Sensors and Actuators B 120 (2006) 63–68 High H 2 sensing performance in hydrogen trititanate-derived TiO 2 Hyun-Su Kim, Won-Taek Moon, Youn-Ki Jun, Seong-Hyeon Hong ∗ School of Materials Science and Engineering and Nano Systems Institute-National Core Research Center, Seoul National University, Seoul 151-742, Republic of Korea Received 15 November 2005; received in revised form 26 January 2006; accepted 26 January 2006 Available online 28 February 2006 Abstract Two types of H 2 Ti 3 O 7 powders were prepared by ion exchange and hydrothermal methods, and the H 2 sensing properties of the H 2 Ti 3 O 7 -derived TiO 2 sensors were examined. In the ion exchange method, Na 2 Ti 3 O 7 was first synthesized via a solid-state reaction, and H 2 Ti 3 O 7 was obtained from Na + /H + exchange on Na 2 Ti 3 O 7 .H 2 Ti 3 O 7 was also prepared via a hydrothermal reaction of TiO 2 powder in a NaOH solution. The morphology, size, and phase evolution of H 2 Ti 3 O 7 were found to be strongly dependent on the preparation methods. The TiO 2 sensors fabricated by the H 2 Ti 3 O 7 powders calcined at 700 ◦ C (ion exchanged) exhibited an excellent gas response (S = 30,000) to 1.0% H 2 /N 2 at 500 ◦ C, which was three orders of magnitude higher than that of the hydrothermally synthesized powder and commercial anatase powder even though its specific surface area was smaller. The higher H 2 gas response in the TiO 2 sensor derived from the ion exchanged H 2 Ti 3 O 7 is discussed in terms of the metastable ␤-TiO 2 and anatase phases. © 2006 Elsevier B.V. All rights reserved. Keywords: H 2 Ti 3 O 7 ;TiO 2 ;H 2 gas sensor; Phase evolution 1. Introduction Hydrogen has attracted a great deal of attention as a clean, efficient, and sustainable energy source [1], which can be used directly for the combustion or as a fuel in fuel cells. For such applications, a reliable hydrogen sensor is needed to detect a leakage from the storage and transporta- tion as well as to monitor the concentration over a wide range. It has been recently reported that TiO 2 thin films with well-dispersed sub-micron pores fabricated by the anodic oxi- dation of a Ti plate exhibited a gas response greater than 10 3 to 1.0% H 2 [2,3]. Varghese et al. [4] obtained the high gas response (∼10 4 ) to 1000 ppm H 2 in a well-defined TiO 2 nano-tube array formed by anodic oxidation. Jun et al. [5] showed an extremely high gas response (1.2 × 10 6 ) to 1.0% H 2 in thermally oxidized TiO 2 films consisting of short cracks (non-continuous) and continuous cracks. These results sug- gest that a hydrogen sensor with a high gas response can be ∗ Corresponding author. Tel.: +82 2 880 6273; fax: +82 2 883 8197. E-mail address: shhong@plaza.snu.ac.kr (S H. Hong). achieved using TiO 2 with various nano-dimensional architec- tures. Recently, low-dimensional nano-structured TiO 2 materials (nano-tube, nano-fiber, and nano-wire) have been prepared by a hydrothermal reaction of TiO 2 powders in an alkaline solu- tion [6]. Among them, the nano-tube has an extremely high specific surface area (>200 m 2 /g), and was identified to be a hydrogen trititanate (H 2 Ti 3 O 7 ) [7].H 2 Ti 3 O 7 is known to have a ramsdellite structure [8] and was considered to be a potential solid oxide fuel cell electrolyte due to its appreciable protonic conductivity [9]. The nano-tubes were sintered into nano-rods after calcination but the diameter of the nano-tube was nearly unchanged [6], indicating the preservation of the high surface area even after calcination at high temperatures. H 2 Ti 3 O 7 was also synthesized by ion exchange from Li 2 Ti 3 O 7 or Na 2 Ti 3 O 7 , and was reported to transform into rutile TiO 2 through a defec- tive and hydrated form or ␤-TiO 2 [8,10]. The high surface- to-volume ratio of the H 2 Ti 3 O 7 -derived powders appears to make their electrical response extremely sensitive to the species adsorbed on the surface, but no attempt has been made to confirm this. The present study was aimed to prepare the H 2 Ti 3 O 7 powders by hydrothermal and ion exchange methods, and investigate the phase evolution of H 2 Ti 3 O 7 powders as well as the H 2 sensing 0925-4005/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2006.01.043 64 H S. Kim et al. / Sensors and Actuators B 120 (2006) 63–68 properties of the sensors fabricated from these H 2 Ti 3 O 7 -derived TiO 2 powders. 2. Experimental 2.1. Preparation and characterization of H 2 Ti 3 O 7 powder Two types of H 2 Ti 3 O 7 powders were prepared by ion exchange and hydrothermal methods. In the ion exchange method, Na 2 Ti 3 O 7 was first synthesized by a solid state reac- tion of Na 2 CO 3 (Aldrich) and TiO 2 (Degussa Co.) at 800 ◦ C, and H 2 Ti 3 O 7 powder was then obtained by hydrolyzing Na 2 Ti 3 O 7 in a 0.5 M HCl aqueous solution at 60 ◦ C for 3 days. Additional pro- cessing details are described elsewhere [10]. In the hydrothermal synthesis, 1 g of commercial TiO 2 powder (Aldrich) was added to 50 ml of a 10 M NaOH aqueous solution and hydrothermally treated in an autoclave at 130 ◦ C for 72 h. The reaction product was washed thoroughly with distilled water and a 0.1 M HCl aqueous solution until the pH of the washing solution became lower than 7 in order to remove sodium (Na) [6,7]. The pow- ders prepared by ion exchange and hydrothermal methods are referred to as i- and h-, respectively, and the calcination tem- perature will follow the hyphen. For example, i-700 designates the powder synthesized by ion exchange method and calcined at 700 ◦ C. The phases of the as-dried and calcined powders were deter- mined by X-ray diffraction (XRD, Cu K␣, λ = 1.5406 ˚ A). For the phase evolution, the as-prepared H 2 Ti 3 O 7 powders were calcined at temperatures from 200 to 1000 ◦ C with an interval of 100 ◦ C for 1 h. In addition, thermal analysis was conducted by differential thermal analysis (DTA) and thermogravimet- ric analysis (TGA). The specific surface area of the powders was measured by a BET method (ASAP 2010, Micromeritics). Scanning electron microscope (SEM) and transmission electron microscope (TEM) were used to observe the morphology of the powders. 2.2. Fabrication and characterization of sensors For the electrical measurements, comb-like Pt electrodes were formed by sputtering Pt on an alumina substrate through a mask, and Pt lead wires were attached to them using a Pt paste [11]. The sensors were fabricated by printing the slurry, which was made by mixing the calcined H 2 Ti 3 O 7 powders (>600 ◦ C) with a 1 wt.% cellulose aqueous solution, on the electroded substrate and heat-treating at 600 ◦ C for 1 h. For comparison, sensors were also fabricated using commercial anatase powder and synthesized Na 2 Ti 3 O 7 powder. The H 2 and CO sensing properties were determined by measuring the changes in the electric resistance between 200–10,000 ppm H 2 (or CO) balanced with N 2 (or air) and ultra pure N 2 (or air) at 500–550 ◦ C. The electrical resistance was measured using a multimeter (2000 Multimeter, Keithley, USA). The magnitude of the gas response (S) is defined as the ratio (R o /R g ) of the resistance in N 2 (or air) (R o ) to that in the sample gas (R g ). The response time (t 90% ) is defined as the time required for the sensor to reach 90% of its final signal. Fig. 1. XRD patterns of the H 2 Ti 3 O 7 powders prepared by the ion exchange method: (A) as-dried and calcined at (B) 400 ◦ C, (C) 600 ◦ C, and (D) 800 ◦ C. 3. Result and discussion 3.1. Characterization of H 2 Ti 3 O 7 powder The phase evolution of the H 2 Ti 3 O 7 powders prepared by the ion exchange method is shown in Fig. 1 as a function of the calcination temperature. The XRD pattern of the as-dried pow- der showed very sharp diffraction peaks (Fig. 1(A)), indicating a well-crystallized structure, and this pattern was in good agree- ment with that of the monoclinic H 2 Ti 3 O 7 (JCPDS Card #41- 1092) except several minor peaks. These peaks were identified to be those of residual, unsubstituted Na-containing compound (Na 2 Ti 6 O 13 ). The Na content in the as-dried powder was 1 wt.%, as determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES). When heated, H 2 Ti 3 O 7 was trans- formed into ␤-TiO 2 , which was accompanied by a 7% weight loss at approximately 250 ◦ C on the TGA curve correspond- ing to the dehydration of H 2 Ti 3 O 7 . As a result, only ␤-TiO 2 phase was observed at 400 ◦ C(Fig. 1(B)), which existed up to 700 ◦ C. ␤-TiO 2 was first prepared by the hydrolysis of K 2 Ti 4 O 9 and subsequent heat treatment [12]. Its structure is less compact than that of the other forms of TiO 2 and it slowly transformed into anatase between 600 and 700 ◦ C [12]. With further heat- ing, peaks for anatase were observed in the specimen calcined at 600 ◦ C(Fig. 1(C)) and kept present up to 900 ◦ C. Rutile began to appear at 800 ◦ C(Fig. 1(D)) and all the powders completely transformed into rutile at 1000 ◦ C. The phase evolution was fur- ther supported by thermal analysis. Two endothermic peaks at 250 and 560 ◦ C and one exothermic peak at 720 ◦ C were found in the DTA curve (not shown here). Based on the XRD results, these peaks were attributed to the transformation from H 2 Ti 3 O 7 to ␤-TiO 2 , from ␤-TiO 2 to anatase, and from anatase to rutile, respectively. On the other hand, very broad diffraction peaks were observed for the hydrothermally synthesized powder (Fig. 2(A)) H S. Kim et al. / Sensors and Actuators B 120 (2006) 63–68 65 Fig. 2. XRD patterns of the H 2 Ti 3 O 7 powders prepared by the hydrothermal method: (A) as-dried and calcined at (B) 400 ◦ C and (C) 800 ◦ C. and the high-resolution transmission electron microscopy (HRTEM) image revealed the nano-tubular morphology (not shown here). In accordance with a previous report [7], the H 2 Ti 3 O 7 nano-tubes were formed during the hydrothermal treat- ment. With heating, H 2 Ti 3 O 7 was directly transformed into anatase without formation of ␤-TiO 2 above 400 ◦ C(Fig. 2(B)), which was much lower than the former case. In the DTA curve (not shown here), only one big endothermic peak was observed around 120 ◦ C, which corresponded to the dehydra- tion of H 2 Ti 3 O 7 . No evidence of ␤-TiO 2 formation was found in this study and the previous report [6]. It is speculated that a scroll structure of a single sheet of titanium oxide inhibits the transformation into ␤-TiO 2 in the hydrothermally synthe- sized H 2 Ti 3 O 7 . The phase transition from anatase to rutile occurred at a comparable temperature to the ion exchanged case (Fig. 2(C)). The morphologies of the calcined powders at 700 ◦ C are shown in Fig. 3. In the ion exchange method, a micron-sized H 2 Ti 3 O 7 powder was initially produced and particle coarsening (or growth) was not significant with heat treatment, resulting in the plate-like particles of 1–2 ␮m size after calcination at 700 ◦ C (Fig. 3(A)). In contrast, nano-tubular particles were originally obtained in the hydrothermal method and calcination resulted in the elongated particles of 20–30 nm wide and 100–200 nm long (Fig. 3(B)). Indeed, HRTEM images indicated that the nano- tubular structure changed to a nano-rod structure with a circular cross-section as in the earlier report [6]. The specific surface areas of the calcined powders (700 ◦ C) determined by BET were 3.6 and 34.5 m 2 /g for the ion exchange and hydrothermal cases, respectively. 3.2. Characterization of sensors Fig. 4(A) shows a response transient to 1.0% H 2 balanced with N 2 of the i-700 sensor measured at 500 ◦ C. Upon injecting a 1.0% H 2 /N 2 sample gas, the resistance rapidly decreased by more than four orders of magnitude. The recovery was slightly Fig. 3. SEM micrographs of the calcined powders (700 ◦ C) prepared by (A) ion exchange and (B) hydrothermal method. Fig. 4. (A) Response transient of the i-700 sensor to 1.0% H 2 /N 2 at 500 ◦ C and (B) magnitude of gas response (S) and response time (t 90% ) in the sensors prepared by i-700, h-700, anatase, and Na 2 Ti 3 O 7 . 66 H S. Kim et al. / Sensors and Actuators B 120 (2006) 63–68 slow but the sensing signal was quite stable and reversible even after switching the gases several times. The magnitude of the gas response (S) for the i-700 sensor was estimated to be 30,000, which was significantly higher than that of commer- cial anatase powder (S = 3.3, Fig. 4(B)) and comparable to the recently reported values achieved in anodically oxidized or ther- mally oxidized TiO 2 films [2–5]. The response time of the i-700 sensor was also very short (t 90% = 1 s). However, the H 2 sens- ing performance of the h-700 sensor was quite poor (S = 34.5) compared with the i-700 sensor even though the specific surface area was 9–10 times higher (34.5 versus 3.6 m 2 /g). It was speculated that the superior sensing performance of the i-700 sensor might be due to the composition or crystal struc- tures (phases) of the i-700 powder. The i-700 powder contains a considerable amount of sodium, and thus the sensor was fab- ricated from Na 2 Ti 3 O 7 powder alone. As shown in Fig. 4(B), the magnitude of the gas response for Na 2 Ti 3 O 7 sensor was 6, which suggests that Na inclusion had a negligible effect on the H 2 sensing properties, considering the gas response of 30,000 for the i-700 sensor. Another highly possible cause was considered to arise from the crystal structure of the calcined powders. The XRD results indicated that the i-700 powder consisted of ␤-TiO 2 and anatase while the h-700 powder was composed of anatase. Based on this comparison, the presence of ␤-TiO 2 appears to be responsible for the high H 2 sensing performance in the i-700 sensor. In order to determinethe role of ␤-TiO 2 , the powders obtained by the ion exchange method were calcined at different temper- atures (600–800 ◦ C) and the H 2 sensing properties were deter- mined. More ␤-TiO 2 phase was present in the powder calcined at 600 ◦ C, and ␤-TiO 2 phase completely disappeared at 800 ◦ C, which resulted in a mixture of anatase and rutile. The electrical resistances and the magnitude of gas response to 1.0% H 2 /N 2 are shown in Fig. 5 as a function of the calcination temperature. Contrary to our expectation, the i-600 sensor had much lower gas response than the i-700 even though it contained more ␤-TiO 2 phase. It appears that the high H 2 sensing performance is not directly related to the amount of ␤-TiO 2 present. As expected, the i-800 sensor exhibited the lowest gas response possibly due to the low gas response of the rutile phase. Fig. 5. Electrical resistance (R o in N 2 gas and R g in sample gas (1% H 2 )), magnitude of gas response, and response time at 500 ◦ C in the i-600, i-700, and i-800 sensors. A close examination revealed that the resistances in N 2 gas (R o ) were in the similar range irrespective of the calcination temperatures but the resistance in the sample gas (R g ) was lower by three orders of magnitude for the i-700 sensor. The notice- able decrease of resistance in the sample gas might be found in the characteristics of ␤-TiO 2 . ␤-TiO 2 is known to transform slowly into anatase between 600 and 700 ◦ C [12]. Based on this meta-stability, it was postulated that ␤-TiO 2 might be unstable in certain atmosphere (H 2 or N 2 ) and transform into a more stable phase. This transformation might contribute the decrease of the resistance in the i-700 sensor. To confirm this hypothe- sis, the i-700 powder was annealed under the sensor operation conditions (500 ◦ C, N 2 and 1.0 H 2 /N 2 ) and XRD analysis was conducted at room temperature. Indeed, the relative peak inten- sities of the ␤-TiO 2 and anatase phases were varied depending on the atmosphere used (Fig. 6). The phase transformation was reversible, which was demonstrated by the repeated atmospheric changes. The peak intensity of ␤-TiO 2 was slightly reduced after annealing for 40 h in air, indicating that the phase transforma- tion from ␤-TiO 2 to anatase at 500 ◦ C is quite slow and that the thermally induced transformation is not significant compared to the atmospheric change. From this observation, the high H 2 sensing performance of the i-700 sensor can be attributed to the reversible phase transformation of the ␤-TiO 2 phase with atmo- sphere and the accompanying change in resistance. The i-600 sensor exhibited the similar behavior with atmosphere change but the peak intensity difference was very small, which might reflect the lower gas response of i-600 sensor. At present, the mechanisms for the phase transformation and resistance change with atmosphere are not well understood and in situ phase anal- ysis under sensor operation conditions is further required. The concentration dependence of the H 2 gas response in the i-700 sensor at 550 ◦ C is shown in Fig. 7(A). The magnitude of the gas response increased almost linearly with increasing H 2 concentration from 200 to 10,000 ppm. The sensor exhib- ited excellent sensing properties over a wide range of the H 2 concentrations. As a sensing mechanism in a N 2 atmosphere, Fig. 6. XRD patterns of the i-700 powders annealed at 500 ◦ C in different atmo- spheres: (A) 1% H 2 /N 2 and (B) pure N 2 . H S. Kim et al. / Sensors and Actuators B 120 (2006) 63–68 67 Fig. 7. (A) H 2 concentration dependence of the gas response and (B) H 2 selec- tivity against CO in the i-700 sensor. Varghese et al. [4] suggested the chemisorption of the spilled- over hydrogen atoms and the consequent creation of an electron accumulation layer on the TiO 2 surface, which enhanced the electrical conductance. The suggested mechanism was different from the operating principle of semiconductor-type sensors in an oxidizing atmosphere. The selectivity between H 2 and CO gases and the effect of the balance gas in the i-700 sensor were further investigated at 500 ◦ C(Fig. 7(B)). The magnitude of the gas response to 1.0% CO balanced with N 2 was ∼15, which was 2000 times lower than that for H 2 . Thus, the i-700 sensor exhibited an excellent selectivity toward H 2 gasinaN 2 atmosphere. However, the gas response was extremely low in air. The magnitude of the gas response to 1.0% H 2 balanced with air was only 3 and even lower with CO gas. In the TiO 2 sensor with Pd electrode, the lower gas response in air atmosphere was attributed to the par- tial oxidation of Pd into PdO and the resultant decrease of H atom dissolution in the Pd electrode [13]. However, the appli- cation of this mechanism to Pt electrode of this experiment is hardly conceivable. Further studies are needed to determine the sensing mechanism at different atmospheres and to improve the gas response in the presence of oxygen for the practical applications. 4. Conclusion Micron-sized, plate-like H 2 Ti 3 O 7 was obtained in the ion exchange method and it transformed into ␤-TiO 2 , anatase, and rutile in sequence with heat treatment. Trititanante nano-tubes were obtained in the hydrothermal method and directly trans- formed into anatase without formation of ␤-TiO 2 . The i-700 sen- sor, which was composed of ␤-TiO 2 and anatase phases, showed a quick response (t 90% = 1 s) and an excellent gas response (S = 30,000) to 1.0% H 2 /N 2 at 500 ◦ C. However, the H 2 gas response was three orders of magnitude lower in the h-700 sen- sor, which was composed of anatase phase, even though the specific surface area was 9–10 times higher than i-700. The higher H 2 sensitivity in the i-700 sensor is speculated to be due to the reversible phase transformation between ␤-TiO 2 and anatase depending on the atmosphere used and the resultant change in resistivity. The i-700 sensor exhibited the linear con- centration dependence over a wide range of H 2 concentrations (20–10,000 ppm) and the high H 2 selectivity against CO. How- ever, the gas response in air was extremely small and needs to be further improved for practical applications. Acknowledgment This work was supported by the Nano Systems Institute- National Core Research Center (NSI-NCRC) program of Korea Science and Engineering Foundation (KOSEF), Korea. References [1] M.S. Dresselhaus, I.L. Thomas, Alternative energy technologies, Nature 414 (2001) 332–337. [2] Y. Shimizu, N. Kuwano, T. Hyodo, M. Egashira, High H 2 sensing perfor- mance of anodically oxidized TiO 2 film contacted with Pd, Sens. Actuators B 83 (2002) 195–201. [3] T. Iwanaga, T. Hyodo, Y. Shimizu, M. Egashira, H 2 sensing properties and mechanism of anodically oxidized TiO 2 film contacted with Pd electrode, Sens. Actuators B 93 (2003) 519–525. [4] O.K. Varghese, D. Gong, M. Paulose, K.G. Ong, E.C. Dickey, C.A. Grimes, Extreme changes inthe electrical resistance of titania nanotubes with hydro- gen exposure, Adv. Mater. 15 (2003) 624–627. [5] Y K. Jun, H S. Kim, J H. Lee, S H. Hong, High H 2 sensing behavior of TiO 2 films formed by thermal oxidation, Sens. Actuators B 107 (2005) 264–270. [6] Z Y. Yuan, B L. Su, Titanium oxide nanotubes, nanofibers and nanowires, Colloids Surf. A 241 (2004) 173–183. [7] Q. Chen, W. Zhou, G. Du, L M. Peng, Trititanate nanotubes made via a single alkali treatment, Adv. Mater. 14 (2002) 1208–1211. [8] A. Le Bail, J.L. Fourquet, Crystal structure and thermal behavior of H 2 Ti 3 O 7 : a new defective ramsdellite form from Li + /H + exchange on Li 2 Ti 3 O 7 , Mat. Res. Bull. 27 (1992) 75–85. [9] D.J.D. Corcoran, D.P. Tunstall, J.T.S. Irvine, Hydrogen titanates as poten- tial proton conducting fuel cell electrolytes, Solid State Ionics 136-137 (2000) 297–303. [10] H. Izawa, S. Kikkawa, M. Koizumi, Ion exchange and dehydration of layered titanates, Na 2 Ti 3 O 7 and K 2 Ti 4 O 9 , J. Phys. Chem. 86 (1982) 5023–5026. [11] K. Nomura, H. Shiozawa, T. Takada, H. Reuther, E. Richter, Gas-sensor properties of SnO 2 films implanted with gold and iron ions, J. Mater. Sci. Mater. Electron. 8 (1997) 301–306. [12] R. Marchand, L. Brohan, M. Tournoux, TiO 2 (B) a new form of titanium dioxide and the potassium octatitanate K 2 Ti 8 O 17 , Mater. Res. Bull. 15 (1980) 1129–1133. [13] H. Miyazaki, T. Hyodo, Y. Shimizu, M. Egashira, Hydrogen-sensing prop- erties of anodically oxidized TiO 2 film sensors: effects of preparation and pretreatment conditions, Sens. Actuators B 108 (2005) 467–472. 68 H S. Kim et al. / Sensors and Actuators B 120 (2006) 63–68 Biographies Hyun-Su Kim studied materials science and engineering and received his BS in 2003 at Seoul National University in Korea. He is currently studying for MS degree at Seoul National University. His research interests is semiconducting gas sensor and nano-materials. Won-Taek Moon studied materials science and engineering and received his BS in 2003 at University of Seoul in Korea. He is currently studying for MS degree at Seoul National University. His research interests include semiconducting gas sensors and thin film deposition. Youn-Ki Jun studied materials science and engineering and received his BS, MS in 1999 and 2001, respectively, at Seoul National University in Korea. He is currently studying for PhD degree at Seoul National Uni- versity. His research interests are semiconducting gas sensor and ceramic processing. Seong-Hyeon Hong has been anassistant professor at Seoul National University since 1998. He received MS degree in 1990 from Seoul National University and PhD degree in 1996 from Pennsylvania State University. His research interest includes the processing of inorganic materials. . high H 2 sensing performance in the i-700 sensor. In order to determinethe role of ␤-TiO 2 , the powders obtained by the ion exchange method were calcined. Sensors and Actuators B 120 (2006) 63–68 High H 2 sensing performance in hydrogen trititanate-derived TiO 2 Hyun-Su Kim, Won-Taek Moon, Youn-Ki