Article pubs.acs.org/JPCC
When the Sequence of Thin Film Deposition Matters: Examination of Organic-on-Organic Heterostructure Formation Using Molecular Beam Techniques and in Situ Real Time X‑ray Synchrotron Radiation E. R. Kish,† R. K. Nahm,† A. R. Woll,‡ and J. R. Engstrom*,† †
School of Chemical and Biomolecular Engineering and ‡Cornell High Energy Synchrotron Source, Cornell University, Ithaca, New York 14853, United States S Supporting Information *
ABSTRACT: We have examined the growth of bilayers and superlattices of pentacene and perylene derivatives (PTCDICn) using in situ real time X-ray synchrotron radiation techniques and ex situ atomic force microscopy. We find that the growth of PTCDI-Cn layers on 1 monolayer (ML) of pentacene is initially 2D layer-by-layer (LbL), eventually transitioning to a mode of growth that is more 3D after several monolayers have been deposited. We find that the extent of 2D LbL growth depends on the length of the alkyl end chains, Cn: the smoothest films are formed with PTCDI-C5, while the roughest are formed with PTCDI-C13. These observations reflect a difference in the Ehrlich−Schwoebel barrier for step-edge crossing with alkyl end-chain length. When the sequence of deposition is reversed, we observe spectacular changes in the evolution of surface roughness for the growth of pentacene thin films on 1 ML of PTCDI-Cn. The growth is immediately 3D, while still remaining crystalline. The morphology of these thin films indicates significant reorganization of the deposited pentacene during and/or subsequent to growth, producing 3D islands that can only form if “uphill” transport is operative. Surface energy is driving this process, where the growth of high surface energy layers (pentacene) on low surface energy materials (PTCDI-Cn) is not favored, resulting in significant reorganization. Well-ordered superlattices of pentacene and PTCDI-Cn cannot be grown, consistent with our results on the bilayers. We find that thin film roughness increases abruptly with the deposition of the second ML of pentacene or the formation of the first pentacene-on-PTCDI-Cn interface. Our results indicate that the successful growth of superlattices of small molecular organic thin films will require matching of surface energies to minimize the driving forces for reorganization.
I. INTRODUCTION Complex conjugated molecules have been extensively studied for applications in thin film electronics and photovoltaics due to their electronic properties and ability to form highly ordered films at relatively low temperatures. Of particular interest is the challenge of integrating both p-type or donor (e.g., pentacene) and n-type or acceptor small molecule organic semiconductors into the same device structure. This is necessary for the fabrication of devices such as small molecule based photovoltaics, field effect transistors, ambipolar devices, and complementary inverters.1 Heterostructures composed of inorganic semiconductors have long been studied for a variety of applications.2 So-called superlattices composed of alternating layers of two different inorganic semiconductors have been fabricated from a variety of materials3 for photonic4 and photovoltaic devices.5 Such inorganic superlattice structures, when examined by X-ray diffraction,6 display distinct diffraction peaks, demonstrating a well-ordered superstructure. However, constructing such superlattices from organic materials is likely to be considerably more difficult, since organic molecules are bound to each other by weak van der Waals interactions rather than the strong covalent © XXXX American Chemical Society
bonds present in the heteroepitaxial growth of inorganic semiconductors.7 And while roughness in inorganic heterojunctions is often driven by lattice mismatch,8 for organic− organic heterostructures, the vacuum/thin film surface energies and interactions at the interface between the two organic components pose a serious concern.9−12 In the fabrication of organic photovoltaic devices, the interface between layers of n-type and p-type semiconductors is of particular interest, as it plays a vital role in charge separation. Heterostructures more complex than simple bilayer junctions are desirable for organic photovoltaic devices due to the fact that the efficiency of a bilayer device is limited as charge separation only occurs near the donor−acceptor interface.13 The desirable thickness of an active layer in such a device is limited by the relatively short diffusion lengths of excitons in organic semiconductors, typically on the order of 10 nm.14 This, combined with the desire for high interfacial area to facilitate charge separation, has led to investigation of more Received: February 19, 2016 Revised: February 24, 2016
A
DOI: 10.1021/acs.jpcc.6b01717 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C complex structures, such as codeposited films and multilayer stacks.15−18 To examine the complicated structures of such films, X-ray scattering has been shown to be a useful tool.19 In addition to such structures based on small molecule organic semiconductors,20 bulk heterojunctions composed of semiconducting polymers have been extensively studied.21 In this study, we examine directly the vapor phase growth of heterojunctions composed of the p-type organic semiconductor pentacene and a group of structurally similar n-type organic semiconductor molecules. Each n-type molecule consists of an aromatic perylene core and aliphatic side chains of different lengths, namely, N,N′-ditridecylperylene-3,4,9,10-tetracarboxylic diimide (PTCDI-C13), N,N′-dioctylperylene-3,4,9,10-tetracarboxylic diimide (PTCDI-C8), and N,N′-dipentylperylene3,4,9,10-tetracarboxylic diimide (PTCDI-C5). The structures of these three molecules are illustrated in Figure 1. Both PTCDI-
described in detail elsewhere.29 Briefly, the system consists of four separately pumped chambers: a main scattering chamber, a source and antechamber, which act to produce the supersonic beam, and a fast entry load-lock. All chambers are pumped by high-throughput turbomolecular pumps. The base pressure of the chamber was typically ∼4 × 10−9 Torr, and samples were loaded via the load-lock chamber, which was evacuated to ∼10−7 Torr prior to sample transfer into the main chamber. Substrates were Si(100) wafers (Wacker-Siltronic, p-type, 100 mm diameter, 500−550 μm thick, 38−63 Ω·cm) subject to a SC-1 clean, 15 s HF dip, and a SC-2 clean followed by growth of ∼300 nm thick SiO2 films by wet thermal oxidation at 1100 °C. Next, these wafers were cleaned and degreased by sonication for 15 min in anhydrous CHCl3 solution (99%+), sonicated in deionized (DI) water for 15 min, washed with DI water, dried with N2, and exposed to UV-ozone for 15 min. These processes provided a clean and reproducible hydrophilic surface. Supersonic molecular beams of PTCDI-C13, PTCDI-C8, and PTCDI-C5 (all 99.8% Sigma-Aldrich Corp.) were generated by passing a carrier gas (He, 99.999%) through a temperaturecontrolled container (the evaporator) containing these species located upstream of the nozzle (150 μm orifice). The doubly differentially pumped beam passed through a trumpet shaped skimmer into an antechamber and through an aperture that produced a well-defined beam spot on the substrate surface. The mean kinetic energy of the molecules in the supersonic molecular beam can be controlled by adjusting the flow rate of the carrier gas. During deposition the substrate temperature was kept at Ts ∼ 40 °C, and in all cases the supersonic beam was incident along the surface normal. In addition to the supersonic molecular beam source, we also make use of a more conventional thermal effusion source (CreaTec Fischer & Co. GmbH) to generate essentially thermal energy incident fluxes of pentacene. This source possesses a 10 cm3 crucible constructed of pyrolytic BN, and it is fitted with a pneumatically controlled shutter. In our system, the source is mounted directly to the main scattering chamber of UHV system (angle of incidence is 45° off the substrate surface normal, 10 cm from the substrate surface), which also houses the sample. For deposition from the thermal source a translatable shadow mask, possessing a square 15 × 4 mm2 opening, ∼5 mm from the substrate surface, was used to define a beam spot on the sample. During these experiments, the thermal effusion source was heated to a temperature (ca. 105− 110 °C) to achieve the desired flux, the shadow mask was moved into place, and the shutter was opened and then closed to produce the desired exposure. Time-resolved and in situ measurements of the intensity of the scattered X-rays were made using a silicon avalanche photodiode detector (APD, Oxford Danfysik, Oxford, UK). During PTCDI-Cn and pentacene thin film growth the intensity was monitored at the anti-Bragg position (001/2; qz = qBragg/2), which is an effective monitor of the nature of growth, i.e., layerby-layer (LbL) vs 3D islanded growth.30 Following deposition and X-ray analysis, the samples were removed for ex situ analysis using atomic force microscopy (AFM), conducted in tapping mode using a DI 3100 Dimension microscope. The X-ray data at the anti-Bragg position were fitted using a modified version31,32 of the mean-field rate equation model of growth first proposed by Cohen and co-workers.33 Briefly, the equations for the coverage of individual layers (θn) are given by
Figure 1. Space-filling models for the molecules of interest here: pentacene and N,N′-dipentyl-, N,N′-diocttyl-, and N,N′-ditridecylperlyene-3,4,9,10-tetracarboxylic diimide (PTCDI-C5, PTCDI-C8, and PTCDI-C13).
C13 and PTCDI-C8 have been examined previously, including studies by our group22 and by others concerning their behavior in thin films and as the active material in devices. For example, heterojunctions of pentacene or other p-type organic semiconductors and PTCDI-C8 or PTCDI-C13 have been used to fabricate photovoltaic and ambipolar devices.1,23−25 PTCDI-C5 is comparatively less studied,26−28 but it has been shown to form an ordered crystalline thin film when deposited using vapor phase deposition.26,28 Here we report on the thin film growth of PTCDI-Cn on ultrathin films of pentacene and the inverse, the thin film growth of pentacene on ultrathin PTCDI-Cn films, using a combination of in situ and ex situ surface sensitive techniques. In addition to bilayer structures, we also examine more complex multilayer structures consisting of alternating monolayers of pentacene and PTCDI-Cn. In this work, we deposit thin films of PTCDI-Cn in ultrahigh vacuum (UHV) using a collimated supersonic molecular beam, while pentacene thin films are grown using a more conventional thermal evaporator. Most importantly, we make use of in situ, real time synchrotron X-ray scattering to monitor the dynamics of thin film growth. In selected cases we also employ ex situ analysis of thin film morphology using atomic force microscopy. We will find that the application of these two complementary techniques is decisive concerning the study of these systems.
II. EXPERIMENTAL PROCEDURES The experiments were carried out in the G3 station of the Cornell High Energy Synchrotron Source (CHESS) in a custom-designed UHV system fitted with Be windows that is B
DOI: 10.1021/acs.jpcc.6b01717 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C dθn = Sn − 1F[(θn − 1 − θn) − αn − 1(θn − 1 − θn)] dt + SnFαn(θn − θn + 1)
(1)
where n = 0 represents the substrate, n = 1 the first molecular layer, etc., Sn is the probability of adsorption for molecules incident on the nth layer, F is the incident molecular flux (ML s−1), and αn is the fraction of molecules that after initially impacting and landing on top of the nth layer then drop down and become part of the nth layer via some mechanism (e.g., overcoming the Ehrlich−Schwobel barrier at a step edge). In this model we also assume that there are two values for the probability of adsorption: one for adsorption on the substrate (S0) and one for that on previously existing molecular layers, independent of their thickness (S1 = S2 = S3 ...). Once layer coverages have been calculated by integrating eq 1, these can then be used to calculate the scattered X-ray intensity as a function of time.30−33 The intensity of the scattered beam (I) depends upon the layer population, θn(t), according to the relationship ∞
I(t ) = |rsubse−iφ + rfilm ∑ θn(t )e−iqzdn|2 n
(2)
where rsubs and rfilm are the scattering amplitudes of the substrate and the film, ϕ is the phase change upon reflection, qz is the out-of-plane scattering vector, and d is the out-of-plane interplanar spacing. At the anti-Bragg position, qzd = π, which results in a change in the sign of the thin film terms in the summation. If each layer fills sequentially, such as in perfect LbL growth, an oscillation in the intensity results.
Figure 2. (a) A 3 × 3 μm2 atomic force micrograph of ∼1 ML pentacene deposited on SiO2 at room temperature. (b) Histogram of the surface heights of the micrograph shown in (a). Here the zero of height is the mean height, which in this case is approximately the height of the first monolayer.
III. RESULTS A. Growth of PTCDI-Cn on Pentacene. Here we will describe the growth of PTCDI-Cn thin films on previously deposited thin films of pentacene. The thin films of pentacene of one monolayer (ML) in thickness were grown on SiO2 via thermal evaporation as described in section II (growth rate ∼0.01 ML s−1). Prior to the growth of this thin film, we made use of X-ray scattering measurements at the anti-Bragg condition to determine precisely the exposure needed to form one monolayer. We chose this thickness for the pentacene thin films such that the resulting surface would be essentially just as smooth as the starting substrate, which we now demonstrate. In Figure 2a we show an atomic force (AF) micrograph (3 × 3 μm2) of a pentacene thin film of approximately 1 ML in thickness deposited on SiO2. As can be seen, the surface consists mostly of molecules present in the first layer, with very small fractions representing the bare surface and a second and third layer. Two small protrusions (∼0.070 μm diameter, ∼0.015−0.020 μm in height, representing 0.5 μm. Features (islands) in the upper layers of this thin film also appear to be anisotropic and possess relatively straight edges. Features for the other two thin films, in contrast, are more irregular, and it is more difficult to identify large areas representing a single thickness/coverage. The surface height histograms also reflect these differences. For the thin film of PTCDI-C8 the occupancy is dominated by one layer (the fourth, at ∼64%), with smaller contributions by the two adjacent layers. The large, very flat terraces (local RMS is ∼0.17 nm) also make it easy to identify these contributions. For the thin films of PTCDI-C5 or PTCDI-C13 we see that two adjacent layers contribute similarly, and for the latter, the lack of easily identifiable terraces does not result in clearly resolved peaks in the histograms. We note that in previous work on the growth of PTCDI-C13 on a self-assembled monolayer (HMDS) at a coverage of ∼4 ML the surface height histogram was similar to that observed here for the growth of a thin film of PTCDI-C8 on 1 ML of pentacene.22 In Figure 5 we plot the RMS surface roughness for thin films of PTCDI-Cn deposited on 1 ML of pentacene (shown as
Figure 6. Specular X-ray reflectivity of thin films of (a) PTCDI-C5 (45 nm), (b) PTCDI-C8 (30 nm), and (c) PTCDI-C13 (25 nm) grown on a predeposited monolayer of pentacene at room temperature.
deposited on 1 ML of pentacene. These data have been collected ex situ, following deposition in the chamber in the G3 station, using a separate setup in the G2 station of CHESS. These films correspond to those we considered above in Figure 3. As may be seen, we observe strong diffraction in all three cases, verifying that crystalline thin films are formed. Given the range of reciprocal space analyzed, we observe Bragg peaks up to second order for PTCDI-C5, third for PTCDI-C8, and fourth for PTCDI-C13. Well-defined Laue oscillations, indicative of a well-ordered lamellar structure, are observed in all cases, and the coherent thicknesses implied by these data, D (D = 2π/Δq, where Δq is the period of the Laue oscillations about the 00l Bragg peaks), are given by (a) 45, (b) 30, and (c) 25 nm. These values compare well to those implied by the fits to the antiBragg oscillations given in Figure 3, namely (a) 40, (b) 28, and (c) 21 nm. In general, diffraction features shown in Figure 6 appear as expected based on the previously reported dz-spacings for all three molecules, and they are given by (a) 1.83, (b) 2.03, and (c) 2.68 nm. In Figure 6a, we also show the expected position for diffraction from both the thin film (dz = 1.83 nm) and bulk phase of PTCDI-C5 (dz = 1.63 nm) as well as that from pentacene (dz = 1.54 nm).26 While there seems to be a small contribution from the bulk phase [(001)] vs the thin film
Figure 5. Thin film RMS surface roughness as a function of thin film thickness for (a) PTCDI-C5, (b) PTCDI-C8, and (c) PTCDI-C13 grown on a predeposited monolayer of pentacene as produced by the fit to the X-ray data shown in Figure 3. Solid circles represent roughness obtained directly from AF micrographs. In (d) we compare the roughness produced by the fits to the X-ray data for all three cases.
smooth lines) as a function of thickness (total coverage) as produced by the fits to the X-ray data displayed above in Figure 3. For comparison, we also plot the values obtained from ex situ AFM (solid circles) for a series of thin film thicknesses (cf. Figure 4 and Supporting Information). To convert the values obtained from AFM to MLs, we use monolayer thicknesses of (a) 18.3 ± 0.1 Å for PTCDI-C5, (b) 20.3 ± 0.1 Å for PTCDIC8, and (c) 26.8 ± 0.1 Å for PTCDI-C13. As may be seen, the agreement between the two methods is in general good, where the largest deviations are observed for the PTCDI-C13 thin films, at low to moderate coverages. For each molecule, the difference between the final RMS measured by AFM and that produced by modeling the X-ray scattering is
