Formation of crystallized titania nanotubes and their transformation into nanowires

Transcription

1 INSTITUTE OF PHYSICS PUBLISHING Nanotechnology 16 (2005) NANOTECHNOLOGY doi: / /16/9/086 Formation of crystallized titania nanotubes and their transformation into nanowires BPoudel 1,3,WZWang 1,CDames 2,JYHuang 1,SKunwar 1, DZWang 1,DBanerjee 1,GChen 2 and Z F Ren 1,3 1 Department of Physics, Boston College, Chestnut Hill, MA 02467, USA 2 Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA poudel@bc.edu and renzh@bc.edu Received 25 May 2005, in final form 4 July 2005 Published 28 July 2005 Online at stacks.iop.org/nano/16/1935 Abstract Gram quantities of titania (TiO 2 )nanotubes, with a typical outside diameter about 9 nm, wall thickness of about 2.5 nm, and length of about 600 nm, were synthesized from anatase nanopowder and micropowder using the hydrothermal method. The crystallization, structure, and phase stability of the nanotubes at high temperatures were studied. A morphology change from nanotube to nanowire was observed at 650 C. The as-prepared nanotubes were usually contaminated with sodium impurities and were poorly crystallized, but under optimized synthesis conditions the impurity phase was completely removed, resulting in highly crystallized nanotubes. The volume filling fraction of the autoclave as well as the concentration of the acid treatment were found to be particularly important for controlling the purity and crystallinity of the resulting nanotubes. The various TiO 2 -derived nanotube phases (sodium titanate and hydrogen titanate) reported previously by different groups were also observed under different synthesis conditions, resolving the contradiction among the previous results. 1. Introduction Low dimensional titania (TiO 2 ) nanostructures such as nanotubes, nanowires, and nanobelts have attracted much attention in recent years because they have potential applications in electronics, optics, catalysts, sensors, and energy conversion [1 7]. For example, titania nanotubes have shown more than a twofold increase in short-circuit current density compared to nanoparticles in thin film dye-sensitized solar cells [8]. Nanotubes have large specific surface area available for the absorption of photons compared to bulk material and they also provide channels for enhanced electron transfer, thereby helping to increase the efficiencies for solar cells, electrolysis, and photocatalysis. Also, the TiO 2 -derived nanotubes such as sodium titanate (Na x H 1 x Ti 3 O 7, x 0.75) and hydrogen titanate (H 2 Ti 3 O 7 ) have useful optical and magnetic properties such as room 3 Authors to whom any correspondence should be addressed. temperature photoluminescence [9]. For all of these applications, purity, crystallinity, and stability at elevated temperatures are particularly important. Titania nanotubes of various dimensions have been synthesized by different groups using techniques like anodization [10], sol gel [5], and molecular assembly [8]. However, these methods are either not suitable for large scale production or not able to yield very low dimensional, well separated, crystallized nanotubes without one doing postfabrication annealing. Asimple and cost-effective hydrothermal method for the large scale production of pure titania nanotubes with small diameters has been introduced by Kasuga et al [1, 2]. There is currently much discussion about the formation mechanism and phases of the resulting nanotubes. In their pioneering work, Kasuga et al have concluded that the as-prepared nanotubes were titania and were formed during the process of washing after the hydrothermal treatment. Following a similar /05/ $ IOP Publishing Ltd Printed in the UK 1935

2 BPoudel et al procedure, Yao et al [11]have alsoobserved titania nanotubes. Sun and Li [9] have obtained nanotubes with both titania and sodium titanate phases, while concurring that the washing procedure was the key to nanotube formation. Du et al [12] have found that the resulting nanotubes were hydrogen titanate (H 2 Ti 3 O 7 ) rather than titania. Yoshikazu et al [13], however, have recently reported that the as-prepared nanotubes were composed of hydrated hydrogen titanate (H 2 Ti 3 O 7 nh 2 O (n < 3)). The structures of such nanotubes have been discussed in the reports by Chen et al [14, 15] andatransformation from hydrogen titanate to anatase or rutile under hydrothermal treatment in acidic solution has been reported in various literature [9, 16]. The nanotubes in these reports have all exhibited poor crystallinity. Wang et al [3] havereported slightly better crystallinity with nanotubes in the anatase phase, while concluding that the washing process was not important for nanotube formation. However, a careful inspection of their energy dispersion x-ray analysis (EDX) spectra reveals that their nanotubes were contaminated with sodium impurities. As we discuss below, these impurities are more apparent after heating above 550 C. In this context, we have studied the synthesis conditions in order to determine optimal parameters for the growth of pure crystalline anatase nanotubes by the hydrothermal method. By studying the alkali treatment and acid washing steps, we have identified conditions leading to each of the various phases reported previously by different groups, and established the optimal conditions for the formation of pure crystalline titania nanotubes. Because many technological applications operate at elevated temperatures, we have also studied the structure and phase stability of the nanotubes at temperatures up to 1000 C. 2. Experimental details The starting materials were anatase nanopowder (99.8%, 100 nm, Aldrich) or micropowder (99.9%, 44 µm, Alfa Aesar), NaOH (97%, Aldrich), and HCl (35%, Aldrich). In the optimized synthesis using micropowder, 3 g of anatase powder was first mixed with 168 ml of 10 M NaOH, stirred for 1 h in a beaker, and transferred into a 200 ml autoclave (84% filling fraction). After 30 h of reaction in the sealed autoclave at 120 C, the product was treated with 1 M HCl solution for at least 2 h, washed several times with distilled water, anddried in an oven for several hours at 110 C(sample C). To observe the effect of acid treatment on the crystallinity and purity of the nanotubes, some of the product after the hydrothermal treatment was taken out and washed only with water (sample A), and another part of the same product was treated with 0.1 M HCl and washed with water (sample B). Similarly, to observe the effect of the filling fraction of the autoclave, the autoclave was filled with different volumes of the solution, and after the hydrothermal step the product was treated with different concentrations of HCl and washed with water. The samples obtained from 50% (100/200 ml) filling fraction of the autoclave and treatment with 0.1 and 1 M HCl were denoted as D and E, respectively. The synthesis conditions and properties of the resulting nanotubes of all of these samples are summarized in table 1. To study the stability of the nanotubes at high temperature, samples were annealed for 2 h in a tube furnace either Table 1. Synthesis conditions, crystallinity, and impurities for various nanotube samples. Autoclave Impurities filling HCl fraction concentration Hydrogen Sample (%) (M) Crystallinity Sodium titanate A 84 0 Good Present Present B Good Present Present C 84 a 1 a Excellent None None D Poor Present Present E 50 1 Poor Present None a Represents the optimal condition. in vacuum or in oxygen at various temperatures up to 1000 C. For the annealing, the furnace was first ramped up at 40 Cmin 1 to the target temperature, then held for about 45 min, and finally cooled down to room temperature inside the furnace with the power turned off. The overall phase of the nanotube samples was analysed using an x-ray diffractometer (XRD, Bruker-AXS, G8 GAADS) with Cu Kα radiation, and the phase of an individual nanotube or a small region of the nanotubes was determined using selected-area electron diffraction (SAED). The morphology of the samples was examined using scanning electron microscopy (SEM, JEOL-6340F) and high resolution transmission electron microscopy (HRTEM, JEOL-2010F), while the atomic composition was determined using an EDX spectrometer attached to the TEM. 3. Results and discussion 3.1. Optimization of synthesis conditions for pure crystalline titania nanotubes As explained further below, synthesis of pure crystalline titania nanotubes requires optimization of the temperature of the autoclave during the hydrothermal treatment, the volume of solution in the autoclave (filling fraction), and the concentration of the acid during the wash step. SEM (figure 1(a)) and TEM (figure 1(c)) images of a sample from optimal synthesis conditions (sample C) clearly show the abundance of highly crystallized titania nanotubes. The typical length of these nanotubes is 600 nm, the outside diameter is 9 nm,andthewall thickness is 2.5 nm. The spacing between layers of the wall is usually 0.70 to 0.75 nm (figure 1(c) inset), significantly less than that observed by Suzuki and Yoshikawa [13], who reported 0.92 and 0.79 nm for H 2 Ti 3 O 7 nh 2 O (n < 3) and H 2 Ti 3 O 7,respectively, and by Wang et al [3], who reported 0.93 nm for titania. This may be because of the excess H 2 Oand/or hydrogen titanate leading to larger layer spacing. Also, we conclude that the radial direction cannot be the c-axis of the crystal lattice as suggested by Wang et al because the observed spacing is smaller than the usual 0.93 nm interplanar spacing along the c-axis. For all of the nanotube samples from optimal synthesis conditions, EDX studies show that the molar ratio of Ti and O is 1:2, consistent with the standard TiO 2 source nanopowder as described in the experimental details. Figures 2(b) (d) show the XRD patterns of the as-prepared nanotubes from 1936

3 Formation of crystallized titania nanotubes and their transformation into nanowires (a) (b) (c) (d) Figure 2. XRDpatterns of nanotubes synthesized under optimal autoclave conditions and treated with (b) 0 M (sample A), (c) 0.1 M (sample B), and (d) 1 M HCl (sample C) showing that all the samples, with or without acid treatment, exhibit strong anatase peaks. (d) Also shows that hydrogen titanate peaks can be removed by 1 M acid treatment. Figure 1. (a) SEM and (c) TEM images of the nanotubes synthesized under optimal autoclave conditions and treated with 1MHCl (sample C), showing good uniformity and crystallinity, (b) SEM image of the sample from the same synthesis but without acid treatment (sample A), showing similar morphology, and (d) TEM image of the 1 M HCl treated nanotubes synthesized under non-optimal autoclave conditions at 50% filling fraction (sample E), showing poor crystallinity. the optimal autoclave conditions treated with various acid concentrations, as compared to the original starting anatase micropowder (figure 2(a)), confirming the anatase phase of the nanotubes. To our surprise, the autoclave filling fraction had a strong influence on the crystallinity of the resulting nanotubes, with different optimal filling fractions for micropowders and nanopowders. For the micropowder in the standard hydrothermal protocol at 120 Cfor30 h reaction time, it was necessary to fill the autoclave to at least 84% by volume (168 ml in a 200 ml autoclave) to achieve excellent crystallinity. At lower volume fractions, varying the temperature between 110 and 135 Cdid not improve the crystallinity, as shown in a TEM picture (sample C, figure 1(d)) for 50% volume fraction. For nanopowder, it was necessary to fill the autoclave to at least 90% to achieve good crystallinity. Similar requirements of high autoclave filling fraction were seen for experiments using asmaller autoclave vessel(125 ml) or in an argon environment, ruling out the possible effect on the crystallinity caused by the presence of atmospheric gases. One possible explanation for this unusual observation is that the pressure inside the vessels during the autoclave step is strongly dependent on the filling fraction above 80% because of the thermal expansion of the different volumes of liquid water: although we have not measured the pressure directly, simple calculations suggest that the pressure inside an 84% filled vessel exceeds that of a 50% filled vessel byabout 50 kpa (392 versus 342 kpa). For the 90% filling fraction required for nanopowder, the calculated change is 136 kpa (478 versus 342 kpa). This observation also suggests that the nanopowder needs more pressure to form crystalline nanotubes compared to micropowder. More work is necessary to identify the true mechanism linking crystallinity with high filling fractions. We found that acid treatment was necessary and identified an optimal concentration between 0.5 and 1.5 M for high yield of pure nanotubes. Concentrations above 2 M destroyed the nanotubes leaving only 100 nm clumps, while concentrations below 0.5 M failed to remove the sodium impurities. However, even at these optimal concentrations the sodium impurities could be removed only when the autoclave step was performed close to the optimal volume faction ( 84% for micropowder, and 90% for nanopowder). This is evident from figure 3 which shows the XRD spectra of several nanotube samples made from micropowder after annealing at650 C(asdiscussed below, annealing at this temperature causes any sodium impurities to crystallize into sodium titanate). The samples with high autoclave filling fraction can be stripped of any sodium impurities for acid concentrations above 0.5 M (sample C, figure 3(d))but retained for concentrations below 0.5 M (sample B, figure 3(c)). Samples with lower autoclave filling fraction retain sodium impurities for both high (sample E, figure 3(b)) and low (sample D, figure 3(a)) acid concentrations (although samples treated with acid concentrations exceeding 1.5 M are free of sodium impurities for all autoclave filling fractions between 33% and 90%, the resulting nanotubes are partially or completely destroyed). Treatment with nitric acid gave a similar result. This observation is further supported by comparing the EDX spectra in figure 4, whichshowsthatfor 84% autoclave filling fraction, sodium remains after washing with water (sample A, figure 4(a)) or treating with 0.1 M acid (sample 1937

4 BPoudel et al Figure 3. XRD patterns of nanotube samples annealed at 650 C, synthesized under the conditions of (a) 50% autoclave filling fraction and 0.1 M acid treatment (sample D), (b) 50% and 1 M (sample E), (c) 84% and 0.1 M (sample B), and (d) 84% and 1 M (sample C), the only condition for complete removal of the sodium impurity phase. Figure 4. EDXpattern of nanotube samples synthesized under optimal autoclave conditions and treated with (a) 0 M (sample A), (b) 0.1 M (sample B), and (c) 1 M HCl (sample C), the only condition for complete removal of the sodium impurities, and (d) samples synthesized under non-optimal autoclave conditions at 50% filling fraction, showing that the sodium impurities remain even after 1 M HCl (sample E). Cu and C peaks are from the TEM grid. B, figure 4(b)), but can be removed completely after treating with 1M acid (sample C,figure 4(c)). At 50% autoclave filling fraction, however, sodium remains even after treating with 1 M acid (sample E, figure 4(d)). These results suggest that good crystallinity prior to the acid treatment may cause the sodium atoms to be more weakly bound, and more easily removed by the acid. The strong anatase peaks of figure 2(b) clearly show that crystalline nanotubes form prior to any acid treatment, consistent with Wang et al [3], and contrary to the mechanism originally proposed by Kasuga et al [2] wheremetastable anatase [17] does not begin to appear until the acid treatment. An SEM image of the sample prior to the acid treatment step (sample A, figure 1(b)) also shows that the nanotubes are formed before the treatment. Apparently the primary effect of the acid treatment is only to remove impurities. For example, small hydrogen titanate peaks are seen in the XRD spectra for acid treatment up to 0.1 M (figures 2(b) and (c)) but are gone after acid treatment at 1 M (figure 2(d)). Similarly, sodium peaks are seen in the EDX spectra for acid treatment up to 0.1 M (figures 4(a) and (b)) but are gone after acid treatment at 1 M (figure 4(c)). Thus, we conclude that the nanotube crystallinity is controlled primarily by the autoclave step, while the impurities are controlled primarily by the acid treatment Observation and explanation of other phases An additional outcome of the systematic study outlined above is that we have identified synthesis conditions that consistently result in the other titanium oxide phases that have been previously reported, including pure sodium titanate, pure hydrogen titanate, and mixtures of either of these with titania. The XRD spectra of these phases are shown in figure 5 and discussed further below. Sodium impurities are apparently always present prior to the acid treatment regardless of the autoclave temperature and filling fraction. However, for these impurities to crystallize into pure sodium titanate (figure 5(c)), we have found that the autoclave filling fraction should be less than 84% Figure 5. XRD patterns of nanotubes synthesized in different synthesis conditions: (a) 50% autoclave filling fraction at 110 C and 0.1 M HCl with no annealing, (b) 84%, 120 C, 0.1 M,andno annealing, (sample B), (c) 50%, 120 C, 0.1 M, and650 C annealing, (sample D), (d) 84%, 120 C, 0.1 M, and650 C annealing, (sample B), and (e) 84%, 120 C, 1 M, and no annealing. (168/200 ml), the acid concentration should be less than 0.5 M, and the annealing temperature should be above 550 C. If the autoclave filling fraction is higher and the acid concentration is lower, the resulting nanotubes will be a mixture of hydrogen titanate and anatase at room temperature (figure 5(b)), and become a mixture of sodium titanate and anatase after annealing at temperatures above 550 C(figure 5(d)). If the acid concentration is higher, all thesodium will be removed and no sodium titanate phase will appear after the annealing (figure 5(e)). If the annealing temperature is too low, the sodium will not be able to crystallize into sodium titanate. Poorly crystallized hydrogen titanate phases have been widely reported at different steps in the synthesis. For example, Suzuki and Yoshikawa [13]recently reported hydrogen titanate after the acid treatment of nanotubes synthesized by reacting 1938

5 150 mg of TiO 2 powder with 50 ml of 10 M NaOH for 72 h in an autoclave at 110 C. Sun and Li [9]reported similar phases but used 500 mg of TiO 2 in 40 ml of 10 M NaOH for 48 h between 100 and 180 C. Wang et al [3] synthesized nanotubes by a slightly different method without an autoclave, and their XRD pattern also shows a mixture of anatase and hydrogen titanate. We also observe this hydrogen titanate phase (figure 5(a)) when the acid concentration is 0.1 M and the autoclave filling fraction is less than 80%, for synthesis from micropowder at 110 C. Similar results are obtained for nanopowder when the volume fraction is less than 90%. With additional work it may also be possible to optimize the synthesis conditions for pure, bettercrystallized hydrogen titanate. Formation of crystallized titania nanotubes and their transformation into nanowires 3.3. Phase stability of the as-prepared nanotubes at elevated temperatures We have also studied the temperature stability of nanotubes prepared at optimal autoclave conditions and 0.1 M acid treatment (sample B), including the crystallization of the sodium titanate phase, the transformation of anatase to rutile, and an unexpected change from nanotube to nanowire morphology. As mentioned previously, the nanotubes prepared at these conditions contain sodium impurities. Figure 6(a) shows the XRD spectra of nanotube samples annealed at different temperatures in vacuum, which clearly shows that sodium titanate begins to crystallize at 600 Cand is stronger by 650 C. Peaks corresponding to rutile begin to appear at 800 C, well below the transformation temperature of 925 C we observed for the bulk nanopowder, but well above the transformation temperature of 580 C reported for titania nanotubes prepared by anodization [10]. When similar nanotube samples were annealed in an oxygen environment, XRD patterns (figure 6(b)) show that astrong sodium titanate phase is present even at 600 C, and the rutile peaks at 800 Care also stronger than for the vacuum annealedsamples. This shows that the nanotubes are less stable in oxygen compared to vacuum, although still more stable than the nanotubes prepared by anodization Morphology change of nanotubes into nanowires Although the original morphology is preserved for nanotubes treated with 0.1 M acid and annealed at temperatures up to 550 C, the SEM image in figure 7(a) shows that the nanotubes annealed at 650 Chavechanged into shorter rods with larger diameters. HRTEM (figure 7(c)) shows that these shorter rods are, in fact, nanowires with diameters more than 20 nm. Similar morphology changes are also observed for nanotubes treated with 1 M acid (sample C, figures 7(b) and (d)). The phase of the latter nanowires is still a pure anatase as seen in XRD patterns (figure 5(e)) and reconfirmed by SAED, while the phase of the nanowires obtained by treating only with 0.1 M acid is mixed with sodium titanate. SEM observation of the samples annealed at 800 Cshowsthat nanotubes collapse into an irregularly shaped morphology losing their one-dimensional structures. To better understand the formation, phase, and morphology change during heat treatment, we are performing a systematic study using in situ Figure 6. XRDpatterns of nanotubes synthesized under optimal autoclave conditions, treated with 0.1 M HCl (sample B), and annealed at different temperatures in (a) vacuum, showing weak sodium titanate peaks at 600 Cand weak rutile peaks at 800 C, and in (b) oxygen ambient, showing strong sodium titanate peaks at 600 C, strong rutile peaks at 800 C, and nearly complete transformation of anatase to rutile at 870 C. high temperature TEM observations, which will be reported in a forthcoming paper. 4. Conclusion We have identified conditions for synthesizing highly crystallized titania nanotubes, with no sodium impurities and no hydrogen titanate phase, from the hydrothermal method starting from either nanopowders or micropowders of bulk anatase. We conclude that the nanotube crystallinity is controlled primarily by the autoclave step, while the impurities are controlled primarily by the acid treatment. When starting from micropowder, it is necessary to fill the autoclave to more than 84% by volume and treat with 0.5 to 1.5 M HCl to obtain nanotubes with high purity and crystallinity. When starting from nanopowder, it is necessary to increase the autoclave filling fraction to 90%. The anatase phase of the asprepared nanotubes is stable up to 700 C, butthe morphology changes from nanotubes to nanowires above 550 C. If sodium impurities are present, they typically crystallize into sodium titanate above 600 C. Synthesis conditions have also been identified for obtaining pure hydrogen titanate and sodium titanate nanotubes, thus clarifying some of the current controversy in the literature about the phases of nanotubes grown with the hydrothermal method. 1939

6 BPoudel et al (a) (b) (c) (d) Figure 7. SEM (a) and TEM (c) images of nanotubes synthesized under optimal autoclave condition and treated with 0.1 M HCl (sample B), and annealed at 650 C; (b) SEM and (d) TEM images of similar samples but treated with 1 M HCl (sample C) showing the morphology changes into nanowires. Acknowledgments We thank Professor Jefferson W Tester for helpful discussions. The work performed by Boston College is supported by the US Army Research Development and Engineering Command, Natick Soldier Center, under grant DAAD and NASA. The work performed by MIT is supported by DOE, under grant FG02-02ER References [1] Kasuga T, Hiramatsu M, Hoson A, Sekino T and Nihara K 1998 Langmuir [2] Kasuga T, Hiramatsu M, Hoson A, Sekino T and Nihara K 1999 Adv. Mater [3] Wang W, Varghese O, Paulose M and Grimes C A 2004 J. Mater. Res [4] Wang W, Grimes C A, Varghese O and Paulose M 2003 unpublished [5] Lei Y, Zhang L D, Meng G W,LiGH,ZhangXY, Liang C H, Chen W and Wang S X 2001 Appl. Phys. Lett [6] Mor G K, Carvalho M A, Varghese O K, Pishko M V and Grimes C A 2004 J. Mater. Res [7] Tian Z R, Voigt J A, Liu J, McKenzie B and Xu H 2003 J. Am. Chem. Soc [8] Adachi M, Murata Y, Okada I and Yoshikawa S 2003 J. Electrochem. Soc. 150 G488 [9] Sun X and Li Y 2003 Chem. Eur. J [10] Varghese K, Gong D, Paulose M, Grimes C A and Dickey E C 2003 J. Mater. Res [11] Yao B D, Chan Y F, Zhang X Y, Zhang W F, Yang Z Y and Wang N 2003 Appl. Phys. Lett [12] Du G H, Chen Q, Che R C, Yuan Z Y and Peng L M 2001 Appl. Phys. Lett [13] Suzuki Y and Yoshikawa S 2004 J. Mater. Res [14] Chen Q, Du G H, Zhang S and Peng L M 2002 Acta Crystallogr. B [15] Chen Q, Zhou W, Du G H and Peng L M 2002 Adv. Mater [16] Zhu H Y, Lan Y, Gao X P, Ringer S P, Zheng Z F, Song D Y and Zhao J C 2005 J. Am. Chem. Soc [17] Feist T and Davies P K 1992 J. Solid State Chem