LEED, STM, and TDS Studies of Ordered Thin Films of the Rhombus

Rhombus-Shaped Polycondensed Aromatic Hydrocarbon ... polycondensed aromatic hydrocarbon “rhombus-C54”, C54H22, on MoS2 and graphite (0001) and ...
0 downloads 0 Views 884KB Size
656

Langmuir 2005, 21, 656-665

LEED, STM, and TDS Studies of Ordered Thin Films of the Rhombus-Shaped Polycondensed Aromatic Hydrocarbon C54H22, on MoS2, GeS, and Graphite Christian Gu¨nther, Norbert Karl, and Jens Pflaum* 3. Physikalisches Institut, Universita¨ t Stuttgart, D-70550 Stuttgart, Germany

Rainer Strohmaier, Bruno Gompf, and Wolfgang Eisenmenger 1. Physikalisches Institut, Universita¨ t Stuttgart, D-70550 Stuttgart, Germany

Markus Mu¨ller and Klaus Mu¨llen Max-Planck-Institut fu¨ r Polymerforschung, Ackermannweg 10, D-55128 Mainz, Germany Received August 6, 2004. In Final Form: October 19, 2004 Low-energy electron diffraction (LEED), scanning tunneling microscopy (STM), and thermal desorption spectroscopy (TDS) are used to study vacuum vapor-deposited molecular thin films of the rhombus-shaped polycondensed aromatic hydrocarbon “rhombus-C54”, C54H22, on MoS2 and graphite (0001) and on GeS (010) substrates. It is found that this compound forms well-ordered incommensurate superstructures of the closest packed flat-lying molecules in well-defined azimuthal orientations to the substrate. These films are thermally remarkably stable. By TDS, a monolayer binding energy on graphite of 2.3 eV was derived, whereas the molecules in the second layer were found to be less strongly bound (1.9 eV). This difference allows the preparation of monolayers by desorbing multilayers at the appropriate temperature. Apparently, this molecule is a promising candidate for further studies aiming at applications in organic electronics such as organic field effect transistors or light emitting displays.

Introduction Recently, there has been rapidly growing activity, both in academic and in industrial research, to develop organicbased thin film electronic devices. Organic electronic devices are expected to, first of all, meet a rapidly growing demand for inexpensive low level electronic circuits that can be used in intelligent electronic labels, identification tags, smart cards, or sensors. Yet also high standard devices, such as flat, high pixel-resolution, full color image displays, have not only been demonstrated to be feasible, but are presently already close to commercialization. It is generally expected that by using organic electronic materials fabrication costs can be kept low because of the possibility of using cheap thin film deposition processes such as (comparatively low temperature) vapor deposition, spin casting, or ink jet printing. Concepts of future microminiaturization down to the level of individual molecules, acting as conductive pathways, switches, and single bit memories, do not appear unreasonable.1,2 There are numerous problems, however, that must be solved. Suitable molecules have to be designed, synthesized, purified, assembled, and addressed, circuits with the required electrical functions have to be developed, and device performance has to be optimized. It is largely recognized that a concerted interdisciplinary cooperation is necessary between physics, chemistry, materials research, and electronic engineering. * Corresponding author. Phone: ++49 (0)711 685-5228. Fax: ++49 (0)711 685-5281. E-mail: [email protected]. (1) Bachtold, A.; Hadley, P.; Nakanishi, T.; Dekker, C. Science 2001, 294, 1317. (2) McEuen, P. L.; Fuhrer, M.; Park, H. IEEE Trans. Nanotechnol. 2002, 1, 78.

Suitable organic molecules should combine a number of desirable properties. Their highest occupied and lowest unoccupied energy levels should be located in an energy range and be separated by an energy gap suitable for the envisaged electronic applications, a feature offered by molecules with π-bonds. The room-temperature vapor pressure should be very low, but sublimation without decomposition at sufficiently elevated temperature should be possible, to allow vapor deposition under clean vacuum conditions. For efficient electrical transport (by electrons and holes), extended intermolecular overlap of π-electron wave functions is necessary. These criteria are met by many planar aromatic and heteroaromatic molecules with conjugated π-electron systems, if their geometrical shape is suitable for high-density space filling. For the ultimate, at least academic, goal, epitaxial growth of organic thin films on inorganic crystal surfaces, the ability of the material to grow in sheet and stack crystal structures (like, e.g., PTCDA,3 see also ref 4, Figure 1) would appear optimal. On the other hand, nonepitaxial zipper-like selforganization of periodic structures which, for example, occurs for diindenoperylene5 may require different packing patterns. The intrinsic thermal stability of a molecule is based on strong intramolecular bonds and high activation barriers for isomerization. Therefore, the molecular stability is high if multivalent atoms are linked in networks, if weakly bound ligands are avoided, and if the molecular symmetry is high. To meet the requirements of low vapor pressure (3) Forrest, S. R. Chem. Rev. 1997, 97, 1793. (4) Umbach, E.; Sokolowski, M.; Fink, R. Appl. Phys. A 1996, 63, 565. Alonso, M. I.; Garriga, M.; Karl, N.; Osso´, J. O.; Schreiber, F. Org. Electron. 2002, 3, 23. (5) Du¨rr, A. C.; Schreiber, F.; Mu¨nch, M.; Karl, N.; Krause, B.; Kruppa, V.; Dosch, H. Appl. Phys. Lett. 2002, 81, 2276.

10.1021/la048009s CCC: $30.25 © 2005 American Chemical Society Published on Web 12/10/2004

LEED, STM, and TDS Studies of Ordered Thin Films

Figure 1. Molecular structure of the molecule rhombus-C54, C54H22.

and of the thermal stability mentioned above, it turns out that molecular weight should roughly be between ca. 200 and 500 amu. With these goals in mind, we have studied in detail the synthesis and thin film formation capabilities of the large, highly symmetric polybenzoic molecule rhombus-C54 (tribenzo[hi,o,uv]triphenyleno[2,1,12,11,-bcdef]ovalene, C54H22), 1,6 see Figure 1. In brief, it has been shown that, starting from 1,4dibromo-2,5-di-trans-styrylbenzene, reaction with a suitable aromatic bromonic acid leads to a p-terphenyl derivative that already has the carbon framework of 1. A double intramolecular Diels-Alder cycloaddition reaction leads with high yield to the precursor product 1,2,10,11-tetraphenyltetrabenzo[a,c,h,j]anthracene consisting of different stereoisomers that can all be dehydrogenated by reaction with DDQ to form the desired aromatic hydrocarbon molecule 1. Confirmation of the expected molecular weight was obtained by mass spectrometry; the expected planar rhombic shape of the molecule could directly be confirmed by scanning tunneling microscopic imaging (STM) of an ordered monolayer film, vapor-deposited onto a cleaved MoS2 (0001) crystal surface, a method which proved to be a useful new tool for analyzing large essentially insoluble molecules.6,7 In this paper, we present results of thermodynamic and complementary structural studies of ordered rhombusC54 thin films on the layered inorganic semiconductors MoS2, GeS, and graphite with weak, presumably van der Waals type interactions at the interfaces, but different surface periodicity intervals and surface symmetry. We mainly focus on the first molecular layer, as it determines the possibility and the kind of ordered epitaxial film growth. A high degree of molecular order is a principal condition for obtaining high electronic switching speed, low loss, and low heat dissipation in organic electronic devices, properties that are based on high charge carrier mobilities.8,9 The geometrical structures of the thin films prepared are analyzed by diffraction of low-energy electrons (LEED) and scanning tunneling microscopy (STM), two complementary methods, the first to yield information on the long-range order in reciprocal space, while the second enables direct imaging of the molecular arrangement on a nanometer scale. Furthermore, thermal desorption spectroscopy (TDS) allows one to draw conclusions on the respective binding energies. Experimental Section Purification by Fractional Sublimation. The dark-colored synthetic raw product was purified via the temperature step sublimation method. This method consists of placing the sample (6) Mu¨ller, M.; Petersen, J.; Strohmaier, R.; Guenther, Ch.; Karl, N.; Mu¨llen, K. Angew. Chem. 1996, 108, 947; Angew. Chem., Int. Ed. Engl. 1996, 35, 886. (7) Strohmaier, R. STM-Bildkontrast und Monolagenkristallographie planarer Aromaten. Ph.D. Thesis, University of Stuttgart, 1997. (8) Karl, N.; Marktanner, J. Mol. Cryst. Liq. Cryst. 1998, 315, 163.

Langmuir, Vol. 21, No. 2, 2005 657 into a fused silica tube which then is connected at its open end (opposite the sample) to a high vacuum pump (ca. 10-4 Pa). Fractionation is achieved by a stepwise increase of the temperature of a surrounding tubular heating furnace combined with a simultaneous stepwise pulling apart of the furnace to expose a clean (cooled-down) part of the tube in each step for condensation of the respective next fraction (cf., ref 10, Figure 6.6a). The main fraction sublimes at a heater temperature of ca. 600 °C in 2 h, forming a bright orange-red sublimate with high fluorescence quantum yield. A subsequent increase of the heater temperature to 650 °C gave another (smaller and final) fraction in about 1 h, see Figure 2. At those rather high temperatures required to evaporate this large molecule, gradual decomposition of the raw material took place under formation of a loose black nonvolatile residue. The yield in this first sublimation step amounts to only 15% of the mass of the pristine material, unfortunately, while a fast second sublimation of the orange-red product gave a considerably higher mass yield (up to 90% of the starting material). Slow sublimation in a small temperature gradient was attempted to obtain crystals from the vapor phase; however, it failed and only black material was left back. We conclude from this observation that the high loss in the first sublimation step was mainly caused by polymerization, perhaps catalytically enhanced by a residue of the CuCl2 and AlCl3 catalyst used in the synthesis for the last cyclodehydrogenation step (the final product, C54H22, is essentially insoluble in all conventional solvents; hence prepurification by extraction could not be applied). Upon fast heating, a large fraction of the molecules sublime intact, but the material is obviously not thermally stable enough to withstand extended exposure to the sublimation temperature without decomposition. Advantageously, the main decomposition product is of a higher molecular weight, does not sublime, and hence remains in the crucible, whereas the undecomposed molecule evaporates. In this context, it is remarkable that an organic molecule can sustain temperatures of dark red glow, at least on a short time scale. It was checked by mass spectrometry after sublimation that the 600 and 650 °C fractions were still C54H22. Very slow recrystallization from molten pyrene (mp 156 °C) turned out to be possible, too. A mixture of 1 mg of rhombus-C54 and 1 g of zone-refined pyrene in an evacuated conical fused silica ingot was kept for about 24 h at 200 °C with occasional rotation and stopping for mixing. After programmed cooling at 0.4°/min, tiny crystals, as shown in Figure 3, with yellow/red dichroism in polarized light could be isolated after dissolution of the pyrene matrix in xylene. Only presublimed orange material gave this result; otherwise, the solidified melt was black. For the prepared rhombus-C54 crystals, an estimation of the crystallographic structure will be carried out and reported in a forthcoming paper. Preparation of Thin Films: LEED, TDS, and STM Techniques. Preparation of thin films by vacuum-vapor deposition turned out to be somewhat problematic, because, at a temperature where sublimation became detectable by an oscillating quartz crystal balance, beginning decomposition of the prepurified material was observed, leading to a black residue in the crucible. Therefore, it was necessary to work at a crucible temperature that gave only very low deposition rates of 0.01-0.03 nm/min. Only a small number of mono- or low number multilayer samples could thus be obtained from one charge. The substrates used, single-crystalline MoS2, GeS, and singlecrystalline graphite, and highly ordered pyrolytic graphite (HOPG), were cleaned by peeling off the topmost layers with an adhesive tape and then immediately were transferred into an ultrahigh vacuum (10-8 Pa) preparation chamber, and heated at 600 °C for several minutes for graphite, at 400 °C for 1 h (or 450 °C for several minutes) for MoS2, and at 200 °C for 24 h (or 300 °C for 12 h) for GeS. After these preparation steps, clear LEED pictures of the substrate surfaces could be obtained, cf., ref 11. All three surfaces remained unreconstructed; the MoS2 (0001) and graphite single-crystal (0001) cleavage planes displayed (9) Karl, N. In Organic Electronic Materials; Farchioni, R., Grosso, G., Eds.; Springer: Berlin, 2001; Chapter 8, p 283. (10) Karl, N. In Organic Electronic Materials; Farchioni, R., Grosso, G., Eds.; Springer: Berlin, 2001; Chapter 6, p 215. (11) Karl, N.; Gu¨nther, Ch. Cryst. Res. Technol. 1999, 34, 243.

658

Langmuir, Vol. 21, No. 2, 2005

Gu¨ nther et al.

Figure 2. Purification of rhombus-C54 by temperature-step sublimation under vacuum. The source material was at the left side (outside the figure), and the vacuum pump was connected at the right side of the fused silica tube (smaller diameter ca. 8 mm).

p6mm symmetry, with reciprocal lengths in accordance with the literature lattice parameters a ) 0.316 and 0.246 nm, respectively; the GeS (010) cleavage plane displayed p2mm symmetry with a ) 0.429 nm and c ) 0.364 nm;12 HOPG, due to its composition of coplanar, but in the (0001) plane azimuthally disordered microcrystallites, gave the corresponding ring pattern. Prior to evaporation of C54H22, the crucible with the source material was outgassed at a temperature of 200 °C overnight. The absolute thickness of the deposited films was determined from the frequency shift of a calibrated quartz crystal balance covered simultaneously in the deposition cycle. An alternative, more precise, but only relative method consisted of an integral mass spectrometric detection of the yield of thermally desorbed molecules. As it was checked and confirmed, there was a linear correlation among the results of both methods between 0 and 1.5 nm film thickness, that is, up to 5 monolayer thickness. The LEED equipment consisted of an Omicron SPECTALEED system with the option of an extended mechanical shift facility, allowing an appropriate approach to the sample that was mounted on a translation-rotation manipulator in the center of a spherical 30 cm diameter UHV chamber. The hemispherical LEED screen was either (version I) imaged by a high aperture photocamera lens onto an electrostatic image intensifier (System MD 8100, Electrophysics, Nutley, NJ) and photographed in a conventional way (see ref 13) or, in an advanced phase of this work, imaged by a f/1,4-16 mm focal length camera lens onto the thermoelectrically cooled 509 × 739 pixel CCD chip of a HamamatsuC5985 video camera system, the latter connected to a PC (version II). According to “version II”, the digital images were processed using the software packet MARK I by FRT Co. Both versions allow one to work at a substantially reduced electron beam current of less than 1 µA, which permits longer

adjustment and working phases before the image contrast fades due to electron beam-induced degradation of the organic layers. In the former case, considerable image distortions had to be taken into account and to be compensated by careful calibrations with substrates of known surface lattice periodicity parameters, which limited the evaluation accuracy for unknown LEED patterns to (5% in length and (0.5° in angle. The advantage of the latter equipment is that geometrically fixed pixel information is formed at an early stage. It is, however, slower than the image intensifier system, as the drawback of using a high sensitivity Si chip is a considerably large integration time. After a nontelecentric imaging of the LEED screen was corrected for by using surfaces with well-known surface structure for calibration, an absolute reading accuracy of 0.4% could be routinely achieved.14 Taking into account that the large lattice constant of the organic adsorbant layers requires rather low LEED energies of around 10 eV, the single domain commensurate superstructure of hexabenzocoronene on GeS with long-range order11 was especially suited for calibration in that it allowed indexing higher order superstructure reflections up to an electron energy of 60 eV. In contrast, the typically small lattice constants of the inorganic substrates used in this work do not allow for calibration in the low-energy (long wavelength) regime. Alternatively, the high anisotropic (010) cleavage plane of natural orthorhombic stibnite (Sb2S3, a ) 1.130 nm, c ) 0.3839 nm12) can be used as well.11 In these calibration steps, it was necessary to correct the electron energy reading by a constant value of -1.2 eV, probably due to a “built-in” potential difference caused by work function offsets between the electron source and the grids of the LEED screen. As the LEED optics can be adjusted at an appropriate distance from the sample, an additional source of error might originate from existing stray fields of charged parts. All effects together were estimated to result in about (2% uncertainty in the evaluated lattice dimensions. TDS measurements were carried out in a second similar UHV chamber connected to the one described. Therefore, after deposition, the sample had been transferred to the TDS setup without breaking the UHV. For programmed heating, a tantalum wire inserted in an Al2O3 ceramic tube was used in conjunction with a Pt-100 temperature sensor attached to the sample support. The mass spectrometer detector (Balzers QMG 512) with a mass range extended to m/e ) 1024 was set to the expected mass and a crossbeam ion source was placed as close as possible to the surface to be studied. After its detection was confirmed, the spectrometer was adjusted to half of this mass, to detect doubly ionized molecules that gave considerably higher sensitivity. Detection of desorption from the main sample holder was thus essentially excluded geometrically; moreover, thermal isolation between the sample support and the sample holder gave sufficient heating delay of the latter such that essentially no (less than 2%) desorption interferences occurred. The temperature of the sample support could be measured with an accuracy of better than 2 °C in thermal equilibrium; however, ramping at a heating rate of 30 °C per minute, the temperature reading lagged behind by up to 6 °C. STM investigations were made by a home-built video frequency STM16 that allowed scanning at 16 frames per second at a resolution of 128 × 128 pixels, or 4 frames/s at 256 × 256 pixels, which greatly improved the signal-to-noise ratio. Fast data acquisition, in addition, allowed the operator to immediately react

(12) Wyckoff, R. W. G. Crystal Structures; J. Wiley: New York, 1963; Vol. 2. (13) Zimmermann, U.; Karl, N. Surf. Sci. 1992, 268, 296.

(14) Gu¨nther, Ch. Organische Molekularstrahlepitaxie: Ordnungsprinzipien grosser Aromaten auf Schichthalbleitern. Ph.D. Thesis, University of Stuttgart, 1998.

Figure 3. Small rhombus-C54 single crystals obtained from molten pyrene (field of view ca. 100 × 100 µm2).

LEED, STM, and TDS Studies of Ordered Thin Films

Langmuir, Vol. 21, No. 2, 2005 659

Figure 4. Rhombus-C54 film on MoS2 (0001). (a) LEED pattern at 12.7 eV of a ca. three molecular layers thick film; (b) calculated LEED pattern and unit cells of one domain as expected from the structural model (O symbols indicate the missing (11 h ) spot; the 2-D lattice parameters are listed in Table 1); and (c) structural model of one of the six substrate-induced domains. on slight changes of parameters during a scan. This is advantageous because with weakly bound organic adsorbates even a small reduction of the appropriate distance of the tunnel tip can lead to destruction of an ordered layer. Substrate and sample preparation for the STM measurements were done in situ at a pressure of 3 × 10-8 Pa. Fresh surfaces of graphite and MoS2, prepared ex situ and immediately transferred via a load lock to the UHV system, were heated to 200 °C to desorb possible contaminants prior to vapor-deposition of the rhombus-C54 molecules. The sample was then transferred under vacuum to the tunnel microscope, mounted in an attached separate vacuum chamber. To reduce the noise level during the tunnel measurements, the mechanical vacuum pumps were switched off while vacuum was maintained by an ion getter pump. The polarity of the tunnel current as indicated in the figure captions is referred to as the voltage applied to the substrate.

Results LEED Results. On all three substrates used, GeS, MoS2, and graphite, evaluable LEED images of rhombusC54 layers were obtained, however, only for electron energies