Surface Area Characterization of Obliquely Deposited Metal Oxide

Dec 11, 2009 - Electrical and Computer Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2V4. ‡ National Research ..... Alberto Palm...
8 downloads 11 Views 3MB Size
Download PDF
pubs.acs.org/Langmuir © 2009 American Chemical Society

Surface Area Characterization of Obliquely Deposited Metal Oxide Nanostructured Thin Films )

Kathleen M. Krause,*,† Michael T. Taschuk,† Ken D. Harris,‡ David A. Rider,§ Nicholas G. Wakefield,† Jeremy C. Sit,† Jillian M. Buriak,‡,§ Matthias Thommes, and Michael J. Brett*,†,‡ Electrical and Computer Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2V4, ‡National Research Council (NRC), National Institute for Nanotechnology (NINT), Edmonton, Alberta, Canada T6G 2M9, §Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2, and Quantachrome, 1900 Corporate Drive, Boynton Beach, Florida 33426 )

†

Received September 11, 2009. Revised Manuscript Received November 19, 2009 The glancing angle deposition (GLAD) technique is used to fabricate nanostructured thin films with high surface area. Quantifying this property is important for optimizing GLAD-based device performance. Our group has used highsensitivity krypton gas adsorption and the complementary technique of cyclic voltammetry to measure surface area as a function of deposition angle, thickness, and morphological characteristics for several metal oxide thin films. In this work, we studied amorphous titanium dioxide (TiO2), amorphous silicon dioxide (SiO2), and polycrystalline indium tin oxide (ITO) nanostructures with vertical and helical post morphologies over a range of oblique deposition angles from 0 to 86. Krypton gas sorption isotherms, evaluated using the Brunauer-Emmettt-Teller (BET) method, revealed maximum surface area enhancements of 880 ( 110, 980 ( 125, and 210 ( 30 times the footprint area (equivalently 300 ( 40, 570 ( 70, and 50 ( 6 m2 g-1) for vertical posts TiO2, SiO2, and ITO. We also applied the cyclic voltammetry technique to these ITO films and observed the same overall trends as seen with the BET method. In addition, we applied the BET method to the measurement of helical films and found that the surface area trend was shifted with respect to that of vertical post films. This revealed the important influence of the substrate rotation rate and film morphology on surface properties. Finally, we showed that the surface area scales linearly with film thickness, with slopes of 730 ( 35 to 235 ( 10 m2 m-2 μm-1 found for titania vertical post films deposited at angles from 70 to 85. This characterization effort will allow for the optimization of solar, photonic, and sensing devices fabricated from thin metal oxide films using GLAD.

Introduction Glancing angle deposition (GLAD) is an established singlestep physical vapor-deposition technique for fabricating nanostructured thin films from a full range of organic, semiconductor, and dielectric materials.1-4 GLAD employs substrate rotation and obliquely incident vapor flux to create characteristic columnar structures that impart devices with unique and useful properties. Such devices include optical filters, photonic structures, and antireflection coatings.5,6 Also, helical nanosprings, microfluidic devices, solar cells, liquid crystal scaffolding, and humidity sensors have been fabricated using the GLAD technique.7-11 For this work, we were interested in assessing the surface *Corresponding authors. E-mail: [email protected], [email protected]. (1) Robbie, K.; Brett, M. J. J. Vac. Sci. Technol., A 1997, 15, 1460–1465. (2) Lakhtakia, A.; Messier, R. Sculptured Thin Films: Nanoengineered Morphology and Optics; SPIE Press: Bellingham, WA, 2004. (3) Messier, R.; Venugopal, V. C.; Sunal, P. D. J. Vac. Sci. Technol., A 2000, 18, 1538–1545. (4) Hawkeye, M. M.; Brett, M. J. J. Vac. Sci. Technol., A 2007, 25, 1317–1335. (5) van Popta, A. C.; Sit, J. C.; Brett, M. J. Proc. SPIE-Int. Soc. Opt. Eng. 2004, 5464, 198. (6) Xi, J.-Q.; Schubert, M. F.; Kim, J. K.; Schubert, E. F.; Chen, M.; Lin, S.-Y.; Liu, W.; Smart, J. A. Nat. Photonics 2007, 1, 176–179. (7) Dice, G. D.; Brett, M. J.; Wang, D.; Buriak, J. M. Appl. Phys. Lett. 2007, 90, 253101. (8) Bezuidenhout, L. W.; Brett, M. J. J. Chromatogr., A 2008, 1183, 179–185. (9) Chhajed, S.; Schubert, M. F.; Kim, J. K.; Schubert, E. F. Appl. Phys. Lett. 2008, 93, 251108. (10) Kennedy, S. R.; Sit, J. C.; Broer, D. J.; Brett, M. J. Liq. Cryst. 2001, 28, 1799–1803. (11) Steele, J. J.; Taschuk, M. T.; Brett, M. J. IEEE Sens. J. 2008, 8, 1422–1429.

4368 DOI: 10.1021/la903444e

characteristics of GLAD films fabricated with several different morphologies and from several different materials in order to provide information to the community for the optimization of devices. The characterization of the fundamental properties of GLADbased devices is crucial to their optimization. To address this need, simulation and experimental work has been done by several groups. Simulation results by Suzuki et al. provide a 3D ballistic deposition and surface diffusion model for predicting the surface area as a function of deposition angle.12 Others have experimentally determined GLAD film surface areas for a range of materials. For example, Kiema et al. used a surface acoustic wave (SAW) technique to estimate the surface area enhancement of carbon GLAD films, deposited at 85, to be 63 m2 m-2.13 Organic films have also been studied by researchers such as Demirel who predicted a surface area enhancement of 392 m2 m-2 for an obliquely deposited poly(p-xylylene) film.15 Li et al. are among those who have studied metal nanostructured films; they determined that titanium GLAD films, deposited at oblique angles of 65 to 80, had a maximum specific surface area of 25 m2 g-1.16 Additionally, Broughton et al. used a SAW technique to measure the surface area enhancement of manganese GLAD films to be up (12) Suzuki, M.; Taga, Y. J. Appl. Phys. 2001, 90, 5599–5605. (13) Kiema, G. K.; Brett, M. J. J. Electrochem. Soc. 2003, 150, E342–E347. (14) Broughton, J. N.; Brett, M. J. Electrochem. Solid-State Lett. 2002, 5, A279– A282. (15) Demirel, M. C. Colloids Surf., A 2008, 321, 121–124. (16) Li, C.-C.; Huang, J.-L.; Lin, R.-J.; Lii, D.-F.; Chen, C.-H. J. Vac. Sci. Technol., A 2007, 25, 1373–1380.

Published on Web 12/11/2009

Langmuir 2010, 26(6), 4368–4376

Krause et al.

to 9 m2 m-2.14 Finally, the Kay and Mullins groups have used a temperature-programmed desorption (TPD) technique to estimate the maximum surface area of ballistically deposited slanted post Pd, TiO2, TiC, MgO, and amorphous solid water (ASW) films to be 120, 100, 840, 1000, and 3000 m2 g-1, respectively.17-21 The work by Kay and Mullins was performed at ultrahigh vacuum with no substrate rotation but with temperature controlled substrates. In this current paper, we build upon previous work by providing surface area characterization for metal oxide GLAD films that have both vertical and helical post morphologies. We compare the densities and surface areas of amorphous titania (TiO2), amorphous silica (SiO2), and polycrystalline indium tin oxide (ITO) fabricated over a full range of deposition angles. Using krypton gas sorption and cyclic voltammetry, we explore the relationship between substrate rotation rate (i.e., slanted vs vertical vs helical posts) and film thickness on surface area. This is the first report on the surface characteristics of metal oxide vertical post and helical GLAD films. We also believe that this is the first account of krypton gas sorption measurements of GLAD thin films. This work will provide valuable insight into the structure of obliquely deposited thin films. The materials that we chose for this study are commonly used to create GLAD-based devices for photonic, sensing, and display applications. Silica is a common insulator that has been used as nanostructured scaffolding for liquid crystals as well as for microfluidic and sensing applications.7,8,10 Titania is a high refractive index material used for optical interference filters and antireflection coatings.6,22,23 Also, because it is hydrophilic, titania is an excellent choice for humidity sensors.11 ITO is a transparent conductor that is used as the transparent electrode in thin film solar cells.24-29 In addition, obliquely deposited ITO has been used to align liquid crystals.29 Finally, both ITO and TiO2 are highly transparent following postdeposition annealing and have interesting optical birefringent and retardation properties when structured with morphologies such as helices, zigzags, and anisotropic columns.5,29 Because these materials are used in so many applications, our surface area findings should allow for improvements in many device applications. To measure the surface characteristics of the GLAD metal oxide films studied here, we used high-sensitivity krypton gas physical adsorption. Physical gas adsorption is a standard method (17) Kim, J.; Dohnalek, Z.; Kay, B. D. Surf. Sci. 2005, 586, 137–145. (18) Flaherty, D. W.; Dohnalek, Z.; Dohnalkova, A.; Arey, B. W.; McCready, D. E.; Ponnusamy, N.; Mullins, C. B.; Kay, B. D. J. Phys. Chem. C 2007, 111, 4765–4773. (19) Flaherty, D. W.; Hahn, N. T.; Ferrer, D.; Engstrom, T. R.; Tanaka, P. L.; Mullins, C. B. J. Phys. Chem. C 2009, 113, 12742–12752. (20) Dohnalek, Z.; Kimmel, G. A.; McCready, D. E.; Young, J. S.; Dohnalkova, A.; Smith, R. S.; Kay, B. D. J. Phys. Chem. B 2002, 106, 3526–3529. (21) Stevenson, K. P.; Kimmel, G. A.; Dohnalek, Z.; Smith, R. S.; Kay, B. D. Science 1999, 283, 1505–1507. (22) Lin, S.-S.; Chen, S.-C.; Hung, Y.-H. Ceram. Int. 2009, 35, 1581–1586. (23) Celo, D.; Post, E.; Summers, M.; Smy, T.; Brett, M. J.; Albert, J. Opt. Express 2009, 17, 6655–6664. (24) Zotti, G.; Schiavon, G.; Zecchin, S. Langmuir 1998, 14, 1728–1733. (25) Li, J.; Wang, L.; Liu, J.; Evmenenko, G.; Dutta, P.; Marks, T. J. Langmuir 2008, 24, 5755–5765. (26) Donley, C.; Dunphy, D.; Paine, D.; Carter, C.; Nebesny, K.; Lee, P.; Alloway, D.; Armstrong, N. R. Langmuir 2002, 18, 450–457. (27) Liau, Y.-H.; Scherer, N. F.; Rhodes, K. J. Phys. Chem. B 2001, 101, 3282– 3288. (28) Armstrong, N. R.; Carter, C.; Donley, C.; Simmonds, A.; Lee, P.; Brumbach, M.; Kippelen, B.; Domercq, B.; Yoo, S. Thin Solid Films 2003, 445, 342–352. (29) Harris, K. D.; van Popta, A.; Sit, J. C.; Broer, D. J.; Brett, M. J. Adv. Funct. Mater. 2008, 18, 2147–2153. (30) Lowell, S.; Shields, J. E.; Thomas, M. A.; Thommes, M. Characterization of Porous Solids and Powders: Surface Area, Pore Size, and Density; Springer: Dordrecht, The Netherlands, 2006.

Langmuir 2010, 26(6), 4368–4376

Article

employed to measure the surface area and pore size distribution of powders, thin films, and other porous structures.30 One advantage of this technique is that it is ex situ: characterization can be performed on samples outside of the deposition chamber. Annealing, functionalization, and aging tests are therefore possible. Also, each film under investigation can be fully characterized in terms of not only its absolute surface area but also its specific surface area because its mass, footprint area, and thicknesses can all be measured following deposition. By employing highly accurate volumetric adsorption equipment, it is possible to measure absolute surface areas as low as 0.5-1 m2 with nitrogen (at 77 K) or argon (at 87 K). To measure even lower surface areas, the number of molecules trapped in the void volume of the sample cell needs to be reduced.30 This can be achieved by applying krypton adsorption at liquid-nitrogen temperature for the surface area analysis. Krypton at ∼77 K is 38.5 K below the triple-point temperature (Tr = 115.35 K), and it sublimates (i.e., P0,solid) at ∼ 1.6 Torr (at 77.3 K). However, it has become customary to adopt the saturation pressure of supercooled liquid krypton for the application of the BET equation.30-32 In this case, one assumes that despite the fact that the sorption measurement is performed far below the bulk triple-point temperature the adsorbed krypton layer is liquidlike.33,34 The saturation pressure of supercooled liquid krypton is 2.63 Torr (i.e., the number of molecules in the free space of the sample cell is significantly reduced (to 1/300th) compared to the conditions of nitrogen adsorption at liquid-nitrogen temperature). Hence, krypton adorption at ∼77 K is much more sensitive and can be applied to assess surface areas down to at least 0.05 m2. We used this method to determine the surface areas of GLAD TiO2, SiO2, and ITO films and compared the results to those found in the literature. To extend the findings provided by krypton adsorption, we have also employed cyclic voltammetry to probe the conducting ITO thin film samples.24-28 Cyclic voltammetry is a common technique used to determine surface characteristics of a conducting layer of material. In this work, ferrocene-based molecules were chemisorbed as a self-assembled monolayer (SAM) to the GLAD ITO surface. Immersed in an electrolytic solution, single-electron oxidation reactions occurred for each bound ferrocene molecule on the ITO electrode, allowing us to determine the surface coverage of the structured ITO film. We then inferred the specific surface area of GLAD ITO films using this technique and found agreement with the BET method trend.

Experimental Methods To achieve a sufficient total surface area for krypton adsorption measurements, we prepared samples with large footprint areas. For TiO2 and SiO2 films, a total of half of a 4 in. (100 mm) diameter factory-polished silicon wafer (p-type Æ100æ, University Wafer) was cleaved into pieces of 1 cm  1.5 cm prior to deposition. For ITO films, a 1  1 in.2 ITO-coated glass substrate, also cleaved into small pieces, was used as the deposition substrate (8-12 Ω/0, Delta Technologies Ltd.). Prior to deposition, (31) Thommes, M. A. Physical Adsorption Characterization of Ordered and Amorphous Mesoporous Materials. In Nanoporous Materials: Science and Engineering; Lu, G. Q., Zhao, X. S., Eds.; Imperial College Press: London, 2004; Chapter 11, pp 317-364. (32) Thommes, M.; Nishiyama, N.; Tanaka, S. In Recent Progress in Mesostructured Materials; Proceedings of the 5th International Mesostructured Materials Symposium; Zhao, D., Qiu, S., Tang, Y., Yu, C., Eds.; Elsevier: Amsterdam, 2006; pp 551-554. (33) Thommes, M.; Kohn, R.; Froba, M. Appl. Surf. Sci. 2002, 196, 239–249. (34) Chiu, C.-Y.; Chiang, A. S. T.; Chao, K.-J. Microporous Mesoporous Mater. 2006, 91, 244–253.

DOI: 10.1021/la903444e

4369

Article we individually measured these sample pieces on a Sartorius MC5 microbalance and then attached them to a flat circular deposition chuck. We then photographed the chuck and sample pieces with a calibration ruler for use in subsequent footprint area calculations. Once sample preparation was complete, we carried out depositions on an electron-beam physical vapor-deposition evaporation system (Axxis, Kurt J Lesker). This system is computer-controlled and allows for the manipulation of two substrate-orientation motors: one to set the deposition angle (R) of the substrate chuck with respect to the impinging evaporant flux and another to rotate the substrate (φ) as a function of the deposition rate. Computer control relies on the deposition rate information provided by a quartz crystal oscillator (QCM) (Maxtex, SC-105 Aluminum at 6 MHz). For these depositions, we used TiO2 (Cerac, 99.9% pure rutile), SiO2 (Cerac, 99.99% pure), and ITO (Cerac, 91:9 In2O3/SnO2 99.99% pure). To promote the formation of stoichiometric TiO2, we evaporated TiO2 under the partial pressure of O2 gas; by varying the O2 gas flow, the pressure was typically kept at 8  10-3 Pa. In contrast, SiO2 and ITO were deposited without the addition of O2 gas and were deposited at pressures of 5  10-4 and 2  10-3 Pa. In all cases, the system base pressure was below 1  10-4 Pa. Substrates were not dynamically temperature controlled during the deposition process and reached steady state at 70 C during deposition.35 The morphology of the deposited films was determined by the deposition angle, rate of incident flux, and rotation speed. We deposited 0.5- to 5-μm-thick vertical post films at R from 0 to 86 with a deposition rate of approximately 1 nm s-1 and a substrate rotation rate of 0.1 rev nm-1 of deposited film. In contrast to the vertical post films, a substrate rotation rate of 0.0033 rev nm-1 of deposited film was used to achieve helical films. Following deposition, we annealed TiO2 and ITO samples to ensure stoichiometric TiO2 films with stable optical properties and ITO films with stable optical and electrical properties. We annealed the TiO2 films at 100 C for 24 h, a low enough temperature that we did not change the crystallinity from amorphous to polycrystalline or crystalline.5 Annealing was also used to enhance and stabilize the optical transmission of the film, an indication that oxygen-deficient sites are eliminated and stoichiometry has been achieved.36,37 Additionally, ITO samples were annealed at 300-350 C for 2 to 3 h, with a 10 C min-1 temperature ramp up and down. These recipes are standard practice in our laboratory. Immediately following deposition in the case of SiO2 and after annealing in the case of ITO and TiO2, we weighed the final mass of each sample piece in order to calculate the total film weight. We estimate that the impact of humidity on these postdeposition mass measurements results in a worst-case 6% negative error term in the total film mass on the basis of the calculation of water coverage extrapolated from the work done by Steele et al.38 In addition to this humidity error term, a measurement repeatability error of 0.02 mg/sample piece was taken into account. To calculate film densities, we needed to measure not only the film mass but also the sample footprint area and film thickness. Sample footprint areas were determined using the freeware ImageJ software package (1.36b)39 and the calibrated sample chuck photograph. A repeatability error of 5% per sample piece was taken into account. We measured witness sample thicknesses using a Hitachi field emission S-4800 scanning electron microscope (SEM). As with the samples prepared for surface area analysis, witness samples for TiO2 and SiO2 films were deposited (35) Kupsta, M. University of Alberta, Edmonton, Alberta, Canada. Personal Communication, 2009. (36) Rao, K. N. Opt. Eng. 2002, 41, 2357–2364. (37) Wang, W.-H.; Chao, S. Opt. Lett. 1998, 23, 1417–1419. (38) Steele, J. J.; van Popta, A. C.; Hawkeye, M. H.; Sit, J. C.; Brett, M. J. Sens. Actuators, B 2006, 120, 213–219. (39) Rasband, W. S. ImageJ, version 1.36b; National Institutes of Health: Bethesda, MD, 2006.

4370 DOI: 10.1021/la903444e

Krause et al. on silicon substrates and ITO films were deposited on ITO-coated commercial substrates. A conductive coating was required for imaging of the SiO2 samples, but none was required for the TiO2 and ITO samples. We calculated the uncertainty in film thickness on the basis of the maximum and minimum thicknesses of the topof-film envelope profile. Film uniformity was estimated to be 4% for our 8 cm chuck on the basis of measurements performed by our group.40 In addition to SEM analysis, we also used transmission electron microscopy (TEM) to interrogate individual GLAD columns. For this exercise, we placed GLAD columns, shaved off of sample substrates and sonicated in isopropyl alcohol (IPA), onto conductive grids and then used a JEOL 2200FS TEM for imaging. We corroborated calculated film densities using the optical Bruggeman effective medium optical model based on measurements taken by Mueller matrix ellipsometry.41,42 The required optical data was collected in reflection between 400 and 1700 nm at 45, 55, and 65 angles of incidence using a variable-angle spectroscopic ellipsometer (VASE, J.A. Woollam Co. Inc.) and modeled using the provided software (WVASE32, v3.517, J.A. Woollam Co. Inc.). Once all required density measurements had been performed, we transferred sample pieces from a single deposition to 12-mmdiameter sample bulbs (Quantachrome) in order to carry out physical adsorption tests. Prior to performing an adsorption test, the glass holder containing the sample pieces was degassed at 150 C for 16-24 h under vacuum in one of the degassing stations of the Quantachrome Autosorb-1MP (with cold trap cooled by liquid N2). The level of liquid in the coolant bath was detected using a sensor (Quantachrome, 00080-AR-RTD and 00080-LN2RTD for liquid Ar and N2, respectively). To acquire data for BET analysis, Kr adsorption (Praxair, 99.999% pure) at liquid-nitrogen (Praxair) temperature (77 K) was acquired over pressures from 0.05po to 0.3po, where po = 2.63 mmHg (the saturation pressure of the supercooled liquid). In some cases, full adsorption and desorption isotherms were also collected up to 0.55po (above which the bulk saturation, i.e., sublimation, pressure is reached). To verify that the total film surface areas were low enough to warrant the use of Kr gas, N2 (Praxair, 99.9999% pure) sorption isotherms at 77 K were taken from 0.05po to 1po, where po was approximately 720 mmHg. We processed captured adsorption data using Quantachrome Autosorb software (v1.52) and custom algorithms written in Matlab. To complement the BET analysis, we used cyclic voltammetry on ITO GLAD samples functionalized with monololayers of ferrocene dicarboxylic acid. The ferrocene dicarboxylic acid (Fc(COOH)2) was used as received from Aldrich. Immediately prior to surface modification, we cleaned the ITO GLAD film samples using air plasma (Harrick model PDC-32G) for 10 min, a step that has no obvious impact on the sample structure. The ITO GLAD films were then immediately transferred to a 1 mM solution of Fc(COOH)2 in pure ethanol for 10 min, followed by rinsing with copious amounts of ethanol and acetonitrile. We carried out the required cyclic voltammetric investigations using a potentiostat (Princeton Applied Research, model 2273) employing the Fc(COOH)2-coated GLAD ITO films as the working electrodes (substrate area = 5.25 cm2) in a standard three-electrode electrochemical cell using a Ag/Agþ reference electrode and a Pt wire counter electrode. Potentiodynamic scans from -0.50 to 1.75 V were acquired with a scan rate of 0.10 V s-1 using 0.1 M tetrabutylammonium hexafluorophosphate as the electrolyte. Finally, we determined the charge transferred for the one-electron oxidation of surface-absorbed Fc(COOH)2 by integration of the anodic peak area in a cyclic voltammogram using Princeton Applied Research software (Power Suite v. 2.58). (40) Wakefield, N. G.; Sit, J. C. To be submitted for published. (41) Aspnes, D. E. Thin Solid Films 1982, 89, 249–262. (42) Gospodyn, J.; Sit, J. C. Opt. Mater. 2006, 29, 318–325.

Langmuir 2010, 26(6), 4368–4376

Krause et al.

Article

Figure 2. Side-view SEM of a representative TiO2 GLAD helical sample at (a) R = 70, (b) R = 75, (c) R = 80, and (d) R = 85.

Figure 3. Side-view SEMs of (a) SiO2 and (b) ITO GLAD vertical post films at R = 85.

Figure 1. SEMs of representative TiO2 GLAD vertical post samples. (a) R = 45 side, (b) R = 45 top, (c) R = 60 side, (d) R = 60 top, (e) R = 70 side, (f) R = 70 top, (g) R = 85 side, and (h) R = 85 top.

Results and Discussion We fabricated a number of structured thin films for this study and analyzed their density, porosity, and surface characteristics. The basic properties of these films, such as deposition angle, thickness (tfilm), footprint area (FA), and mass (m), are provided in the Supporting Information for this article. In our study, most SiO2 and TiO2 films were fabricated with a thickness of approximately 1 μm and the ITO films were fabricated with a thickness of 0.5 μm. The exception was films fabricated for the thickness study, where thicknesses ranged from 0.5 to 5 μm. To illustrate the nature of these films, top-down and side-view SEM images of representative TiO2 vertical post GLAD films are shown in Figure 1, and side views of TiO2 helical thin films are presented in Figure 2. These images reveal that as the deposition angle increases, the spacing and mesoporosity between the GLAD columns also increases, as does the diameter of the columns. At first glance, films deposited at low deposition angles appear to be quite dense; however, an analysis of top-down SEM images reveals that porosity and surface roughness are present above a deposition angle of 60. Equally important is the column broadening that is apparent at high deposition angles; this broadening, along with porosity, plays into the establishment of the specific surface area of the sample.43 As illustrated in Figure 3a, SiO2 GLAD films do not appear to be fundamentally different from TiO2 and exhibit similar columnar properties as a function of deposition angle. In contrast, the ITO films are quite different: rougher column walls and bifurcations are characteristics of these structures as seen in Figure 3b. Further comparisons of the structure of the resultant GLAD films for the three oxides provide insight into the differences (43) Buzea, C.; Beydaghyan, G.; Elliott, C.; Robbie, K. Nanotechnology 2005, 16, 1986–1992.

Langmuir 2010, 26(6), 4368–4376

among TiO2, SiO2, and ITO films. TEM images of representative GLAD columns are shown in Figure 4, with inset electron diffraction patterns. These electron diffraction patterns show that TiO2 is not crystalline but ITO is polycrystalline. Furthermore, individual TiO2 and SiO2 columns are made up of amorphous filaments with features of dimensions on the order of 10 nm, and ITO columns are made up of crystalline 5-10-nm-sized particles. Figure 5 presents the calculated and expected thin film densities for all samples. The measured densities of vertical post films fabricated from each of the three different materials are shown in Figure 5a (with TiO2, SiO2, and ITO thicknesses of roughly 1, 1, and 0.5 μm, respectively). As discussed above, we calculated these film densities on the basis of the measured film masses, footprint areas, and thicknesses. We estimated the density uncertainties on the basis of uncertainties in the mass, footprint area, and thickness of the samples. These TiO2 and SiO2 film densities have also been corroborated using the optical Bruggeman effective medium optical model41 based on measurements taken by Mueller matrix ellipsometry.42 To parametrize the GLAD film densities, we have applied the Poxson method.44 Poxson et al. propose an analytical model that predicts the porosity of nanoporous vertical post films grown using the GLAD technique. A single fitting parameter (c) takes into account the geometry of the nanostructure and, along with the deposition angle, predicts the porosity (P) of the deposited thin film. The porosity, a unitless quantity, indicates that the fraction of void in the film is given by Poxson’s original equation: P ¼

R tan R c þ R tan R

ð1Þ

The normalized density of a film is the fraction of material in a film. The absolute density (δ) is given by δ ¼ δbulk

c c þ R tan R

ð2Þ

(44) Poxson, D. J.; Mont, F. W.; Schubert, M. F.; Kim, J. K.; Schubert, E. F. Appl. Phys. Lett. 2008, 93, 101914.

DOI: 10.1021/la903444e

4371

Article

Krause et al.

Figure 4. TEMs of (a) TiO2, (b) SiO2, and (c) ITO GLAD columns, with insets showing electron diffraction pattern for TiO2 and ITO films.

The fitting parameter describes the relationship among the volume of material in a nanocolumn (Vm), the cross-sectional area of the nanocolumn (Acs), and the height of initial nucleation sites (hn): πVm c ¼ 2Acs hn

ð3Þ

For our three sets of vertical post films, the application of eq 1 results in Poxson fitting parameters of 7.01 ( 0.21 (R2 = 0.985), 4.44 ( 0.10 (R2 = 0.991), and 8.70 ( 0.85 (R2 = 0.712) for TiO2, SiO2, and ITO, respectively. In Poxson’s original paper, density fitting parameters of 3.17 ( 0.11 (R2 = 0.993) and 3.55 ( 0.21 (R2 = 0.995) were given for SiO2 and ITO, respectively. The higher-fitting parameters required for the data presented are due to somewhat higher densities, especially in the mid-deposition angle range of 50-75. These higher densities are most likely due to differences in deposition conditions, such as substrate temperature, crucible-to-substrate distance, and deposition rate. The substrate rotation and film thickness rate may also play roles in the density trend, as discussed below. Film thickness and broadening may explain some of the differences between our density fitting parameters and those provided by Poxson. In the original Poxson paper, all film thicknesses were 200 nm whereas ours are at least twice as thick. Columnar competition and extinction are dominant factors in shorter films. Also, thicker GLAD films, especially the oxide films, exhibit film broadening as thickness increases. Although film densities appear to be fairly uniform with thickness, as shown in Figure 5b for TiO2 GLAD films of increasing thickness, there may be some density dependence on film thickness when they are less than 1 μm thick. In addition to evaluating the density trends of vertical post films, we are also interested in how the densities of helical GLAD films track with the deposition angle and how well the Poxson theory applies. Figure 5c shows the density profile of TiO2 helical GLAD films deposited over a range of deposition angles. The application of eq 2, Poxson’s original density fit, results in the trend termed Poxsono, which does not closely predict the measurement trend. This is because helical films are denser at moderate deposition angles but less dense at the very oblique angle of 85. This higher density at moderate deposition angles may provide a clue as to why our fitting parameters for the densities of the vertical post films shown in Figure 5a are different from those described by Poxson. Rotation rates may have been significantly different, leading to differences in film morphology.45 An attempt to fit the helical density data to the original (linear) Poxson model results in a fitting parameter of 6.54 ( 0.66 with an R2 value of 0.920. To fit the data better, the (45) Dick, B.; Brett, M. J.; Smy, T. J. Vac. Sci. Technol., B 2003, 21, 23–28.

4372 DOI: 10.1021/la903444e

original model is extended with an added quadratic term, as described by δq δq ¼

1 1 1 1 þ R tan R þ ðR tan RÞ2 c k

ð4Þ

with c and k being fitting parameters of 11.2 and 14.0. This added term provides a better fit with an R2 value of 0.997. This fit is shown in Figure 5c as Poxsonq. The successful application of eq 4 to the density trend of helical GLAD films suggests that an extension to the Poxson model is appropriate for these as well as perhaps slanted post films. The physical basis for this extension is under investigation. In addition to studying the densities of our GLAD films, we also wanted to determine their surface area. It appeared that (as expected) the sensitivity of N2 adsorption was not sufficient and Kr gas adsorption was required to obtain a meaningful analysis of the small surface area films presented here. Selected 77 K krypton gas adsorption and desorption isotherms are shown in Figure 6. Vgas indicates the amount of gas adsorbed onto the film at each pressure (p) relative to the vapor pressure (po) of krypton at 77 K. Sublimation of krypton within the pores of the thin film occurs at approximately 0.6 vapor pressure. Adsorption isotherms manifest differently depending on the characteristics of the porous sample under investigation. The International Union of Pure and Applied Chemistry (IUPAC) has classified six reference sorption isotherms.46 A comparison of a given experimental isotherm with the standard isotherms allows for the conditions leading to the formation of the given isotherm to be straightforwardly determined. For example, the type I sorption isotherm rises to a plateau and exhibits no hysteresis, indicating that the material being tested is microporous (pore diameters of