Pablo Castro-Latorre*a,
Sebastián Miranda-Rojas*b and
Fernando Mendizabal*a
aDepartamento de Química, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile. E-mail: hagua@uchile.cl
bDepartamento de Ciencias Químicas, Facultad de Ciencias Exactas, Universidad Andres Bello, Avenida República 275, Santiago, Chile. E-mail: sebastian.miranda@unab.cl
First published on 23rd January 2020
Here we aim to explore the nature of the forces governing the adsorption of gold–phthalocyanine on gold substrates. For this, we designed computational models of metal-free phthalocyanine and gold–phthalocyanine deposited over a gold metallic surface represented by cluster models of different sizes and geometries. Thereby, we were able to determine the role of the metal center and of the size of the substrate in the interaction process. For this purpose, we worked within the framework provided by density functional theory, were the inclusion of the semi-empirical correction of the dispersion forces of Grimme's group was indispensable. It has been shown that the interaction between molecules and surfaces is ruled by van der Waals attractive forces, which determine the stabilization of the studied systems and their geometric properties. Their contribution was characterized by energy decomposition analysis and through the visualization of the dispersion interactions by means of the NCI methodology. Moreover, calculations of Density of States (DOS) showed that the molecule-surface system displays a metal–organic interface evidenced by changes in their electronic structure, in agreement with a charge transfer process found to take place between the interacting parts.
![]() | ||
Fig. 2 Structure of the interaction mode of AuPc with model1 to model4. In orange are the Au atoms with which the metal center interacts. |
System | Method | Au–Au(1)a | Au–Au(2)b | Au–M–Au | Mol.–surf.c |
---|---|---|---|---|---|
a Closest Au–Ausurf. distance.b Second closest Au–Ausurf. distance. In the case of H2Pc, this magnitude is the distance between the H atoms in the center of the molecule and the closest Au atoms in the surface.c Distance from the surface's plane to the molecule's plane. | |||||
AuPc-model1 | PBE | 3.86 | 3.90 | 43.7° | 3.34 |
PBE-D3 | 3.54 | 3.60 | 46.8° | 3.18 | |
H2Pc-model1 | PBE | 3.74 | 3.80 | 3.46 | |
PBE-D3 | 3.45 | 3.53 | 3.32 | ||
AuPc-model2 | PBE | 3.81 | 4.04 | 43.1° | 3.60 |
PBE-D3 | 3.52 | 3.77 | 45.8° | 3.30 | |
H2Pc-model2 | PBE | 3.71 | 3.97 | 3.53 | |
PBE-D3 | 3.32 | 3.56 | 3.27 | ||
AuPc-model3 | PBE-D3 | 3.61 | 3.63 | 47.0° | 3.34 |
H2Pc-model3 | PBE-D3 | 3.53 | 3.54 | 3.27 | |
AuPc-model4 | PBE-D3 | 3.54 | 3.73 | 46.6° | 3.37 |
H2Pc-model4 | PBE-D3 | 3.38 | 3.43 | 3.30 |
The effect of the inclusion of the dispersion correction was assessed using model1 and model2. According to the results, after its incorporation there is a decrease in the interaction distances between the molecule and the surface, evidencing the relevance of dispersion forces in this interaction. The results obtained for model2 with PBE-D3 showed good agreement with the experimental STM images and theoretical results using a periodic approach available for CuPc, with distances between 3 and 3.25 Å.14–16,19 The shortening in the metal-gold distance has been observed in MPc–gold systems (M = Fe, Cu, Co) previously studied with gold clusters of the same size. The only difference is that MPc–Au distances are shorter than systems with AuPc–Au.16
The interaction energies for the systems optimized without dispersion contribution are listed in Table S1 in ESI,† results that reveal a repulsive character for this interaction. The only system with a minor stabilizing contribution was AuPc–Au58 using PBE with an interaction energy of −6.3 kcal mol−1. These results point out the relevance of the proper description of the dispersion contribution in order to properly represent the stabilizing forces that lead to the formation of these complexes. The interaction energies calculated after the incorporation of the dispersion correction are listed in Table 2 together with the respective EDA results (model1 results are in Table S2†). The interaction energies between AuPc and model2 was of −89.0 kcal mol−1. This is in line with results from periodic calculations that quantify the interaction energy for CuPc in approximately 80 kcal mol−1 using PBE functional with vdWsurf corrections.19 The fact that the dispersion term is higher than the interaction energies with both levels of theory, points out that the dispersion forces are the main stabilizing contribution to the formation of these complexes. This also explains the decrease in the interaction distances after the incorporation of the dispersion correction as detailed above. Regarding the role of the metal center in the interaction strength, the comparison between AuPc-model2 and H2Pc-model2 showed an increase between 17.3 kcal mol−1. According to these results, we calculated a contribution of ca. 19% of the interaction energy provided by the inclusion of Au as the metal center. These results are comparable with the MPc complexes (M = Fe, Co, Cu) on Aun (n = 26, 58). The results showed that dispersion forces rule the MPc–gold interaction, with binding strengths ranging between 61 and 153 kcal mol−1.16 For complexes with MPc (M = Fe, Co and Cu) the interaction energy is higher compared to the AuPc on the same gold cluster. See the Table S3† in ESI.† Such a difference occurs because the MPc complexes have shorter distances over the gold surface. Comparing the resulting interaction energies of AuPc with model2 and model3, an increase of 12.3 kcal mol−1 is observed in the interaction energies. A minor part of that increase comes from the dispersion energy that is related to the effective area of interaction, while the rest of the difference in the interaction comes from a great reduction in the electrostatic and polarization terms of the interaction.
System | ΔETOT | ΔEELECT | ΔEORB–RELAX | ΔEDISP | ΔEXC | ΔEEXCH–REPUL |
---|---|---|---|---|---|---|
AuPc-model2 | −89.0 | 50.8 | 0.0 | −86.2 | −76.4 | 22.8 |
H2Pc-model2 | −71.7 | 114.1 | 0.0 | −83.3 | −79.4 | −23.2 |
AuPc-model3 | −101.3 | 26.4 | −0.1 | −88.5 | −75.0 | 35.9 |
H2Pc-model3 | −76.8 | 98.9 | −6.3 | −86.7 | −76.7 | −6.0 |
AuPc-model4 | −100.1 | 24.0 | −0.1 | −89.4 | −74.9 | 40.3 |
H2Pc-model4 | −76.6 | 104.6 | −0.1 | −87.8 | −76.2 | −17.1 |
The results obtained from the EDA calculations corroborate the importance of weak interactions, as can be seen from the dispersion term showed in Table 2. The ex-rep and OICT terms from eqn (1) presented in the Theoretical and computational details section are not included since the interaction does not involve covalent bonds and therefore, the value of such terms according to this EDA approach is 0. Our results reinforce the principle in which the stabilization of systems involving physisorption over a metallic surface is a process driven by weak van der Waals interactions. As a consequence, most of the stabilization is additive in nature and depends on the size of the extended area of contact between the interacting parts. Thereby, the size of the model representing a Au(111) system (model2) plays an important role regarding metal–π interactions, since model2 capture to its full extent the interaction between the molecule and the surface. This can be evidenced from the analysis of the increase in interaction strength after increasing the cluster size from model1 to model2 (results for model1 are shown in Table S4†). Within the context of the used EDA scheme, the electrostatic term represents the classical electronic repulsion, in line with the expected electronic repulsion added after the inclusion of the metal center that causes a decrease both in the electrostatic repulsion and the polarization stabilization. The model3 exhibit only a minor increase for the dispersion term, while the difference in the interaction energy with the layered model comes mainly from the drastic decrease in the electrostatic and polarization terms, thus revealing that edge effects have no influence on the interaction energies.
We analyzed the differences in the charge transfer process by using the natural population analysis (NPA), results listed in Table 3. Since the interacting parts are neutral, this analysis gives information about the charge transfer between the subunits after forming the complex. Interestingly, this only occurred when the Au was incorporated at the center of the Pc scaffold. These results reveal the role of Au as the metal center in the binding nature as an indispensable chemical block able to assist the charge transfer process with direction from the molecule to the gold substrate and are in agreement with experimental evidence of CuPc37 and previous theoretical results21 regarding the direction of the process. For other metal centers such as Fe or Ni, the charge transfer process goes from the gold substrate towards the molecule, showing that AuPc with its metal center with d9 electronic configuration follows the trend of CuPc in the charge transfer process.
System | Method | Au-cluster | Au | Pc |
---|---|---|---|---|
AuPc | PBE-D3 | 1.21 | −1.21 | |
AuPc-model2 | PBE-D3 | −0.71 | 1.27 | −0.57 |
H2Pc-model2 | PBE-D3 | −0.01 | 0.01 | |
AuPc-model3 | PBE-D3 | −0.88 | 1.27 | −0.38 |
H2Pc-model3 | PBE-D3 | −0.10 | 0.10 | |
AuPc-model4 | PBE-D3 | −0.87 | 1.27 | −0.40 |
H2Pc-model4 | PBE-D3 | −0.04 | 0.04 |
In general, the interaction between AuPc and the four models is characterized by a donor–acceptor behavior, with the substrate (Au cluster) as the electron acceptor. According to this, the decrease in the electrostatic repulsion observed for AuPc-model2 is a consequence of the transfer of electronic charge between the fragments that allows to decrease the electronic overlapping between the subunits, which also leads to a decrease in the polarization of the fragments. To understand the nature of the charge transfer process that seems to be the second main source of stabilization besides the dispersion contribution, we calculated the fragmental electronic chemical potential as described in the methodology section. This chemical descriptor provides a measure of the electron escape tendency from the system, being a suitable property to understand the directionality of charge transfer process. Being the μ defined as the partial derivative of the energy with respect to the change in the number of electrons at constant external potential, the more negative the value of μ, the larger the change-decrease (negative slope) in energy under the addition of an electron; and therefore, the higher the stabilization towards receiving an electron. Thus, the differences between the μ of the subunits of the complex not only provides the tendency towards charge transfer, but its sign provide information about the direction of the process. We arbitrarily defined Δμ as the difference between the electronic chemical potential from the substrate (μSUBS) and the ligands (μAuPc and μH2Pc). A simple comparison between AuPc and H2Pc from the data listed in Table 4 (results for model1 are in Table S5†) shows that the former has a higher tendency towards donating electronic charge. Then, after comparing the μ of AuPc and H2Pc with the μSUBS of the substrate, the largest value of Δμ was obtained with AuPc, indicating that the AuPc–gold cluster systems will carry out more charge transfer than the H2Pc–gold cluster complex. The negative sign of Δμ in both types of complexes indicates that the gold substrate has a higher μSUBS than both ligands, thus being the most suitable electron acceptor of the complex. Actually, for the complexes with H2Pc the charge transfer was close to zero indicating that probably there is a threshold in which the difference between the electronic chemical potentials may trigger the electron transfer process. According to our results, PBE-D3 level of theory is able to provide a proper description of the electronic phenomenon involved in the AuPc–gold interaction.
System | Method | μAu-substrate | μAuPc | Δμ |
---|---|---|---|---|
AuPc-model2 | PBE-D3 | −5.11 | −3.71 | −1.40 |
H2Pc-model2 | PBE-D3 | −5.10 | −4.30 | −0.80 |
AuPc-model3 | PBE-D3 | −5.41 | −3.70 | −1.71 |
H2Pc-model3 | PBE-D3 | −5.41 | −4.29 | −1.12 |
AuPc-model4 | PBE-D3 | −5.29 | −3.70 | −1.59 |
H2Pc-model4 | PBE-D3 | −5.29 | −4.30 | −0.99 |
To characterize the nature of the non-covalent interactions arising at the interface between AuPc/H2Pc and the surface of the gold substrate, we calculated the non-covalent interaction index (NCI), results shown in Fig. 3. This index allows visualizing different types of interactions in real space through a color code. As identified by the green areas from the NCI surface, the main stabilizing contribution to the non-covalent part is located at the interatomic space between the AuPc/H2Pc and the gold surface atoms. While most of the stabilizing interactions are close to the center region of the interface; at the outer region of the macrocycle, where the electron delocalization takes part, there is a slightly destabilizing interaction located between the delocalized π-bonds from the phenyl rings and the gold surface. The addition of the gold atom at the center of the Pc incorporates a stabilizing interaction with the gold surface that accounts for the increase in the dispersion contribution after the inclusion of the gold center. Another important aspect is that the contribution of the gold atom of AuPc is related to the dispersion term, a characteristic feature of aurophilic interaction, although in this case is between a gold center of oxidation state +2 and a neutral metallic Au surface, which is 2.9 kcal mol−1 with PBE-D3 by model 2. As stated above, the van der Waals force dominates the interaction between these molecules and the surface, and it is related to the area that the molecules occupy over the surface.
![]() | ||
Fig. 3 NCI index surface representation (isovalue = 0.5) of the (a) H2Pc–Au58 and (b) AuPc–Au58 complexes. |
In order to study the electronic structure of the metal–organic interface,2,18,37–43 density of states (DOS) calculations were carried out to determine the effects of the surface–molecule interaction. In Fig. 4 is displayed a DOS diagram of the interacting system AuPc–Au58 (green line) along with a diagram of sub-systems AuPc (blue line) and Au(111) surface (red line). Also is displayed the sum of the diagrams of the separate sub-systems (yellow line) to compare with the DOS of the interacting system. The zero at the diagram is set to be the Fermi level (EF) of the Au(111) surface cluster model, indicated by the vertical black line. The broad peak of the Fermi level of AuPc is due to the degeneracy of a d9 electronic configuration in a square–planar coordination environment. It is observed that the peak corresponding to the Fermi level of the AuPc (indicated by the arrow) is shifted by ∼0.9 eV towards the Fermi level of the Au(111) in the interacting system. This is appreciated by a decrease in the DOS around the EF of AuPc when comparing the sum of the isolated sub-systems with the interacting system, and an increase in the DOS at the EF of the Au(111) substrate in the interacting system. The Energy Level Alignment43 is a characteristic feature of an organic molecule or polymer interacting with a metallic substrate, in which the electronic structure of the interface and it's principal properties such as charge transfer character is determined by electrostatic and polarization components of the interaction.44 This is useful to rationalize the magnitudes of these terms provided by the EDA calculation discussed above, since they provide a wave-function based description of the electronic structure of the interacting systems. It is an interesting feature that this system exhibit such a displacement on the density of states given that is a weakly interacting one, on a physisorption regime based on π-molecules on a clean metal surface where its stability is driven mainly from van der Waals forces.45 However, this interaction is not easily compassed on basic interface models such as the integer charge-transfer model (ICT) or induced density of interfacial states (IDIS)46 because of the presence of d orbitals from the metal center. Nevertheless, the Fermi level shift in the diagram is in accordance with the calculated difference of electronic chemical potential as discussed in Table 4, related to the direction of charge transfer, thus showing that the energy level alignment is related to the equilibration of the chemical potential, as has been suggested by previous work on the field.47
A DOS diagram of AuPc–Au58 using the flat cluster was also drafted (Fig. S1, ESI†) and it highlights the importance of a layered cluster model to reproduce the electronic structure of the interface. This can be explained by the fact that the cluster model fails to account for the electrostatic and polarization terms (Table 2) that define the metal–organic interface and the energy level alignment. The importance of those terms in the interface are also described by charge rearrangements and cumulative charge transfer in periodic calculations,19 and is supportive of the need of a layered cluster to accurately describe the interface electronic structure.
The DOS diagram of the layered model2 (Fig. S2†) resembles the one obtained with periodic methods of calculation.48,49 Another important aspect is that the dispersion correction does not modify the DOS diagram. This is an expected result since the dispersion correction is semi-empirical and it does not modify the wave function of the systems being considered and thus, the energy level alignment displayed by this system may be regarded as a feature that arise from the geometry of the Au(111) cluster rather than the consideration of van der Waals forces, thus showing that the cluster model is able to reproduce some important properties regarding molecule–surface interactions.
All calculations were carried out within the Density Functional Theory framework, using GGA PBE(Perdew–Burke–Ernzenhof)52 exchange–correlation functionals as implemented in the Turbomole package.54,55 This functional was chosen to be consistent with previous publications regarding surface chemistry and related phenomena. The meta-GGA TPSS (Tao, Perdew, Staroverov, Scuseria)53 functional was also used to obtain a comparison in the description of surface chemistry with the more broadly used and more validated PBE functional.21,56 The comparison is provided in the ESI.† Since we are interested in the supramolecular assemblies that this molecules form over a surface, the inclusion of dispersion interactions is fundamental to accurately describe the non-covalent interaction between the macrocycle and the Au(111) surface. This is incorporated using the “DFT-D3” methodology.57 For Au, Stuttgart pseudo relativistic effective core potentials (ECP) were used, with 19 valence electrons.58 Two f-type polarization functions were added Au (αf = 0.20, 1.19).21 All atoms were treated with Ahlrichs type def2-TZVP basis set,59 adding one d-type polarization function for the optimizations, and subsequently the energies were corrected by single point calculations using the def2-TZVP basis set.
The counterpoise correction has been used to avoid the basis set superposition error (BSSE) in the calculation of interaction energies, through single point calculations, based on the optimized geometry obtained.60 To gain more insight on the interaction between the macrocycle and the surface, energy decomposition analysis (EDA) calculations were performed.61,62 This methodology expresses the interaction energy in different physical components, comprising an electrostatic term (ES), the exchange–repulsion term (EXR), a term labelled orbital interaction and charge transfer (OICT), polarization contribution (POL) and the dispersion term (DISP), as expressed in eqn (1).
ΔE = EES + EEXR + EOICT + EPOL + EDISP | (1) |
Also, density of states (DOS) calculations were performed to obtain information regarding the electronic structure of the systems. DOS has been built via a Gaussian distribution of the one-electron eigenstates obtained by the DFT methods described above, and considering a bandwidth of 0.27 eV.
To characterize the intermolecular interactions, the non-covalent interaction index (NCI) was used.63 This is a qualitative tool for describing attractive and repulsive interactions in real space, based on the reduced density gradient (RDG) and can be interpreted by color coded isosurfaces that represent those interactions. Hydrogen bonds are labelled as strong interactions and are represented by blue color. Weak van der Waals interactions are shown in green color, while repulsive steric interactions are displayed in red color. Natural Population Analysis (NPA)64 calculations were used to study the charge transfer processes that may be involved in these systems. Finally, the fragmental electronic chemical potential was calculated according to the following equation:
μ = (I + A)/2 = (εLUMO + εHOMO)/2 | (2) |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9ra07959a |
This journal is © The Royal Society of Chemistry 2020 |