP. Acetonitrile has a large dipole moment of 3.9 D, whereas nitromethane and acetone have dipole moments of 3.5 and 2.9 D, respectively.
Hence, we modelled the dipole–dipole contribution to bonding of an acetonitrile dimer. We found that acetonitrile can form several dimer configurations on the surface. The most favourable was predicted to be an antiparallel configuration with the two central carbon atoms about 3.38 Å apart and two N⋯H contacts at a distance of about 2.5 Å (Fig. 5). The dimerization energy Edimer (i.e. energy of the adsorbed dimer compared to the energy of two adsorbed monomers) was 4.6 kcal mol−1 in this configuration. There are no reports of experimental measurements of the interaction energy of an acetonitrile dimer in the literature. However, an elaborate quantum-chemical study reported a dimerization energy of 4.9 kcal mol−1 calculated for an isolated acetonitrile dimer at the MP2 level.52 According to the latter study, the antiparallel configuration was the most favourable for an isolated dimer, and the reported bond distances for the central carbon atoms (3.5 Å) and two N⋯H contacts (2.6 Å) were in good agreement with our values. The dimer matched well with the surface potential of BP; when we removed the dimer from the surface and relaxed its geometry again, the energy gain (i.e. deformation energy due to surface adsorption) was negligible (<0.1 kcal mol−1).
![]() | ||
Fig. 5 Geometry and dimerization energy of the two most favourable configurations of an acetonitrile dimer on the surface of BP. |
Therefore, in summary, dipole–dipole interactions of adsorbed polar molecules contribute to the molecule–surface bonding and explain why the magnitude of the calculated adsorption enthalpies of isolated molecules (e.g. −7.2 kcal mol−1 for acetonitrile) were considerably lower than that of the respective measured values (−10.9 kcal mol−1 for acetonitrile). The calculated adsorption enthalpy of acetonitrile in a dimer configuration was −11.3 kcal mol−1, in excellent agreement with the experimental value. It should be noted that we assumed that the difference between the adsorption energy and enthalpy did not depend on the cluster size, i.e. it was the same for the monomer and dimer. In our previous study of clustering of ethanol on graphene, we found that the thermodynamic difference between the adsorption energy and enthalpy did not depend on the cluster size n up to n = 5 because the increase of zero-point energy corrections with n was compensated by the decrease of thermal correction with n arising from the increased rigidity of the clusters.42
The role of dipole–dipole interactions depends on the orientation of the molecular dipole moments with respect to a surface, as derived by Kokalj from an electrostatic model describing the adsorption of polar molecules.53 Adsorbed molecules with dipoles oriented perpendicular to the surface prefer to stay separated owing to repulsive electrostatic interactions.53 In contrast, for dipoles aligned parallel to the surface, as is the case of acetonitrile, nitromethane and acetone on BP, accumulation of molecules on the surface is anticipated, in line with the dimerization of acetonitrile predicted from our calculations.
It remains to be clarified why a similar effect of dipole–dipole interactions was not observed in previous studies of acetonitrile or nitromethane adsorbed on graphene.30,47 As Fig. 4A shows, graphene has a nearly flat surface potential and the surface diffusion barrier is not high enough to hinder thermal motion of the inspected molecules at room temperature. Thus, the molecules can freely move and rotate in two dimensions, and this behaviour was actually observed experimentally as thermally activated Brownian motion.51 Effectively, the molecules on graphene behave like a two-dimensional gas and the dipole–dipole interaction energy from randomly oriented dipoles (i.e. Keesom energy) is very low and does not contribute to the molecule–surface bonding. The clustering of molecules occurs even on graphene42 but only when the dimerization (clustering) energy is comparable to the adsorption energy of a molecule, which is clearly not the case for acetonitrile and nitromethane on BP. So the surface potential of BP, which traps the molecules, hinders their motion and promotes their local ordering, is the reason for the increased molecule–surface interaction of polar molecules.
To corroborate this explanation of molecular behaviour, we performed MD simulation of acetonitrile and nitromethane on BP and graphene at a small surface coverage of ∼10%. At 100 K, the molecules on the BP surface formed pronounced cluster chains across the individual valleys of BP (Fig. 6). The molecules were highly aligned with respect to each other. The distribution of the orientation of dipole moments for nitromethane molecules showed one pronounced peak at an angle of 45 degrees (Fig. 7). The distribution obtained for acetonitrile revealed two preferred mutual alignments of the molecules, at 45 degrees and 60 degrees. The orientation of dipole moments coincided with the geometries of molecular dimers obtained from the DFT calculation (Fig. 5). Both acetonitrile and nitromethane formed predominantly large clusters (>12 molecules) with a minor portion of small clusters (of preferentially five and six molecules, respectively). On graphene, the molecules formed one large cluster owing to the ease of movement, but there was no preferential orientation of molecules within the cluster (Fig. 7). Such behaviour supports the idea of dipole–dipole interactions of randomly oriented dipoles on graphene presented above.
At 300 K, both acetonitrile and nitromethane were predicted to move vigorously over the surface, forming temporarily small clusters. The preferential orientation of molecules on BP was significantly less pronounced than that at room temperature, indicating that the diffusion barriers were not high enough to hinder thermal motion at 300 K. However, it should be noted that although the diffusion barrier on BP was high enough to resist thermal motion at room temperature according to our DFT calculations (see above), the barrier was calculated for the ideal case (the adsorbed molecule was always in its local minima) and did not take into account the rotational degrees of freedom of the molecule diffusing over the surface.
We further probed the surface by determining the isosteric adsorption enthalpies of six organic molecules to BP as a function of the surface coverage. The adsorption enthalpies of non-polar molecules (benzene, dioxane, and cyclohexane) were very similar to those measured on graphene, whereas polar molecules (acetonitrile, nitromethane, and acetone) had significantly higher affinity to BP than to graphene. This discrepancy was explained by noting that the barrier against surface diffusion of molecules on BP was predicted to be up to four times higher than the barrier on graphene. Thus, the surface potential of BP traps molecules and hinders their in-plane motion, which allows mutual orientation of molecular dipoles and increases the contribution of dipole–dipole interaction to the molecule–surface bonding.
This journal is © The Royal Society of Chemistry 2018 |