GT

DD

a

ARRAA

KTACDI

1

tdlaswiiada

tfio

0h

Electrochimica Acta 83 (2012) 420– 429

Contents lists available at SciVerse ScienceDirect

Electrochimica Acta

jou rn al h om epa ge: www.elsev ier .com/ locate /e lec tac ta

rowth mechanism and morphology control of double-layer and bamboo-typeiO2 nanotube arrays by anodic oxidation

ongsheng Guan, Paul J. Hymel, Ying Wang ∗

epartment of Mechanical Engineering, Louisiana State University, Baton Rouge, LA 70803, USA

r t i c l e i n f o

rticle history:eceived 28 May 2012eceived in revised form 7 August 2012ccepted 8 August 2012vailable online xxx

eywords:iO2 nanotubesnodic oxidationurrent transientouble-layer

a b s t r a c t

We have synthesized multilayer and bamboo-type TiO2 nanotube arrays via alternating-voltage anodiza-tion steps in hydrous ethylene glycol (EG) containing NH4F and investigated their growth mechanismsusing experimental and theoretical approaches. Current transients are recorded to study real-time mor-phological evolution of anodic TiO2 films during anodization at high and low voltages (Vhigh, Vlow). Currentchanges after each voltage ramp to Vhigh are observed along with sequential origination of pits, pores andtubes in a compact barrier layer at the base of oxide film. Two anodization steps at Vhigh separated byone step at Vlow with equal holding time yield double-layer smooth-walled TiO2 nanotubes. However,repetition of this sequence does not produce nanotubes of more layers, but makes lower-layer nanotubeslonger and induces ridges on their walls to form bamboo-type tubes. Formation mechanisms of double-layer TiO2 consisting of smooth-walled or bamboo-type nanotubes are explored. A proper holding time of

on diffusion low-voltage anodization is required for ridge formation, but ridge spacing is determined by high-voltageanodization time. The ridge spacing increases linearly with the high-voltage anodization time, and canbe theoretically calculated for lower-layer bamboo-type nanotubes formed in EG electrolytes with 5 vol%H2O. Less water (2 vol%) in electrolyte results in larger ridge spacing, while more water (10–15 vol%) notonly reduces the ridge spacing, but also causes instability during growth of multilayer TiO2 nanotubearrays and eventually leads to formation of a disordered porous TiO2 structure.

© 2012 Elsevier Ltd. All rights reserved.

. Introduction

Self-assembled TiO2 nanotube arrays formed by anodic oxida-ion of Ti were firstly reported in 2001, and soon after they haverawn tremendous attention due to their well-ordered structure,

arge surface area and broad applications [1]. In the past decade,nodic TiO2 nanotubes have been widely used in dye-sensitizedolar cells (DSSCs) [2,3], photo catalysis [4,5], gas sensing [6–8],ater splitting [9,10], biomedical materials [11,12], and lithium-

on batteries (LIBs) [13,14]. One advantage of these TiO2 nanotubess their easy fabrication via anodic oxidation. The length, diameternd wall features of tubes can be adjusted by tuning synthesis con-itions, such as anodization voltage or current (constant [15–17] orlternating [18,19]), electrolyte composition and anodization time.

More recently, there have been some research efforts to fur-her increase the surface area of anodic TiO2 nanotubes, such as

abrication of multilayer TiO2 nanotube arrays with extra porousnterlayers [20,21] or bamboo-type TiO2 nanotubes with ridges onuter tube walls by anodizing Ti under alternating-voltage (AV)

∗ Corresponding author. Tel.: +1 225 578 8577; fax: +1 225 578 5924.E-mail address: ywang@lsu.edu (Y. Wang).

013-4686/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.electacta.2012.08.036

conditions [18,19]. These new TiO2 nanostructures provide largersurface area due to extra interfaces or rough walls, while retainingthe vertically ordered one-dimensional nanostructure. In addition,they enable more light scattering when used in DSSCs, and providemore flexibility for dimension tuning and morphology engineering.For example, our group synthesized and explored growth mech-anism of double-layer and sandwich-structured smooth-walledTiO2 nanotube arrays via modified AV anodization conditions [22].Double-layer TiO2 nanotube arrays can also be prepared by usingstepping-voltage anodization [23], or by anodizing Ti in alternatingaqueous and organic electrolytes [24,25]. Regarding bamboo-typeTiO2 nanotube arrays, Zhang et al. [26] achieved ordered bamboo-type TiO2 nanotubes in electrolytes with different viscosity andwater content. Li et al. [27,28] employed AV anodization to synthe-size ordered TiO2 nanotubes with bamboo-shaped upper sectionand smooth-walled lower section. Schmuki and co-workers [29]integrated bamboo-type TiO2 nanotube arrays into DSSCs; the newDSSCs yielded a photo conversion efficiency 55% higher than that ofDSSCs based on smooth-walled nanotubes with identical film thick-

ness of 8 �m, due to larger surface area of bamboo-type nanotubesfor more dye loading and enhanced light scattering for more pho-ton absorption. In addition, Xie et al. [30] synthesized bamboo-typeTiO2 nanotubes with various ridge densities on tube walls, which

imica Acta 83 (2012) 420– 429 421

feauL

ni(atatmnebtotetwects

lea(onlTataTbtbactto

2

wmacENrpP3pos

t1 t3

t2

Applied Potent ial/V

60

10

Time/min

1V s-1

t5 t7 t9

t4 t6 t8

D. Guan et al. / Electroch

ound applications in DSSCs for up to 49% higher photo conversionfficiencies than smooth-walled nanotubes. Consequently, muchttention is focused on the bamboo-type TiO2 nanotube due to itsnique advantages and potential applications in photo catalysts,IBs and DSSCs [31].

To date, there are only a few papers about bamboo-type TiO2anotube arrays as summarized above. Syntheses in these reports

nvolve electrolytes containing hazardous HF, high voltage pairs120 V/40 V), and slow TiO2 nanotube growth rate (e.g., 12 nm/minnd 37 nm/min) [18,29]. Though formation process of bamboo-ype TiO2 nanotubes is briefly outlined [18], there is no reportbout fundamental explorations of growth mechanism underlyinghe synthesis and no clear understanding of factors that control

orphological features such as ridge spacing on bamboo-typeanotubes. Ridge spacing provides a simple way for measuringlectrochemical growth rate of anodic TiO2 nanotubes. It shoulde noted that the growth rate cannot be calculated directly fromhe final tube length divided by total anodic time, since the lengthf nanotubes is decided by both the growth process of their rootsowards substrate and the dissolution process of their tops intolectrolyte. However, ridge spacing on bamboo-type TiO2 nano-ubes is the intact outcome from the electrochemical growth lengthithin a given time period, and thus it can be used to calculate the

lectrochemical growth rate. On the other hand, the ridge spacingan be calculated from the growth rate of TiO2 nanotubes or elec-rochemical etching rate of Ti, which has been briefly mentioned inome early reports [32,33].

In the present work, we report facile synthesis of double-ayer and bamboo-type TiO2 nanotube arrays using non-toxiclectrolytes (ethylene glycol containing NH4F and H2O) and rel-tively lower voltage pairs (60/10 V) with faster tube growth rate60–70 nm/min). We synthesize and explore growth mechanismf double-layer TiO2 nanostructures composed of smooth-walledanotubes in the upper layer and bamboo-type nanotubes in the

ower layer. For comparison purposes, single-layer smooth-wallediO2 nanotubes and double-layer smooth-walled TiO2 nanotubesre fabricated as well. Current measurements are used to studyhe real-time growth process and morphological evolution of thesenodic TiO2 nanostructures. Formation mechanism of double-layeriO2 nanotubes with or without a bamboo-type layer is proposedased on ion diffusion-controlled process inside tubes and inter-ube cavities. Fundamental factors that affect the morphology ofamboo-type TiO2 nanotubes are studied with both experimentalpproaches and theoretical calculations and by manipulating waterontent in electrolyte and adjusting high or low voltage anodizationime. To the best of our knowledge, so far there is no report abouthe effect of water content and low-voltage anodization durationn ridge formation.

. Experimental

Ti foils (99.5 wt%) (10 mm × 10 mm × 0.25 mm) in this studyere purchased from Alfa Aesar. Prior to any electrochemical treat-ent, Ti foils were degreased and rinsed by sonicating in ethanol

nd deionized water. A two-electrode cell with a Pt mesh as theounter electrode was assembled for electrochemical anodization.lectrolytes were anhydrous ethylene glycol (EG) with 0.3 wt%H4F and 2–15 vol% H2O. All the solutions were prepared from

eagent grade chemicals and deionized water. The voltage was sup-lied by a DC power supply with digital display (Model 1623A,K Precision). A Data Acquisition/Data Logger Switch Unit (Aglient

4970A) was employed to record real-time anodic current. To pre-are multilayer TiO2 nanotubes, the anodization process consistsf several alternating high and low-voltage anodization steps, ashown in Fig. 1. The voltage is first increased from zero to Vhigh with

Fig. 1. Anodization sequence for the formation of double-layer and bamboo-typeTiO2 nanotube arrays.

a ramp rate of 1 V s−1 and remains for a time t1, then drops to Vlowand is kept at Vlow for a time t2, followed by increasing to Vhigh witha rate of 1 V s−1 and being kept at Vhigh for a time t3. Such voltagealteration is repeated for different TiO2 nanotube structures. Forcomparison purpose, single-layer TiO2 nanotubes were preparedby anodizing Ti sheets at 60 V for 20 min. All the experiments werecarried out at room temperature. After anodization was completed,the samples were immediately rinsed in deionized water and driedin air. A FEI Quanta 3D FEG scanning electron microscope (SEM)was used to characterize the morphology of TiO2 nanotube arraysformed on the front side of Ti foils.

3. Results and discussion

3.1. Morphology evolution of TiO2 nanotube arrays

Table 1 summarizes various anodization conditions and mor-phological features of resultant TiO2 nanotube arrays. It can beseen that the TiO2 tubular structure evolves from single layer todouble-layer structure of smooth nanotubes, and further to double-layer structure composed of smooth tubes above and bamboo-typetubes below, under constant voltage (CV) anodization and differentalternating high and low-voltage anodization conditions. Clearly,number of anodizing steps and their holding time affects the quan-tity and spacing of bamboo ridges. The formation mechanism ofthese nanostructures and factors affecting their morphologicalfeatures are discussed with both SEM observations and electro-chemical analyses in the following sections.

Fig. 2 shows SEM images of TiO2 nanotube arrays synthesizedunder CV or AV conditions in EG electrolytes containing 0.3 wt%NH4F and 5 vol% H2O. The direct anodization at 60 V leads to for-mation of single-layer TiO2 nanotube arrays (CV-NT, Fig. 2a). Thetubular structure is composed of hollow nanotubes (Fig. S1a) with ahemispherical closed bottom. These tubes have smooth walls withan average length of ∼3.10 �m and their outer diameter is slightlyincreased from the top (120 nm) to bottom (152 nm).

Our previous results demonstrate that the 1:1 ratio of high-voltage anodization time (t1) to low-voltage anodization time (t2)is a critical parameter for synthesis of double-layer TiO2 nan-otube arrays by using three high/low-voltage anodization steps inEG electrolytes containing 0.3 wt% NH4F and 5 vol% H2O at roomtemperature [22]. This in-situ preparation of double-layer TiO2nanotubes is much more convenient than the two-step anodizationmethod using alternating electrolytes with different compositionsreported in literature [24,25]. Fig. 2b shows AV-NT1 composed ofsmooth-walled nanotubes in both the upper layer and lower layer.

The upper layer reaches a thickness of ∼1.6 �m where the nano-tubes have an outer diameter of ∼152 nm, and the lower layer is1.7 �m thick and the outer tube diameter becomes a little larger(∼172 nm). SEM image in Fig. S1b shows these upper-layer tubes

422 D. Guan et al. / Electrochimica Acta 83 (2012) 420– 429

Table 1Single-layer and double-layer TiO2 nanotubes synthesized in EG electrolytes containing 0.3 wt% NH4F and 5 vol% H2O under different voltage conditions.

Applied potential (V) 60 10 60 10 60 10 60 10 60 Layer structure of films Number of ridges Ridge spacingTime (min) t1 t3 t5 t7 t9

t2 t4 t6 t8

CV-NT 20 SAV-NT1 10 10 10 S/S 0AV-NT2 20/3 20/3 20/3 20/3 20/3 S/B 1 1050AV-NT3 5 5 5 5 5 5 5 S/B 2 800AV-NT4 4 4 4 4 4 4 4 4 4 S/B 3 645AV-NT5 10 10 10 10 10 10 10 10 10 S/B 1 1600

S: single-layer smooth nanotubes; S/S: double-layer structure composed of smooth nanotubes in both layers; S/B: double-layer structure composed of smooth nanotubes inthe upper layer and bamboo-type nanotubes in the lower layer.

Fig. 2. Cross-sectional SEM images of different TiO2 nanotube arrays synthesized in EG electrolytes containing 0.3 wt% NH4F and 5 vol% H2O: (a) CV-NT, (b) AV-NT1, (c)AV-NT2, (d) AV-NT3, (e) AV-NT4 and (f) AV-NT5. (Insets) The whole single-layer or double-layer structure of TiO2 nanotubes.

imica

wos

mA2ncb(iLuitntrtleiitdesfinf

a1nulitIbtfihagisbrt

3

ds[gtwptpo

D. Guan et al. / Electroch

ith round open entrances. In addition, it is noted that Ti is fullyxidized during formation of either single-layer or double-layertructures by anodization, yielding TiO2 nanotubes [22].

The high/low-voltage anodization steps are then repeated forore times to explore morphological evolution of TiO2 nanotubes.nodic oxidation of Ti with three anodization steps at 60 V for0/3 min separated by two steps at 10 V for 20/3 min yields aew type of double-layer nanotube arrays (AV-NT2) consisting oflosed-bottom smooth-walled nanotubes in the upper layer andamboo-type nanotubes (one ridge on each tube) in the lower layerFig. 2c). This double-layer structure is ∼3.0 �m thick and the spac-ng between the bamboo ridge and tube base is about 1050 nm.ikewise, AV-NT3 is composed of smooth-walled nanotubes in thepper layer and bamboo-type nanotubes (two ridges on each tube)

n the lower layer (Fig. 2d), with a total thickness of ∼2.9 �m; andhe spacing between the two neighboring ridges on the lower-layeranotubes is ∼800 nm. AV-NT4 consists of smooth-walled nano-ubes in the upper layer and bamboo-type nanotubes with threeidges on each tube in the lower layer (Fig. 2e). In this structure,he top layer is composed of small closed-bottom nanotubes with aength of ∼0.5 �m and an outer diameter of ∼94 nm, which cov-rs a layer of bamboo-type nanotubes that are larger (∼160 nmn outer diameter) and much longer (∼2.6 �m). The ridge spac-ng (or the length of tube sections) is ∼645 nm, a little longer thanhe length of top-layer nanotubes, since the top layer suffers fromissolution by electrolyte during the anodization process. Openntrances can be also observed in top-layer TiO2 nanotubes, ashown in its top-view SEM image (Fig. S1c). It is observed that theseve different nanotube arrays in Fig. 2a–e have similar total thick-esses (2.9–3.3 �m), since the total high-anodization voltage time

or growing these nanotubes is the same (20 min).Furthermore, if we increase both high-voltage and low-voltage

nodization time (e.g. from 4 to 10 min) while keeping t1/t2 as in the multiple-step anodization sequence, double-layer TiO2anotube arrays are still resulted. For instance, in AV-NT5 thepper-layer nanotubes are smooth and closed-bottom with a

ength of ∼1.3 �m and the lower-layer bamboo-type nanotubess ∼5.7 �m long (Fig. 2f). Open entrances are also observed inhe top-view SEM image of upper-layer nanotubes (Fig. S1d).nterestingly, some lower-layer nanotubes only have one bam-oo ridge on their outer walls. The spacing from this single ridgeo tube top (or the length of 1st section) is ∼1.6 �m, almost oneourth of the total length of lower-layer nanotubes, and thus its assumed that two ridges disappear in subsequent alteration ofigh/low-voltage anodization steps. Such morphological discrep-ncies suggest changes in the local anodization conditions for tuberowth beneath a thicker upper layer and a longer tube section. Its ensured that the closely-packed long tubes hinder inward diffu-ion of ionic species (e.g. F− ions and H2O) from reaching the tubease, which creates a different electrolyte environment there, withespect to anodization for AV-NT4 that produces shorter upper-ubes and tube sections.

.2. Anodic current versus time transients

It is well known that current–time transients during anodic oxi-ation of Ti under CV or AV conditions reveal details about differenttages of the film growth process and corresponding morphologies34–37]. Fig. 3a and b shows current–time transients for CV-NTrown in agitated electrolyte. When the applied voltage goes upo 60 V at the beginning, the initial anodic current is increased asell and reaches to its maximum. In this process, water is decom-

osed fast at anode to produce O2− ions and Ti is oxidized quicklyo form a compact oxide film. The compact film hinders ion trans-ort and eventually causes an abrupt current drop. The formationf an initial compact film is commonly regarded as the stage I of

Acta 83 (2012) 420– 429 423

TiO2 nanotube growth, as noted in the current plot in Fig. 3b. Later,some species, especially F− ions, aggregate in regions with highsurface energy (e.g. micro cracks) and selectively dissolve oxidesthere to originate tiny pits and pores in the compact film [38].The development of pits and pores converts the film into a porousstructure and attenuates the film thickness, which facilitates iontransport through this film to slow down the current drop (or usu-ally to increase the anodic current slightly during anodization of Tiin static electrolytes) [36,37]. Formation of the porous film is thegrowth stage II of TiO2 nanotube arrays. As the pores grow deeper,the inter-pore sites become high-surface-energy regions attractingF− ions for faster dissolution of oxides there [38,39]. As a result,cavities emerge between the pores, and turn the porous film intoa tubular structure. An array of parallel nanotubes appears with athin compact barrier layer underneath, and their primary develop-ment in depth continues to attenuate the oxide film and retard thecurrent drop, but later their steady-state growth yields relativelystable current as shown in Fig. 3b. Formation of the tubular film isthe growth stage III of anodic TiO2 nanotubes. Hence, it can be con-cluded that the current transient measurement provides insight forexploring the morphological evolution of anodic oxide film grownon Ti substrate under constant voltage steps.

Fig. 3c and d displays the current–time plots recorded duringgrowth of AV-NT1. Similarly, a current peak appears after the ini-tial voltage ramp to 60 V, due to the formation of a compact oxidefilm (stage I), and then the current drops at a reducing rate due tothe origination of pores or tubes in the oxide film (stage II and III).Within the tube growth, the voltage is quickly altered to 10 V at∼660 s, yielding an instant current drop by ∼5.8 times. However,the current quickly recovers a little and then drops (inset of Fig. 3d),suggesting that the oxide film becomes a little thinner first and thenthicker. The temporarily thinning oxide film is due to weaker oxi-dation of Ti and slower movement of barrier layer at 10 V, whileH+ and F− ion profiles achieved at 60 V continue to dissolve oxidesof barrier layer at a fast speed. After a short while, these speciesare mostly consumed and their profiles are finally adjusted accord-ing to anodization conditions at 10 V. The barrier layer grows to bethicker, yielding a small current drop, but the subsequent pit or poreformation in this layer yields an increased current. If the anodiza-tion step at 10 V is performed for a sufficient time (e.g. ≥30 min),small tubes can be formed under the barrier layer, but the hold-ing time of 10 min here only allows origination of tiny pores, asdescribed in our earlier work [22]. When the voltage is tuned to be60 V again, the current increase/decrease/increase sequence vividlyindicates the three growth stages of a new layer of nanotubes underthe first one [34,35]. Surprisingly, this current variation sequencerepresents typical current–time behaviors of anodizing Ti in statichydrous electrolytes [35,37], indicating that ion diffusion processdominates the mass transfer inside the nanotube layer, rather thanthe convection process in stirred bulk electrolyte [22]. The final cur-rent plateau tells a steady growth state of the new-layer nanotubesat 60 V, and its height is lower than the small current peak beforeit.

Fig. 3e and f presents the current–time plots recorded duringgrowth of AV-NT4. The first two high-voltage anodization stepsproduce similar current changes as in Fig. 3d, suggesting formationof double-layer TiO2 nanotube arrays. However, the subsequentthree anodization steps at 60 V yield different current–time tran-sients featured by their current plateaus higher than the peakbefore it. For example, the current starts to rise at ∼1108 s andreaches a peak, as the voltage goes back to 60 V, but later dropsa little due to thickened barrier layer under the lower-layer nano-

tubes. Then the pit or pore origination attenuates the barrier layer,which could only cause the current to increase until reaching aplateau the same as the previous one. However, the current keepsincreasing to form a plateau higher than the peak right before it.

424 D. Guan et al. / Electrochimica Acta 83 (2012) 420– 429

120010008006004002000-5

0

5

10

15

20

25

30

35

40

45

I/m

A

t/s

b

Stage I

Stage III

Stage II

200017501500125010007505002500-5

0

5

10

15

20

25

30

35

40

45

I/m

A

t/s

d

I II III

I II III

2700240021001800150012009006003000-5

0

5

10

15

20

25

30

35

40

45

I/m

A

t/s

f

6000500040003000200010000-5

0

5

10

15

20

25

30

35

40

45

I/m

A

t/s

h

c

10 10

10

60

10

Time/min

1V s-1

Applied Potential/V

a

20 60

Time/min

1V s-1

Applied Potential/V

e

4 4

4

60

10

Time/min

1V s-1

4 4 4

4 4 4

Applied Potential/V

g

10 10

10

60

10

Time/min

1V s-1

1010 10

10 10 10

Applied Potential/V

Fig. 3. Anodization sequences and current–time curves recorded during anodic growth of different TiO2 nanotube arrays in stirred EG electrolytes containing 0.3 wt% NH4Fand 5 vol% H2O: (a and b) CV-NT, (c and d) AV-NT1, (e and f) AV-NT4, (g and h) AV-NT5.

imica

Iibtwistiotdrltener

icsdbvtrbcmtfros

3n

scaaodapvdmct

ppnalwtboto

D. Guan et al. / Electroch

t is assumed that growth of bigger pores in the barrier layernvolves after the voltage returns to 60 V, and further attenuates thearrier layer. In the presence of formed tube section above, theseubes and inter-tube cavities are most likely to be the bigger pores,hich are much larger than nascent pits originated at 60 V. Besides,

dentical current–time plots appear in the final two high-voltageteps as well, suggesting that the same morphological features forhe tubes emerge in the three anodization steps. At this point, its understood that there is a competitive relation between growthf former tube sections and nascent pits along the vertical direc-ion to the Ti substrate. Since ion diffusion to the bottom is slowedown by the closely arrayed nanotube layer and a compact thin bar-ier layer above, pit development is difficult inside the base barrierayer, which is incomparable to the re-started growth of formedubes and cavities. Therefore, the newly-born pits are consumed oraten, which greatly reduces the probability of developing a newanotube layer from these pits. Moreover, the temporarily thick-ned barrier layer leaves some debris on tube walls to generateidges, yielding bamboo-type tubes [22].

Fig. 3g and h exhibits current–time transient recorded dur-ng growth of longer nanotubes AV-NT5. It can be seen that thisurrent–time plot is much like that shown in Fig. 3f, suggestingome similarities in the growth of double-layer TiO2 nanotubesuring multi-voltage alterations. However, one distinct differenceetween the two plots is the current changes in the final twooltage switches from 10 V to 60 V. The current increases afterhe voltage is back to 60 V, then drops slightly, and soon afterises to a current plateau higher than the current peak rightefore it. The reduced current drop indicates that the barrier layerhanges very little in thickness at the voltage switch, thanks touch fewer ions reaching the base barrier layer through fur-

her thicker tube layer and tube sections above. Consequently, pitormation tends to be slowed down or even suppressed in bar-ier layer, which is easier to be replaced by re-started growthf former tubes and inter-tube cavities, yielding smooth tubeections.

.3. Growth mechanism of double-layer and bamboo-type TiO2anotubes

Our previous work reports in detail that the growth ofingle-layer TiO2 nanotubes is highly dependent on stable ion con-entration profiles (e.g. pH gradient) established inside nanotubesnd cavities under constant-voltage anodizing conditions (Fig. 4a),nd efficient variations of the concentration profiles on the switchf high and low anodic voltages will induce their evolution to aouble-layer structure with smooth tubes in both the upper layernd lower layer (Fig. 4b and c) [22]. It should be noted that tinyits and pores emerge and grow in the barrier layer when the low-oltage anodization proceeds, as shown in Fig. 4b. Here we furtheriscuss the formation of ridges on the lower-layer nanotubes whenore alternating-voltage steps are applied, by still taking H+ ion

oncentration (or pH gradient) profile inside tubes as an exampleo simplify our model.

In the second high-voltage anodization process, water decom-osition at the tube bottom yields H+ ions and a stable pH gradientrofile is established to allow the steady growth of lower-layeranotubes (Fig. 4d). If the voltage is switched to a lower value (10 V)gain, fewer H+ ions are produced and the pH gradient profile inower-layer nanotubes becomes less steep (Fig. 4e). After a short

hile, tiny pits and pores are originated in the barrier layer underhe low voltage, but their growth is relatively slow since the closed-

ottom upper-layer tubes with a barrier layer below hinder ionsutside (e.g. F− ions) from diffusion to the bottom of lower-layerubes, and ions inside (e.g. H+ ions) from diffusion to the entrancef upper-layer tubes. Hence, within the holding time of low voltage

Acta 83 (2012) 420– 429 425

that is equal to time of previous high voltage step, the pH gradientprofile in the lower layer fails to have sufficient variation for initia-tion of a third tube formation at the subsequent high-voltage step.A subsequent step back to a high voltage (60 V) would thicken thebarrier layer near the substrate, which then is attenuated by growthof existing pits and pores. However, the restarted growth of formertubes and inter-tube cavities is faster than them, and thus they areconsumed. This helps to attenuate the barrier layer and facilitateion transport through it, and further increase the anodic currentas shown in current–time plots in Fig. 3f. In particular, the concavecavity deepens relatively fast towards the convex triangular pillar ofmetal among three neighboring tube bases, leaving some residuesof thickened barrier layer on walls of two neighboring tubes to formridges, yielding bamboo-type features (Fig. 2e). Thereby, the firsttube section of bamboo-type nanotubes is generated in the lowerlayer. More ridges are formed in the same way during the followinghigh/low-voltage anodization steps, since the formed upper-layertubes and ridge layers retard ion diffusion through the whole tubu-lar layer. Hence, the anodic current transients for the final threevoltage switches from 10 V to 60 V are the same, as shown in Fig. 3f.

So far it has been known that ion diffusion inside nanotubes andcavities is very significant not only for initiation of new nanotubelayer below, but also for formation of bamboo ridges on the lower-layer tubes. In some cases, high-voltage anodization proceeds fora long period, resulting in long upper-layer tubes and long tubesections with large spacing between ridges shown in Fig. 2f. Thesenanotubes are closely packed, and their gaps become narrow whennear to the bottom, which makes ion diffusion much more diffi-cult. As the voltage is altered back to a high value, thickness of thebarrier layer changes very little in which tiny pits possibly emergelater. However, these pits grow so slowly that the re-grown tubesand cavities consume them immediately. As a result, formationof other bamboo ridges after the first one is suppressed, yield-ing long smooth tube walls with fewer ridges. At this point, it isassumed that the smooth tube section is easier to be produced inorganic solvent-based electrolytes with less water and slower iondiffusion.

3.4. Role of low-voltage anodization

According to the growth model above, a sufficiently longlow-voltage anodization step is critical for growth of double-layer nanotubes, and even for origination of an extra layerof small nanotubes between the upper layer and lower layer[22]. Fig. 5b shows SEM image of double-layer nanotubesformed by four anodization steps at 60 V for 4 min separatedby three steps at 10 V for 4 min, 10 min and 20 min, respec-tively (Fig. 5a). The resulted structure is also composed of shortclosed-bottom nanotubes in the upper layer and bamboo-typenanotubes with two ridges on each in the lower-layer. Thespacing between bamboo ridges is ∼640 nm, which is closeto the ridge spacing in AV-NT4, indicating that different low-voltage anodization time has no effect on the ridge spacing.Moreover, it is noted earlier that the total thicknesses of fivedifferent TiO2 nanotube arrays in Fig. 2a–e are quite close(2.9–3.3 �m) due to the same total time (20 min) of high-voltageanodization steps, though their total low-voltage anodizationtime ranges from 0 min to 16 min. These results indicate thatthe ridge spacing and nanotube length are dependent on high-voltage anodization time, rather than low-voltage anodization.This can also be understood from the current–time plots inFig. 3f and h showing much smaller anodic current at low-

voltage steps than at high-voltage steps. Theoretically, withinthe same time period, much larger quantity of electric chargeis transferred at a high-voltage step than at a low-voltage step,and Ti is mostly oxidized to form oxides during high-voltage

426 D. Guan et al. / Electrochimica Acta 83 (2012) 420– 429

Fig. 4. Schematic showing growth of double-layer structure containing smooth-walled TiO2 nanotubes or bamboo-type TiO2 nanotubes via anodic oxidation: (a) pH gradientprofile in nanotubes during their steady growth at a high voltage, (b) much less steep pH gradient and tube formation when a low voltage is kept for a sufficiently long period(t1/t2 ≤ 1), (c) formation of double-layer smooth TiO2 nanotubes, (d) pH gradient profile in double-layer nanotubes during their steady growth at a high voltage, (e) less steeppH gradient in the lower-layer nanotubes and pit or pore formation in the barrier layer after the voltage is altered to a low value, (f) formation of double-layer structure withl

al

intb(hntmt

ong bamboo-type nanotubes in the lower layer.

nodization to achieve nanotube growth, according to Faraday’saw [40].

On the other hand, a proper low-voltage anodization step isndispensable for ridge formation on the lower-layer bamboo-typeanotubes. Fig. 5d shows SEM image of double-layer TiO2 nano-ubes formed by four anodization steps at 60 V for 4 min separatedy three steps at 10 V for 4 min, 2 min and 1 min, respectivelyFig. 5c). The first low-voltage anodization step (10 V for 4 min)elps initiation of a new tubular layer under short upper-layer

anotubes. The second low-voltage step (2 min) facilitates forma-ion of a ridge on each nanotube in the lower layer. However, no

ore ridges are observed in the SEM image in Fig. 5d, indicatinghat the third low-voltage step (1 min) is too short to induce ridge

formation. This can be explained by the ion-diffusion-controlledin-situ growth mode of double-layer and bamboo-type TiO2 nano-tubes above. Inside the lower-layer tubes, ion diffusion is so slowthat the ion concentration profile established at previous high-voltage anodization (60 V for 4 min) can change little within tooshort time (1 min), and thus pit formation cannot be induced here.Once the high voltage is applied again, only growth of tubes andinter-cavities is re-started, yielding smooth tube sections. Hence, aproper ratio of high/low-voltage anodization time is critical for wall

features of lower-layer nanotubes in double-layer structures. Insummary, low-voltage anodization time needs to be long enough toinduce ridge formation on bamboo-type nanotubes, but overly longlow-voltage anodization time has no effect on the ridge spacing.

D. Guan et al. / Electrochimica Acta 83 (2012) 420– 429 427

Fig. 5. Anodization sequences and cross-sectional SEM images of TiO2 nanotubes formed in EG electrolytes containing 0.3 wt% NH4F and 5 vol% H2O under different conditions:(a and b) four anodization steps at 60 V for 4 min separated by three steps at 10 V for 4 min, 10 min and 20 min; (c and d) four anodization steps at 60 V for 4 min separatedb

3o

tewtsfFbacbi1∼Ftmoos

vhtf

y three steps at 10 V for 4 min, 2 min and 1 min, respectively.

.5. Effects of high-voltage anodization time and water contentn ridge spacing

It has been reported in literature that bamboo-type TiO2 nano-ubes with dense ridges have larger surface area and thus shownhanced DSSC efficiency as photo-anodes compared to smooth-alled TiO2 nanotubes [29,30]. Our study above has demonstrated

hat the ridge spacing is dependent on high-voltage anodizationteps. Therefore, we can manipulate and optimize morphologicaleatures of bamboo-type TiO2 nanotubes for device applications.ig. 6b shows SEM image of double-layer TiO2 nanotubes formedy five anodization steps at 60 V for 2 min separated by four stepst 10 V for 2 min (Fig. 6a). This structure contains one layer of shortlosed-bottom nanotubes (∼185 nm long) covering a layer of longeramboo-type nanotubes with three ridges on each. The ridge spac-

ng is ∼330 nm. When the anodization time is further reduced to min (Fig. 6c), the ridges become much denser with a spacing of165 nm (Fig. 6d), while the total thickness of the nanotube array inig. 6b (1.43 �m) remains similar to that in Fig. 6d (1.48 �m) due tohe same total time of high-voltage anodization steps (10 min). As

ore high/low-voltage anodization steps are applied, dissolutionf upper-layer nanotubes becomes more severe, yielding a layerf directional nanowires on the top of bamboo-type nanotubes ashown in Fig. 6d.

Fig. 7 summarizes the correlation between ridge spacing of

arious nanotubes observed in SEM images presented earlier andigh-voltage anodization time and water content used for their syn-heses. It can be clearly seen that the ridge spacing increases linearlyrom 165 nm to 1600 nm, with the high-voltage anodization time

from 1 min to 10 min. The theoretical ridge spacing can be derivedfrom nanotube length (L) grown within a given time as follows:

L = QM

(Fnı)(1)

where Q is the circulated charge (C cm−2), M is molecule weight ofTiO2 (79.9 g mol−1), F is the Faraday constant (96,500 C equiv−1), nis the number of electrons involved in the reaction and ı is densityof TiO2 (3.8–4.1 g cm−3) [41]. The circulated charge is calculatedby integrating the corresponding current–time plot. For example,the current–time plot corresponding to the fourth high-voltageanodization (1650–1950 s in Fig. 3f) gives a total circulated chargeof 2.160 C cm−2, and leads to a theoretical oxide growth length of∼1120 nm. However, the oxide film appears on both the front sideand back side of Ti substrate, and simultaneous growth of nano-tubes at the two sides is lopsided: tube length is longer at thefront side than at the back side [42]. For anodization in EG elec-trolyte with 0.3 wt% NH4F and 2 vol% H2O at 60 V, the tube lengthis ∼29% bigger at the front side than at the back side [42] andthus we estimate the theoretical ridge spacing of our sample tobe ∼650 nm, very close to the actual spacing of 645 nm betweenridges on AV-NT4 (Fig. 2e). Correspondingly, the theoretical elec-trochemical growth rate of these nanotubes is ∼162 nm per minute.Furthermore, it can be seen from Fig. 3f and h that the partial cir-

culated charge mostly depends on width of current plateaus ofhigh-voltage anodization steps. Since the plateaus are parallel tothe time axis, the linear relationship between ridge spacing andhigh-voltage anodization time can be anticipated.

428 D. Guan et al. / Electrochimica Acta 83 (2012) 420– 429

Fig. 6. Anodization sequences and cross-sectional SEM images of double-layer TiO2 nanotube arrays synthesized in EG electrolytes containing 0.3 wt% NH4F and 5 vol% H2Ou y fourw parat(

bf52sw

FsT

nder different conditions: (a) five anodization steps at 60 V for 2 min separated bhole double-layer nanotube array); (c) ten anodization steps at 60 V for 1 min se

inset: the whole nanotube array).

In addition, water content also affects the ridge spacing inamboo-type nanotubes. For example, the ridge spacing is 645 nmor double-layer TiO2 nanotubes grown in EG electrolyte with

vol% H2O, and 1100 nm for those formed in EG electrolyte with vol% H2O (Fig. S2), obtained via the same anodization sequencehown in Fig. 3e. The discrepancy can be explained as follows: lessater in electrolytes yields thinner barrier layers [43] at the Ti

109876543210

200

400

600

800

1000

1200

1400

1600

1800

11001050

460165

1600

800

645

330

Sp

acin

g b

etw

ee

n r

idg

es/n

m

Time/min

2 vol% H2O

5 vol% H2O

10 vol% H2O

ig. 7. Relationship between ridge spacing and time for high-voltage anodizationtep used for synthesis of double-layer structures with lower-layer bamboo-typeiO2 nanotubes in EG electrolytes containing 0.3 wt% NH4F and 2–10 vol% H2O.

steps at 10 V for 2 min, (b) bamboo-type nanotubes in the lower layer (inset: theed by nine steps at 10 V for 1 min, (d) bamboo-type nanotubes in the lower layer

substrate because the donation of oxygen becomes more difficultand thus it is harder to form oxides [44]. Ionic transport across thebarrier layer is enhanced, which thereby accelerates the motion ofthis layer toward substrate, resulting in faster growth of nanotubes.On the other hand, a thicker barrier layer is formed to hinder ionictransport and thus reduce the growth rate of nanotubes when watercontent is increased. As a result, the ridge spacing becomes shorter,as evidenced by 645–460 nm for bamboo-type nanotubes formedby the same anodization sequences in EG electrolytes with increas-ing water content from 5 vol% to 10 vol% (Figure S3). We also foundthat a further increase of water content in EG electrolyte to 15 vol%under the same anodization sequence yields disordered porous orspongeous layers, due to growth instability induced by excessivewater for fierce anodic reactions, which is in agreement with ourprevious findings [22].

4. Conclusions

We have synthesized double-layer smooth-walled TiO2 nano-tubes and double-layer TiO2 nanotubes with bamboo-typenanotubes in the lower layer via alternating-voltage anodizationof Ti in NH4F-containing ethylene glycol electrolytes. Anodic cur-rent is recorded during growth of various TiO2 nanotube arrays, inorder to evaluate their morphological evolutions at different stages.With voltages applied, the increase/decrease/increase sequences

of current represent formation of oxide barrier layer, pores andnanotubes, respectively. A plateau of current appears upon sta-ble growth of TiO2 nanotubes. Based on these findings, growthmechanisms of double-layer structures containing smooth-walled

imica

onthltstrtitfwcble

fetdisfrod

A

(FmirC

A

i2

R

[

[

[[

[

[[

[

[

[

[

[[[[

[[

[

[[

[

[[[

[

[[

[

[

[

[[

D. Guan et al. / Electroch

r bamboo-type nanotubes are proposed and ion diffusion inanotubes is considered to be a dominant factor. A sufficient varia-ion of ion concentration profiles in upper-layer nanotubes duringigh/lower-voltage alteration can induce formation of a new tube

ayer underneath. The morphology of lower-layer nanotubes can beuned by repeating the high-voltage and low-voltage anodizationteps, and thus bamboo-type nanotubes are obtained. Sufficientime of low-voltage anodization step is required for formation ofidges, but the spacing between neighboring ridges relies on theime of high-voltage anodization step. In particular, the ridge spac-ng is found to be increased linearly from 165 nm to 1600 nm, withhe time of high-voltage anodization step from 1 min to 10 min,or lower-layer bamboo-type nanotubes formed in EG electrolytesith 0.3 wt% NH4F and 5 vol% H2O. In addition, decreasing water

ontent in electrolytes results in larger ridge spacing on lower-layeramboo-type nanotubes, owning to origination of a thin barrier

ayer, while higher water content causes smaller ridge spacing andven instability for growth of double-layer nanotubes.

Double-layer and bamboo-type TiO2 nanotubes have larger sur-ace area than single-layer nanotubes with the same length, due toxtra layers or interfaces present between two neighboring nano-ubes. Their surface area can be further enlarged by producingenser ridges on longer bamboo-type tubes, simply by repeat-

ng high-voltage and low-voltage anodization steps. Controllableynthesis of double-layer and bamboo-type TiO2 nanotubes andundamental understanding of their growth characteristics in thiseport provide direct paths to optimize dimensions and morphol-gy of various TiO2 nanotube arrays for maximizing TiO2-basedevice performances such as DSSCs.

cknowledgements

This work is supported by BP – Gulf Mexico Research InitiativeGRI) fund, LABOR – RCS fund, and LSU College of Engineering –IER grant. D.S. Guan acknowledges LSU Graduate School Supple-entary Award. The authors would like to thank Prof. Wen Jin Meng

n Department of Mechanical Engineering at LSU for electrical cur-ent measurements, and also thank the Materials Characterizationenter at LSU for the use of SEM.

ppendix A. Supplementary data

Supplementary data associated with this article can be found,n the online version, at http://dx.doi.org/10.1016/j.electacta.012.08.036.

eferences

[1] D. Gong, C.A. Grimes, O.K. Varghese, Journal of Materials Research 16 (2001)3331.

[2] O.K. Varghese, M. Paulose, C.A. Grimes, Nature Nanotechnology 4 (2009)592.

[3] M.D. Ye, X.K. Xin, C.J. Lin, Z.Q. Lin, Nano Letters 11 (2011) 3214.

[[

[

Acta 83 (2012) 420– 429 429

[4] Q.D. Zhao, X.Y. Li, N. Wang, Y. Hou, X. Quan, G.H. Chen, Journal of NanoparticleResearch 11 (2009) 2153.

[5] Y.B. Xie, D.G. Fu, Journal of Applied Electrochemistry 40 (2010) 1281.[6] O.K. Varghese, D. Gong, M. Paulose, K.G. Ong, C.A. Grimes, Sensors and Actuators

B 93 (2003) 338.[7] E. Sennik, Z. Colak, N. Kilinc, Z.Z. Ozturk, International Journal of Hydrogen

Energy 35 (2010) 4420.[8] H.F. Lu, F. Li, G. Liu, Z.G. Chen, D.W. Wang, H.T. Fang, G.Q. Lu, Z.H. Jiang, H.M.

Cheng, Nanotechnology 19 (2008) 405504.[9] Z.H. Zhang, P. Wang, Energy & Environmental Science 5 (2012) 650.10] K. Shankar, G.K. Mor, H.E. Prakasam, S. Yoriya, M. Paulose, O.K. Varghese, C.A.

Grimes, Nanotechnology 18 (2007) 065707.11] S. Rani, S.C. Roy, M. Paulose, O.K. Varghese, G.K. Mor, S. Kim, S. Yoriya, T.J.

LaTempa, C.A. Grimes, Physical Chemistry Chemical Physics 12 (2010) 2780.12] D. Losic, S. Simovic, Expert Opinion on Drug Delivery 6 (2009) 1363.13] H.T. Fang, M. Liu, D.W. Wang, T. Sun, D.S. Guan, F. Li, J.G. Zhou, T.K. Sham, H.M.

Cheng, Nanotechnology 20 (2009) 225701.14] D.S. Guan, C. Cai, Y. Wang, Journal of Nanoscience and Nanotechnology 11

(2011) 3641.15] K. Yasuda, P. Schmuki, Electrochimica Acta 52 (2007) 4053.16] S. Kaneco, Y.S. Chen, P. Westerhoff, J.C. Crittenden, Scripta Materialia 56 (2007)

373.17] T. Stergiopoulos, A. Valota, V. Likodimos, T. Speliotis, D. Niarchos, P. Skeldon,

G.E. Thompson, P. Falaras, Nanotechnology 20 (2009) 365601.18] S.P. Albu, D.H. Kim, P. Schmuki, Angewandte Chemie International Edition 47

(2008) 1916.19] Y.J. Ji, K.C. Lin, H.G. Zheng, J.J. Zhu, A.C.S. Samia, Electrochemistry Communica-

tions 13 (2011) 1013.20] H.Y. Li, J.S. Wang, K. Huang, G.S. Sun, M.L. Zhou, Materials Letters 65 (2011)

1188.21] X.Y. Wang, S. Zhang, L.D. Sun, Thin Solid Films 519 (2011) 4694.22] D.S. Guan, Y. Wang, Nanoscale 4 (2012) 2968.23] J.T. Ding, Z.W. Xiao, H.S. Fan, Physica Status Solidi (RRL) 6 (2012) 166.24] J.M. Macak, S.P. Albu, D.H. Kim, I. Paramasivam, S. Aldabergerova, P. Schmuki,

Electrochemical and Solid-State Letters 10 (2007) K28.25] Y. Yang, X.H. Wang, L.T. Li, Materials Science and Engineering B 149 (2008) 58.26] F. Zhang, S.G. Chen, Y.S. Yin, C. Lin, C.R. Xue, Journal of Alloys and Compounds

490 (2010) 247.27] S.Q. Li, G.M. Zhang, D.Z. Guo, L.G. Yu, W. Zhang, Journal of Physical Chemistry

C 113 (2009) 12759.28] S.Q. Li, J.B. Yin, G.M. Zhang, Science China Chemistry 53 (2010) 1068.29] D.H. Kim, A. Ghicov, S.P. Albu, P. Schmuki, Journal of the American Chemical

Society 130 (2008) 16454.30] Y.L. Xie, Z.X. Li, H. Xu, K.F. Xie, Z.G. Xu, H.L. Zhang, Electrochemistry Communi-

cations 17 (2012) 34.31] Q.F. Zhang, G.Z. Cao, Nano Today 6 (2011) 91.32] J.M. Macak, K. Sirotna, P. Schmuki, Electrochimica Acta 50 (2005) 3679.33] J.M. Macak, H. Tsuchiya, P. Schmuki, Angewandte Chemie International Edition

44 (2005) 2100.34] L.V. Taveira, J.M. Macak, H. Tsuchiya, L.F.P. Dick, P. Schmuki, Journal of the

Electrochemical Society 152 (2005) B405.35] H. Liu, L. Tao, W.Z. Shen, Nanotechnology 22 (2011) 155603.36] D.S. Guan, H.T. Fang, H.F. Lu, T. Sun, F. Li, M. Liu, Progress in Chemistry 20 (2008)

1868.37] C.B. Cao, J.L. Li, X. Wang, X.P. Song, Z.Q. Sun, Journal of Materials Research 26

(2011) 437.38] J.L. Tao, J.L. Zhao, C.C. Tang, Y.R. Kang, Y.X. Li, New Journal of Chemistry 32

(2008) 2164.39] S. Berger, S.P. Albu, F. Schmidt-Stein, H. Hildebrand, P. Schmuki, J.S. Hammond,

D.F. Paul, S. Reichlmaier, Surface Science 605 (2011) L57.40] R.G. Ehl, A.J. Ihde, Journal of Chemical Education 31 (1954) 226.41] G.F. Ortiz, I. Hanzu, P. Knauth, P. Lavela, J.L. Tirado, T. Djenizian, Electrochimica

Acta 54 (2009) 4262.

42] L.D. Sun, S. Zhang, X.W. Sun, X.Y. Wang, Y.L. Cai, Langmuir 26 (2010) 18424.43] S. Berger, J. Kunze, P. Schmuki, A.T. Valota, D.J. LeClere, P. Skeldon, G.E. Thomp-

son, Journal of the Electrochemical Society 157 (2010) C18.44] H. Foll, S. Langa, J. Carstensen, M. Christophersen, I.M. Tiginyanu, Advanced

Materials 15 (2003) 183.