We have used plane-wave density functional theory methods to explore the adsorption modes and configurations of uracil molecules on a gold surface to gain insight into the rational design of surface functionalization. We have investigated at the molecular level, the interactions of the RNA pyrimidine base uracil molecule isolated on the single crystal (100) surface of the gold substrate to determine the structure of uracil and orientation. Our calculations have shown that the most stable adsorbate structure is the enol tautomer of uracil, which adsorbs flat onto the gold surface through one of its carbonyl atoms. This configuration, which is compatible with previous experimental findings, is thermodynamically preferred over the adsorbed keto structure by approximately 0.23 eV (22.2 kJ mol−1).
Supramolecular films on surfaces are attracting increasing interest from the scientific community because they are used in a varied range of applications in functionalized surface-based technologies (Swalen et al. 1987; Ullman 1991; Barth et al. 2005). Indeed, since thin organic films can modify the chemical and structural properties of surfaces they have become relevant in a large range of technologies, including electronic and optical materials (Swalen et al. 1987), electrocatalysts (Baizer 1991) and biological and chemical sensors (Murray 1992). Moreover, organic molecules can be used to protect reactive surfaces, acting as corrosion inhibitors (Parkins et al. 1980; Janata et al. 1994), and they are often used as additives in planting baths (Plieth 1992).
Supramolecular chemistry is the study of molecules assembled by non-covalent weak interactions, i.e. hydrogen bonding (H-bonding) and van der Waals (vdW) forces. In materials science, in particular, supramolecular chemistry is important in the attachment of monolayer films on solid surfaces that subsequently can be used for technological processes (Kasemo 1998; Barlow & Raval 2003).
In biological systems, non-covalent interactions are crucially important and determine highly specific functions, i.e. molecular recognition, as well as the formation of bio-interfaces such as DNA, RNA or proteins on surfaces. In other words, molecular recognition could be translated to a surface by DNA/RNA base immobilization.
DNA is the store of genetic information, whose structure consists of a right-handed double helix stabilized by H-bonding between purine and pyrimidine bases in complementary antiparallel strands, and by base-stacking interactions between base layers (Watson & Crick 1953). A correct matching between base pairs ensures the information passage from the parent DNA strand to the daughter RNA strand, although eight other possible pairings lead to DNA mismatches that can sometimes generate tumours. Purine and pyrimidine bases are the building blocks of DNA, RNA and several other relevant biological macromolecules such as ATP. In this context, an effective approach to the characterization of the recognition between the DNA/RNA bases can consist of the use of model systems made of molecular adducts of nucleobases, for example, the RNA base uracil. The uracil molecule is fundamental to the investigation of DNA/RNA pairing and mispairing as it has a role in the process of mutation, where nucleobase tautomerism can be responsible for mutations when non-standard base pairs are formed (Topal & Fresco 1976).
We suggest that the interaction and role of uracil in complex systems could become better understood if we study its adsorption on surfaces where its immobilization in the two-dimensional plane will enable easier investigation of individual interactions which are difficult to isolate otherwise. In this, our first paper on the surface adsorption of uracil, we therefore aim to explore the interaction of an isolated uracil molecule with a gold surface to contribute to our understanding of surface functionalization.
We have chosen gold as the substrate for our study as this metal has been studied intensively, particularly in electrochemistry, and consequently experimental data are available in the literature. For example, Dretschkow et al. (1997) have used scanning tunnelling microscopy (STM) to study uracil adsorption on gold where, depending on the experimental conditions, long-range ordered films of physisorbed and chemisorbed molecules were shown to form on single-crystal electrodes of Au(100) and Au(111). Physisorbed films appeared as two-dimensional patterns of individual uracils which are compatible with molecules lying flat on the surface. Chemisorbed geometries were suggested to coordinate the gold surface through deprotonation of one of the two nitrogen atoms, in particular, the N3H function has been shown to deprotonate on Au(111) (Li et al. 1999; Cunha et al. 2001).
Moreover, the adsorption of uracil has been studied both experimentally and theoretically on other surfaces such as copper and silicon (Rocco et al. 1989; Nakagawa et al. 1997; Lopez et al. 2002; Seino et al. 2003; Schmidt et al. 2006). Uracil adsorption on copper surfaces was studied by ultraviolet photoelectron spectroscopy (UPS) which demonstrated an upright molecular orientation on the Cu (110) substrate (Rocco et al. 1989). STM images of uracil on the Cu (111) surface showed that molecules interact weakly with the surface and that in this case the formation of the films in particular is mainly owing to the electrostatic interaction of the H-bonding between different uracil molecules on the surface (Nakagawa et al. 1997). STM, HREELS and ab initio modelling studies of uracil films adsorbed on silicon surfaces have indicated that the orientation of the molecular plane is again upright and that bonding occurs through the carbonylic oxygen of the enol tautomer of uracil (Lopez et al. 2002; Seino et al. 2003; Schmidt et al. 2006).
In this work, we have applied computational techniques based on plane-wave density functional theory (DFT) to investigate at the molecular level the interaction of an isolated uracil molecule with a single-crystal (100) surface of gold as the substrate, with particular reference to the surface–adsorbate structure and orientation. At this level of methodology we have focused on the electrostatic interactions (covalent and hydrogen-bonds) to the surface, as we do not expect a significant contribution from the vdW interactions where only isolated uracil molecules are concerned (Calzolari et al. 2010; Liu et al. 2010). Since uracil contains a C=C double bond, two N–H bonds and two carbonyl bonds (C=O) (table 1), a number of different reaction scenarios may occur during adsorption. Consequently, we have explored adsorption geometries with the molecule as both keto and enol tautomers, as well as dissociative adsorption to the surface through cleavage of either the OH or NH bonds.
2. Computational methods
Geometry optimizations were performed using the VASP code (Kresse & Hafner 1994; Kresse & Furthmuller 1996). In the periodic DFT framework used, the Kohn–Sham equations have been solved by means of the generalized gradient approximation (GGA-PW91) proposed by Perdew & Wang (1992) and Perdew et al. (1992). The eigenstates of the electron wave functions were expanded on a plane-waves basis set, using pseudopotentials to describe the electron–ion interactions within the projector-augmented waves (PAW) approach (Blochl 1994). The k-point grid was set to 8×8×8 for the bulk cell with 14 Au atoms, with an energy convergence achieved at 10−2 eV. The lattice parameter after geometry optimization is 4.078 Å, which is in good agreement with the experimental value of 4.0782 Å (Haynes 2010–2011).
We have investigated the (100) surface with a slab of four layers and a 15 Å vacuum space between slabs to avoid interaction between the surfaces along the z-axis. The lower two layers in the slab were kept frozen at their bulk optimized position, whereas the top two layers and the adsorbed uracil molecule were allowed to move freely during the optimization. We have used a 2×2×1 k-mesh for a 3×3×2 supercell, which is large enough to accommodate the uracil molecule. We have tested different forms of the adsorbed uracil molecule (as shown in table 1) in different orientations to the surface (parallel, perpendicular and tilted) and we have also considered different bonding interactions to the gold atoms for the single molecule on the surface (through the carbonyl oxygens, O1 and O2; the hydrogens of the amino groups, N1(H) and N3(H); the nitrogen atoms N1 and N3, or the enol hydroxyl groups, O1(H) and O2(H); and their combinations). We have first calculated different uracil conformers in the gas phase, where we considered a molecule in a box of 15×15×15 Å with an energy cutoff of 500 eV on the Γ point. The results are reported in table 1 which show that the intact uracil molecule in the keto form is the most stable configuration in the gas phase.
In our molecular dynamics (MD) calculations, the atom positions were first relaxed at constant temperature (300 K). The time-step was set at 1.5 fs and the geometries were sampled up to 1.5 ps, using a micro-canonical ensemble to obtain a reliable image of the equilibrium geometry at 300 K. A finite number of trajectories have been considered for the keto form only. As starting points, we considered several modes of adsorption to explore all possible combinations of binding to the surface. This preliminary work allowed us to investigate local minima on the potential energy surface and to gain insight into as many realistic adsorption geometries as possible. Following the MD simulations, the most stable adsorption modes were geometry optimized to obtain the lowest energy configurations and adsorption energies. However, we found that during the MD simulations, the uracil changed neither in position nor adsorption mode or configuration from the initial configuration and it was only during energy minimization that relaxation led to changes in the adsorption structures and geometries. As such, the adsorbed enol forms of the uracil were also analysed on the basis of the MD simulations of the keto form. All the initial and final configurations, as well as the MD movies, were visualized with the Modelview software (MODELVIEW software).
The energies of adsorption (Eads) of uracil (keto and enol forms) to the Au(100) surface were calculated as follows: 2.1where E(U) and E(Au) are the total electronic energies of uracil (depending on the forms as reported in table 1) and the Au(100) surface, respectively, after separate geometry optimization, and E(U, Au) is the energy of the surface–adsorbate system when the molecule is adsorbed intact.
We have also considered NH or OH cleavage of the uracil molecule with the formation of a uracil radical, as experimentally deprotonation has been found to occur upon adsorption on gold electrodes, albeit in a solution phase: 2.2where now, for completion of the calculated energy cycle, we have calculated the interaction energy with respect to isolated uracil and hydrogen molecules, as follows: 2.3where E(Au, U•) is the total energy of the optimized geometry of the dissociated uracil on Au, E(Au) is the energy of the clean Au(100) slab after geometry optimization, E(U) is the energy of the uracil molecule and E(H2) is the energy of an isolated H2 molecule after geometry optimization which is calculated to be −6.6835 eV. Only the uracil radical with N• or O• was adsorbed on the Au surface as experimentally H is often observed to become detached from gold substrates at moderate temperatures (Sault et al. 1986; Kartusch & van Bokhoven 2009). As we have not considered solvent in this work, the desorbed uracil and hydrogen molecules are considered as isolated gas phase molecules, as is common protocol in calculations of surface adsorption processes (Irrera & Costa 2008; Jones & Jenkins 2008).
In this work, we have not considered long-range dispersion (vdW) interactions in the calculations of the gold–uracil systems, as we have considered isolated molecules adsorbed on the surface, where the surface–uracil interactions dominate and the results should not be affected by the missing dispersion energy terms. However, very recently, functionality has been implemented in the VASP code through the DFT-D2 method of Grimme (2006), to consider the vdW interactions. We have therefore also carried out pilot calculations with the DFT-D2 functionality on the lowest energy structures of the enol and keto tautomers to assess the effect of the dispersion term on the adsorption structures and energies. These pilot calculations have confirmed that when an energy correction for the dispersion forces is included the absolute adsorption energies shift upwards, but this shift is constant and does not affect either the adsorbed structures or, indeed, the differences in adsorption energies between the enol and keto tautomers. As we are interested primarily in the relative stabilities of the surface–adsorbate structures, which remain the same, the calculations without the extra vdW interaction term are therefore suitable for this study.
We have performed a systematic analysis of the adsorption of uracil on the Au(100) to elucidate the chemical bonding, the geometries and the orientations of the uracil with respect to the surface atoms, where we have aimed to reproduce and interpret a previous STM imaging study undertaken by Dretschkow et al. (1997).
First, we have analysed the intact molecule as the keto and enol tautomers in several configurations on the surface to describe the physisorbed phase, whereas the dehydrogenized molecule through NH or OH cleavage represents the chemisorbed species. The results are summarized in table 2 with the optimized adsorption geometries shown in figures 1–4, where we have used the labels presented in table 1 to identify the different atoms.
(a) Adsorption of the keto form of uracil
Our calculations show that five low-energy adsorption modes are possible for the keto-uracil on Au(100); the optimized geometries are shown in figure 1, while structural and energetic details are illustrated in table 2. Analysis indicates that an upright uracil molecule, bonding through oxygen O1 on top of the gold atoms with the N3H amino group facing the surface, is the most stable coordination mode of the keto uracil, with an adsorption energy of −0.70 eV (figure 1a). In this mode, the uracil molecule is positioned vertically to the surface and aligned in the  direction, with the O1 bonding to a gold atom on top at d=2.68 Å. The other four geometries are very close in energy and hence they can be considered competitive. Virtually equally stable, with an adsorption energy of −0.69 eV (where the difference in energy is below the accuracy of the calculations) is the configuration with the uracil tilted at roughly 60° to the surface, with the oxygen O1 at d=2.70 Å on top of a gold atom (figure 1b). When the uracil is lying flat on the surface, the adsorption energy is −0.65 eV with the closest distance of the molecule to Au at d=3.40 Å (figure 1c). The vertical position with N1H and C6H facing the substrate, with the nitrogen atom at d=3.64 Å, has an adsorption energy equal to −0.64 eV (figure 1d). Finally, when only O1 interacts with the surface (figure 1e) the adsorption energy is −0.61 eV and the oxygen O1 is found bridging two gold atoms at d=3.33 and 3.08 Å.
(b) Adsorption of the enol form of uracil
The flat geometry shown in figure 2a (N1, O2H) is the most stable structure and the lowest energy configuration of all the flat and vertical geometries of both the enol as well as the keto forms, with an adsorption energy equal to −0.93 eV. Binding to the surface occurs through the carbonyl oxygen O1 at an O1–Au distance of 2.60 Å. We have analysed the amount of charge transfer between uracil and gold for this configuration, where the molecule donates approximately 0.04 e− to the surface. Although, this donation does not indicate a covalent bond, it reveals a non-negligible interaction. The charge plot is visualized in figure 3, where the interaction of the uracil molecule’s carbonyl oxygen with the surface is shown through its overlap with a gold atom.
The same uracil form (N1, O2H) adsorbs vertically with an energy of −0.67 eV, with d(N–Au)=3.29 Å and d(O–Au)=3.88 (figure 2b). The (N3, O1H) uracil molecule adsorbs vertically with d(N–Au)=2.60 Å and Eads=−0.68 eV (figure 2c), whereas uracil in a double enol form (O1H, O2H) also adsorbs vertically with Eads=−0.66 eV at distance (figure 2d). These modes are very close in energy to, and hence competitive with, the keto forms in figure 1a,b. Geometry 2e with uracil in the form (N3, O2H) lies flat onto the surface and binds through O1 at d=2.58 Å with Eads=−0.59 eV. We note that the adsorption geometries of all enol and three out of the five keto uracils have the oxygen O1 either facing the surface or binding through it, strongly suggesting that this function is the preferred link between uracil and gold.
(c) Adsorption of dehydrogenated form of uracil
Finally, we have considered NH and OH cleavage to describe chemisorption of uracil as keto and enol tautomers to the surface, which could interpret STM images measured at sufficiently positive electrode potentials.
The two lowest energy adsorption modes are shown in figure 4 and details of the geometries and interaction energies are summarized in table 2. NH cleavage has been considered at both nitrogen atoms in the uracil, as suggested in Dretschkow et al. (1997). Cleavage of N3H between carbonyls (O1N3 ⋅ O2) and adsorption to the surface is athermic (figure 4a) with d(N–Au)=2.20 Å, d(O–Au)=2.33 and 2.43 Å, while N1H cleavage (N1 ⋅ O1) is endothermic at 0.19 eV (figure 4b) with d(N–Au)=2.29 Å and d(O–Au)=2.18 Å.
OH cleavage has also been calculated, since it was found to be the preferred way to adsorb on silicon (Lopez et al. 2002; Seino et al. 2003; Schmidt et al. 2006), but here it is found to be strongly endothermic on gold. For example, the adsorption of the (O1 ⋅ N1HN3) species was calculated to cost Ei=0.73 eV. Similarly, NH cleavage of the enol form (N1 ⋅ O1HN3) has been calculated to cost Ei=1.51 eV.
We have performed ab initio molecular dynamics simulations and energy minimization calculations to investigate the adsorption on the Au(100) surface of uracil in its tautomeric and radical forms, where we have studied the energetic and structural details of the single isolated molecule on a sufficiently large surface.
In conclusion, the intact enol uracil depicted in figure 2a is found to be the most stable surface–adsorbate structure with a low, but still relevant, charge transfer. As low potential STM images have identified that the uracil molecules lie flat on the surface (Dretschkow et al. 1997), we suggest that this adsorption geometry is compatible with the experimental findings, when the potential applied to the surface is not positive enough to induce chemisorption.
Our findings also suggest that the chemisorbed species identified by the STM measurements should be the molecule after cleavage of the amino group N3H. Even though the lowest energy chemisorbed geometry for this radical is calculated to be only athermic, it is still significantly more stable than the radical obtained by OH cleavage. Experimental findings have identified this chemisorbed species on Au(111) (Cunha et al. 2001), and on the basis of our calculations, we suggest that the same species would be present on the Au(100), although this has not as yet been confirmed from experiment.
We note from the experimental data in the literature that STM images show a significant change in the adsorption modes in correspondence to the variation of the potential applied, although our approach has not considered any electric field to attempt to reproduce the tip potential. With this proviso in mind, we consider that, among all configurations investigated, the one reported in figure 4a may be expected to be the closest to experiment. However, we aim to confirm this suggestion in the future, by extending this study to include an electric field to determine whether any structural or energetic changes would occur to the chemisorbed phase identified in this work. Indeed, the investigation of the adsorption mechanism of a single uracil on Au(100) is only our first step in the computational characterization of this system and consequently we have not as yet considered the intermolecular lateral interactions between uracils which are responsible for the ordering of the uracils in the self-assembly process on the surface. These systems are currently under study, including the implementation of the long-range dispersion interaction correction which will be essential to model the intermolecular interactions correctly.
S.I. acknowledges the Newton International Fellowship scheme for a fellowship and HECToR High Performance Computing resources via membership of the EPSRC-funded Material Chemistry Consortium (grant no. EP/D504872). The authors are grateful to Prof. G. Portalone, D. Costa, A. Roldan-Martinez and G. Jones for useful discussions. B. Diawara is acknowledged for generous use of Modelview software.
One contribution to a Special feature ‘High-performance computing in the chemistry and physics of materials’.
- Received December 17, 2010.
- Accepted January 11, 2011.
- This journal is © 2011 The Royal Society