LB Films of Rodlike Phthalocyanine Aggregates: Specular X-ray

Niranjani Kumaran, Carrie L. Donley, Sergio B. Mendes, and Neal R. Armstrong ... Carrie L. Donley, Ryan M. Hernandez, Chet Carter, Michael D. Carducci...
5 downloads 0 Views 584KB Size
Download PDF
7998

Langmuir 2004, 20, 7998-8005

LB Films of Rodlike Phthalocyanine Aggregates: Specular X-ray Reflectivity Studies of the Effect of Interface Modification on Coherence and Microstructure Wei Xia, Britt A. Minch,† Michael D. Carducci, and Neal R. Armstrong* Department of Chemistry, University of Arizona, Tucson, Arizona 85721 Received May 21, 2004 We present here X-ray specular reflectivity (XRR) characterization of the ordering of Langmuir-Blodgett films of the liquid crystalline phthalocyanine (Pc) 2,3,9,10,16,17,23,34-octakis(2-benzyloxyethoxy)copper(II) phthalocyanine, 1, on Si(100) wafers with a native oxide layer and these same substrates modified with monolayers of varying percentages of methyl- and phenyl-terminated silanes. The central copper atom in these Pc’s provides for high contrast in X-ray reflectivity for single-bilayer films of 1. The XRR data are modeled as arising from multiple layers of organic material, above and below the rows of copper atoms in the aligned Pc cores; variations in the total thickness of these films, and the spacing between the rows of copper atoms, are attributed to changes in the interaction with the substrate, and changes in the Pc orientation and side chain packing. The most pronounced effect on these parameters comes from variations in the ratio of the phenyl-silane versus methyl-silane content of the substrate modifiers and annealing of these films past their crystalline (K) f liquid crystalline (LC) mesophase transition temperature. Transfer of multiple bilayers of this Pc leads to additional changes in the thickness of each layer, eventually forming a hexagonal close-packed array, reminiscent of bulk fibers of this material, as revealed by X-ray diffraction (XRD).

Introduction Discotic mesophase materials (phthalocyanines, triphenylenes, hexabenzocoronenes, etc.) are currently of interest as the active electronic material in emerging organic field-effect transistor (OFET) and organic photovoltaic (OPV) technologies.1-19 Considerable effort has been extended to orient these disklike molecules, and * To whom correspondence should be addressed. E-mail: nra@ u.arizona.edu. † Present address: Max-Planck Institute fu ¨ r Polymerforschung, Mainz, Germany. (1) Piris, J.; Debije, M. G.; Stutzmann, N.; van de Craats, A. M.; Watson, M. D.; Mu¨llen, K.; Warman, J. M. Adv. Mater. 2003, 15, 1736. (2) Adam, D.; Schumacher, P.; Simmerer, J.; Haussling, L.; Siemensmeyer, K.; Etzbach, K. H.; Ringsdorf, H.; Harrer, D. Nature 1994, 371, 141. (3) Schouten, P. G.; Warman, J. M.; Dehaas, M. P.; Vanderpol, J. F.; Zwikker, J. W. J. Am. Chem. Soc. 1992, 114, 9028. (4) Chouten, P. G.; Warman, J. M.; Dehaas, M. P.; Vannostrum, C. F.; Gelinck, G. H.; Nolte, R. J. M.; Copyn, M. J.; Zwikker, J. W.; Engel, M. K.; Hanack, M.; Chang, Y. H.; Ford, W. T. J. Am. Chem. Soc. 1994, 116, 6880. (5) van de Craats, A. M.; Warman, J. M.; Fechtenko¨tter, A.; Brand, J. D.; Harbison, M. A.; Mu¨llen, K. Adv. Mater. 1999, 11, 1469. (6) Tchebotareva, N.; Yin, X. M.; Watson, M. D.; Samori, P.; Rabe, J. P.; Mu¨llen, K. J. Am. Chem. Soc. 2003, 125, 9734. (7) Tracz, A.; Jeszka, J. K.; Watson, M. D.; Pisula, W.; Mu¨llen, K.; Pakula, T. J. Am. Chem. Soc. 2003, 125, 1682. (8) van de Craats, A. M.; Stutzmann, N.; Bunk, O.; Nielsen, M. M.; Watson, M.; Mu¨llen, K.; Chanzy, H. D.; Sirringhaus, H.; Friend, R. H. Adv. Mater. 2003, 15, 495. (9) Kubowicz, S.; Pietsch, U.; Watson, M. D.; Tchebotareva, N.; Mu¨llen, K.; Thu¨nemann, A. F. Langmuir 2003, 19, 5036. (10) Laursen, B. W.; Norgaard, K.; Reitzel, N.; Simonsen, J. B.; Neilsen, C. B.; Als-Nielsen, J.; Bjornholm, T. Langmuir 2004, 20, 4139. (11) Kubowicz, S.; Thu¨nemann, A. F.; Geue, T. M.; Pietsch, U.; Watson, M. D.; Tchebotareva, N.; Mu¨llen, K. Langmuir 2003, 19, 10997. (12) Torsi, L.; Dodabalapur, A.; Rothberg, L. J.; Fung, A. W. P.; Katz, H. E. Science 1996, 272, 1462. (13) Katz, H. E.; Bao, Z. J. Phys. Chem. B 2000, 104, 671. (14) Smolenyak, P.; Peterson, R.; Nebesny, K.; To¨rker, M.; O’Brien, D. F.; Armstrong, N. R. J. Am. Chem. Soc. 1999, 121, 8628. (15) Zangmeister, R. A. P.; Smolenyak, P. E.; Drager, A. S.; O’Brien, D. F.; Armstrong, N. R. Langmuir 2001, 17, 7071. (16) Donley, C. L.; Xia, W.; Minch, B. A.; Zangmeister, R. A. P.; Drager, A. S.; Nebesny, K.; O’Brien, D. F.; Armstrong, N. R. Langmuir 2003, 19, 6512.

related organic semiconductors, in structures which favor the preferred conduction axes of these devices. In the case of OFET technologies, the optimal orientations in most small molecule organic semiconductors place the plane of the molecule (disks) perpendicular to the substrate plane, with close cofacial overlap between adjacent molecules, thereby favoring charge mobility parallel to the substrate plane.1-11,14-17,20,21 We have recently shown that octa-substituted phthalocyanines (Pcs), such as 2,3,9,10,16,17,23,24-octakis(2benzyloxyethoxy)phthalocyaninato copper(II) (1, Figure 1), self-organize into columnar aggregates and form rigid (∼56-67 Å thickness) bilayers on a Langmuir-Blodgett trough, which provides for horizontal transfer of highly coherent films to various substrates.14-18 Coherent aggregates typically show a column-column spacing (lateral periodicity) of ∼27 Å (inferred from 2D fast fourier transform (FFT) analysis of the atomic force microscopy (AFM) image in Figure 1). In many of these films, we have seen coherence in the rodlike aggregates of up to ∼2500 Å and root-mean-square (rms) roughness along the top surface of the rod of ∼6 Å. Both the field-effect (hole) mobilities and the direct current (dc) conductivities in these films are highly anisotropic, with transport along the column axis versus transport perpendicular to the column axis favored by more than 10× when measured on micron length scales and 100-1000× when measured on submicron length scales.14-18 Recent studies showed that the most coherent Pc films are obtained on Si(100) (17) Donley, C. L.; Zangmeister, R. A. P.; Xia, W.; Minch, B. A.; Drager, A.; Cherian, S. K.; LaRussa, A.; Kippelen, B.; Domercq, B.; Mathine, D. L.; O’Brien, D. F.; Armstrong, N. R. J. Mater. Res. 2004, 19, 20872099. (18) Donley, C. L. Ph.D. Dissertation, University of Arizona, 2003; and in preparation. (19) Flora, W.; Mendes, S.; Doherty, W.; Saavedra, S. S.; Armstrong, N. R. Langmuir, submitted for publication. Flora, W. Ph.D. Dissertation, University of Arizona, 2004. (20) Kazmaier, P.; Hoffman, R. J. Am. Chem. Soc. 1994, 116, 9684. (21) Cornil, J.; Beljonne, D.; Calbert, J. P.; Bredas, J. L. Adv. Mater. 2001, 13, 1053.

10.1021/la048736n CCC: $27.50 © 2004 American Chemical Society Published on Web 08/06/2004

LB Films of Rodlike Phthalocyanine Aggregates

Langmuir, Vol. 20, No. 19, 2004 7999

Figure 1. (a) A Langmuir-Blodgett pressure-area isotherm of Pc 1 and a schematic of its structure. (b) 100 nm × 100 nm AFM (friction) image of a single bilayer of Pc 1 on Si(100), hydrophobized with a 50:50 mixture of 1,1,1,3,3,3-hexamethyldisilazane (forming M-silane surface modification) and 1,3-diphenyl-1,1,3,3-tetramethyldisilazane (forming P-silane surface modification), showing an average column-column separation distance of d ) ∼27 Å (refs 12 and 25) and a rms roughness along the top of each Pc column of ∼6 Å. (c) The bilayer of Pc 1 modeled as five discrete layers in the X-ray reflectivity study, with the modified substrate treated as two discrete layers on silicon.

substrates with (a) native oxide surface layers hydrophobized with a mixture of 1,1,1,3,3,3-hexamethyldisilazane (M-silane) and 1,3-diphenyl-1,1,3,3-tetramethyldisilazane (P-silane) (50:50 in volume) or (b) thermally grown oxide layers (500-1000 Å) with the same modifiers.15 Other studies from this group have focused on determining the tilt angle of the Pc core in such LangmuirBlodgett (LB) films, both as single bilayers and as multilayers, on modified gold and IR-transparent silicon (Fourier transform infrared (FT-IR), reflection-absorption infrared spectroscopy (RAIRS), and transmission experiments) and on surface modified transparent broad-band attenuated total reflection (ATR) elements, for visible wavelength spectroscopy.16,18,19 The FT-IR studies focused on the use of orthogonal vibrations (in-plane stretching (νPc-O-C) bands and out-of-plane bending (δC-H) vibrations) for the determination of the compound tilt angles of the Pc core. The visible wavelength ATR studies have used the dichroism in the Q-band absorbance of the Pc to estimate similar angles. Both of these recent studies have concluded that for bilayer and multilayer films of Pc 1 the Pc core is nearly upright with respect to the substrate plane (with the tilt angle away from the normal axis being ∼3° in the RAIRS experiments and ∼10° in the ATR/UVvis studies).16-19 In this study, we focus on characterization of the film thickness and column packing in bilayer and multilayer films of Pc 1 as a function of the surface modification of silicon oxide substrates by using X-ray reflectivity (XRR) and X-ray diffraction (XRD). The ordering of the initially deposited Pc film has been studied as a function of interface properties, tuned by substrate chemical modification with M- and P-silanes in differing ratios and by annealing of the Pc films from the crystalline to liquid crystalline (K f LC) transition temperatures and back to room temperature. XRR has been extensively used to probe the surface and interface properties in ordered organic thin films; monolayer coverages are typically characterized

with synchrotron X-ray reflectivity systems, whereas greater film thicknesses can be characterized with XRR systems with more conventional X-ray sources.11,22-25 There have been some significant successes in the characterization of monolayer and multilayer phthalocyanine-based thin film materials.26-30 Measurement of the reflectivity in a specular geometry as a function of the incident angle provides information about the refractive index depth profile, namely, the electron/mass density profile along the normal axis direction, total and/or individual layer thickness, and interface roughness.24,25,31,32 The presence of copper in the Pc disk provides a critical enhancement in density contrast within each Pc bilayer, which leads to reasonable signal-to-noise in the XRR data, with reasonable data acquisition times, and the possibility of modeling the XRR data to find the effective thickness of the Pc layers to within (1 Å. These thicknesses are themselves a reflection of the average tilt angles of the Pc core and the space-filling and average tilt angles of the side chains. (22) Asmussen, A.; Riegler, H. J. Chem. Phys. 1996, 104, 8151. (23) Cowley, R. A.; Ryan, T. W. J. Phys. D: Appl. Phys. 1987, 20, 61. (24) Malik, A.; Lin, W.; Durbin, M. K.; Marks, T. J.; Dutta, P. J. Chem. Phys. 1997, 107, 645. (25) van der Boom, M. E.; Evmenenko, G.; Yu, C. J.; Dutta, P.; Marks, T. J. Langmuir 2003, 19, 10531. (26) Valkova, L.; Menelle, A.; Borovkov, N.; Erokhin, V.; Pisani, M.; Ciuchi, F.; Carsughi, F.; Pinozzi, F.; Pergolini, M.; Padke, R.; Bernstorff, S.; Rustichelli, F. J. Appl. Crystallogr. 2003, 36, 758. (27) Locklin, J.; Shinbo, K.; Onishi, K.; Kaneko, F.; Bao, Z. N.; Advincula, R. C. Chem. Mater. 2003, 15, 1404. (28) Gregory, B. W.; Vaknin, D.; Gray, J. D.; Ocko, B. M.; Cotton, T. M.; Struve, W. S. J. Phys. Chem. B 1999, 103, 502. (29) Souto, J.; RodriguezMendez, M. L.; Penacorada, F.; Reiche, J. Mater. Sci. Eng. C: Biomimetic Mater., Sens. Syst. 1997, 5, 59. (30) Gregory, B. W.; Vaknin, D.; Gray, J. D.; Ocko, B. M.; Stroeve, P.; Cotton, T. M.; Struve, W. S. J. Phys. Chem. B 1997, 101, 2006. (31) Parratt, L. G. Phys. Rev. 1954, 95, 359. (32) Nevot, L.; Croce, P. Rev. Phys. Appl. 1980, 15, 761.

8000

Langmuir, Vol. 20, No. 19, 2004

Xia et al.

Figure 2. XRR of a single-bilayer film (as-deposited) of Pc 1 on Si(100), hydrophobized with a 50:50 mixture of M- and P-silanes. The inset schematics include a still model of Pc 1 and bilayer sample models derived from XRR measurement and simulations (schematics I and II). Schematic I is in closest agreement to FT-IR (RAIRS) and visible wavelength ATR studies which provide estimates for the tilt angle of the Pc core away from the normal axis (refs 15 and 16).

Experimental Section The synthesis of the compound CuPc(OCH2CH2OBz)8 [2,3,9,10,16,17,23,24-octakis(2-benzyloxyethoxy)phthalocyaninato copper(II)] has been reported elsewhere.33 Si(100) substrates were soaked in Klean AR (93% sulfuric acid and 0.4% chromium trioxide, Mallinckrodt) for 15 min, followed by rinsing with a copious amount of 18 MΩ water (Millipore) prior to use. Solutions of 1,1,1,3,3,3-hexamethyldisilazane (M-silane) and 1,3-diphenyl-1,1,3,3-tetramethyldisilazane (P-silane) (Aldrich, used as received) at various volumetric ratios (100:0, 75:25, 50:50, 25:75, and 0:100) were made in CHCl3 (HPLC grade, EMD, no further purification).15 The substrate surfaces were hydrophobized by ultrasonication of these disilazane solutions with heat for 30 min, rinsed with a copious amount of chloroform, and blow-dried with nitrogen. A Riegler & Kerstein RK3 Langmuir-Blodgett (LB) trough was used to produce LB films. Pc solutions in chloroform were applied onto the air/water interface, allowing the solvent to evaporate before compression with two parallel barriers. A Wilhemy balance was positioned at the center of the trough to measure the surface pressure as the surface area of the Pc film was reduced during compression. Compression was stopped at ∼50 Å2 per molecule, where stable bilayer structures are formed.14-17 Films were then segmented into 15 individual squares by removing some of the water at the subphase until films fell onto the metal baffle positioned underneath. Horizontal (Schaefer) transfer of these bilayer films was then carried out to appropriately modify the Si(100) wafers.14-18 Multiple transfers can be carried out to achieve multilayer films, since transferring from one section did not disturb adjacent sections. Residual water on the substrates, between transfer steps and after the final transfer, was removed with a stream of nitrogen. Annealing of these LB films was done at 120 °C, a temperature higher than that of the liquid crystal (LC) transition (∼70 °C) under vacuum (∼10-6 Torr), for 4 h to remove any residual water and enhance the ordering within the films. (33) Drager, A. S.; O’Brein, D. F. J. Org. Chem. 2000, 65, 2257.

AFM images were taken with a multimode Nanoscope III system (Digital Instruments, Santa Barbara, CA). Oxide sharpened silicon nitride probes with a nominal force constant of 0.38 N/m were used for imaging and were ozone cleaned for ∼1 h prior to use. Pc molecular columns were recorded in contact mode in solution (1 mM KCl in 18 MΩ Millipore water) and displayed in the friction channel. Specular X-ray reflectivity (XRR) measurements of LB films on Si(100) substrates were conducted with a PANalytical X’Pert PRO MPD system with copper (KR) radiation (λ ) 1.54 Å) at 45 kV (40 mA target current). The X-ray mirror was used in the incident beam optics with a beam divergence of 1/32°. Parallel plates were used on the diffracted beam side, with receiving slits of 0.27 mm. Continuous scans along omega/2theta (ω-2θ) were obtained at step sizes of 0.005°, with 40 s of data acquisition per step. An angular resolution of 0.02° can be achieved with this setup at a primary beam intensity of ∼6.0 × 107 counts/ s. The reflected beam intensity was recorded as a function of the wave vector transfer (q ) krefl - kinc; k ) 2π/λ) along the substrate normal (qz) in a specular configuration, as shown in the inset schematic of Figure 2. The wave vector transfer (qz) is directly related to the incident angle (θi) (qz ) sin θi × 4π/λ). No offspecular/background scattering has been subtracted, and the intensities (counts per second) are given in arbitrary units, since rescaling was involved to facilitate the comparison of multiple reflectivity curves. Full pattern simulations were performed with X’Pert Reflectivity 1.0, PANalytical, where a layered sample model was constructed to generate a simulated reflectivity curve, given the initial mass density, thickness, and interface roughness parameters. Simulations were followed by a nonlinear fitting to refine the initial parameters and minimize the deviation between the experimental and simulated reflectivity curves. Three algorithms (segmented fit, genetic algorithm, and a combined fitting model) were used in the trial-and-error process with best fits chosen only when results from these algorithms agreed among themselves. X-ray diffraction (XRD) measurements of these films were conducted with the same instrument at 50 kV (40 mA target current). A fixed divergent slit of 1/4° was used followed by an

LB Films of Rodlike Phthalocyanine Aggregates antiscattering slit of 1/2°. X’Celerator, a RTMS (real time multiple strip) detector, was used in the diffracted beam optics, allowing faster data collection than conventional detectors without compromising the resolution. XRD data analysis was done with X’Pert Plus 1.0, PANalytical.

Results and Discussion XRR Studies of the Si(100)/Native Oxide Substrate and the As-Deposited Pc Films. XRR has been previously used to examine silicon wafers with various thicknesses of oxide and oxide modifiers,23-25,34 and accurate information on the oxide layer thickness, electron density, and roughness of both oxide/air and substrate/oxide interfaces has been successfully derived. Our characterization of the native oxide layers on our Si(100) substrates is detailed in the Supporting Information for this paper. With the silicon wafers used in this study, no significant density contrast at the substrate/oxide interface was observed. Our XRR data indicate that the rms roughness at the oxide/air interface is ∼6 Å, consistent with the rms roughness values determined by AFM studies;15,16 the thickness of the surface oxide layer is ∼20 Å. The silane monolayers which were the targets of this study were successfully differentiated from the underlying oxide layer with XRR, even though they possessed an interface density contrast smaller than 2:1. The critical wave vector transfer (qc), determined by the electron number density (linearly related to the mass density) of the uppermost layer (qc ∝ Fe1/2),35 was ∼0.028 Å-1 for all substrates, indicating only small density variations among all of the silane layers explored, regardless of the silane mixing ratio. The apparent density of silicon and silicon oxide stayed close to the nominal values with little variation in roughness at the substrate/oxide and oxide/silane interfaces, supporting the validity of our two-layer model of hydrophobized substrates. The apparent thickness of the nonreacted oxide layer, after modification with M- and/or P-silanes, decreased from ∼20 Å to ∼5-10 Å, which is reasonable considering that some of the oxide is converted to an interfacial layer (with low density contrast) as a result of the reaction between the oxide and the disilazane modifiers. XRR was next used to investigate ordering within singlebilayer films of Pc 1 on these hydrophobized substrates (Figure 2). For modeling purposes, the high electron/mass density contrast along the Pc rod axis, arising from the central copper atoms, provides for the possibility to divide up the Pc bilayer into five separate layers (Figure 1). Proceeding from the film/air interface, layer 1 represents the upper side chains and one-half of the Pc core region above the top row of copper atoms, which are defined as layer 2. Layer 3 is the thickest layer, representing the lower half of this uppermost Pc rod along with the lower side chains in this rod, the upper side chains in the lower Pc layer, and half of the Pc core from this layer. Layer 4 represents the row of copper atoms in the bottom Pc layer, while layer 5 represents the lower half of this Pc layer, and its side chains, interfaced to the silane modifier layer. The hydrophobized substrate represents two additional layers, oxide and silane modification layers built upon silicon wafers (see the Supporting Information). The XRR data on a single, as-deposited bilayer film of Pc 1 on Si(100), hydrophobized with a 50:50 mixture of Mand P-silanes, were first simulated (Figure 2) with this five-layer model. The thickness of a single-bilayer film (34) Kobayashi, S.; Nishikawa, T.; Takenobu, T.; Mori, S.; Shimoda, T.; Mitani, T.; Shimotani, H.; Yoshimoto, N.; Ogawa, S.; Iwasa, Y. Nat. Mater. 2004, 3, 317. (35) Als-Nielsen, J.; McMorrow, D. Elements of Modern X-ray Physics; Wiley: New York, 2001; pp 70-78.

Langmuir, Vol. 20, No. 19, 2004 8001 Table 1. XRR Fitting Parameters of a Single Bilayer of Pc 1 on Si(100), Hydrophobized with 50:50 Mixed M- and P-Silanes: Best Fit of Figure 2a layer 1 layer 2 layer 3 layer 4 layer 5 silazane SiO2 Si

density (g/cm3)

thickness (Å)

roughness (Å)

1 3.5 1.47 4.04 1.23 1.53 2.41 2.24

25.3 1.9 23.9 1.4 14.7 7.5 8.5

13 5.7 4.9 5.5 8.1 3.4 5.6 0.1

a It is possible to have a large film/air interface roughness if domains/islands exist at the micron scale.

would be ∼76 Å if the Pc molecules stand edge-on to the substrate with fully extended side chains. A total thickness of ∼67 ( 6 Å was derived instead from the fitting of the XRR data. The deviation at each organic layer in thickness is ∼10%, 5% at each copper layer, as a conservative estimate. Individual layer thicknesses derived from the fitting process were the following: layer 1, 25.3 Å; layer 2, 1.9 Å; layer 3, 23.9 Å; layer 4, 1.4 Å; layer 5, 14.7 Ås with additional silane (7.5 Å), SiO2 (8.5 Å), and substrate layers (Table 1). The uncertainty in the overall film thickness from the fitting of the XRR data in Figure 2 is therefore ∼5-10%. In these calculations, it was assumed that the Pc cores are planar, with diameters of ∼14 Å, with copper atoms at their center with diameters of ∼2 Å, and that the Pc cores are surrounded by side chains of ∼12 Å in length, if fully extended. The copper layer density which provided the best fit for the simulated data (∼3.54.0 g/cm3, Table 1) is less than half of the bulk density of copper, an expected result given that the cofacially stacked Pc cores represent a single “row” of copper atoms spaced apart by 3.5-4.0 Å. The densities of the organic portions of these layers which produced the fit in Figure 2 were consistently found to be close to 1 g/cm3. The Pc film thicknesses estimated from this XRR data provide for at least two options for describing the tilt angles of the Pc core and of the side chains. Using a “top-down” approach to analyze the individual layer thicknesses, we start by assuming that layer 1 achieves its thickness from a nearly upright Pc core and extended side chains (schematic I in Figure 2). Layer 3 achieves its thickness with the Pc cores in both layers being upright, with layers 2 and 4 ∼24 Å apart, and the side chain regions from both the top and bottom Pc layers modeled as having an effective tilt angle of ∼60° away from the normal axis. The thickness of the bottom organic layer (∼15 Å) suggested a tilt of those side chains of ∼40° away from the normal axis, at the Pc/silane interface. If a “bottom-up” approach was taken in the analysis of the film thicknesses derived from the XRR data (schematic II in Figure 2), the thickness of layer 5 can be modeled by extended side chains and a Pc core tilted ∼60° from the normal axis. Layer 3 must then have an effective side chain tilt of ∼40°, and the thickness of layer 1 is modeled by a tilt of the Pc core of ∼60° and fully extended side chains. This model, however, is not consistent with recent FT-IR (RAIRS) and visible wavelength ATR characterization of the microscopic orientation within bilayers of this same Pc, after annealing, which suggests that the Pc cores are only tilted a few degrees away from the normal axis.16,18,19 In unannealed films, there is likely to be a significant distribution of orientations of the Pc cores, and there are other orientation combinations which might fit the thickness profile, giving some freedom to the tilt angles of both the Pc cores and the side chains. In the subsequent studies described below, however, we assumed Pc core

8002

Langmuir, Vol. 20, No. 19, 2004

Xia et al.

Figure 3. XRR of a single-bilayer Pc film (as-deposited) on Si(100), hydrophobized with a mixture of M- and P-silanes at various ratios: (a) M/P ) 0:100; (b) M/P ) 50:50; (c) M/P ) 100:0. The insets are schematics of M- and P-silane surface moieties with their lengths labeled. Table 2. XRR Fitting Parameters of Hydrophobized Si(100) with M- and P-Silanes Mixed at Various Volumetric Ratios: Best Fit of Figure 3 density (g/cm3)

thickness (Å)

roughness (Å)

100MOP silazane SiO2 Si

1.3 2.4 2.3

silazane SiO2 Si

0.93 2.4 2.3

silazane SiO2 Si

1.3 2.5 2.3

silazane SiO2 Si

1.1 2 2.2

silazane SiO2 Si

0.78 2.4 2

0.3 9.1

16 5.2 0.1

5.9 4.5

4.7 11 0.5

0.1 9.7

18 5.3 0.01

5.6 5

12 3.8 0.5

8.2 4.5

4.2 5 0.1

75M25P

50M50P

25M75P

OM100P

orientations close to upright, to compare the influence of surface modification on Pc microstructures. XRR Studies of Pc Bilayers on Surfaces Modified with Different Ratios of M- and P-Silanes. XRR measurements were next considered for single bilayers of Pc 1, as-deposited on substrates modified with M- and P-silanes at various mixing ratios (Figure 3). The density, thickness, and roughness profiles throughout these modified samples were extracted via the layered sample and the substrate models discussed above, and the parameters leading to the best fits are available in Table 2. The variations in X-ray reflectivity for these different substrates are manifestations of alterations in layer thicknesses and/or reconstruction of one of the interfaces as the substrate surface composition is changed. The observed densities of both the silicon and silicon oxide layers in these experiments varied within (20% of their nominal values, with little variation in the thickness of the oxide (∼10 Å) and silane layers (∼7-9 Å), regardless of the M-silane/P-silane ratios.

Figure 4 summarizes the models used to rationalize the XRR data in Figure 3, for each of the silane mixing ratios. It is most instructive to compare the 100% methylterminated surfaces with the 100% phenyl-terminated surfaces. The 100M0P surfaces produced a bilayer of a total thickness of 49.4 Å, which suggests the side chains at the Pc film/silane interface (layer 5) are tilted ∼60° away from the normal axis. The same degree of side chain tilt could be inferred for layer 3, with higher side chain tilt angles (∼78°) in layer 1. The 0M100P modification produced a Pc bilayer with a total thickness of 76.0 Å (there is ∼10% uncertainty in the thickness value). Assuming nearly upright Pc cores in these as-deposited films suggests that the side chains in layers 1, 3, and 5 were fully extended, that is, that a very different packing structure in these Pc films is introduced with a 100% phenyl-silane modifier. Intermediate degrees of side chain tilt were inferred for the 25M75P and 75M25P substrates. The Pc side chains in layer 5 appear to change their average tilt angles in a systematic fashion, from 0° on a 0M100P surface to ∼40° on a 50M50P surface and ∼60° tilt on a 100M0P surface. The tilt angles of the side chains in layer 3 (∼50-60°) were essentially independent of surface composition. A comparison of the XRR data for a single-bilayer film before and after annealing, on both unmodified and modified oxide/Si(100) (50M50P) wafers, is shown in Figure 5 (models for these data are shown in schematics III and IV in Figure 4). Of most interest are the changes in the Pc bilayer after annealing on the modified surfaces (thickness of ∼46.4 Å for the annealed films vs thickness of 67.2 Å for the unannealed films). Annealing decreased the overall thickness of the Pc bilayer, and is modeled as an increase in the average tilt angle of all of the side chains, with no change in the Pc core orientation; even layer 1 (the upper portion of the side chains) had to be modeled with a significant effective tilt of the side chains, as in the interior layers. These apparent changes in tilt angle are consistent with an overall densification of the thin film material and a certain degree of interdigitation of the benzyl-terminated side chains with the phenyl groups on the modified surface (see below). Pc films deposited on unmodified substrates showed a significantly higher critical angle (as-deposited) which was most likely due to the exposure of surface oxide to the bilayer as a result of poor film transfer from the LB trough, making the effective density of the surface layer higher.15 There was also a dramatic decrease in the density of the copper layer (∼1.5 g/cm3, Table 4 with fitting parameters available in the Supporting Information) relative to that of the films deposited on modified surfaces. The density alteration was most likely due to the formation of lateral domains and islands of Pc 1, which has been confirmed in previous AFM studies which showed a poor quality of film transfer between the LB bilayer of Pc 1 and the bare oxide substrate.15 Layer-by-layer thickness analysis on the as-deposited film of Pc 1 on unmodified substrates suggested (schematic V in Figure 4a) that both top and bottom side chains are fully extended along the normal. Side chains sandwiched between the upright Pc cores, however, were tilted ∼70° away from the normal. Annealing introduced a significant compression to layer 5 (an effective side chain tilt of ∼85°). Density contrasts at the copper/organic interfaces were much smaller than those in the modified set of films, suggesting substantial reorganization at both the Pc/Pc and Pc/substrate interfaces due to film/substrate incompatibility. XRR Studies of Pc Multilayers. Multilayer films were constructed through a bilayer-by-bilayer deposition approach that resulted in a copper-copper “superlattice”

LB Films of Rodlike Phthalocyanine Aggregates

Langmuir, Vol. 20, No. 19, 2004 8003

Figure 4. Schematics of single-bilayer models for horizontally transferred LB films of Pc 1 on Si(100), both bare (b) and hydrophobized with M- and P-silanes mixed at various volumetric ratios (a); model construction based on XRR results from Figures 3 and 5.

Figure 5. XRR of single-bilayer films of Pc 1 on Si(100): (a) an as-deposited film on bare Si(100); (b) an annealed film on bare Si(100); (c) an as-deposited film on a 50M50P/Si(100) substrate; (d) an annealed film on a 50M50P/Si(100) substrate.

(inset schematic, Figure 6). The strong registry in the copper-copper layer spacing in these annealed multilayers along the normal was apparent in both the first- and second-order Bragg reflections, as shown in Figure 6. Bragg’s law dictates a copper-copper layer spacing of ∼24 Å (Table 4, available in the Supporting Information) with corresponding reflections independent of the number of bilayers (n). Models for Pc 1 multilayers were presented in repeating units, as shown in Figure 7, where the top half of the Pc core and the top side chains from layer A together with the bottom side chains and the bottom half of the Pc core from layer (A + 1) constitute a repeating unit. The number of repeating units (N) in a multilayer system was determined by the number of bilayers (n) deposited: N ) 2n - 1. Therefore, the two-bilayer film

Figure 6. XRR of annealed multi-bilayer Pc 1 films on Si(100), hydrophobized with a 50:50 mixture of M- and P-silanes: (a) 14 bilayers; (b) 4 bilayers; (c) 2 bilayers. The inset is a schematic of the copper superlattice present in multilayer films deposited from a bilayer-by-bilayer approach.

was treated as a superlattice with three repeating units. Similarly, 7 repeating units were fitted for the 4-bilayer films and 27 repeating units for the 14-bilayer films. Film thicknesses (T values), estimated from the frequency of the Kiessig fringes and calculated from the spacing between the first two adjacent fringes (T ∝ qz-1),35 are given in Table 4, which is available in the Supporting Information. Models of multilayer films of Pc 1 at various thicknesses suggested a tilt of ∼50-60° away from the normal in the bottom side chains and much the same tilt (∼65°) in the side chains sandwiched between the Pc cores which were set upright (Figure 7). The very top layers of these films were fitted individually, as they did not make

8004

Langmuir, Vol. 20, No. 19, 2004

Xia et al.

Figure 7. Schematics of models for horizontally transferred LB multilayer films of Pc 1 (annealed) on Si(100), hydrophobized with a 50:50 mixture of M- and P-silanes; model construction based on XRR results from Figure 6.

Figure 8. (I) XRD of annealed multilayer LB Pc 1 films on Si(100), hydrophobized with a 50:50 mixture of M- and P-silanes: (a) one bilayer; (b) two bilayers; (c) three bilayers; (d) four bilayers. (II) Schematic of Pc columns in a hexagonal close-packed geometry. (III) Powder XRD of Pc 1.

a complete repeating unit, and were essentially fully extended to meet the thickness requirement. Multilayer models favored upright Pc cores along the normal and took approximately the same amount of tilt in the sandwiched side chains between the Pc cores. XRD studies showed Bragg reflections corresponding to the copper superlattice within the multilayer film with a spacing (d) of ∼23 Å along the normal, which were present in films as thin as two bilayers and became well-defined at three bilayers (Figure 8(I)), suggesting an extremely rigid ordering in the system. If a hexagonal close-packed packing geometry is assumed, one can derive the lateral periodicity (D) from the vertical spacing (d) (D ) d/sin 60°), as demonstrated in schematic II of Figure 8. The column

diameter of Pc 1 was ∼27 Å, consistent with the lateral periodicity reported from AFM studies (Figure 1).14-16 XRD measurements were also performed on well-ground powders of Pc 1 to acquire all the possible reflections in its bulk phase. Only a few reflections were observed (Figure 8(III)), typical for liquid crystalline (LC) and/or polycrystalline materials.36-38 The dominant Bragg reflection around 3.8° in 2θ corresponded to a spacing of ∼23 Å and (36) Beattie, D. R.; Hindmarsh, P. J.; Goodby, J. W.; Haslam, S. D.; Richardson, R. M. J. Mater. Chem. 1992, 2, 1262. (37) Blanton, T. N.; Chen, H. P.; Mastrangelo, J.; Chen, S. H. Adv. X-Ray Anal. 2001, 44, 18. (38) Ban, K.; Nishizawa, K.; Ohta, K.; Shirai, H. J. Mater. Chem. 2000, 10, 1083.

LB Films of Rodlike Phthalocyanine Aggregates

was attributed to the intercolumn spacing (d), as shown in the inset of Figure 8(III) with Pc molecules in the bulk structure aggregated in well-aligned columnar structures, the preferred orientation for Pc 1 in the bulk phase. Reflections were also observed at a large angle regime (∼26° in 2θ) corresponding to a characteristic spacing (d′′) of ∼3.5 Å.38 The d′′ spacing was close to the van der Waals distance and was therefore considered as the intracolumn or intermolecular spacing, which is depicted in the inset as well. Both intercolumn and intracolumn reflections were highlighted and fitted with multiple peaks, as indicated in Figure 8(III). These broad Bragg reflections were expected in these LC materials, since imperfections are present and ordering may only be sustained in one or two dimensions over distances of < ∼100 nm. The width of a Bragg reflection usually indicates the correlation/ coherence in the corresponding order; the correlation length can be estimated with the Scherer formula (L ) 0.9λ/fwhm cos θ).39 Correlation lengths of up to ∼400 Å were found in the dominant reflection at a d spacing of ∼23 Å, considered the lower end of such ordering. Correlation with regard to the intercolumn ordering is expected to be infinite in an ideal case where multilayer Pc films are built through a bilayer-by-bilayer deposition procedure. The intracolumn ordering (d′′) seemed to extend its coherence up to ∼250 Å in the powder phase. Better coherence in d′′ has been observed in columnar aggregates in AFM studies of Pc 1.15,16 Correlation lengths as large as ∼2500 Å have been seen in LB films of these Pc materials with further optimization likely through additional substrate modifications and reduction of vibrations during horizontal film transfer. Conclusions The organization of an organic semiconductor thin film at the oxide/organic interface in an organic field-effect transistor is predicted to play a critical role in the transport of charge and in the device characteristics of the OFET, especially in the saturation region, where majority carriers accumulate at this interface and provide the conductive pathway.12,13,40 It is clear that there can be substantial variability in the coherence and thickness of a single monolayer or bilayer of LB films of a discotic mesophase material like Pc 1, and it is expected that this variability extends to other thin film discotic mesophase materials regardless of the deposition method.1-11,14-18 We have shown here that, if there is sufficient electron number density contrast in the material, XRR with conventional X-ray sources can provide adequate signal-to-noise, with reasonable data acquisition times, allowing for structural characterization of the first-deposited monolayer or bilayer of material. It is further clear that, for our LC Pcs with benzyl-terminated side chains, the interaction between the termini of these side chains and the substrate modifiers can be critical, determining not only the total thickness of the Pc film, which is related to the packing of the side chains, but also to some extent the orientation of the Pc disks within the bilayer film. The studies presented here confirm earlier micron-scale AFM studies of these same materials which showed that a silicon oxide surface modified with a 50:50 mix of methyl- and phenylterminated silanes provided the optimum interaction with single bilayers of these Pc films and the most coherent films (in x, y, and z).15,16,18,19 In those earlier studies, we suggested that the spacing of the phenyl groups in the (39) Cullity, B. D. Elements of X-ray Diffraction; Addison-Wesley: New York, 1967; pp 99-101. (40) Cherian, S.; Donley, C. L.; Mathine, D.; Xia, W.; LaRussa, L.; Armstrong, N. R. J. Appl. Phys., submitted for publication.

Langmuir, Vol. 20, No. 19, 2004 8005

surface modifier (∼7.1 Å average separation distance in the 50M50P modified surface) would accommodate the interdigitation of the terminal benzyl groups in the side chains of the Pc with these phenyl surface modifiers, in an orientation which favors “edge-to-cofacial” interactions, which are believed to be a preferred type of interaction between phenyl rings.41,42 These interactions, if reinforced for sufficient coherence lengths in an organic film, can lead to a kind of “commensurate epitaxy”, proposed by Ulman and Scaringe,43 which reinforces the coherence in a thin film of molecular aggregates held together by purely weak (van der Waals) forces. The XRR studies here suggest that the 100P modified surface does not allow for that type of interaction and creates an as-deposited thin film which is thicker, has more extended side chains, and has a lesser coherence than the 50M50P surface. The 100M surface provides for better packing of the side chains in the as-deposited films but still does not result in this type of interdigitation and the coherence seen in the thin films deposited on the 50M50P modified surfaces. These studies also show that horizontally transferred Langmir-Blodgett multilayer films are not in their most stable packing architectures but are closest to that point for films deposited on the 50M50P surfaces. Prolonged annealing at temperatures above the liquid crystal (LC) transition of these discotic materials produces Pc 1 multilayers with a true copper superlattice and side chains packed to an optimal extent. Hexagonal close-packed (hcp) packing models showed agreement between XRD and AFM studies on the lateral periodicity in these films. Powder XRD studies indicated the dominance of Pc columnar aggregates even in the bulk phase material. Correlation lengths, estimated from the widths of Bragg reflections, were essentially the lower limits to their corresponding order in thin films. Extended coherence was observed in LB films of this material with further optimization to be achieved in films deposited on properly modified substrates. These studies are now being extended to further thin film processing conditions which can extend the coherence of these aggregates to greater length scales and to new discotic mesophase systems which form both hexagonal and distorted rectangular columnar assembles.44 There is also substantial interest in determining how the electrical properties of these assemblies are influenced by the coherence lengths in these aggregates, where those electrical properties are measured primarily on submicron length scales. Acknowledgment. Research support is gratefully acknowledged from the National Science Foundation, Chemistry-0211900, and the NSF-STC-Center for Materials and Devices for Information Technology, DMR0120967, Pfizer Pharmaceuticals (graduate fellowships to W.X.). We are also grateful to Dr. Joachim Woitok (PANalytical, Almelo, The Netherlands) for helpful discussions. Supporting Information Available: Figure showing XRR measurements of bare Si(100) and Si(100) hydrophobized with a mixture of M- and P-silanes at various ratios and tables showing XRR fitting parameters. This material is available free of charge via the Internet at http://pubs.acs.org. LA048736N (41) Whitten, D. G.; Chen, L. H.; Geiger, H. C.; Perlstein, J.; Song, X. D. J. Phys. Chem. B 1998, 102, 10098. (42) Geiger, H. C.; Perlstein, J.; Lachicotte, R. J.; Wyrozebski, K.; Whitten, D. G. Langmuir 1999, 15, 5606. (43) Ulman, A.; Scaringe, R. P. Langmuir, 1992, 8, 894. (44) Minch, B. A.; Xia, W.; Donley, C. L.; Hernandez, R. M.; Carter, C.; Carducci, M. D.; Dawson, A.; O’Brien, D. F.; Armstrong, N. R. Chem. Mater., submitted for publication.