16179
2006, 110, 16179-16184 Published on Web 07/28/2006
Anodic Growth of Highly Ordered TiO2 Nanotube Arrays to 134 µm in Length Maggie Paulose,† Karthik Shankar,‡ Sorachon Yoriya,§ Haripriya E. Prakasam,‡ Oomman K. Varghese,†,| Gopal K. Mor,| Thomas A. Latempa,† Adriana Fitzgerald,‡ and Craig A. Grimes*,‡,§,| SentechBiomed Corporation, 200 InnoVation BouleVard, State College, PennsylVania 16803, Department of Electrical Engineering, The PennsylVania State UniVersity, UniVersity Park, PennsylVania 16801, Department of Materials Science and Engineering, The PennsylVania State UniVersity, UniVersity Park, PennsylVania 16801, and Materials Research Institute, The PennsylVania State UniVersity, UniVersity Park, PennsylVania 16801 ReceiVed: June 27, 2006; In Final Form: July 18, 2006
Described is the fabrication of self-aligned highly ordered TiO2 nanotube arrays by potentiostatic anodization of Ti foil having lengths up to 134 µm, representing well over an order of magnitude increase in length thus far reported. We have achieved the very long nanotube arrays in fluoride ion containing baths in combination with a variety of nonaqueous organic polar electrolytes including dimethyl sulfoxide, formamide, ethylene glycol, and N-methylformamide. Depending on the anodization voltage, pore diameters of the resulting nanotube arrays range from 20 to 150 nm. Our longest nanotube arrays yield a roughness factor of 4750 and lengthto-width (outer diameter) aspect ratio of ≈835. The as-prepared nanotubes are amorphous but crystallize with annealing at elevated temperatures. In initial measurements, 45 µm long nanotube-array samples, 550 °C annealed, under UV illumination show a remarkable water photoelectrolysis photoconversion efficiency of 16.25%.
Tubular inorganic nanostructures offer great potential for use in sensing, heterojunction solar cells, water photolysis, fuel cells, molecular filtration, and tissue engineering.1,2 Highly ordered vertically oriented TiO2 nanotube arrays fabricated by potentiostatic anodization constitute a material architecture that offers a large internal surface area without a concomitant decrease in geometric and structural order. In contrast to random nanoparticle systems where slow electron diffusion typically limits device performance,3,4 the precisely oriented nature of the crystalline (after annealing) nanotube arrays makes them excellent electron percolation pathways for vectorial charge transfer between interfaces.4,5 Furthermore, the nanotube-array architecture offers the ability to influence the absorption and propagation of light through the architecture by precisely designing and controlling the architectural parameters including nanotube pore size, wall thickness, and length.6 For applications where the use of vertically oriented titania nanotubes has been studied, these advantages have manifested themselves in an extraordinary enhancement of the extant TiO2 properties. For example, hydrogen sensors based on the use of titania nanotube arrays of single-micron length exhibit an unprecedented 50 000 000 000% change in electrical resistivity upon exposure to 1000 ppm of hydrogen gas at room temperature.7,8 In their * Corresponding author. E-mail:
[email protected]. † SentechBiomed Corporation. ‡ Department of Electrical Engineering, The Pennsylvania State University. § Department of Materials Science and Engineering, The Pennsylvania State University. | Materials Research Institute, The Pennsylvania State University.
10.1021/jp064020k CCC: $33.50
use as hydrogen sensors, the TiO2 nanotube arrays possess such excellent photocatalytic properties that they are able to selfclean from environmental contamination with exposure to ambient light.9 TiO2-nanotube-array-based photoanodes have been reported to enhance the photocleavage of water,10,11 have been reported to perform as a superior electrocatalyst for methanol oxidation,12 and have been successfully used as photoanodes in dye-sensitized solar cells (DSSCs).5,13 Among one-dimensional architectures, nanotube arrays have a higher surface area than nanowires due to the additional surface area enclosed inside the hollow structure. For a given pore diameter and wall thickness, the internal surface area increases almost linearly with nanotube length. As initially reported by Gong and co-workers in 2001,14 the first generation titania nanotube arrays fabricated by anodization using an aqueous HF-based electrolyte could be grown to a length of about 500 nm.14,15 The nanotube-array length was subsequently increased to about 7 µm by control of the anodization electrolyte pH (higher values while remaining acidic) which reduced the chemical dissolution of TiO2 during anodization.16,17 Here, we report for the first time on a new generation of vertically oriented TiO2 nanotubes with lengths up to, to date, 134 µm, by the use of various nonaqueous organic18-20 electrolytes. We note the technique is in its infancy, and one could expect further improvements in nanotube-array length following the synthesis procedures described herein. Titanium foils with a thickness of ≈250 µm (99.7%; SigmaAldrich) were cleaned with ethanol prior to anodization. The foils were subjected to potentiostatic anodization in a two© 2006 American Chemical Society
16180 J. Phys. Chem. B, Vol. 110, No. 33, 2006 electrode electrochemical cell connected to a dc power supply. In all cases, a platinum foil was used as the counter electrode. All of the experiments were done at about 20 °C. The anodization current was monitored using a Keithley (model 2000) digital multimeter interfaced with a computer. In combination with either HF, KF, or NaF to provide fluoride ions, we have obtained nanotube arrays up to approximately 134 µm in length using a variety of organic electrolytes including dimethyl sulfoxide (DMSO), formamide (FA), ethylene glycol, and N-methylformamide (NMF). In general, the key to successfully achieving very long nanotube arrays is to minimize water content in the anodization bath to less than 5%. With organic electrolytes, the donation of oxygen is more difficult in comparison to water, thus reducing the tendency to form oxide.18 At the same time, the reduction in water content reduces the chemical dissolution of the oxide in the fluorine containing electrolytes and hence aids the longer-nanotube formation. Illustrative electrolyte compositions include the following: (1) A 0.25 wt % NH4F (98+% ACS reagent) and ethylene glycol (99.8% anhydrous) solution; samples anodized at 60 V for 17 h achieved a nanotube array of 134 µm. The average inner diameter of the tube was 110 nm and outer diameter 160 nm. (2) A dimethyl sulfoxide (DMSO; 99.6%) and 2% hydrofluoric acid (HF; 48% aqueous solution) mixture. Samples that were preanodized in 0.5% HF in water at 20 V followed by anodization in 2% HF containing DMSO electrolyte at 40 V showed superior growth characteristics per growing substantially faster. (3) Formamide (FA; 99%) and/or N-methylformamide (NMF; 99%) solutions containing 1-5 wt % of deionized water and 0.3-0.6 wt % NH4F (98%). While to date we have observed lengths up to 134 µm, our studies show that a further increase in length is possible by optimizing the process parameters. Furthermore, the chemistry appears quite flexible, with substitution between the fluoride containing acids and organics generally resulting in the nanotube-array architecture of several tens of microns. Transmission electron microscopy (TEM) images with their corresponding diffraction patterns show the as-fabricated nanotubes to be amorphous. Consistent with our earlier studies regarding the nanotube-array crystallization and phase transformation, samples annealed at 550 °C showed only the anatase phase.21 Crystallization of the nanotubes initially occurs in the anatase phase, with constraints imposed by the wall limiting the phase transformation of anatase to rutile on annealing the sample at high temperatures.21 However, the anatase crystallites at the base of the nanotubes are transformed to rutile due to the greater degree of freedom for phase transformation. Figure 1 shows FESEM images of samples prepared in a 0.25% NH4F and ethylene glycol solution at an anodization potential of 60 V for 17 h. The nanotube array is 134 µm long, with a roughness factor (the real surface area per unit geometric surface area is called the roughness factor, a unitless quantity22) of 4750 (hexagonal packing, 160 nm outer diameter, 25 nm wall thickness) and a length-to-width aspect ratio of ≈835 (length, 134 µm; average tube outer diameter, 160 nm). Figure 2 shows FESEM images of samples prepared in DMSO containing 2% HF at 40 V for 69 h. These nanotubes have a length of approximately 45 µm, a pore diameter of ≈120 nm, and a wall thickness of ≈15 nm. The roughness factor for these nanotubes is ∼1800. The top surface of the as-prepared samples is typically covered with broken tubes and other debris from the anodization bath. The pores appear clogged in such cases, although a complete clogging does not take place, as this prevents the electrolyte species from reaching the bottom of the tubes, thus terminating the anodization process. The chemical
Letters etching rate of the oxide by the fluoride is low in nonaqueous organic electrolytes, allowing the reaction products to stay at the surface of the nanotubes. Figure 2c shows the surface after removing the debris using sonication; the tubes are vertically oriented and distinct. Anodization in the range 20-60 V yielded a regular, well-aligned nanotube-array architecture. Figure 3 shows FESEM images of a sample fabricated at 60 V, for 70 h, in the DMSO electrolyte containing 2 vol % HF; the resulting nanotube-array length is 93 µm. For a 70 h anodization at, respectively, anodization potentials of 20 and 60 V, an increase in length from about 10 to 93 µm was observed. The resulting pore diameters were, respectively, ∼50, 120, and 150 nm for 20, 40, and 60 V potentials. The annealed 60 V 93 µm long tubes have an outer diameter of approximately 200 nm, and hence a roughness factor of ≈3200, with a lengthto-outer-diameter aspect ratio of ≈465. As the HF concentration varied from 1 to 4%, the length of the nanotubes grown at 20 V increased from 4.4 to 29 µm. The titanium foil when subjected to anodization at 20 V in 0.5% HF in deionized water before anodizing at 40 V in DMSO containing 2% HF solution yielded 82 µm long nanotubes, almost a factor of 2 increase in length from that obtained without using the preanodization step. Figure 4 shows nanotubes nearly 70 µm long grown in a FAbased electrolyte by anodization for 48 h at a constant potential of 35 V. The average outer diameter of these nanotubes was determined to be 180 nm, wall thickness ≈24 nm, resulting in an aspect ratio of ≈390. Lower anodization potentials result in shorter nanotubes with smaller diameters. We attribute the increase in nanotube length with larger anodization voltage to the increased driving force for ionic transport through the barrier layer at the bottom of the pore resulting in faster movement of the Ti/TiO2 interface into the Ti metal. In the FA/NMF electrolytes, we observe an increase in the outer nanotube diameter as the anodization voltage is increased, an observation that agrees with the reported behavior of nanotubular TiO2 as well as other anodically formed metal oxides. All else being equal, nanotubes with a smaller pore diameter and approximately 10% greater length are obtained using a NMF rather than FA electrolyte; the reason for this has yet to be elucidated; however, certainly, this is as yet an unoptimized synthesis approach. In comparison to aqueous electrolytes, the range of applied anodization potentials over which nanotube arrays are obtained is significantly extended in the FA/NMF-based electrolytes, with nanotubes formed at voltages between 10 and 50 V (compared to 10-29 V for nanotubes of micron length in KF or NaF electrolytes). At 60 V, the anodization was found to be unstable with sharp fluctuations in the current; we have not attempted anodization at voltages below 10 V. The anodization duration is also an important variable. We observed that the length of the nanotubes increased with time up to a certain maximum length beyond which it declined. For example, increasing the anodization duration at 35 V to 164 h resulted in a nanotube length of 10 µm compared to 30 µm obtained after 88 h, and 70 µm after 48 h. The time to maximum length is a function of the anodization voltage with the nanotubes reaching this maximum sooner at lower potentials. Figure 5 compares the real time potentiostatic anodization behavior of Ti anodized at 40 V in DMSO + 2% HF electrolyte, 25 V in an electrolyte consisting of FA and NMF in the ratio 8:5 (similar responses are obtained using either FA or NMF), and 25 V in aqueous solutions of 0.5 wt % NH4F and KF. The differences are striking, indicative of different growth mechanisms. The formation of nanotube arrays in fluoride containing electrolytes is the result of three simultaneously occurring
Letters
J. Phys. Chem. B, Vol. 110, No. 33, 2006 16181
Figure 1. FESEM images of a nanotube-array sample grown at 60 V in 0.25 wt % NH4F in ethylene glycol: (a-d) cross-sectional views at varying degrees of magnification; (e) view of bottom surface; (f) top surface.
processes: (1) field assisted oxidation of Ti metal to form titanium dioxide, (2) field assisted dissolution of Ti metal ions in the electrolyte, and (3) chemical dissolution of Ti and TiO2 due to etching by fluoride ions. The same three processes govern the anodic formation of several other self-organized nanoporous metal oxides.23-25 With the highly polar FA and NMF electrolytes, the dramatically increased nanotube-array length appears to be due to an accelerated rate of nanotube growth from the substrate. The dielectric constants of FA and NMF are,
respectively, 111 and 182.4, much greater than that of water which has a dielectric constant of 78.39.26 For a given potential, higher electrolyte capacitance induces more charges to be formed on the oxide layer, improving extraction of the Ti4+ ions, while the higher electrolyte polarity allows HF to be easily dissolved, facilitating its availability at the TiO2/electrolyte interface. We are confident that this fabrication approach can be successfully applied to produce high aspect ratio oxides in other valve metals.
16182 J. Phys. Chem. B, Vol. 110, No. 33, 2006
Letters
Figure 3. FESEM images of a nanotube array grown at 60 V in DMSO containing 2% HF: (a) top surface view; (b) cross-sectional view.
Figure 2. FESEM images of a nanotube-array sample grown at 40 V in DMSO containing 2% HF: (a) cross-sectional view; (b) lateral view; (c) view of top surface.
Returning to Figure 5, in aqueous electrolytes, the anodization current drops sharply from a high initial value in the first few seconds of anodization due to initial formation of an insulating oxide layer. Thereafter, pitting of the oxide layer by fluoride ions commences and the anodization current increases to reach a local maximum after about 5000 s. In contrast, the anodization current in the FA/NMF-based organic electrolyte remains relatively constant in the first 100 s of anodization, followed by a region of steeply decreasing current and then a region where the current continues to decrease at a smaller rate. During the initial anodization stage of high current, gas evolution at the anode is observable which, since gas evolution requires electronic charge transfer, is indicative of electronic conduction dominating in the early stages of the anodization. We attribute this to the formation of a thinner initial oxide layer in the nonaqueous electrolyte, high ionic conductivity, and low Ti oxidation rate. The thinner oxide layer allows much greater ionic conduction than in aqueous electrolytes and faster movement of the Ti/TiO2 interface into the Ti substrate, ultimately enabling substantial increases in the nanotube length one can obtain.
Higher anodization potentials provide a greater driving force for both electronic and ionic conduction. Nanotube formation reduces the surface area available for anodization with a correlated decrease in current density, while deepening of the pore occurs in the third and final region of the anodization process seen at approximately 5000 s. In contrast, with the DMSO electrolyte, the increased nanotube-array length appears to be largely due to limited chemical dissolution of titania by the hydrofluoric acid. The formation of nanotubes in an aqueous HF containing electrolyte is described by a localized dissolution model,27 with an oxide layer initially formed via the following reactions:16
2H2O f O2 + 4e- + 4H+ Ti + O2 f TiO2 The pore formation occurs as a result of the localized chemical dissolution of the oxide by F- according to the following reaction:16 TiO2 + 6F- + 4H+ f TiF62- + 2H2O. This leads to a higher field at the bottom of the pore that drives further oxidation and field assisted dissolution where Ti ions come out of the metal and dissolve in solution. The metallic region between the pores also undergoes a similar transition leading to the tube formation. For a given rate of pore formation, the chemical dissolution of the oxide at the pore mouth by Fdetermines the tube length. To control this dissolution reaction, the H+ ion concentration was reduced by limiting the water content to the level of water contained in HF containing solution. This water ensured the field assisted etching of the Ti foil at the pore bottom, and additionally, protophilic DMSO accepts a proton from HF, reducing its activity. This allowed the DMSO nanotubes to grow deep into the titanium foil without any significant loss from the pore mouth. The presence of DMSO
Letters
J. Phys. Chem. B, Vol. 110, No. 33, 2006 16183
Figure 5. Real time observation of the anodization behavior of a Ti foil (99.8% pure) anodized at 25 V in (a) an electrolyte containing 0.5 wt % NH4F and FA/NMF in the ratio 8:5, (b) an aqueous electrolyte containing 0.5 wt % NH4F, (c) an aqueous electrolyte containing 0.5 wt % KF, and (d) a Ti sample anodized at 40 V in DMSO + 2% HF.
Figure 6. Photocurrent density and corresponding photoconversion efficiency generated from ≈45 µm long nanotube arrays, 550 °C annealed, under 95 mW/cm2 320-400 nm UV illumination. The maximum value of 16.25% is achieved at -0.52 V vs Ag/AgCl. Figure 4. FESEM images of TiO2 nanotubes grown in FA-based electrolyte at 35 V for 48 h showing (a) a cross section at lower magnification, (b) a cross section at higher magnification, and (c) a top surface image.
modifies the space charge region in the pores, thereby avoiding the lateral etching as well, leading to the steady pore growth and low chemical etching of the nanotube walls. Figure 6 shows the I-V characteristics and corresponding light energy to chemical energy conversion (photoconversion) efficiency of a typical ≈45 µm long nanotube-array (of Figure 2) sample annealed at 550 °C in dry oxygen, in a 1 M KOH solution under 320-400 nm UV illumination, 100 mW/cm2. We see a remarkable, and as yet unoptimized, maximum photoconversion efficiency of 16.25%. The photoconversion efficiency, η, is calculated as28
η(%) ) [(total power output electrical power output)/light power input] × 100 ) jp[(Erev° - |Eapp|)/I0] × 100 where jp is the photocurrent density (mW/cm2), jpErev° is the total power output, jp|Eapp| is the electrical power input, and I0 is the power density of incident light (mW/cm2). Erev° is the standard reversible potential which is 1.23 V/NHE, and the
applied potential is Eapp ) Emeas - Eaoc, where Emeas is the electrode potential (vs Ag/AgCl) of the working electrode at which photocurrent was measured under illumination and Eaoc is the electrode potential (vs Ag/AgCl) of the same working electrode under open circuit conditions, under the same illumination, and in the same electrolyte. The voltage at which the photocurrent becomes zero was taken as Eaoc. As shown in Figure 6, the dark current was very low, almost 4 orders of magnitude less than the photocurrent magnitude. Acknowledgment. Support of this work by the Department of Energy under grant DE-FG02-06ER15772 is gratefully acknowledged. S.Y. gratefully acknowledges support under the Royal Thai Government Scholarship provided by the National Metal and Materials Technology Center, the National Science and Technology Development Agency, Thailand. References and Notes (1) Hueso, L.; Mathur, N. Nature 2004, 427, 301. (2) Tenne, R.; Rao, C. N. R. Philos. Trans. R. Soc. London, Ser. A 2004, 362, 2099. (3) Law, M.; Greene, L. E.; Johnson, J. C.; Saykally, R.; Yang, P. D. Nat. Mater. 2005, 4, 455. (4) Frank, A. J.; Kopidakis, N.; van de Lagemaat, J. Coord. Chem. ReV. 2004, 248, 1165.
16184 J. Phys. Chem. B, Vol. 110, No. 33, 2006 (5) Mor, G. K.; Shankar, K.; Paulose, M.; Varghese, O. K.; Grimes, C. A. Nano Lett. 2006, 6, 215. (6) Ong, K. G., Varghese, O. K., Mor, G. K., Grimes, C. A. J. Nanosci. Nanotechnol. 2005, 5, 1801. (7) Paulose, M.; Varghese, O. K.; Mor, G. K.; Grimes, C. A.; Ong, K. G. Nanotechnology 2006, 17, 398. (8) Varghese, O. K.; Yang, X.; Kendig, J.; Paulose, M.; Zeng, K.; Palmer, C.; Ong, K. G.; Grimes, C. A. Sensor Lett. 2006, 4, 120. (9) Mor, G. K.; Paulose, M.; Varghese, O. K.; Grimes, C. A. Sensor Lett. 2003, 1, 42. (10) Mor, G. K.; Shankar, K.; Paulose, M.; Varghese, O. K.; Grimes, C. A. Nano Lett. 2005, 5, 191. (11) Varghese, O. K.; Paulose, M.; Shankar, K.; Mor, G. K.; Grimes, C. A. J. Nanosci. Nanotechnol. 2005, 5, 1158. (12) Macak, J. M.; Barczuk, P. J.; Tsuchiya, H.; Nowakowska, M. Z.; Ghicov, A.; Chojak, M.; Bauer, S.; Virtanen, S.; Kulesza, P. J.; Schmuki, P. Electrochem. Commun. 2005, 7, 1417. (13) Mor, G. K.; Varghese, O. K.; Paulose, M.; Shankar, K.; Grimes, C. A. Sol. Energy Mater. Sol. Cells 2006, 90, 2011. (14) Gong, D.; Grimes, C. A.; Varghese, O. K.; Hu, W. C.; Singh, R. S.; Chen, Z.; Dickey, E. C. J. Mater. Res. 2001, 16, 3331. (15) Mor, G. K.; Varghese, O. K.; Paulose, M.; Grimes, C. A. AdV. Funct. Mater. 2005, 15, 1291. (16) Cai, Q. Y.; Paulose, M.; Varghese, O. K.; Grimes, C. A. J. Mater. Res. 2005, 20, 230.
Letters (17) Macak, J. M.; Tsuchiya, H.; Schmuki, P. Angew. Chem., Int. Ed. 2005, 44, 2100. (18) Christophersen, M.; Carstensen, J.; Voigt, K.; Foll, H. Phys. Status Solidi A 2003, 197, 34. (19) Harraz, F. A.; Kamada, K.; Kobayashi, K.; Sakka, T.; Ogata, Y. H. J. Electrochem. Soc. 2005, 152, C213. (20) Melody, B.; Kinard, T.; Lessner, P. Electrochem. Solid-State Lett. 1998, 1, 126. (21) Varghese, O. K.; Gong, D. W.; Paulose, M.; Grimes, C. A.; Dickey, E. C. J. Mater. Res. 2003, 18, 156. (22) Trasatti, S.; Petrii, O. A. Pure Appl. Chem. 1991, 63, 711. (23) Parkhutik, V. P.; Shershulsky, V. I. J. Phys. D: Appl. Phys. 1992, 25, 1258. (24) Sieber, I.; Hildebrand, H.; Friedrich, A.; Schmuki, P. Electrochem. Commun. 2005, 7, 97. (25) Tsuchiya, H.; Schmuki, P. Electrochem. Commun. 2005, 7, 49. (26) Izutsu, K., Electrochemistry in Nonaqueous Solutions; WileyVCH: 2002. (27) Mor, G. K.; Varghese, O. K.; Paulose, M.; Mukherjee, N.; Grimes, C. A. J. Mater. Res. 2003, 18, 2588. (28) Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B., Jr. Science 2002, 297, 2243.
