11180
J. Phys. Chem. C 2010, 114, 11180–11184
Structural Transition and Thermal Stability of a Coronene Molecular Monolayer on Cu(110) Lei Zhang,† Dongxia Shi,† Shixuan Du,† Lifeng Chi,‡ Harald Fuchs,‡ and Hong-Jun Gao*,† Institute of Physics, Chinese Academy of Sciences, P.O. Box 603, Beijing 100190, P. R. China, and Physics Institute and Center for Nanotechnology (CeNTech), UniVersity of Muenster, Muenster D-48149, Germany ReceiVed: March 02, 2010; ReVised Manuscript ReceiVed: May 27, 2010
A series of alternating structural transitions of coronene monolayers on Cu(110) surfaces are observed using in situ molecular beam epitaxy with low-energy electron diffraction when varying substrate temperature after deposition. The first transition is observed around 293 K, and the second one happens at 360 K. By employing molecular mechanics and density functional theory calculation, it is found that the first transition is due to the transition from physical adsorption to chemical adsorption, while the second transition can be explained by the adsorbed molecules diffusing from a metastable site to a thermodynamically more stable site. Moreover, the unchanged diffraction pattern in the subsequent cooling and reheating process indicates that the final monolayer structure has a high thermal stability. These transitions were compared with the reversible transitions of coronene on Ag(110). 1. Introduction Organic semiconductors, due to their promising application potential for organic field effect transistors1 and optoelectronic devices,2 have drawn increasing attention. Thin films of organic π-conjugated molecules on metal surfaces play an important role to realize both applications because they have high charge carrier mobility3 and luminescence efficiency.4 Since these properties are remarkably governed by the thin film structures,5 improving the quality of thin films can optimize device performance. As a result, there has been great interest in the structural control of molecular thin films on metal surfaces.6-14 The structures of molecular monolayers can be successfully tuned by modifying growth parameters, like molecule species, substrate surfaces, deposition rate, growth temperature, or lateral nonfunctional alkyl. Take coronene and perylene monolayers for example: Seidel et al.15 investigated monolayer structures on several metal surfaces by using low-energy electron diffraction (LEED) and scanning tunneling microscopy (STM) and found that the structures varied with respect to spacing efficiency and the symmetry of the substrate. By varying the deposition rate of perylene onto the silicon oxide surface, Casu et al.16 observed a transition from island morphology at a low rate to homogeneous film morphology at a high rate, and in another paper,17 they investigated the influence of the growth parameters on molecular orientation in thin films. Sohnchen et al.18 studied the influence of substrate temperature on the monolayer structure during deposition, and they observed two distinct structures which formed at 380 and 450 K. A systematic study about the role of the lateral nonfunctional alkyl was conducted by Shi.19 All of these investigations focused on the morphology of molecular monolayers; however, it is not only the morphology but also the thermal stability of monolayers that determine the performance of organic devices. Since organic devices are supposed to work over a wide range of temperatures, the study * To whom correspondence should be addressed. Tel.: 0086-1082648035. Fax: 0086-10-62556598. E-mail:
[email protected]. † Chinese Academy of Sciences. ‡ University of Muenster.
of thermal stability is extremely important. Several experiments have been performed so far to investigate the thermal stability of monolayer structures.17,20,21 Casu et al.17 studied the thermal and vacuum stability of perylene thin film on the Al2O3/ Ni3Al(111) substrate. By employing molecular beam epitaxy (MBE) together with LEED, Shi et al.20 investigated the structural evolution of coronene on a Ag(110) surface, in which a reversible transition was observed. What the important factors are that will influence the thermal stability of a thin film on a metal surface is still of great interest. In this paper, we report both experimental and theoretical investigations, such as in situ MBE-LEED, molecular mechanics (MM), and density functional theory (DFT) calculations, on the fabrication of coronene overlayers on Cu(110) substrate at 185 K and the structural evolution during subsequent temperature variations. We found a series of irreversible structural transitions (S1fS2fS1) by in situ MBE-LEED as substrate temperature increased after growth. Furthermore, MM and DFT calculations were employed to understand the underlying physics associated in these transitions. It was found that at 185 K the significant interaction between deposited molecules and the substrate is van der Waals (vdW) force which leads to the initial S1. The first transition, S1fS2, could be explained by the interaction between molecules and substrate changing from physical to chemical as temperature increased. The second transition can be viewed as the molecules diffusing from a metastable adsorption site to a thermodynamically more stable site. Moreover, the unchanged LEED pattern in the cooling and reheating process indicates that the final S1 structure has a high thermal stability. 2. Experimental and Theoretical Methods The experiments were carried out in an ultrahigh-vacuum (UHV) chamber with a base pressure below 3.0 × 10-10 mbar. The chamber was equipped with an in situ MBE-LEED. Cu(110) crystals were cleaned by standard sputtering with argon ions for 15 min at 5.0 × 10-6 mbar followed by annealing at 620 K for 3 min. After that, the Cu(110) surface was checked
10.1021/jp101865j 2010 American Chemical Society Published on Web 06/07/2010
Coronene Molecular Monolayer on Cu(110)
J. Phys. Chem. C, Vol. 114, No. 25, 2010 11181
Figure 1. LEED patterns of coronene adsorbed on Cu(110) at 185 K. (A-C) LEED patterns with increasing coverage. (D) LEED pattern when coverage reaches one monolayer.
by LEED. The sublimation temperature of coronene was set at 408 K. The coronene molecules were thoroughly degassed below sublimation temperature before deposition. Cu(110) crystals were kept at a constant temperature of 185 K during the molecular deposition. The molecular aggregation was monitored in situ by LEED. After deposition, the substrates were heated by an inner filament in the sample holder at a rate of 2.5 K/min. The structural changes of the molecular adsorbate during the substrate heating period were monitored in situ by LEED as well. All LEED patterns were taken at an electron beam energy of 13 eV. Theoretical calculations based on density functional theory were carried out to investigate the contribution of chemical interaction on the adsorption energy of the monolayer structures on Cu(110). The Perdew-Wang exchange-correlation functional with generalized-gradient corrections22 and plane waves, as implemented in the VASP code, were used.23,24 Supercells were built with four Cu layers and 2.1 nm of vacuum. An energy cutoff of 400 eV was used after convergence tests. Each molecule was placed on one side, and all atoms except for the bottom two Cu layers were relaxed until the net force on every atom was smaller than 0.02 eV/A. MM calculation was conducted to evaluate the contribution of vdW interaction on the adsorption energy of monolayer structures on Cu(110). Here, we used the MM+ force field,19,20 which is improved from the MM2 force field. In this method, the configuration of the model system in equilibrium was obtained by minimizing the energy. Besides, only the molecules were allowed to relax during the structure optimization. To explain the structural transitions of this molecular monolayer, we calculated the temperature dependence of the Helmholtz free energy. Since the temperature range is below 0.04 eV, which is much lower than the typical energy scale for
electron excitations, the Helmholtz free energy contains the energy of ground state (adsorption energy) and the entropy contributed by the excitation of atomic vibrations. Consequently, the Helmholtz free energy (F) can be derived by the following equation, which is a function of independent volume (V) and temperature (T)25
F(V, T) ) Eground(V) + Fvib ) Ead(V) + kBT
{
∑ ln j,q
1 2
[
∑ pωj(q) + j,q
1 - exp -
pωj(q) kBT
]}
(1)
where Eground is the static contribution to the free energy, including molecule-substrate interactions and molecule-molecule interactions, which can be obtained by the standard DFT and MM calculations mentioned above. Fvib represents the vibrational contribution to the Helmholtz free energy, and ωj(q) is the frequency of the jth phonon mode at the wave vector q in the Brillouin Zone (BZ). The vibrational frequencies were calculated by a direct force constant approach implemented in VASP.26 Systematic independent atomic displacements of 0.02 Å, from the equilibrium in x, y, and z directions, were applied to obtain the Hellmann-Feynman forces of these distorted structures.27,28 The dynamic matrix was constructed only for the coronene molecule as an approximation due to the limited computer time. Vibrational modes for motion between the molecule atoms and the substrate atoms were not taken into account in view of the relatively weak molecule-substrate interaction and the relatively large mass of substrate Cu atoms. These modes do not significantly influence the energetic contrast between the two structures.
11182
J. Phys. Chem. C, Vol. 114, No. 25, 2010
Zhang et al.
3. Results and Discussion During deposition at 185 K, the LEED pattern changed as coverage increased (see Figure 1). Figure 1a is a diffused diffraction pattern at a low coverage which presents a halo-like structure and corresponds to a gas-like state. With the coverage increasing, the diffraction patterns in Figure 1b,c gradually decay into six hexagonal spots. When the coverage reaches one monolayer, as in Figure 1d, the six diffuse spots become sharp, representing a two-dimensional crystalline structure. The geometric analysis of the pattern shows that there is only one selfassembled molecular structure, labeled S1, which can be specified by
( ) 3 2 0 5
and is commensurate with the substrate. We heated the substrate after the coverage had reached one monolayer to determine the influence of temperature on the surface structure and its thermal stability. Interestingly, the resulting LEED patterns showed that the surface structure went through an alternating transition in the period of temperature change. The evolution of LEED patterns is presented in Figure 2. The diffraction pattern shows there is only one structure until the substrate temperature reaches 293 K (Figure 2a), when two weak spots appear beside each of the six bright spots corresponding to S1. At about 307 K, the 12 side spots become as bright as the six central spots, and the diffraction pattern indicates that another crystalline surface structure comes into existence (Figure 2b). This pattern corresponds to
(
3 -1 2 3.5
)
labeled S2, which is incommensurate with the substrate. As the temperature increases further, the six central spots become weaker and weaker (Figure 2c), and at 323 K the central spots totally disappear, leaving only 12 bright side spots (Figure 2d). When the substrate temperature reaches 360 K, the central spots reappear gradually (Figure 2e), and at a temperature of 365 K, they become as bright as the side spots (Figure 2f). As the substrate temperature was heated as high as 373 K, the side spots become weaker and gradually disappear (Figure 2g), which means S2 changes to S1. Figure 2h shows that the diffraction pattern stays the same in the subsequent cooling and reheating process, which means no structural transition is observed and the final S1 is the most stable monolayer structure. This result is different from our previous work with a coronene monolayer on Ag(110), where the structural transition was found to be reversible.21 In the silver case, the first transition was attributed to the free energy difference between the two structures, and the second transition was attributed to the competition between mol-mol interaction and mol-sub interaction. Since both of the two attributions were sensitive to temperature, it was easy to understand how the alternating structural transition was reversible. However, in the copper case, we found the transition to be irreversible; that is to say, after the formation of final S1 at 373 K, there can be no more structure transition with varying temperature. This raises two questions: whether the final S1 is the same as the initial S1 and what is the physics of the structural transition. It is paradoxical to say that the final S1 is exactly the same as the initial S1. If this is really true in every sense, S1 should transform into S2 in the second heating process, which has not been observed. However, the LEED patterns show they have the same lattice. Therefore, the two S1 should have the same
Figure 2. Evolution of LEED patterns during the heating and cooling process. In the heating process (A-G), the evolution shows an S1fS2fS1 structural transition, while in the cooling process (H), the LEED pattern remains the same, indicating that the structural transition is irreversible.
supercell with respect to the substrate, but the configurations of the molecule on the substrate could be subtly different. For aromatic compounds on metallic substrates, the interaction between the molecule and the substrate can vary between weak (vdW interaction, such as benzene on Au) and rather strong (chemical interaction, such as benzene on Ni).29 For coronene on Cu(110), the interaction is in between, so both vdW and chemical interactions contribute to the total adsorption energy. Patrick et al.30 showed that at an extremely low temperature, 4 K, benzene forms a 2D gas on Au(111) substrate. To investigate the overlayer structure of perylene on Cu(110), Witte et al.31 cooled the sample to as low as 110 K to prevent perylene from diffusing on Cu(110). Besides, Seidel et al.15 demonstrated that the diffused LEED pattern represents a 2D gas and fluid phase. In our experiment, coronene has structures similar to perylene,
Coronene Molecular Monolayer on Cu(110)
J. Phys. Chem. C, Vol. 114, No. 25, 2010 11183
Figure 3. Calculation superlattices and adsorption sites of coronene on Cu(110). The solid and dotted lines represent the lattices of S1 and S2, respectively. Four adsorption sites as well as two molecular orientations are considered in the calculation.
and the LEED patterns are diffused diffraction patterns at a low coverage (Figure 1a, b). Thus, the interaction between coronene molecules and Cu(110) should be a vdW-like interaction at 185 K. When increasing temperature, the compactly arranged molecules formed chemical interactions with the copper substrate through full relaxation. As a result, the different environment and interactions lead to the difference between the initial S1 and the final S1. As is a fact, vdW interaction cannot be well described by the traditional DFT.32 Therefore, we took two steps to get the total adsorption energy for the final S1 and S2 structure. First, by performing DFT calculation, we get adsorption energy (Echem) which comes from the chemical interaction. Second, by using MM calculation, we get the contribution made by vdW interaction (EvdW). Finally, by simply adding these two parts together, we got the total adsorption energy (Ead). That is to say: Ead ) Echem + EvdW. For the initial S1, since the adsorption energy is contributed mainly by vdW interaction, we have Ead ) EvdW. Besides the detailed calculation settings which have been discussed in the Experimental and Theoretical Methods section, two supercells are built for S1 and S2, respectively (Figure 3). The one for S1 is commensurate with the substrate and contains one molecule, while the one for S2 is incommensurate, so it has been doubled along the b axis and there are two molecules in it. In addition, for S1 we have considered eight adsorption configurations with respect to the center of the molecule on the substrate (a-d in Figure 3) and the molecular orientation (orientation 1 and 2 in Figure 3); e.g., configurations a and e are both top-sited but with different orientations. Since the supercell for S2 is much larger, the number of atoms in the model for S2 is nearly twice as much as that in S1. As a result, after considering the result for S1, we calculate only the two most likely configurations for S2. Depending on the result obtained by MM calculation, it is found that site b, a hollow-sited configuration with orientation
1, is the most stable site. Therefore, in the initial S1 structure, molecules adopt a hollow site. By combining the result from MM with that from DFT, we found that site c, a bridge-sited configuration with orientation 1, is the most stable one. So, molecules adopt a bridge site in the final S1 structure. Interestingly, the initial S1 configuration becomes less stable when chemical interaction between the π-orbital and the substrate is taken into account. This result confirms our assumption that the different environment and interactions lead to the difference between the initial S1 and the final S1. To answer the second question of what is the physics of the structural transition, the Helmholtz free energy (F) was calculated. Combining the total adsorption energy and vibration, an F-T graph is charted, as presented in Figure 4, which shows the temperature dependence of Helmholtz free energy (F) per molecule. It is clear that, in the temperature range from 0 to 500 K, bridge-sited S1 (final S1) is energetically favorable, followed by S2, and the hollow-sited S1 (initial S1) is less favorable. Up to now, we can explain the two structure transitions as follow: First, at low temperature, physical interaction leads to an initial S1 which is confirmed by MM calculation. Second, as the temperature goes up, chemical interaction becomes important and makes the initial S1 less stable as confirmed by the free energy calculation. It changes to a metastable state S2. Therefore, the change of interaction before and after substrate heating attributes to the first structure transition. Third, as is presented in the F-T graph, S2 is thermodynamically less stable than the final S1. As a result, with the temperature further rising, molecules can get enough energy to overcome diffusion barriers and hop into a more stable site, causing the second structure transition. At last, since bridge-sited S1 (final S1) is the most favorable structure throughout the whole temperature range, it is understandable that, once the surface structure relaxes into the final S1, no structural transition can be found in the subsequent cooling and reheating processes.
11184
J. Phys. Chem. C, Vol. 114, No. 25, 2010
Zhang et al. References and Notes
Figure 4. Calculated temperature dependence of Helmholtz free energy (F). At any given temperature below 500 K, bridge-sited S1 has the lowest free energy, which accounts for the irreversibility of the structural transition.
On the other hand, as for the application of an organic device, thermal stability is very important. By comparing this experiment with the case of coronene on Ag(110), it is clear that the coronene monolayer structure on the Cu(110) substrate is more stable than that on the Ag(110) substrate. This is because the Cu(110) substrate is more active and can have a stronger interaction with adsorbate. 4. Conclusions We report in situ MBE-LEED, MM+, and DFT investigation of coronene monolayers on Cu(110) fabricated at a substrate temperature of 185 K and the structural evolution during the subsequent temperature variation. A series of irreversible structural transitions (initial S1fS2ffinal S1) were observed by in situ MBE-LEED as substrate temperature was raised after deposition. By employing theoretical calculations, it was found that, when deposited at 185 K, the principal interaction between molecules and substrates is vdW force, which leads to a selfassembled molecular structure of initial S1. The first transition was due to the interaction between molecules and substrate and changed from a physical one to a chemical one as temperature increased, and the second transition happened because of the thermal stability of final S1 configuration. With the calculated temperature dependence of free energy, the irreversible transition can be explained. Acknowledgment. This work was partially supported by the Natural Science Foundation of China (NSFC), the MOST 973 projects of China, the Chinese Academy of Sciences (CAS), Shanghai Supercomputer Center, and the Collaborative Research Centre TRR 61.
(1) Horowitz, G. AdV. Mater. 1998, 10, 365. (2) Ho, P. K. H.; Thomas, D. S.; Friend, R. H.; Tessler, N. Science 1999, 285, 233. (3) Katz, H. E.; Lovinger, A. J.; Johnson, J.; Kloc, C.; Siegrist, T.; Li, W.; Lin, Y. Y.; Dodabalapur, A. Nature 2000, 404, 478. (4) Burroughes, J. H.; Bradley, D. D. C.; Brown, A. R.; Marks, R. N.; Mackay, K.; Friend, R. H.; Burns, P. L.; Holmes, A. B. Nature 1990, 347, 539. (5) Di Carlo, A.; Piacenza, F.; Bolognesi, A.; Stadlober, B.; Maresch, H. Appl. Phys. Lett. 2005, 86, 263501. (6) Gao, H. J.; Sohlberg, K.; Xue, Z. Q.; Chen, H. Y.; Hou, S. M.; Ma, L. P.; Fang, X. W.; Pang, S. J.; Pennycook, S. J. Phys. ReV. Lett. 2000, 84, 1780. (7) Shi, D. X.; Song, Y. L.; Zhu, D. B.; Zhang, H. X.; Xie, S. H.; Pang, S. J.; Gao, H. J. AdV. Mater. 2001, 13, 1103. (8) Du, S. X.; Gao, H. J.; Seidel, C.; Tsetseris, L.; Ji, W.; Kopf, H.; Chi, L. F.; Fuchs, H.; Pennycook, S. J.; Pantelides, S. T. Phys. ReV. Lett. 2006, 97, 156105. (9) Gao, L.; Deng, Z. T.; Ji, W.; Lin, X.; Cheng, Z. H.; He, X. B.; Shi, D. X.; Gao, H. J. Phys. ReV. B 2006, 73, 075424. (10) Ji, W.; Lu, Z. Y.; Gao, H. Phys. ReV. Lett. 2006, 97, 246101. (11) Gao, L.; Liu, Q.; Zhang, Y. Y.; Jiang, N.; Zhang, H. G.; Cheng, Z. H.; Qiu, W. F.; Du, S. X.; Liu, Y. Q.; Hofer, W. A.; Gao, H. J. Phys. ReV. Lett. 2008, 101, 197209. (12) Gao, H. J.; Gao, L. Prog. Surf. Sci. 2010, 85, 28. (13) Stepanow, S.; Lin, N.; Barth, J. V.; Kern, K. Chem. Commun. 2006, 20, 2153. (14) English, W. A.; Hipps, K. W. J. Phys. Chem. C 2008, 112, 2026. (15) Seidel, C.; Ellerbrake, R.; Gross, L.; Fuchs, H. Phys. ReV. B 2001, 64, 195418. (16) Casu, M. B.; Yu, X.; Schmitt, S.; Heske, C.; Umbach, E. Chem. Phys. Lett. 2009, 479, 76. (17) Casu, M. B.; Scholl, A.; Bauchspiess, K. R.; Hubner, D.; Schmidt, T.; Heske, C.; Umbach, E. J. Phys. Chem. C 2009, 113, 10990. (18) So¨hnchen, S.; Ha¨nel, K.; Birkner, A.; Witte, G.; Wo¨ll, C. Chem. Mater. 2005, 17, 5297. (19) Shi, D. X.; Ji, W.; Lin, X.; He, X.; Lian, J.; Gao, L.; Cai, J.; Lin, H.; Du, S.; Lin, F.; Seidel, C.; Chi, L.; Hofer, W.; Fuchs, H.; Gao, H. J. Phys. ReV. Lett. 2006, 96, 226101. (20) Wang, Y. L.; Ji, W.; Shi, D. X.; Du, S. X.; Seidel, C.; Ma, Y. G.; Gao, H. J.; Chi, L. F.; Fuchs, H. Phys. ReV. B 2004, 69, 075408. (21) Shi, D. X.; Ji, W.; Yang, B.; Cun, H. Y.; Du, S. X.; Chi, L. F.; Fuchs, H.; Hofer, W. A.; Gao, H. J. J. Phys. Chem. C 2009, 113, 17643. (22) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Phys. ReV. B 1992, 46, 6671. (23) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892. (24) Kresse, G.; Furthmuller, J. Phys. ReV. B 1996, 54, 11169. (25) Wallace, D. C. Thermodynamics of Crystals; Wiley: New York, 1972. ¨ uller, J.; Hafner, J. Europhys. Lett. 1995, 32, 729. (26) Kresse, G.; FurthmA (27) Hellmann, H. EinfA¨uhrung in die Quantenchemie; Franz Deuticke: Leipzig, 1937; p 285. (28) Feynman, R. P. Phys. ReV. 1939, 56, 340. (29) Witte, G.; Woll, C. J. Mater. Res. 2004, 19, 1889. (30) Han, P.; Mantooth, B. A.; Sykes, E. C. H.; Donhauser, Z. J.; Weiss, P. S. J. Am. Chem. Soc. 2004, 126, 10787. (31) Witte, G.; Hanel, K.; Busse, C.; Birkner, A.; Woell, C. Chem. Mater. 2007, 19, 4228. (32) Kohn, W.; Meir, Y.; Makarov, D. E. Phys. ReV. Lett. 1998, 80, 4153.
JP101865J