By Maja Remkar,* Ale Mrzel, Adolf Jesih, Janez KovacÏ, Hagai Cohen, Rosendo SanjinØs, and Francis LØvy Two-dimensional monolayers of C60 molecules were the subject of many theoretical and experimental studies in the last decade as part of the search for new classes of materials. Three major phases with orientational dependence are known in bulk C60 crystals: the glassy phase below 90 K,[1] the orientationally ordered simple cubic phase between 90 K and 260 K,[2] and the disordered face-centered cubic phase with freely rotating C60 molecules at temperatures up to 300 K.[3] The phase-transition temperatures and the corresponding topological ordering depend strongly on the dimensionality of the system. On average, the transition temperatures in low-dimensional arrangements of C60 molecules are lowered in comparison to those in bulk C60 crystals. For example, the first-order orientational order±disorder phase transition at the (111) surface of a C60 crystal at 225 K is found to be well below the corresponding phase transition of the bulk analogue.[4] A C60 array thermally evaporated onto a self-assembled alkylthiol monolayer at 5 K showed domains with orientationally ordered C60 molecules and abrupt domain boundaries.[5] Theoretical studies using a semiempirical approach confirmed that the experimentally observed orientationally ordered phases are a few among several calculated total energy minima.[6] Monolayers of C60 deposited onto hexagonal planes of different substrates adopt a hexagonal close-packed symmetry. The lattice structure and orientational relationship of those C60 monolayers depend on the lattice mismatch.[7] The intermolecular distance of two-dimensional C60 on self-assembled alkylthiol molecules grown on a gold (111) substrate was 0.995 ± 0.01 nm,[6] while the separation between C60 molecules epitaxially grown on a MoS2 (001) plane was 1.00 ± 0.02 nm.[7] C60 molecules, which possess a strong electron affinity, are also important in photovoltaic applications as organic lightemitting diodes and solar cells. Blends of conjugated polymers and fullerenes achieved quantum yields of photoinduced charge generation close to 100 %.[8]
We report here on MoS2±C60 composite crystals grown by a catalyzed transport reaction involving fullerene C60.[9] Transmission electron microscopy (TEM), electron diffraction (ED), X-ray diffraction (XRD), and photoelectron spectroscopy (XPS) data were used to build a model structure of these layered composites. The crystals of the MoS2±C60 complex represent an alternative material to fullerene±polymer composites for potential use for solar-cell applications. The first evidence that the transported material has been grown as an ordered MoS2±C60 composite material was found by ED, which revealed a superstructure with a reciprocal period of q » 1 / 3 a*, where a* is the (001) unit-cell translation in MoS2 reciprocal space; i.e., 0.2737 nm (Joint Committee on Powder Diffraction Standards Card No. 06-0097). Scanning tunneling microscopy of the top surface of the crystals showed the regular atomic structure of the MoS2 lattice, while the Auger depth profile evidenced the presence of carbon in the crystals inside. The steady-state composition reached after 5 min and retained over the next 30 min of sputtering corresponded to a mixture of MoS2 and approximately (18 ±0.5) at.-% of C. Mass spectrometry analysis detected a strong and time-stable flow of C60 molecules from the transported material at temperatures above 470 K. The agreement of these experimental data provides evidence that the C60 is situated inside the crystals. The ED patterns (Fig. 1) show three cross-sections through the reciprocal lattice of the C60±MoS2 composite material. The [001] zone perpendicular to the layers, shown in Figure 1a, evidences the parallel orientation of the MoS2 and
COMMUNICATIONS
New Composite MoS2±C60 Crystals**
±
[*] Prof. M. Remkar, Dr. A. Mrzel, Dr. A. Jesih, Dr. J. KovacÏ JozÏef Stefan Institute Jamova 39, SI-1000 Ljubljana (Slovenia) E-mail:
[email protected] Dr. H. Cohen Chemical Services Unit, Weizmann Institute of Science Rehovot 76100 (Israel) Dr. R. SanjinØs, Prof. F. LØvy Institut de Physique de la Matiere Complexe Ecole Polytechnique FØdØrale de Lausanne CH-1015 Lausanne (Switzerland) [**] We acknowledge M. Cantoni and P. Stadelmann, Ecole Polytechique FØdØrale de Lausanne for help in electron microscopy, and Z. kraba, JozÏef Stefan Institute, for technical assistance in the first crystal growth experiments.
Adv. Mater. 2005, 17, No. 7, April 4
Figure 1. ED and microscopy images of MoS2±C60 composite materials: a) The [001] zone with intense f110g C60 spots arranged in triplets due to the incommensurate relationship of MoS2 and C60 lattices. b) The [1261] zone indexed in accordance with the model structure; the white circles determine the exact positions of the spots. c) Transmission electron microscopy (TEM) image of strongly bent crystal flake with extinction contours and a high density of edge dislocations caused by stacking faults and scanning electron microscopy (SEM) image of the thin crystals (insert). d) TEM image of unusual diffraction contrast typical for dissociation of dislocation loops.
DOI: 10.1002/adma.200400553
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
911
COMMUNICATIONS 912
C60 monolayers, where [100]C60 is parallel to [100]MoS2, and (001)C60 is parallel to (001)MoS2. The MoS2 and C60 sublattices are in a mutually incommensurate relationship; thus, the diffraction spots belonging to the C60 sublattice are indexed separately and presented in italics, while the main spots corresponding to the reciprocal lattice of MoS2 are not italicized. The intense f110g C60 spots (Fig. 1a) are arranged in triplets. While the spot closest to the transmitted beam results from primary scattering by the f110g C60 planes, the other two spots of the triplet are due to double scattering. The MoS2 {hk0} diffraction spots are surrounded by a high density of satellites due to the secondary scattering. The secondary scattering evidences that the C60 is incorporated within the crystals. The dominant satellite along the directions (marked by a white circle) revealing an 0.316 nm interplanar distance is due to the positional relationship of C60 in relation to the molybdenum layers. The {100} C60 spots are much weaker than the corresponding {110} ones, revealing the prevailing rhombohedral stacking of the C60 layers where the ±h + k + l = 3n rule holds. While the [001] zone (Fig. 1a) can be explained by superposition of separated phases, namely blocks of MoS2 and the hexagonal phase of C60, the [1261] zone (Fig. 1b) evidences the regular composite superstructure of the crystals, and cannot be indexed as pure MoS2 lattice or as hexagonal C60. The diffraction spots of the (120Å), (22Å.12), and (10.12) types indexed in accordance with the model structure could be attributed to the (110), (203), and (103) planes, respectively, of pure MoS2, but the appearance of strong diffraction spots indexed as (116Å) planes provides evidence of the highly extended structure of MoS2 perpendicular to the basal plane. In spite of the incommensurate relationship of the two layer components, the mutual interactions between MoS2 and C60 monolayers are relatively strong. The crystals are found without the rotational disorder otherwise typical for MoS2-layered crystals. Moreover, the intensities of the superlattice spots increase with the electron-irradiation time. This effect is believed to arise from the electrostatic interactions between C60 and the MoS2 host crystal, enabling the partial mobility and reorganization of charged C60 molecules. Strong undulations of the crystals (Fig. 1c) and the formation of the contrast typical for highly dissociated dislocation loops (Fig. 1d) evidence a lattice strain incorporated in the thin crystal flakes. A high density of stacking faults results from the relaxation of the lattice strain. The domains of the short-range-ordered C60 molecules in the high-resolution TEM (HRTEM) image (Fig. 2) show a 3 3 superstructure (right insert) with regard to the MoS2 lattice. The mismatch of the sublattices appears as boundary areas with less-pronounced features. These areas are associated with the diffuse scattering in the XRD patterns, as well as in the Fourier transform of the high-resolution TEM image (Fig. 2, left insert). Note that the real-space image clarifies the nature of the a*/3 diffraction spots in Figure 1. In contrast to the a* spots, which correspond to small and bright dots in the real-space image, the a*/3 diffraction spots correspond to
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 2. HRTEM lattice image of the MoS2±C60 composite material (scale bar: 2 nm). The domains of a 3 3 superstructure (with regard to the MoS2 lattice) caused by the short-range-ordered C60 molecules (right insert, scale bar: 2 nm) are separated by boundary areas with a less-pronounced superlattice. This long-range disorder causes the diffuse scattering in XRD and in the Fourier transform of the lattice (left insert).
the faint and large real-space features. They show a shortrange regular sinusoidal functionality (see, for example, the inset and the top-right section of Fig. 2). The size and brightness of these features strongly support their assignment to the C60 molecules. The XPS study of surface composition is presented here by comparing the MoS2±C60 composite with a reference MoS2 crystal that was grown in the same setup, but without C60 and cleaved before insertion into the vacuum. The atomic concentration of carbon in the composite, (2.6 ± 0.2) at.-%, is low compared to the value expected for the bulk structure. This may arise from the washing procedures, designed to remove surface-adsorbed C60 molecules. Binding energies of the various elements were found to be similar for the MoS2±C60 and the reference MoS2. The overall sample conductivities were good enough in both samples to allow relatively accurate determination (± 0.1 eV at worst) of the energy scale (see Experimental). The extracted binding energies of the composite Mo(3d5/2) and S(2p3/2) lines are 229.9 eV and 162.7 eV, respectively, and 230.1 eV and 162.9 eV in the reference MoS2 crystal. The slight difference in the binding energies is close to the experimental error, and thus should be considered carefully. The C(1s) lines, 284.5(5) eV in the composite and 284.6(8) eV in the reference crystal, retain a similar difference between the two samples. The fine and rich low-energy spectral features characterizing solid C60 have been previously reported. Shake-up satellites of the C(1s) line have been resolved by high-resolution XPS.[10] Here, using a conventional X-ray source, only broad features are observed, as shown in Figure 3. This spectral region overlaps with oxidized carbon states, and hence cannot be analyzed quantitatively. Yet, relying on our complementary evidence for the incorporation of C60 within the composite, one may deduce from the shake-up features that the present phase of the C60 indeed differs from the above-cited pure solid-C60 phases. The shake-up processes in this case involve plasmonic collective excitations, which consist of long-range dipole interactions.[11] Therefore, they are expected to be sensitive to the nanometer-scale environmentÐin the present
http://www.advmat.de
Adv. Mater. 2005, 17, No. 7, April 4
CABCABCABCABC
100
001
Figure 3. Shake-up features of the C(1s) spectral region (the lowest arrow). The zero point of the binding-energy scale refers to the position of the main line.
COMMUNICATIONS
010
Figure 4. A model structure of MoS2±C60 composite material: a) View perpendicular to the layers, where the filled circles correspond to C60; and the empty circles correspond to Mo belonging to the first basal plane. b) A view along the (110) direction at position B in the CABCAB stacking order of MoS2 and the C60 molecules. The positions of the C60 molecules are drawn without the consideration of the incommensurate relationship between the MoS2 and C60 sublattices.
observed by XPS. Despite the fact that the superstructure is case, the MoS2 matrix. Spectral deviations compared with the created during the MoS2±C60 crystallization, indicating that pure C60 solid phases were therefore expected. long-range interactions must be involved, the spectroscopic On the basis of ED and XRD data, we propose a model data suggest only weak, short-range interactions between the structure consisting of MoS2 molecular layers with rhomboheMoS2 and C60 layers. This is a typical characteristic of the asdral stacking alternating with the C60 monolayers (Fig. 4). The sembly of van der Waals' species like C60 molecules and dominant stacking of both particular molecular layers is eviMoS2 into a layered material. On the other hand, these strucdent from the high intensity of the {110} C60 diffraction spots tures may exhibit much stronger interactions in the excited in comparison with the {100} C60 spots. Such an intensity distristate. bution is typical for the rhombohedral stacking, where the The growth mechanism of these complex MoS2±C60 crystals {100} spots are ruled out (h + k + l ¹ 3n), but a high density of is not yet clear. They grow at a relatively high temperature stacking faults causes their appearance, albeit with weak (1030 K). At this temperature, the silica ampoule in which the intensities, in the diffraction pattern. In the basal plane, the latcrystals are synthesized radiates visible (yellow±orange) light. tice parameters of MoS2 (0.316 nm) and C60 (1.004 nm) Considering that MoS2 is a semiconductor,[12] we suggest here are incommensurate. The composite material can be described that the growth mechanism is photo-assisted. Under visible by two unit cells which differ in the basal plane but match light, electron±hole pairs are excited in the MoS2 layers. The perpendicular to the layers: a(MoS2) = 0.316 nm; a(C60) = 1.004 nm; c(C60MoS2,) = 4.833 nm. The superlattice unit cell consists of three MoS2 molecular layers Table 1. Comparison of X-ray and ED data. Assignment for the rhombohedral and three C60 monolayers. One double layer composed lattices: a (MoS 2 ±C 6 0 ) = 0.316 nm; a(C 6 0 ) = 1.004 nm; c(MoS 2 ±C 6 0 , C60) = 4.833 nm. of a MoS2 layer and a single layer of C60 is estimated to be 1.208 nm in thickness. The value is larger than the Measured d [a] C60-MoS2 C60 sublattice C60-MoS2 C60 sublattice sum of both molecular layers considering the van der [nm] Calculated d Assignment[b] Calculated d Assignment[b] Waals' distances of sulfur and C60. The interplanar dis[nm] [nm] tances listed in Table 1 correspond to the measured »1 0.870 100 X-ray data. Large distance maxima are superimposed 0.857 101 over diffuse scattering caused by long-range disorder. diffuse scattering 0.819 102 Some low-index diffraction spots appear with low 0.765 103 0.700 0.704 104 intensity due to the high density of stacking faults, such 0.600 0.604 008 0.604 008 as (100), (102), and (103), although they are prohibited 0.412-diffuse 0.403 00.12 by the ±h + k + L= 3n rule in accordance with the clos0.273 0.274 100, 101 est symmetry group R3m (160, C53v). 0.272 102 The local variations of the relative intensities of the 0.267 0.267 104 0.253 0.260 106 diffraction spots belonging to MoS2 and C60 evidence 0.228 0.226 10.12 the inhomogeneous distribution of C60 molecules. The 0.183 0.186 10.19 variations of the C60 concentration lower the predicted 0.158 0.158 110 value of the stoichiometry from the (C60)xMoS2, x = 1 / 9 in accordance with the model, to the x » 1 / 40 [a] ± 0.001 nm. [b] Miller planes in the form (hkl).
Adv. Mater. 2005, 17, No. 7, April 4
http://www.advmat.de
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
913
COMMUNICATIONS
known electron affinity of C60[13] and the vicinity of the MoS2 layers could result in charge separation, directing further growth of alternative C60 and MoS2 monolayers. The intermolecular distance of C60 molecules measured at room temperature matches the value known for the rhombohedral crystal phase; therefore, one can conclude that the C60 molecules are in a neutral state when growth is finished and the material has cooled down. But, we can only speculate what was happening at 1030 K. In-plane long-range order is obviously obtained by close packing of C60. It is not clear how the C60 molecules ªfeelº each other through MoS2 molecular layers and form a rhombohedral stacking order in the perpendicular direction. Also, it remains an enigma how MoS2 layers stack in regular way considering the incommensurate relationship with the sandwiched C60 layers. Although a weak interlayer interaction is observed at room temperature, it apparently plays an important role in the excited state at elevated growth temperatures, affecting the charge modulations in the process of charge separation and recombination. Based on thermal-expansion data, which are available only for C60[14] and MoS2[15] bulk crystals, the mismatch of both lattices slightly decreases with temperature. Although the agreement is not sufficient for the epitaxial growth and the domain structure that appears, it can explain the strong undulation of C60±MoS2 crystals observed at room temperature. The large distance between two neighboring C60 molecules and their inhomogeneous distribution, together with the incommensurate relationship of the sublattices, can both contribute to the deformation of the nanometer-scale structure forming the dislocation loop assemblies. In conclusion, complex MoS2±C60 crystals have been grown by a chemical-transport reaction and characterized by a variety of electron microscopy and spectroscopic techniques. These samples represent the first case of layered crystals composed of alternating MoS2 and C60 molecular layers. The inherent close proximity of photovoltaic-active MoS2 layers to C60 molecules with strong electron affinities suggests a new kind of material for solar-cell applications with high quantum yields of photoinduced charge generation.
Experimental The MoS2-layered crystals with inserted C60 molecules were grown by a catalyzed transport method [9]. To MoS2 powder in the transport tube, 2 wt.-% of C60 was added. The transport reaction was typically run for 22 days at 1030 K in an evacuated silica ampoule at a pressure of 10±3 Pa with a temperature gradient of 6 K cm±1. Iodine had been used as a transport agent. Approximately 10 wt.-% of the starting material was transported by the reaction to form very thin and strongly undulated layered crystals. The zone of the crystal growth was approximately 2 cm wide and clearly separated from the end of silica ampoule, where bundles of MoS2±x nanotubes grew at a slightly lower temperature (1010 K) [9]. The transported material was subsequently thoroughly washed with toluene to remove C60 molecules that were not incorporated into the crystals. The structure of the transported material was studied using X-ray diffraction, transmission electron microscopy (300 keV Philips CM300, JEOL JEM-2010F) and scanning tunneling microscopy
914
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
(Omicron UHV-STM). Electronic structure was studied by X-ray photoelectron spectroscopy (XPS) using the ESCA±Microscopy beamline at the Elettra synchrotron light source (Trieste, Italy). Additional XPS measurements were performed on a Kratos AXIS-HS setup using a monochromatized Al Ka source and a detection pass energy of 20 eV. In order to reduce charging effects, the powder was pressed on an In foil. By comparing measurements under different charging conditions, namely different X-ray or electron-flood-gun fluxes, determination of the influence of charging effects was achieved [11,16]. In-depth compositions of the MoS2±C60 and MoS2 samples were analyzed by Auger electron spectroscopy in combination with Ar+ sputtering using an Auger microprobe (Physical Electronics, Model 545 A). The sputtering rate was determined to be ~2.0 nm min±1. The analyses were performed using a primary electron beam of 3 keV with a diameter of 30 lm. Received: April 13, 2004 Final version: October 25, 2004
±
[1] W. I. F. David, R. M. Ibberson, T. J. S. Dennis, J. P. Hare, K. Prassides, Europhys. Lett. 1992, 18, 219. [2] W. I. F. David, R. M. Ibberson, J. C. Matthewman, K. Prassides, T. J. S. Dennis, J. P. Hare, H. W. Kroto, R. Taylor, D. R. M. Walton, Nature 1991, 353, 147. [3] R. M. Fleming, T. Siegrist, P. Marsh, B. Hessen, A. R. Kortan, D. W. Murphy, R. C. Haddon, R. Tycko, G. Dabbagh, A. M. Mujsce, M. L. Kaplan, S. M. Tahurak, Mater. Res. Soc. Symp. Proc. 1991, 206, 691. [4] A. Goldoni, C. Cepek, S. Modesti, Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 2890. [5] J. G. Hou, Y. Jinlong, W. Haiqian, L. Qunxiang, Z. Changgan, Y. Lanfeng, W. Bing, D. M. Chen, Z. Qingshi, Nature 2001, 409, 2001. [6] L. F. Yuan, J. Yang, H. Wang, C. Zeng, Q. Li, B. Wang, J. G. Hou, Q. Zhu, D. M. Chen, J. Am. Chem. Soc. 2003, 125, 169. [7] M. Sakurai, H. Tada, K. Saiki, A. Koma, H. Funasaka, Y. Kishimoto, Chem. Phys. Lett. 1993, 208, 425. [8] C. J. Brabec, N. S. Sariciftci, J. C. Hummelen, Adv. Funct. Mater. 2001, 11, 15. [9] M. Remskar, A. Mrzel, Z. Skraba, A. Jesih, M. Ceh, J. Demsar, P. Stadelmann, F. LØvy, D. Mihailovic, Science 2001, 292, 479. [10] J. H. Weaver, J. L. Martins, T. Komeda, Y. Chen, T. R. Ohno, G. H. Kroll, N. Troullier, R. E. Haufler, R. E. Smalley, Phys. Rev. Lett. 1991, 66, 1741. [11] H. Cohen, E. Kolodney, T. Maniv, M. Folman, Solid State Commun. 1992, 81, 183. [12] E. Bucher, in Materials Research Society Symposium Proceedings (Ed: A. Aruchamy), Vol. 14, Kluwer Academic Publishers, Dordrecht, The Netherlands 1992. [13] D. L. Lichtenberger, K. W. Nebesny, C. D. Ray, D. R. Huffman, L. D. Lamb, Chem. Phys. Lett. 1991, 176, 203. [14] W. Vogel, Appl. Phys. A: Mater. Sci. Process. 1996, 62, 295. [15] S. H. El-Mahalawy, B. L. Evans, J. Appl. Crystallogr. 1976, 9, 403. [16] I. Doron Mor, A. Hatzor, A. Vaskevich, T. van der Boom-Moav, A. Shanzer, I. Rubinstein, H. Cohen, Nature 2000, 406, 382.
______________________
http://www.advmat.de
Adv. Mater. 2005, 17, No. 7, April 4
