hierarchically structured anatase nanotubes and membranes

Besides the sub-250 nm diameter tube holes, nitro-gen physisorption also revealed that the tube walls were mesoporous with pore diameters centered around 4 nm with extremely high specific

Hierarchically structured anatase nanotubes and membranes

B Maa,b, G.K.L Goha,*, T.S Zhanga, J Mab

a

Institute of Materials Research and Engineering, A*STAR (Agency for Science, Technology and Research), 3 Research Link, Singapore 117602, Singapore

b

School of Materials Science and Engineering, Nanyang Technological University, Singapore 639798, Singapore

a r t i c l e i n f o

Article history:

Received 14 August 2008

Received in revised form 6 March 2009

Accepted 2 May 2009

Available online 9 May 2009

Keywords:

TiO2

Membranes

AAO

Mesoporous

Photocatalysis

a b s t r a c t

Hierarchically structured anatase membranes containing both macro- and mesopores were synthesized

by liquid phase deposition of porous anodic aluminum oxide (AAO) templates at 50 °C It was observed that titania initially forms nanotubes held together by residual AAO, then takes on completely the shape

of the AAO template by displacing alumina and finally forms ‘capped’ membranes in which a layer of mesoporous titania covers the ends of the template Besides the sub-250 nm diameter tube holes, nitro-gen physisorption also revealed that the tube walls were mesoporous with pore diameters centered around 4 nm with extremely high specific surface areas ranging from 179 to 552 m2/g Active photodeg-radation of aqueous solutions of methylene blue indicated that the membranes were photocatalytically active and the highest degradation constant was observed for AAO templates that had been treated for

5 h by liquid phase deposition

Ó 2009 Elsevier Inc All rights reserved

1 Introduction

Titania can be used as a photocatalyst for the decomposition of

aldehyde gas[1,2], destruction of water borne pollutants[3–5]and

water-splitting for hydrogen generation[6] Titania nanostructures

are incorporated in optical devices[7], gas sensors[8]and

dye-sen-sitized solar cells[9] In addition, titania structures are resistant to

organic solvents, are mechanically stable and self-cleaning under

ultraviolet radiation As such, chemical solution methods of

syn-thesizing and depositing titania films and nanostructures become

important because of the relative simplicity and low costs of such

methods

Crystalline titania structures including porous films, nanorods

and nanoparticles can be formed by chemical methods such as

pre-cipitation, sol–gel processing, liquid phase deposition and

hydro-thermal deposition [10–14] With increasing interest in using

titania membranes not only for photocatalysis but also for

filtra-tion in water purificafiltra-tion, the use of anodic aluminum oxide

(AAO) membranes for forming such bifunctional membranes

be-comes an attractive method Although titania nanotubes can be

synthesized directly by anodic oxidation of titanium, the

underly-ing impermeable substrate renders the assemblage unsuitable for

filtration

Yamanaka and co-workers[15]have reported the use of porous

AAO templates and liquid phase deposition (LPD) to fabricate TiO2

and SnO2nanotubes However, to avoid disruption of the template

structure, the growth temperature was kept low (293 K), hence

resulting in close packed oxide nanotubes that were amorphous and therefore not photocatalytic in the as-synthesized state In the present work, self-supporting titania membranes were success-fully prepared by LPD of AAO templates at slightly elevated temper-atures (50 °C) that resulted in the formation of crystalline anatase membranes without the need for subsequent high temperature heat treatment The present titania porous membrane contains the structure of the ordered hexagonal macropores from the AAO template, and also mesopores in the walls of the membrane The membranes are shown to be photocatalytically active in the photo-degradation of methylene blue (MB), a common dye used in the tex-tile industry that is found as a pollutant in waste water

In the LPD process, the reaction equilibrium in Eq.(1)can be controlled by the addition of boric acid that consumes fluoride ions

to form a more stable complex (Eq.(3))[16]

½TiF62þ nH2O () ½Tif6nðOHÞn2þ nHF ð1Þ

BO33 þ 4Fþ 6Hþ) BF4þ 3H2O ð2Þ

Yamanaka and co-workers [15] have proposed that alumina performs a similar role as boric acid via the formation of [AlF6]3 (Eq.(3))

Al2O3þ 3H2O þ 12F

() 2AlF36 þ 6OH ð3Þ

The fluoride ligand offers a slower and more controllable hydro-lysis because of the use of Fscavengers This method allows TiO2

films to be deposited over large areas, conformally on complex shapes and porous bodies and on temperature sensitive substrates such as organics due to the low deposition temperatures[17]

1387-1811/$ - see front matter Ó 2009 Elsevier Inc All rights reserved.

* Corresponding author.

E-mail address: g-goh@imre.a-star.edu.sg (G.K.L Goh).

Contents lists available atScienceDirect

Microporous and Mesoporous Materials

j o u r n a l h o m e p a g e : w w w e l s e v i e r c o m / l o c a t e / m i c r o m e s o

2 Experimental methods

The precursor solution was prepared by dissolving (NH4)2TiF6

(Aldrich) in distilled water to a concentration of 0.1 M As a

struc-ture director and a starting material, a commercial porous anodic

aluminum oxide disc (anodisc 47, WHATMAN Inc.) with straight

macroporous channels 250 nm in diameter was immersed in

25 ml of the precursor solution and maintained at 323 K for 1–

48 h After reaction, the samples were removed from the solution

and washed with distilled water The samples were dried at

323 K for 24 h before characterization

The morphology of the membranes were examined by

second-ary electron microscopy (JEOL JSM 6300F, Japan) while the crystal

phases were identified by powder X-ray diffraction (XRD,

Shima-dzu) using Cu Ka radiation (50 kV, 40 mA) and a scan speed of

0.5°/min in h–2h mode The divergence, anti-scatter and receiving

slits were fixed at 1o The crystallite size of AlFn(OH)3n (n = 1.5,

1.62 and 1.65) was calculated from the (1 1 1) peak based on

Scherrer’s formula Energy dispersive X-ray (EDX) spectroscopy

was used to confirm the composition of the membranes at

differ-ent locations The anatase nanocrystallites in the nanotubes were

studied by transmission electron microscopy (TEM) using a JEOL

JEM 2010 electron microscope operating at 200 kV after dispersing

the samples in ethanol using an ultrasonic bath for 30 min and col-lecting the fragments on holey carbon copper-supported grids Nitrogen adsorption and desorption isotherms were recorded at

77 K using an Accelerated Surface Area and Porosimetry system after the membranes were vacuum dried at 373 K for 12 h to re-move surface moisture to obtain the mesopore size distribution using the Barrett-Joyner-Halenda (BJH) method

The photocatalytic activity of the treated membranes was eval-uated by the photodegradation of a methylene blue (MB) dye An ultraviolet (UV) reaction chamber (RPR-100) was used to irradiate

a glass vessel (the glass having 98% transmittance in the UV range) such that the intensity at 365 nm in the center of the container was

213lW/cm2 Prior to catalytic testing, 6 ± 0.15 mg of the titania membranes were washed with distilled water and soaked in

50 ml distilled water for 24 h to equilibrate surficial hydroxyl groups After drying at 323 K for 24 h, the membranes were im-mersed in 15 ml MB solution (3.7  105M) in the glass reaction vessel The adsorption equilibrium of the dye by the membranes was established by keeping the container in the dark for 24 h until the concentration of MB stabilized The container was then ex-posed to UV irradiation and the degradation of MB was monitored

by a UV–Vis-NIR scanning spectrophotometer (Shimadzu UV-3101PC) at a wavelength of 664 nm, which is nearly centered at

a strong adsorption band of MB

3 Results and discussion

3.1 Structure

Recently, Shyue and co-workers[18]formed TiO2nanotubes by coating the walls of anodic aluminum oxide (AAO) templates by li-quid phase deposition, utilizing both the (NH4)2TiF6precursor and boric acid By lengthening the deposition period, nanowires were also obtained The diameters of both nanotubes and nanowires ob-tained by them corresponded quite well with the diameter of the nanoholes of the AAO template (250 nm), agreeing with their assertion that the TiO2first nucleated on the walls of the AAO tem-plate and then grew inwards, filling the nanoholes until nanowires were obtained In contrast, this work and an earlier work[15]

2θ (degree)

(a)

(b)

(c)

Fig 1 XRD of (a) untreated AAO membranes, (b) AAO membranes treated by liquid

phase deposition for 3 h with Ti/Al ratio of 1 (closed circles – AlFn(OH)3n; open

circles – anatase TiO2), and (c) AAO membranes treated by liquid phase deposition

for 3 h with Ti/Al ratio of 2.

Fig 2 Cross-section SEM micrograph of AAO membrane treated by liquid phase deposition for 1 h near reaction front (white arrows highlight some AlFn(OH)3n crystals

Table 1 Grain size as determined with Scherrer formula.

port that the AAO template actually dissolves according to Eq.(3),

providing scavengers for the fluoride ion This difference is because

the fluoride ion scavengers in the work by Shyue and co-workers

[18]are provided by the boric acid since it is already in solution and thus there is no significant dissolution of the AAO template X-ray diffraction (XRD) supports the formation of [AlF6]3 in solution, as proposed in Eq (3) that goes on to form a highly crys-talline hydrate, AlFn(OH)3n, on precipitation as shown inFig 1 Scanning electron microscopy (SEM) of cross sections revealed that these AlFn(OH)3nprecipitates (larger particles) are found at the re-gion where the AAO template has not completely dissolved, a tran-sition zone between the unreacted AAO and TiO2 nanotubes, as shown inFig 2 This transition zone is found to be further from the surface of the template, being 6lm from the surface after

1 h, 7.5lm after 3 h  10lm after 5 h and so on This latter obser-vation indicates that the AlFn(OH)3nphase only precipitates in re-gions of low relative Ti4+concentration and its formation can be significantly suppressed by increasing the Ti/Al ratio, as confirmed

by XRD inFig 1c

Application of the Scherrer formula to the XRD data concur with the SEM observations that parts of the AAO template has been con-verted to nanocrystalline anatase TiO2with average particle sizes (seeTable 1) around 4 nm and that the AlFn(OH)3nparticles are

Fig 3 TEM image of wall of AAO membrane treated by liquid phase deposition for

1 h revealing presence of nanocrystalline anatase.

larger, with sizes ranging from 25 to 35 nm, for growth periods of 1

to 12 h The nanocrystalline nature of the nanotubes is further

con-firmed by transmission electron microscopy (TEM), as shown in

Fig 3

Morphological observation of the ends of the AAO template after various periods of growth is shown inFig 4 After 1 h, the ar-ray of holes of the template are replaced by a close-packed arar-ray of tubes (Fig 4b) This is similar to the observation by Yamanaka and co-workers[15]who carried out the growth at 293 K, except that

in the present study (323 K), the TiO2tubes are crystalline, while

in Yamanaka’s case, the tubes only become crystalline after a post growth heat treatment at 773 K The replacement of the array of holes with an array of tubes is further evidence that the AAO tem-plate is dissolving and being replaced by TiO2

Further growth led to the apparent thickening of the tube walls with concurrent narrowing of the holes (Fig 4c) After 7 h growth, the holes begin to be completely filled (Fig 4d).Fig 5a shows that individual TiO2nanotubes can be obtained along the outer edges of the treated AAO templates In fact, when a 0.1 M HCl solution is used to dissolve the AAO component, what is left are individual TiO2tubes when the growth period is less than 5 h, coalesced tubes for 5 h growth and capped tubes for 7 h growth and above, the lat-ter two observations being similar to the morphology before acid treatment, as in Fig 5b and c The capped tubes indicates that the tube holes are not filled with TiO2, as initially assumed from

Fig 4d, but instead a TiO2 layer has formed to cap the ends of the coalesced tubes

This controllable decrease in the diameter of the ends of the tubes and eventual capping introduces a new parameter with which to tailor the design of TiO2 membranes For instance, the AAO/TiO2membrane could be used as a nano-selective reactor that would allow only reactants that have not been filtered out by the variable and controllable diameter of the tube ends and/or decom-posed by the photocatalytically active TiO2, providing both size and chemical selectivity, as schematically illustrated inFig 6 In addi-tion, when the tube ends are capped from 7 h onwards, only chem-ical species would be allowed through the mesoporous end caps

3.2 Porosity

Nitrogen physisorption experiments show that treating the AAO template in the (NH4)2TiF6solution for just 1 h caused the specific surface area to increase by 2 orders of magnitude, as detailed in Ta-ble 2 The adsorption–desorption isotherm inFig 7a reveals that this increase in specific surface area is due to the formation of mes-opores, as evidenced by the presence of a hysteresis loop This means that the TiO2nanotube walls are mesoporous The forma-tion of porous material is not surprising since the AAO material actually dissolves in the acidic fluoride solution according to Eq (3) and then TiO2precipitates back onto the undissolved regions

of the AAO membrane Since the TiO2membrane is formed by con-tinuous precipitation (see last paragraph) and not crystal growth,

Fig 5 Cross-section SEM micrographs of AAO membranes treated by liquid phase

deposition for (a) 1 h showing individual TiO2 nanotubes, (b) 5 h showing coalesced

CHEMICAL SELECTIVITY

Organic species are photocatalytically decomposed

SIZE SELECTIVITY

Varying diameter of entrance limits maximum size of reactant particle

this leads to the presence of mesopores At relative pressures more

than 0.9, the sharp increase in the adsorption volume is due to the

presence of macropores i.e the holes of the nanotubes

The pore size distribution calculated from the adsorption

branch by the BJH method is shown inFig 7b It shows that most

of the mesopores in the TiO2nanotube walls are 4 nm in

diame-ter The comparatively lower amount and more broadly distributed

pores centered around 30–40 nm are believed to be the cylindrical

pores located between the hexagonally packed nanotubes In fact, geometrical considerations based on hexagonal packing of cylin-ders shows that the diameter, d, of a pore located between the nanotubes can be related to the outer diameter, D, of the surround-ing tubes accordsurround-ing to[19],

Therefore, for a nanotube diameter, D, of 250 nm, the pore lo-cated between the surrounding tubes would have a diameter, d,

of 38.7 nm, in excellent agreement with the experimental observation

The specific surface area values presented inTable 2are for the treated AAO template i.e a combination of AAO and TiO2 To get an idea of the specific surface areas of the mesoporous TiO2structures formed after the various soaking times, the volume ratio of TiO2to AAO is determined from the position of the transition zone For example, the volume ratio of TiO2to AAO for a reaction time of

1 h would be 1:4 (the initial thickness of the untreated AAO mem-brane is 60lm while the length of the transition zone after 1 h,

6lm, is multiplied by 2 since the conversion of AAO to TiO2occurs from both ends of the template) Therefore, the volume, V, of TiO2

in 1 g of treated AAO template after 1 h (i.e 3  4V + 3.8  V = 1) is 0.063 cm3and so the mass of TiO2is 0.239 g Since we know that the specific surface area of AAO is 7.8 m2/g, it is easily determined that the specific surface area of the TiO2 only is ((137.8 m2/

g  1 g7.8m2/g  0.761 g)/0.239 g) 552 m2/g Accordingly, the specific surface areas of mesoporous TiO2after 3 and 5 h reactions are 400 and 179 m2/g These are extremely high specific surface areas and also show that the decreasing specific surface area trend from 1 h onwards is very significant

As more of the AAO template is transformed to mesoporous TiO2with increasing reaction time, one would expect the specific surface area to increase and not decrease as observed Although the nanotube hole diameter decreases, the inner surface of the tube/hole does not contribute a significant portion of the specific surface area (maximum 7.8 m2/g if the AAO was made of dense TiO2 instead) Generally, it is the mesoporosity of the tube/hole walls that contributes to the higher specific surface area Also, a re-examination ofFig 2shows that pores within the tube walls can only be accessed via the holes from 5 h onwards Before 5 h, the pores within the tube walls can be accessed from both the in-ner/larger holes and the smaller holes between the tubes If the pores within the tube walls are interconnected all the way to the surface of the walls, then whether the pores in the walls can be ac-cessed from the larger and smaller holes, as compared with access from just the larger holes from 5 h onwards, should not make any difference to the measured specific surface area But Ma and co-workers[11]have shown that film growth of TiO2by liquid phase deposition occurs by continuous nucleation and so new TiO2 parti-cles may precipitate in existing pores and passages (Fig 8), thereby blocking off pores below them from the wall surface If pores could

be accessed from both directions (inner and outer tube walls), then the blockage would not be a problem Therefore, it is likely that a combination of pore blockage from newly precipitated TiO2within the tube walls and access from only the inner tube walls contribute

to the drastic decrease in specific surface area

3.3 Photocatalysis

The photocatalytic activity of the membranes was evaluated by photocatalytic decolorisation of aqueous methylene blue solutions The photodegradation observed pseudo-first-order reaction kinet-ics, as shown inFig 9, in which the degradation rate constant, k, can be determined from the equation ln(Co/C) = kt, where Coand

C are the initial and final concentrations of the methylene blue dye after time t respectively The apparent rate constant of the

Table 2

Specific surface areas and calculated pore diameters for AAO membranes treated by

liquid phase deposition (LPD) for various times.

LPD time (h) Specific surface area (m 2

/g) Pore diameter (nm)

Fig 7 (a) Adsorption–desorption isotherms (adsorption – closed circles;

desorp-tion – open circles), and (b) BJH pore size distribudesorp-tion of AAO membranes treated by

liquid phase deposition for 1 h.

template is the same as that of the blank MB solution under UV

irradiation (Table 3), confirming that the template does not

con-tribute to photocatalytic activity, while all the anatase membranes

accelerated the degradation of MB AAO templates treated for 5 h

in the (NH4)2TiF6precursor solution displayed the highest

degrada-tion rate constant of 0.0146 min1

The major factors contributing to the overall effectiveness of the

photocatalytic activity are grain size, surface area and crystallinity

When titania is illuminated by UV radiation, an electron/hole pair is

created Both the electron and hole travel to the surface of the tita-nia particle where OH

radicals are created that are mainly respon-sible for the degradation of the methylene blue (MB) dye The grain size is important as smaller grain sizes translates to higher specific surface areas and so offer a greater number of surface sites at which adsorption and degradation of the MB dye can take place Decreas-ing grain size also reduces the rate of recombination of the electron/ hole pairs in the bulk of the grain (the volume charge-carrier recombination rate) But there is a critical grain size at which the in-crease in photocatalytic activity with decreasing grain size changes

to a decreasing trend as the rate of electron/hole recombination at the particle surface becomes more dominant (the surface charge-carrier recombination rate) with the ever escalating surface area

to volume ratio[10] In this study, grain size would not occupy a dominant role since the grain sizes do not vary significantly with treatment time, as shown inTable 1

Therefore, in order to explain the increase and then decrease of the degradation constant with treatment time, the surface areas and crystallinities of the membranes are now more closely exam-ined Previously in Section 3.2, the specific surface areas of the TiO2portion of the treated AAO templates were shown to decrease drastically from 552 to 179 m2/g for treatment times between 1 and 5 h But since the weight of the treated AAO template (a com-bination of AAO and TiO2) was kept constant for the photocatalytic tests, the amount of TiO2present in the MB test solutions would be different for the different treatment times since more AAO would have been converted to TiO2with longer treatment times The ac-tual surface areas of TiO2present are easily calculated to be 132,

118 and 68 m2for treatment times of 1, 3 and 5 h, respectively This is still a significant decrease in surface area over which the degradation of the MB dye can take place and thus the variation

in surface areas cannot explain the increase in photocatalytic activ-ity with increasing treatment time from 1 to 5 h

Ma and co-workers[13]pointed out that in liquid phase depo-sition of titania taking place in an oven set at 50oC, the growth solution takes approximately 90 min to reach the temperature of the oven and the maximum crystallinity attained even after 12 h

is only 75% In addition, the crystallinity of the precipitated mate-rial is still increasing from 70% even after 4 h This is because the measured crystallinity is an average of the maximum crystallinity attained after 90 min with the amorphous or very low crystallinity material precipitated before 90 min The presence of amorphous titania is detrimental to the photocatalytic efficiency as it contains

a variety of trap sites and recombination centers that decreases the concentration of electron/hole pairs generated from UV irradiation Therefore, it is most likely that the photocatalytic efficiency of membranes treated from 1 to 5 h increases despite the decrease

in surface area available because of the increase in titania crystal-linity After 5 h, the photocatalytic efficiency decreases as the surface area decreases, as noted inTable 2

4 Conclusions

Anatase membranes containing both macro- and mesopores were synthesized by liquid phase deposition of porous anodic alu-minum oxide (AAO) templates at 50oC Instead of boric acid, the alumina from the AAO template served as the F scavenger re-quired in the conversion of the (NH4)2TiF6precursor to TiO2 As such, the alumina dissolves and is replaced by precipitated TiO2

Fig 8 Illustration of how continuous nucleation of new material can block off

pores and reduce total internal surface area.

Fig 9 Photodegradation kinetics of AAO membranes treated by liquid phase

deposition for various times (open circle – 1 h; filled triangle – 3 h; open triangle –

5 h; filled square – 12 h; closed circle – blank reference).

Table 3

Degradation rate constants for AAO membranes treated by liquid phase deposition (LPD) for various times.

It was observed that TiO2initially forms a close-packed array of

nanotubes held together by residual AAO As the AAO template is

treated for longer periods, a TiO2 membrane is formed having

sub-250 nm holes This membrane is finally covered on both ends

by a layer of mesoporous TiO2 XRD confirms that crystalline TiO2

is present in the membranes for all treatment times while nitrogen

physisorption also revealed that the tube/membrane walls were

mesoporous with pore diameters centered around 4 nm BET

calcu-lations showed that these mesoporous structures had extremely

high specific surface areas starting from 552 m2/g after 1 h

treat-ment in the LPD solution and then decreasing to 179 m2/g after

5 h treatment

The photodegradation of aqueous solutions of methylene blue

observed pseudo-first-order reaction kinetics and the

photocata-lytic activity first increasing and then decreasing The highest

deg-radation constant, k, of 0.0146/min was observed for AAO templates

that had been treated for 5 h It is believed that despite the lower

surface area of the 5 h treated membrane compared to the 1 and

3 h membranes, the higher crystallinity of the 5 h membrane was

the reason for its maximum photocatalytic activity

Acknowledgment

The authors thank T.J White for useful discussions

References

[1] S.H Lee, M Kang, S.M Cho, G.Y Han, B.W Kim, K.J Yoon, C.H Chung, J Photochem Photobiol A 146 (2001) 121.

[2] M.S Hamdy, O Berg, J.C Jansen, T Maschmeyer, J.A Moulijn, G Mul, Chem Eur J 12 (2006) 620.

[3] M Wark, J Tschirch, O Bartels, D Bahnemann, J Rathousky´, Microporous Mesoporous Mater 84 (2005) 247.

[4] G Balasubramanian, D.D Dionysiou, M.T Suidan, I Baudin, J.M Laîné, Appl Catal B 47 (2004) 73.

[5] L.W Miller, M.I Tejedor-Tejedor, M.A Anderson, Environ Sci Technol 33 (1999) 2070.

[6] T Nakahira, T Inoue, K Iwasaki, H Tanigawa, Y Kouda, S Iwabuchi, K Kojima, Makromol Chem Rapid Commun 9 (1988) 13.

[7] C.C Chang, W.C Chen, Polym Sci., Part A: Polym Chem 39 (2001) 3419 [8] S.K Hazra, S Basu, Sens Actuators B 115 (2006) 403.

[9] Y Ohsaki, N Masaki, T Kitamura, Y Wada, T Okamoto, T Sekino, K Niihara, S Yanagida, Phys Chem Chem Phys 7 (2005) 4157.

[10] K.V Baiju, S Shukla, K.S Sandhya, J James, K.G.K Warrier, J Phys Chem C 111 (2007) 7612.

[11] B Ma, G.K.L Goh, J Ma, T.J White, J Electrochem Soc 154 (2007) D557 [12] B Ma, J Ma, G.K.L Goh, J Mater Sci 43 (2008) 4297.

[13] B Ma, G.K.L Goh, J Ma, J Electroceram 16 (2006) 441.

[14] S.M Klein, J.H Choi, D.J Pine, F.F Lange, J Mater Res 18 (2003) 1457 [15] S Yamanaka, T Hamaguchi, H Muta, K Kurosaki, M Uno, J Alloys Compd 373 (2004) 312.

[16] S Deki, Y Aoi, O Hiroi, A Kajinami, Chem Lett 25 (1996) 433.

[17] A Dutschke, C Diegelmann, P Lobmann, J Mater Chem 13 (2003) 1058 [18] J.-J Shyue, R.E Cochran, N.P Padture, J Mater Res 21 (2006) 2894 [19] G.K.L Goh, S.K Donthu, P.K Pallathadka, Chem Mater 16 (2004) 2857.