Asymmetric Response toward Molecular Fluorination in Binary Copper

Jul 15, 2014 - We report a didactic and simple example of the subtleness in the balance of intermolecular and molecule–substrate interactions and it...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/JPCC

Asymmetric Response toward Molecular Fluorination in Binary Copper−Phthalocyanine/Pentacene Assemblies D. G. de Oteyza,*,†,‡ J. M. Garcia-Lastra,§ E. Goiri,† A. El-Sayed,∥ Y. Wakayama,⊥ and J. E. Ortega†,‡,∥ †

Donostia International Physics Center, Paseo Manuel Lardizabal 4, 20018 San Sebastián, Spain Centro de Física de Materiales, Materials Physics Center, Paseo Manuel Lardizabal 5, 20018 San Sebastián, Spain § Department of Energy Conversion and Storage, Technical University of Denmark, DK-4000 Roskilde, Denmark ∥ Departamento de Física Aplicada I, Universidad del Pais Vasco, Plaza Oñate 2, 20018 San Sebastián, Spain ⊥ International Center of Materials Nanoarchitectonics, National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan ‡

S Supporting Information *

ABSTRACT: We report a didactic and simple example of the subtleness in the balance of intermolecular and molecule−substrate interactions and its effect on molecular self-assembly. The study is performed on two closely related molecular blends of copper phthalocyanines and pentacene, in each of which one of the two molecules is fluorinated. Reversing the fluorination brings about changes in the intermolecular hydrogen bonds, as well as in the interactions with the substrate. As a result, on Au(100) substrates one blend assembles into a crystalline structure, whereas the other, displaying weaker intermolecular interactions and a larger corrugation in the molecule−substrate interaction potential, results in a disordered layer. However, the difference between the two blend’s structures vanishes when substrates with less corrugated interaction potentials are used.



INTRODUCTION Organic semiconducting materials self-assembled into engineered functional nanostructures are envisioned as a cheap and promising alternative to the inorganic semiconductors used in conventional optoelectronic devices. The challenge is to reach a sufficient understanding of the self-assembly processes to allow us to synthesize rationally designed supramolecular structures.1,2 Molecular blends provide particularly attractive model systems, since multicomponent assemblies increase the possibilities of the supramolecular systems in terms of growth and structure, as well as of their electronic properties and ultimate functionality.3 Among the different noncovalent intermolecular interactions, hydrogen bonding is probably most widely used to steer the self-assembly of molecular building blocks into a variety of structures, ranging from extended two-dimensional layers4,5 to nanoporous6,7 or linear structures.8,9 A particular family of H-bonded multicomponent molecular blends is that in which fluorinated and nonfluorinated molecules are combined.3,4,10−12 Fluorination increases a molecule’s ionization potential and electron affinity,13 often turning organic semiconductors from p-type to n-type materials. Many of these blends are therefore nanostructured donor−acceptor systems, thus with an inherent relevance not only to fundamental studies but also to applications. In this work we focus on molecular blends comprising copper phthalocyanines and pentacene. In particular, we © 2014 American Chemical Society

compare combinations of perfluorinated copper phthalocyanines (F16CuPc) and pentacene (PEN) with those of copper phthalocyanine (CuPc) and perfluoropentacene (PFP). We use scanning tunneling microscopy (STM) to analyze the effects of reversed fluorination on the resulting self-assembled structures. Density functional theory (DFT) calculations provide a coherent explanation of the experimental findings, invoking the differences in the balance of intermolecular and molecule− substrate interactions between the different studied systems.



RESULTS AND DISCUSSION Figure 1 shows the chemical structure of the involved molecules, as well as STM images of the two different blends with molecular ratios close to a stoichiometric 1:1 mixture at submonolayer coverage upon deposition on a reconstructed Au(100) surface. The CuPc:PFP blend distributes over the surface in a disordered manner. No trace of the original surface reconstruction is observed anywhere, including the fractions of uncovered surface in between molecules (see inset in Figure 1). We conclude that the system presents significant molecule− substrate interactions that lift the surface reconstruction,14 at least at coverages above 0.75 ML as probed here, and comparatively weak intermolecular interactions that are unable Received: June 20, 2014 Revised: July 15, 2014 Published: July 15, 2014 18626

dx.doi.org/10.1021/jp506151j | J. Phys. Chem. C 2014, 118, 18626−18630

The Journal of Physical Chemistry C

Article

Figure 2. Unit cell of the energetically optimized molecular configurations for CuPc:PFP blends (left) and F16CuPc:PEN blends (right) modeled in a free-standing layer. Inset: hydrogen-bond length variation (d1 vs d2) upon reversed fluorination in linear and nonlinear bond geometries.

and of 0.64 eV for CuPc:PFP, showing the F16CuPc:PEN structure to be more stable than CuPc:PFP by about 0.31 eV/ cell.16 This surprising difference can be understood through a simple electrostatic model. We consider each C−F bond as a dipole consisting of two point charges of opposite sign separated by 1.34 Å (the typical C−F bond distance).17 We do the same for each C−H bond, the distance between the point charges being now 1.09 Å.17 We now sum over the electrostatic interactions between all C−F and C−H dipoles in the unit cell and find dipole−dipole interactions are larger by 0.19 eV/cell in the F16CuPc:PEN blend than in CuPc:PFP. This simple model is thus in good qualitative agreement with the DFT results and explains why the F16CuPc:PEN blend is more stable than that of CuPc:PFP. It could seem counterintuitive that a permutation between C−F and C−H bonds changes the magnitude of their dipole−dipole interaction. This is clarified in the inset of Figure 2, showing that the H-bond length, which is inversely proportional to its binding energy,18 always changes under a permutation unless the C−F and C−H bonds are aligned. Nevertheless, the H-bonds in this system are still attractive and lead, for free-standing layers, to a crystalline structure similar to that of F16CuPc:PEN. The reason behind the CuPc:PFP blend’s disorder must therefore additionally involve the molecule−substrate interactions. The largest contribution to the latter is expected to come from the phthalocyanines, which are on the one hand significantly larger than their pentacene counterparts and on the other hand include a central Cu atom that typically contributes significantly to the interactions with the substrate. We therefore calculate and compare F16CuPc and CuPc’s interactions with a Au(100) surface. The molecules have been placed on the surface with their central Cu atom on “top”, “hollow”, and “bridge” sites, and for each of those positions four different orientations are chosen so as to provide the best possible sampling taking into account substrate and molecule symmetry considerations. The results are summarized in Figure 3 and clearly evidence that the adsorption energy and the interaction potential corrugation are significantly higher for the nonfluorinated molecule.19 Excessively strong corrugation in the molecule−substrate interactions can have a negative impact on molecular selfassembly. By forcing the molecules into positions and orientations mainly determined by the substrate, it might

Figure 1. (a) Schematic representation of F16CuPc and PEN, as well as CuPc and PFP. (b) STM images (30 × 30 nm2, top, and 10 × 10 nm2, bottom) of submonolayer molecular blends of F16CuPc with PEN (left) and CuPc with PFP (right) in approximately 1:1 ratios deposited on a reconstructed Au(100) surface. The 3 × 3 nm2 inset images display (partially) uncovered substrate regions. Imaging parameters are in the current and voltage range of I = 80−600 pA and V = 0.1−0.8 V.

to drive the formation of ordered supramolecular structures. When changing to F16CuPc:PEN blends, the scenario is quite different: molecules coalesce into islands displaying long-range crystalline order with unit cell parameters a = 22 ± 2 Å, b = 30 ± 2 Å, and γ = 90 ± 4°. The Au(100) surface reconstruction can be observed in the uncovered substrate regions (left side of large-scale image and inset in Figure 1). Under the molecular islands the reconstruction is no longer evident and possibly lifted, although this is difficult to ascertain from our STM images because of the similar periodicity of the molecular overlayer. To understand why stoichiometric 1:1 blends of F16CuPc:PEN and CuPc:PFP form ordered and disordered blends on Au(100), respectively,15 we have performed DFT calculations on free-standing layers of either blend. In both cases the energetically favored configuration is the same, corresponding to a crystalline structure with a rectangular unit cell comprising four molecules, each of which is surrounded by molecules of the opposite species in such a way that the C−H··· F interactions are maximized (Figure 2). In spite of sharing the same unit cell dimensions (a = 23 Å × b = 30.5 Å, γ = 90°) and configuration of the molecular base, our calculations predict total binding energies per unit cell of 0.95 eV for F16CuPc:PEN 18627

dx.doi.org/10.1021/jp506151j | J. Phys. Chem. C 2014, 118, 18626−18630

The Journal of Physical Chemistry C

Article

Figure 3. Calculated adsorption energies of CuPc (red) and F16CuPc (blue) molecules on Au(100) as a function of adsorption site and molecular orientation. Shaded regions span the corrugation range of the respective molecule−substrate interaction potentials. Angle 0 is referred to an azimuthal molecular orientation with a phthalocyanine diagonal parallel to a nearest-neighbor high-symmetry substrate direction. Figure 4. The 9.5 × 9.5 nm2 STM images of the resulting structures of CuPc:PFP (left) and F16CuPc:PEN (right) blends on Au(111) (top) and Ag(111) (bottom) surfaces. Imaging parameters are I = 640 pA and V = −0.21 V; I = 60 pA and V = 0.07 V; I = 750 pA and V = −0.47 V; I = 109 pA and V = −0.21 V for CuPc:PFP/Au, F16CuPc:PEN/Au, CuPc:PFP/Ag, and F16CuPc:PEN/Ag, respectively.

occur that the intermolecular interactions within the molecular layers become seriously compromised. And because intermolecular interactions are a sine qua non requisite for the growth of ordered structures, formation of crystalline multicomponent assemblies can be hampered for this reason. Thus, comparing both mixes, we find that the lower intermolecular interactions in the CuPc:PFP blend (lower by 30% even under optimized intermolecular interactions in free-standing layers) and the much higher molecule−substrate corrugation of the most interacting molecule (1.19 eV for CuPc vs 0.34 eV for FCuPc) provide an intuitive explanation to the disordered layer structure. Counterexamples are found, e.g., on the Au(111) surface (Figure 4), on which both molecular blends show the same crystalline structure, similar to that of their calculated freestanding layers and of F16CuPc:PEN blends on Au(100).12 On the denser (111) surface structure, the corrugation of the molecule−surface potential is expected to be lower, thereby facilitating a molecular adjustment that promotes intermolecular interactions. However, the presence of the Au(111) herringbone reconstruction, which remains underneath the molecular blend overlayers, complicates the associated calculations. Instead, we do the calculations on the Ag(111) surface, on which the same scenario applies (Figure 4), with both molecular blends displaying analogous crystalline structures to those shown above (measured unit cell parameters all summarized in Table 1 together with the calculated freestanding layer parameters).3,20,21 The calculation results are summarized in Figure 5. In line with the results on Au(100), as well as with evidence from previous studies,3,14 the fluorinated molecules show a weaker interaction with these metallic substrates than the nonfluorinated counterparts. But most importantly we confirm that, despite silver being generally more reactive than gold,22 the interaction potential corrugation of CuPc on Ag(111) is about 60% lower than on Au(100) (0.49 vs 1.19 eV). This explains, in a qualitative and simple way, why the CuPc:PFP blends present a disordered structure on Au(100) and a crystalline structure on Ag(111).

Table 1. Unit Cell Parameters for the Various Measured and Calculated Stoichiometric 1:1 Blends F16CuPc:PEN

CuPc:PFP

a

calcd Au(100) Au(111)a Ag(111) calcd Au(100) Au(111) Ag(111)b

a (Å)

b (Å)

γ (deg)

23 22 ± 2 22.5 ± 2 24 ± 2 23

30.5 30 ± 2 28.5 ± 2 27 ± 2 30.5

90 90 ± 4 90 ± 3 92 ± 5 90

22 ± 2 22 ± 2

29 ± 3 29 ± 1

90 ± 3 89 ± 6

From ref 20. bFrom ref 21.



CONCLUSIONS Altogether, our results provide a didactic example of the role of the delicate balance between intermolecular and molecule− substrate interactions, as well as of the consequences brought about by its modification. By comparing closely related blend systems with slightly different intermolecular interactions on the same substrate we show how the resulting structures can vary from a crystalline to a disordered layer as the intermolecular interactions are lowered and the corrugation of the molecule−substrate interaction potential is increased. By switching to more compact metal surfaces, the impact of the substrate is reduced due to the lower corrugation of its interaction potential and both blends are able to form crystalline layers.



METHODS The Au(100), Au(111), and Ag(111) surfaces were prepared by standard sputtering and annealing cycles, and their cleanliness was checked by STM prior to molecular deposition. Deposition took place from resistively heated Knudsen cells at temper18628

dx.doi.org/10.1021/jp506151j | J. Phys. Chem. C 2014, 118, 18626−18630

The Journal of Physical Chemistry C

Article

Competitiveness under projects FIS2012-30996 and FIS201021282-C02-01 and from the ReLiable project funded by the Danish Council for Strategic ResearchProgramme Commission on Sustainable Energy and Environment (project no. 11116792).



Figure 5. Calculated adsorption energies of CuPc (red) and F16CuPc (blue) molecules on Ag(111) as a function of adsorption site and molecular orientation [energy scale common with that used for Au(100) in Figure 3]. Shaded regions span the corrugation range of the respective molecule−substrate interaction potentials. Angle 0 is referred to an azimuthal molecular orientation with a phthalocyanine diagonal parallel to a nearest-neighbor high-symmetry substrate direction.

atures around 380 and 190 °C for the phthalocyanines and acenes, respectively, onto single-crystal surfaces held at room temperature. Scanning tunneling microscopy measurements were performed with commercial systems (JEOL and Omicron VT) at room temperature and in constant current mode. The analysis of the STM images was performed with the freeware WSxM from Nanotec.23 Density functional theory calculations were performed at the local density approximation (LDA) level, using Perdew− Zunger functional.24 For the molecule adsorption energies, the lateral size of the supercell is 20.2 Å × 20.2 Å, ensuring that the interactions among the molecular replicas are negligible (the minimum distance between two atoms of two different replicas is always larger than 4.5 Å). The surface has been modeled with a slab comprising three atomic layers (of which only the upper layer is relaxed in the calculations upon molecular adsorption). Our choice of the LDA Perdew−Zunger functional is justified from our previous study of polyacenes (and their fluorinated counterparts) adsorbed on metallic surfaces, which showed a better performance by LDA as compared to GGA or van der Waals functionals.25



ASSOCIATED CONTENT

* Supporting Information S

STM images of the pristine Au(100) surface and the resulting structures obtained upon submonolayer deposition of CuPc:PFP and F16CuPc:PEN blends. This material is available free of charge via the Internet at http://pubs.acs.org.



REFERENCES

(1) Elemans, J. A. A. W.; Lei, S.; De Feyter, S. Molecular and Supramolecular Networks on Surfaces: From Two-Dimensional Crystal Engineering to Reactivity. Angew. Chem., Int. Ed. 2009, 48, 7298−7332. (2) Bartels, L. Tailoring Molecular Layers at Metal Surfaces. Nat. Chem. 2010, 2, 87−95. (3) El-Sayed, A.; Borghetti, P.; Goiri, E.; Rogero, C.; Floreano, L.; Lovat, G.; Mowbray, D. J.; Cabellos, J. L.; Wakayama, Y.; Rubio, A.; et al. Understanding Energy-Level Alignment in DonorAcceptor/ Metal Interfaces from Core-Level Shifts. ACS Nano 2013, 7, 6914− 6920. (4) Hipps, K. W.; Scudiero, L.; Barlow, D. E.; Cooke, M. P. A SelfOrganized 2-Dimensional Bifunctional Structure Formed by Supramolecular Design. J. Am. Chem. Soc. 2002, 124, 2126−2127. (5) Chen, W.; Li, H.; Huang, H.; Fu, Y.; Zhang, H. L.; Ma, J.; Wee, A. T. S. Two-Dimensional Pentacene:3,4,9,10-Perylenetetracarboxylic Dianhydride Supramolecular Chiral Networks on Ag(111). J. Am. Chem. Soc. 2008, 130, 12285−12289. (6) Theobald, J. A.; Oxtoby, N. S.; Phillips, M. A.; Champness, N. R.; Beton, P. H. Controlling Molecular Deposition and Layer Structure with Supramolecular Surface Assemblies. Nature 2003, 424, 1029. (7) Perdigao, L. M. A.; Champness, N. R.; Beton, P. H. Surface SelfAssembly of the Cyanuric Acid−Melamine Hydrogen Bonded Network. Chem. Commun. 2006, 538−540. (8) Cañas-Ventura, M. E.; Xiao, W.; Wasserfallen, D.; Müllen, K.; Brune, H.; Barth, J. V.; Fasel, R. Self-Assembly of Periodic Bicomponent Wires and Ribbons. Angew. Chem., Int. Ed. 2007, 46, 1814−1818. (9) Ruiz-Oses, M.; Gonzalez-Lakunza, N.; Silanes, I.; Gourdon, A.; Arnau, A.; Ortega, J. E. Self-Assembly of Heterogeneous Supramolecular Structures with Uniaxial Anisotropy. J. Phys. Chem. B 2006, 110, 25573−25577. (10) Barrena, E.; de Oteyza, D. G.; Dosch, H.; Wakayama, Y. 2D Supramolecular Self-Assembly of Binary Organic Monolayers. ChemPhysChem 2007, 8, 1915−1918. (11) Huang, Y. L.; Chen, W.; Li, H.; Ma, J.; Pflaum, J.; Wee, A. T. S. Tunable Two-Dimensional Binary Molecular Networks. Small 2010, 6, 70−75. (12) Wakayama, Y.; de Oteyza, D. G.; Garcia-Lastra, J. M.; Mowbray, D. J. Solid-State Reactions in Binary Molecular Assemblies of F16CuPc and Pentacene. ACS Nano 2011, 5, 581−589. (13) Tang, M. L.; Bao, Z. Halogenated Materials as Organic Semiconductors. Chem. Mater. 2011, 23, 446−455. (14) Lo, Y.-Y.; Chang, J.-H.; Hoffmann, G.; Su, W.-B.; Wu, C.; Chang, C.-S. A Comparative Study on the Adsorption Behavior of Pentacene and Perfluoropentacene Molecules on Au(111) Surfaces. Jpn. J. Appl. Phys. 2013, 52, 101601−1−6. (15) Although throughout the article we focus on blends with a 1:1 molecular ratio, similar findings are in fact found with pentacene excess (whether fluorinated or not) in a 1:2 ratio (Supporting Information Figure S1). (16) We repeated the study of the free-standing layers using GGA− PBE and van der Waals functionals, obtaining the same difference (∼0.3 eV/cell) in the binding energy as in the LDA calculations. (17) Allen, F. H.; Watson, D. G.; Brammer, L.; Orpen, A. G.; Taylor, R. Typical Interatomic Distances: Organic Compounds. Int. Tables Crystallogr. 2006, C, 790−811. (18) Steiner, T. The Hydrogen Bond in the Solid State. Angew. Chem., Int. Ed. 2002, 41, 48−76. (19) The adsorption energy has been calculated using gas-phase molecules and an unreconstructed Au(100) surface as reference.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +34 943018820. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the Spanish Grant Nos. MAT2010-21156-C03- and PIB2010US-00652 and the Basque Government Grant No. IT-621-13. J.M.G.-L. acknowledges support from the Spanish Ministry of Economy and 18629

dx.doi.org/10.1021/jp506151j | J. Phys. Chem. C 2014, 118, 18626−18630

The Journal of Physical Chemistry C

Article

Therefore, while it is legitimate to compare the energies of both molecules, the absolute adsorption energy values should be lower due to the reconstruction of the pristine Au(100) surface. That is, values in the graph should be rigidly shifted downwards by an amount that corresponds to the Au(100) reconstruction energy. This, however, does not change the interaction corrugation, which is most important for the present work. (20) El-Sayed, A.; Mowbray, D. J.; Garcia-Lastra, J. M.; Rogero, C.; Goiri, E.; Borghetti, P.; Turak, A.; Doyle, B. P.; Dell’Angela, M.; Floreano, L.; et al. Supramolecular Environment-Dependent Electronic Properties of Metal−Organic Interfaces. J. Phys. Chem. C 2012, 116, 4780−4785. (21) Goiri, E.; Matena, M.; El-Sayed, A.; Lobo-Checa, J.; Borghetti, P.; Rogero, C.; Detlefs, B.; Duvernay, J.; Ortega, J. E.; Oteyza, D. G. Self-Assembly of Bicomponent Molecular Monolayers: Adsorption Height Changes and Their Consequences. Phys. Rev. Lett. 2014, 112, 117602−1−5. (22) Hammer, B.; Norskov, J. K. Why Gold is the Noblest of All Metals. Nature 1995, 376, 238−240. (23) Horcas, I.; Fernandez, R.; Gomez-Rodriguez, J. M.; Colchero, J.; Gomez-Herrero, J.; Baro, A. M. WSxM: A Software for Scanning Probe Microscopy and a Tool for Nanotechnology. Rev. Sci. Instrum. 2007, 78, 013705−1−8. (24) Perdew, J. P.; Zunger, A. Self-Interaction Correction to DensityFunctional Approximations for Many-Electron Systems. Phys. Rev. B 1981, 23, 5048−5079. (25) Goiri, E.; García-Lastra, J. M.; Corso, M.; Adb El-Fattah, Z. M.; Ortega, J. E.; de Oteyza, D. G. Understanding Periodic Dislocations in 2D Supramolecular Crystals: The PFP/Ag(111) Interface. J. Phys. Chem. Lett. 2012, 3, 848−852.

18630

dx.doi.org/10.1021/jp506151j | J. Phys. Chem. C 2014, 118, 18626−18630