Maximilian
Dreher
,
Dayeon
Kang
,
Tobias
Breuer
and
Gregor
Witte
*
Molekulare Festkörperphysik, Philipps-University Marburg, D-35032 Marburg, Germany. E-mail: gregor.witte@physik.uni-marburg.de
First published on 24th July 2019
Due to their anisotropic optoelectronic properties, crystalline organic fibers constitute an interesting class of nanoscale materials with great potential for integration into future optoelectronic devices based on organic–inorganic hybrid systems. While chemical synthesis allows for flexible tailoring of electronic molecular properties, structural control of hybrid structures is hampered by the incompatibility of traditional structuring methods such as, e.g. lithography, with these molecular materials. Therefore, other approaches must be used. Here we examine the formation of crystalline fibers of dinaphthothienothiophene (DNTT), a recently synthesized chemically robust organic semiconductor with high charge carrier mobility, upon film growth on noble metal substrates. Based on a comparison of the film growth on a regularly stepped, vicinal surface, we show that substrate steps affect the azimuthal molecular orientation in the seed layer. In particular, they induce a fiber orientation which competes with that of fibers formed on flat terraces and thereby strongly limits the fiber dimensions. We demonstrate a strategy to suppress this parasitical step-induced fiber nucleation by first exposing Ag(111) surfaces to oxygen, or even briefly to ambient conditions, which causes a selective saturation of the active step sites, while subsequent deposition of DNTT yields strongly enlarged fibers that are epitaxially aligned on the (111) surface.
New conceptsSelf-organization is an important route for structuring molecular materials especially for organic/inorganic hybrid architectures. While molecular materials often form crystalline fibers, their length and orientation is limited by surface defects such as steps of the supports that cannot be prevented even on very perfect, single crystalline substrates, thus drastically restricting their use in device applications. Based on the identification of the initial fiber growth, we introduce here a new concept to suppress the defect-driven fiber nucleation at such substrate defects by selective blocking of active step sites using small molecules, so that the fiber formation is only governed by the intrinsic epitaxial growth on ideal, defect-free surface regions. This not only leads to an exclusive growth of epitaxial fibers but also strongly increases the fiber size to several hundreds of microns. This novel approach is quite versatile and allows a distinct improvement of template assisted growth and thereby the quality of organic/inorganic hybrids. |
The formation of crystalline molecular fibers was first observed upon growth of para-phenylene films on alkali halides and has attracted particular attention because these aggregates exhibit distinct luminescence properties8,9 and serve as a model system for microwave-guide applications below the diffraction limit.10 Since then, the formation of crystalline fibers has been extensively studied and is observed for several rod-like molecules upon deposition onto various substrates including silicates, metal oxides, alkali halides and boron nitride.6,11–21 Moreover, by using anisotropic substrates as template or specific electrostatic molecule–substrate couplings, an azimuthal alignment of such fibers has been demonstrated.11,14,15,17,18 Interestingly, while anisotropic metal substrates such as fcc(110) or vicinal surfaces can be used to prepare molecular seed layers with well-defined uniaxial molecular alignment,17,21–24 almost no fiber alignment is found in thicker multilayer films.23,24 Only during initial multilayer growth (essentially bilayers) a preferential in-plane alignment is maintained, while in case of the periodically oxidized Cu(110)–O(2 × 1) surface it is also preserved in thicker films.17 Crystalline fibers are frequently observed upon growth of rod-like molecules such as para-quaterphenyl or pentacene on (111)-oriented noble metal surfaces, but these are rather short and exhibit only limited azimuthal alignment.25,26 Complementary studies on polycrystalline gold surfaces demonstrate further that surface defects such as steps critically affect the nucleation, which hampers the development of fibers of mesoscopic length.27 A transfer of this concept, which also allows a controlled preparation of crystalline fibers of high performing OSCs, has not yet been systematically studied. In this regard, dinaphthothienothiophene (DNTT) is an interesting that combines excellent charge carrier mobility with high chemical robustness,28 which allows the preparation of long-term stable OFETs. In previous work it was shown that DNTT forms a commensurate seed layer on the densely packed Au(111) and Ag(111) surfaces, while subsequent multilayer growth proceeds in a Stranski–Krastanov mode yielding rather short crystalline fibers, that appear azimuthally randomly oriented,29 and therefore not suitable for integration in optoelectronic devices.
In the present work we have combined scanning tunneling microscopy (STM), atomic force microscopy (AFM), X-ray diffraction (XRD) and optical microscopy to analyze the DNTT fiber formation on different length scales, ranging from the initial seed layer structure to the crystalline texture and orientation of mesoscopic fibers. This enabled us in particular to identify nano-scale factors that control the orientation and limit the length of such fibers. Based on these results, we develop and validate various strategies to enhance the mesoscopic sized fibers and their orientation on noble metal surfaces. By using a stepped Au(998) surface we have at first analyzed the influence of steps on the structure of the seed layer and the subsequent fiber growth. This shows that steps induce different molecular alignments compared to the undisturbed (111) surface which results in a competing, different fiber orientation. Similar fiber orientations were also found on the (111) surfaces of Ag and Au and are attributed to defect step induced nucleation which limits the fiber size and orientation on the flat surfaces. In a second step we demonstrate that this undesired step-induced nucleation can be deactivated by exposing the Ag(111) surface to oxygen or air. This causes a selective covering of steps by oxygen and thereby deactivates the step-induced fiber nucleation and yields largely enhanced DNTT fibers that are exclusively epitaxially oriented along the 〈1
0〉Ag azimuth directions.
) planes parallel to the Ag(111) surface. While some fibers show short straight segments, appearing preferentially aligned along the 〈1
0〉Ag azimuth directions, they cluster and form bunches with noticeable azimuthal spread as depicted in Fig. 1g. In our previous work this fiber formation could be rationalized by the lateral structural misfit between the saturated monolayer and the bulk crystal structure of DNTT which prevents the formation of well-ordered crystalline islands of large size,29 but we were unable to further improve the fiber size and alignment.
0〉 azimuth directions (cf.Fig. 2a). At first we analyzed the initial growth of DNTT on Au(998) and compared this arrangement with the reported structures on the (111) surfaces of the noble metals. Fig. 2b shows a typical STM micrograph for a DNTT film with a nominal thickness of 2 Å, which is the expected coverage for one monolayer with flat-lying molecules. In this image one can clearly recognize two gold terraces, where 6 to 7 DNTT molecules are uniformly aligned on each terrace with their long axis parallel to the step edges. Neighboring molecules are slightly shifted and form an oblique unit cell with the vectors b1 = 17 Å and b2 = 8 Å enclosing an angle of 60° on the terraces. This structure is well described by a higher-order-commensurate
phase. This adlayer structure has a molecular surface area of about 110 Å2 which corresponds to a similar packing density as found for the planar DNTT monolayer adopted on the (111) surfaces of silver and gold (∼130 Å2 per molecule), but exhibits a different azimuthal molecular orientation (cf.Fig. 1b). Remarkably, the Au(998) surface appears structurally very stable and shows no reconstruction or faceting upon molecular adsorption, while this is frequently found for other vicinal metal surfaces.38 For slightly larger DNTT coverages a phase transition occurs that parallels the situation found before on the (111) surfaces (cf.Fig. 1d). Fig. 2d shows the beginning of this transformation for a nominal thickness of 2.2 Å on one of the terraces (indicated by the dashed line in Fig. 2d). Although the further developed structure is difficult to image (due to diffusion of admolecules) it allows proposing a structural model. Here, the unit cell is enlarged and exhibits a larger corrugation. This can be understood by adding another molecule and simultaneously levering every second molecule yielding herringbone packing which can be described by a higher-order-commensurate
phase. Interestingly, in this phase the molecules are stacked slightly differently than in the initial phase and their short sides form an angle of about 67° with respect to the step edge direction (indicated by the blue lines in Fig. 2c and d).
While the uniform step direction of the vicinal surface effectively suppresses rotational domains in the DNTT adlayer, mirror-domains still can be seen in the STM data as depicted in Fig. 3a and b. Interestingly, not only the unit cell itself, but also the DNTT molecules appear mirrored due to their pro-chiral structure causing chiral domains upon adsorption on surfaces as schematically shown in Fig. 3c. After characterizing the initial growth of DNTT on Au(998), we analyzed if and how the nanoscopic structure of the seed layer is transferred to thicker films. Interestingly, when increasing the DNTT coverage after completion of the seed layer a spontaneous formation of tall fibers with a height of several 10 nm (corresponding AFM data are presented below) takes place, which hampers their microscopic analysis by means of STM. To examine the morphology of mesoscopic films and the azimuthal fiber orientation, we have used optical microscopy as exemplarily shown in Fig. 3d for a 50 nm thick DNTT film deposited onto the Au(998) surface. It reveals the formation of isolated, elongated straight fibers with a length of 10–20 μm and a width of 0.5–1.0 μm that are oriented at an angle of 65° with respect to the gold step edge direction and occasionally appear at the corresponding mirror direction. Corresponding XRD measurements (see Fig. S3, ESI†) show that these fibers are crystalline and again exhibit (120) or (12
) contact planes like on the (111) metal substrates. Note that the fibers are much longer than the terrace width of the vicinal surface (10 μm vs. 0.004 μm) demonstrating that the fibers are overgrowing several thousand terraces. The accordance of the molecular stacking direction in the densely packed monolayer and the fiber direction thus demonstrates that the seed layer effectively controls the macroscopic fiber orientation. Although we have clearly demonstrated that the vicinal surface allows a lateral alignment of the DNTT fibers, this approach cannot be scaled up because of limitations in preparing these supports on large dimensions.
0〉Ag azimuth directions, that can be recognized in the AFM image as characteristic densely packed step edges (indicated by dashed line in Fig. 4a). A closer inspection reveals additionally branched fibers at an angle of 30° to the preferred fiber directions. This behavior is depicted schematically in Fig. 4b, where the two limiting fiber directions, one parallel and the other at an angle to the step direction are shown in blue and red, respectively. With increasing coverage these limits are filled up yielding bunches with a fan-like fiber distribution.
Next, we have studied a strategy for suppressing this parasitic, step-induced fiber nucleation. To this end, we exploit the fact that adsorption of small molecules leads to a preferred step decoration due to the increased adsorption energy at steps of metal surfaces. Here, we used oxygen, since the rather low sticking coefficient on silver39,40 causes essentially an oxygen adsorption at defects (such as steps) and leaves the defect-free regions largely uncovered. Since molecules experience a reduced substrate interaction on oxides or partially oxidized metal surfaces as compared to clean metals, it is expected that the DNTT fiber formation is decisively determined by the uncovered (111) surface regions. Fig. 4c shows an optical micrograph of a 40 nm DNTT film that was grown at 340 K on a Ag(111) surface which had been exposed to 500 L oxygen before DNTT growth. In contrast to the DNTT film formed on the untreated silver substrate (cf.Fig. 1g), only long fibers running mainly along the 〈1
0〉Ag directions are formed on the oxygen-treated surface while the parasitic fibers oriented with a rotational offset of 30° are suppressed. Our hypothesis that defect-free surface regions remain essentially free of oxygen is corroborated by the fact that the LEED pattern of the (111) surface is virtually unaffected by this oxygen exposure (cf. Fig. S5, ESI†). A systematic variation of the oxygen uptake shows that the most uniform fibers are obtained for a pre-dosing of 100–2000 L (see Fig. S6, ESI†). Moreover, we note that this pretreatment is rather robust since an improved fiber ordering is even observed for samples that are briefly exposed to ambient conditions. This is impressively demonstrated in Fig. 4d which shows an optical micrograph of a DNTT film that was grown on a Ag(111) sample that had been taken out of the UHV-system and exposed to ambient air for about 10 seconds before the sample was transferred back to the vacuum system (using a load lock system) and DNTT was deposited. Clearly, the film consists of long fibers extended over more than 50 μm that are exclusively oriented along 〈1
0〉Ag directions.
) (cf. ESI†). Both of these have rather similar inclinations with respect to the (020) plane as visualized in Fig. 5b and c (31.8° for the (120)-plane and 33.2° for the (12
)-plane). Therefore, the (020) reflections of both fiber out-of-plane orientations are simultaneously detected due to the large acceptance angle of about 5° for the inclination in our set-up which is required to collect sufficient diffraction intensity. Comparing the observed DNTT peak positions with the substrate 〈1
0〉Ag azimuth directions (shown as grey solid lines in Fig. 5a) the different (020)DNTT signals can now be identified as groups of reflection pairs, each appearing symmetrically at ±28° (shown green) and ±10° (red) relative to the 〈1
0〉Ag azimuth directions. To understand these angles we have visualized the intersections between the (020) plane and the contact planes of both fiber orientations. As shown in Fig. 5d and e the intersection lines (shown yellow) include angles of 28° and 10° with the nearest 〈1
0〉Ag azimuth direction for the (120) and (12
) fiber orientation. This reveals that both these signal groups correspond to fibers that are aligned with their stacking direction along the 〈1
0〉Ag azimuths (cf.Fig. 5d and e), which occur also as mirror domains with respect to the stacking direction. Moreover, due to the symmetry of the Ag(111) surface these DNTT domains occur also as 6-fold rotational domains, each with a DNTT-stacking along 〈1
0〉Ag, thus resulting in 12 reflections for both out-of-plane fiber orientations (indicated as green and red peaks in Fig. 5a). In addition much weaker (020)DNTT reflections are observed along directions whose surface projected azimuth is oriented along the 〈1
0〉Ag direction (denoted by black peaks in Fig. 5a). Analyzing corresponding intersections, one finds that these azimuthal positions correspond to (12
)-oriented fibers where the long axis of the molecules is aligned parallel to the step edges. In this regard we note, that this “wrong” molecular alignment occurs in thick films where the template effect of the substrate is limited and other branches are formed as directly found in the corresponding optical micrograph (see Fig. S7, ESI†), while such defect-induced fibers are almost completely excluded for thinner DNTT films (cf.Fig. 4d). However, as mentioned before, such an azimuthal in-plane XRD analysis has not been possible for the thinner films (∼40 nm) due to their low scattering intensity.
Complementarily to the XRD measurements polarization-resolved optical microscopy was used to characterize the azimuthal distribution of the DNTT fibers and their homogeneity. This analysis is illustrated by the example of a 40 nm DNTT film grown at 340 K onto a Ag(111) surface that was first exposed to air. The corresponding unpolarized optical micrograph is depicted in Fig. 6a and shows elongated fibers running exclusively along the 〈1
0〉Ag azimuth directions. For the polarization analysis a series of micrographs was recorded with linearly polarized light, wherein the polarization direction was varied. As a next step the intensity of each pixel was analyzed and the polarization angle of the maximal intensity is represented in a false color plot as shown in Fig. 6b, where the color code defines the polarization angle (cf. inset). The observed polarization contrast is caused by a distinct absorption anisotropy of the fibers. As shown in previous work,41 DNTT solids exhibit the main absorption band in the visible range around 2.75 eV (≙580 nm) due to an excitonic excitation. The corresponding transition dipole (TDM) moment is essentially given by the TDM of the lowest energy transition in DNTT molecules, which is oriented along the long axis of the naphthalene units (as depicted in Fig. 6d). Since all molecules in the solid are stacked with their long axis nearly perpendicular to the stacking direction, resp. long axis of the fibers, of the (120)- and (12
)-oriented fibers (cf.Fig. 6d), these do not absorb light if the E-vector of the illuminating light is oriented along the fiber direction (i.e. perpendicular to the TDM) and thus appear brightest in that case. The distinct fiber orientation is also evident by the polar histogram shown in Fig. 6c where all pixels with the same polarization angle of maximum intensity are summed up and plotted as a function of the azimuthal polarization. While this polarization analysis map in Fig. 6b clearly reflects the distinct azimuthal distribution of the fibers, it also provides information about the homogeneity of the molecular arrangement within the fibers when considering the signal broadness observed in this evaluation. This shows that despite a uniform molecular alignment for the individual azimuthally oriented fibers, they reveal some inhomogeneity along the fibers and sometimes appear broken.
![]() | ||
| Fig. 6 (a) Optical micrograph of a DNTT/Ag(111) film with a nominal thickness of 40 nm deposited at 340 K. (b) Polarization analysis map of image (a), where colors represent the polarization angle of maximal pixel intensity. The polar plot in (c) represents a histogram of the occurrence of pixels in (b) with a certain polarization angle. Panel (d) shows the TDM of the lowest energetic excitation in DNTT,41 together with the molecular arrangement adopted in a (120)-oriented fiber. | ||
While using the regular step array of a vicinal surface enables a mesoscopic alignment of the DNTT fibers as demonstrated for the Au(998) substrate, this approach cannot be scaled up because of limitations in preparing such templates with large surfaces. Yet, we were able to develop a strategy to reliably suppress this defect-driven, parasitic fiber growth, which competes with the intrinsic epitaxial alignment of DNTT adopted on the undisturbed, densely packed (111) surface and, as a result, limits the dimensions of the fibers. Combining X-ray diffraction, AFM and polarization resolved optical microscopy we show that first exposing the Ag(111) substrate to oxygen causes a selective adsorption at steps and thereby deactivates the step induced nucleation. The subsequent DNTT film growth results in the formation of largely enhanced crystalline fibers extending over more than 100 μm that are exclusively oriented along the 〈1
0〉Ag azimuth directions which validates our approach.
Since highly ordered and very smooth (111)-oriented noble metal surfaces can be easily and reliably prepared by epitaxial film growth on mica supports, this approach can be scaled up also to larger surfaces. In a more general scheme, these substrates may serve as template for the fiber growth, while the fibers can be later transferred onto the target substrate. We envision that the presently introduced approach of improving organic film ordering by suppression of unwanted nucleation by initial selective adsorption of small molecules at steps can also be applied to other molecular systems. Therefore, this approach should be helpful for various hybrid systems where structural control on a mesoscopic scale is required.
0〉-oriented step-edges extending over several microns.31 Prior to each organic film deposition, the metal surfaces were cleaned in situ by repeated cycles of Ar+ sputtering (Eion ≈ 800 eV) and annealing (≈730–750 K) until a sharp low energy electron diffraction (LEED) pattern with a low background signal was observed and no traces of contaminations were found by STM. The DNTT (Sigma-Aldrich, purity: 99%) thin films were prepared under ultra-high vacuum (UHV) conditions by organic molecular beam deposition (OMBD) from aluminum crucibles of a resistively heated Knudsen cell. The deposition rate (typical ≈ 4 Å min−1) was monitored by a quartz crystal microbalance (QCM) and the metal substrates were kept at 340 K during deposition. Additional DNTT films were grown onto silver substrates which had been exposed to oxygen (Praxair, purity 6.0) dosages between 100 and 2000 L (1 L ≡ 1.33 × 10−6 mbar s) by backfilling the UHV chamber to a pressure of 1 × 10−6 mbar.
Footnote |
| † Electronic supplementary information (ESI) available: Additional XRD, LEED and AFM data as well optical micrographs. See DOI: 10.1039/c9nh00422j |
| This journal is © The Royal Society of Chemistry 2019 |