The adsorption and reactions of SiClx (x=0-4) on hydroxylated TiO2 anatase (101) surface: A computational study on the functionalization of titania with Cl2Si(O)O adsorbate

(1)

The adsorption and reactions of SiCl

x

(x = 0–4) on hydroxylated TiO

2

anatase (1 0 1)

surface: A computational study on the functionalization of titania with Cl

2

Si(O)O

adsorbate

Wen-Fei Huang

a

, Hsin-Tsung Chen

b,⇑

, M.C. Lin

a,⇑

a

Center of Interdisciplinary Molecular Science, National Chiao Tung University, Hsinchu 300, Taiwan

b

Department of Chemistry, Chung Yuan Christian University, Chungli 32023, Taiwan

a r t i c l e

i n f o

Article history: Received 21 April 2012

Received in revised form 22 May 2012 Accepted 22 May 2012

Available online 5 June 2012 Keywords:

SiClxadsorption

Hydroxylated TiO2

Density functional theory

a b s t r a c t

The adsorption and reactions of the SiClx(x = 0–4) on the hydroxylated TiO2anatase (1 0 1) surface have been investigated by using periodic density functional theory calculations in conjunction with the pro-jected augmented wave (PAW) approach. The adsorption and reactions tend to occur more readily on the ‘Ow’ site derived from water than the ‘Os’ site from TiO2as revealed by the potential energy proﬁles and adsorption energies. The stepwise reactions of SiClxcan be achieved by dehydrochlorination taking place by three paths: Ow-path, cross-path, and Os-path. The Ow-path is the lowest energy path, in which Cl3Si–Ow(a) and Cl2Si–(Ow)Ow(a) are the main products formed by spontaneous reactions. The ready for-mation and the high stability of Cl2Si–(Ow)Ow(a) suggest that it can be employed as a molecular linker for Si and other semiconductor quantum dot growth on titania through its high reactivity towards SiHx rad-icals and metal alkyls, respectively.

1. Introduction

Titanium dioxide (TiO2) has been shown to be a very versatile

material as demonstrated in numerous theoretical and experimen-tal studies because of its promising applications to fabrication of photocatalysts and photoelectrochemical devices[1,2]. The photo-physics of TiO2sensitized by a variety of dyes[3–5], polymers[6,7],

and semiconductors[8–10]have been widely studied in particular for solar energy conversions. One of the well-known semiconduc-tor quantum dots for solar cell applications is silicon, which has been studied extensively [11–20]. In 2004, the growth of Si on TiO2rutile (1 1 0) surface was reported by Abad et al.[21].

To improve the heterogeneous interface and achieve higher photovoltaic efﬁciencies, inorganic linkers have been employed for semiconductor quantum dots growth on TiO2 [22–27].

Inor-ganic linkers for pure TiO2anatase (1 0 1) and rutile (1 1 0) surfaces,

SiHx, have been studied theoretically by Huang et al. [28].

How-ever, the TiO2surface is known to be readily covered with hydroxyl

groups in the water-rich environment[29–33]. SiHx, particularly

SiH4, was found to interact weakly with the hydroxylated TiO2

surface and thus may not be a good linker between silicon and hydroxylated TiO2surface. Based on the known propensity of SiCl4

reactions with HO-containing molecules and their common use as

Si sources in the silica-coating process[34,35], SiClx(x = 1–4) is

ex-pected to be a potential linker for the hydroxylated TiO2surface.

In this work, we study the adsorption of SiClxwith the

hydrox-ylated TiO2anatase (1 0 1) surface as well as their decomposition

reactions by dehydrochlorination employing the density functional theory (DFT). The potential energy proﬁles of the reactions are cal-culated by nudged elastic band (NEB) method. The results of this study should be useful for understanding the mechanism for SiClx

adsorption and decomposition on the TiO2surface for fabrication

of optoelectronic devices and solar cells.

2. Computational methods and models

All calculations were performed by the spin-polarized DFT with the projected augmented wave method (PAW) [36] as imple-mented in the Vienna ab initio simulation package (VASP) [37,38]. The ionic cores were described by the generalized gradient approximation (GGA) with the PW91[39]formulation which had been shown to work well for gas-surface reactions[22,27,40,41]. The electronic orbitals were represented by the plane-wave expan-sion including all the plane waves with their kinetic energies smal-ler than the chosen cut-off energy, hk2/2 m < Ecut(500 eV), which

ensures the convergence. The nudged elastic band (NEB) method [42]was applied to locate the transition states (TSs), up to eight images for each calculated TS. All the transition structures were veriﬁed by the frequency calculations.

http://dx.doi.org/10.1016/j.comptc.2012.05.035

⇑ Corresponding authors.

E-mail addresses: htchen@cycu.edu.tw (H.-T. Chen), chemmcl@emory.edu

(M.C. Lin).

Contents lists available atSciVerse ScienceDirect

Computational and Theoretical Chemistry

(2)

The bulk TiO2was optimized ﬁrst with (4 4 4)

Monkhorst-Pack k-points. Based on the optimized bulk TiO2geometry, the

sur-face super cell consisting of 24 [TiO2] units (three Ti layers, see

Supporting Information) was modeled as periodically repeated slabs, separated by a vacuum space greater than 13 Å, which guar-anties no interactions between the slabs. The lowest layers of each slab were ﬁxed to preserve the calculated bulk parameters, while the remaining layers were fully relaxed to simulate the surface behavior during the calculations. The (4 6 1) Monkhorst-Pack k-points were used for the TiO2 surface calculation. Gas-phase

molecules were simulated in a 20 Å cubic box, which is large en-ough to ignore interactions between each periodic gas molecules.

3. Results and discussion

To verify the reliability of the computational results, we ﬁrst compared the calculated bulk lattice constants with experimental values. The predicted lattice constants of TiO2 anatase are

a = 3.824 Å and c = 9.678 Å, which are in good agreement with the experimental values,[54,55]a = 3.872–3.875 Å and c = 9.502– 9.514 Å. The predicted fractional coordinate is u = 0.208, which also agrees well with the experimental value, u = 0.208[56]. In addition, the predicted adsorption energy of H2O on TiO2anatase

(1 0 1) surface is also in good agreement with experimental results, which will be discussed below. The good prediction of bulk geom-etry and adsorption energy of H2O on TiO2anatase (1 0 1) surface

supports the validity of the surface model. The geometries of gas phase molecules, SiClx(g) (x = 1–4), are also examined, and the

comparison of the experimental and calculated results is made in Table S1 of Supporting Information. Our predicted results are in

oxygen). The two unsaturated sites, Ti5cand O2c, are more reactive

[57,58]. In the hydroxylation reaction, H2O molecule ﬁrstly adsorbs

at the Ti5csite, forming H2O–Ti5c(a), and then dissociates to HO and

H as described below. The H atom binds to the bridge oxygen (O2c)

and co-adsorbs with the HO radical at the Ti5c site forming

HO–Ti5c,H–O2c(a). All the calculated energies at 1 ML are also shown

inFig. 1a. The adsorption energy of H2O–Ti5c(a) at 1 ML is 17.3

kcal/mol, which is in good agreement with calculated (15.9–16.6 kcal/mol) [57,59,60] and experimental (16.6–17.1 kcal/mol)[61] results. The transition state lies 5.0 kcal/mol below the reactants, TiO2(s) + H2O(g); the overall reaction is exothermicity of 10.3

kcal/mol. Hence, the formation of the hydroxylated TiO2is a

sponta-neous reaction, as observed experimentally[29,32,62].

There are two types of hydroxyl groups on the TiO2surface. One

derived from the H2O molecule, and the other one is formed by H

reaction with the O2cof the TiO2csurface. To distinguish the two

types of oxygen sources, we name the O atoms from H2O and

TiO2as ‘Ow’ and ‘Os’, respectively. The calculated geometry of the

fully hydroxylated TiO2 anatase (1 0 1) surface is shown in

Fig. 1b. The bond length and bond angles of H–Owand H–Ow–

Ti5c are 0.971 Å and 122.7°, respectively, while those for H–Os

and H–Os–Ti5care 0.972 Å and 121.3°. A Bader atomic charge

anal-ysis [63] shows the two types of hydroxyl groups have similar charge distribution. The charges of two H atoms are 1.00 e, and those of ‘Ow’ and ‘Os’ are 1.67 e and 1.60 e.

However, the reactivities of these two types of hydroxyl groups are different due to the surface morphology. The ‘Ow’ and ‘Os’

atoms lie above the ﬁrst Ti layer by 1.822 and 0.627 Å respectively. In other words, ‘Ow’ protrudes out of the surface more evidently

than ‘Os’ does. Thus gas molecule can more easily adsorb on the

‘Ow’ site than on the ‘Os’ site as will be reﬂected by the differences

(a)

(b)

Fig. 1. (a) Scheme of the formation of hydroxylated TiO2surface. All the energies are referred to the energy of TiO2(s) + H2O(g). (b) The conﬁguration of fully hydroxylated

(3)

in adsorption energies and potential energy proﬁles discussed below.

3.2. Adsorption and reactions of SiClx(x = 0–4) on the TiO2anatase

(1 1 0) surface

To generalize the nomenclature, ClxSi–(h)P1. . .Py(a) (P1. . .Py=

Owor Os) is named for SiClx adsorbates formed by the Si head

bonding with P1, P2. . .Py sites, which can be one to three sites

(y = 1–3). The preﬁxed ‘h’ denotes an H atom adsorbed concur-rently on the P1site. For example, ClxSi–hOsOw(a) represents SiClx

doubly bonded with Osand Owatoms with a hydrogen atom

simul-taneously bonding with the Osatom.

3.2.1. Dehydrochlorination reactions

Atomic layer growth of metal oxide can occur on a hydroxylated surface with MCl4(M = Si[64]or Ti[65]) by dehydrochlorination. A

general reaction can be illustrated by HO(a) + MCl4(g) ?

Cl3MO(a) + HCl(g). HCl can be formed from the reaction of one of

the Cl atoms of MCl4with one of the H atoms on the

hydroxyl-ated surface. Hence, SiClx decomposition on the hydroxylated

TiO2 surface can be achieved by a similar dehydrochlorination

reaction.

The possible mechanisms of SiClxdecomposition reactions are

shown inFig. 2. Three reaction pathways are considered. The reac-tion following the ‘Ow’ bonding site ﬁrst (the order is Ow–Ow–Os) is

named Ow-path, and the one following the ‘Os’ bonding site ﬁrst

(the order is Os–Os–Ow) is named Os-path. The remaining reaction

is named cross-path, which may have the order of Ow–Os–Owor

Os–Ow–Os.

3.2.2. The lowest energy path

In our calculations, we ﬁnd that the Ow-path is the lowest

en-ergy path, and the Os-path is the highest energy path. The complete

potential energy proﬁles are shown inSupporting Information of

Fig. S2. As shown in the ﬁgure, the highest energy barriers of the Ow-path, cross-path and Os-path are 9.7, 15.9, and 37.6 kcal/mol,

respectively. As mentioned above, the gas molecule can attach to the protruding ‘Ow’ site more readily than to the ‘Os’ site. In the

fol-lowing, we only discuss the Ow-path mechanism.

Fig. 3shows the potential energy proﬁle describing the Ow-path

mechanism. The simpliﬁed geometries depicted inFig. 3are shown inFig. 4. As illustrated inFig. 3a, SiCl4physically associates with

one of the Owatoms on the fully hydroxylated TiO2anatase surface

ﬁrstly with an association energy of 11.5 kcal/mol and the Si. . .Ow

distance of 3.926 Å. Overcoming a small energy barrier of 7.2 kcal/mol (TS1, Fig. 4b), the SiCl4 molecule can chemically bond

with the surface with an Si–Owbond of 1.753 Å, forming Cl4Si–

hOw(a) (Fig. 4c) with an exothermicity of 7.7 kcal/mol. The

sur-rounding H atoms can undergo hydrogen-bonding with the Cl atoms in Cl4Si–hOw(a). Dehydrochlorination occurs when one of

the Cl atoms is attracted by one of the nearest H atoms on the hydroxylated surface; the barrier for the process at TS2 is 14.1 kcal/mol below the reactants (Fig. 4d), giving an H. . .Cl bond-ing complex M1 (Fig. 4e) with an overall exothermicity of 21.8 kcal/mol. The distances of Cl. . .Si and Cl. . .H of M1 are 3.671 and 1.874 Å, respectively. Crossing the small TS3 (Fig. 4f) barrier of 4.6 kcal/mol, one HCl molecule is produced to form a hydrogen bonding complex HCl:Cl3Si–Ow(a) (Fig. 4g) with an overall

exo-thermicity of 24.0 kcal/mol. The physically adsorbed HCl(g) can be released endothermically with 3.6 kcal/mol, producing Cl3Si–

Ow(a) (Fig. 4h) on the surface which has a binding energy as high

as 133 kcal/mol (seeTable 1).

Starting from Cl3Si–Ow(a), the Si atom can doubly bond with

another oxygen atom of ‘Ow(H)’ to form Cl3Si–hOwOw(a) (Fig. 4j)

by overcoming the TS4 barrier of 14.8 kcal/mol; the complex lies 4.5 kcal/mol below the initial reactants. The bond lengths of Si–Owand Si–Ow(H) are 1.643 and 1.896 Å, respectively. The longer

bond length of Si–Ow(H) resulted from the additional bonding

with the hydrogen atom. The over-bonding of the Si atom in

Fig. 2. Scheme of SiClxdecomposition on the hydroxylated TiO2anatase (1 0 1) surface. (For the surface sketch, only the adsorbed sites are shown. The reactions are all in mass

(4)

Cl3Si–hOwOw(a) enhances the hydrogen bonding between one of Cl

atoms with the H attached to the Owatom bonding with Si by

over-coming a small energy barrier of 2.9 kcal/mol (TS5,Fig. 4k) forming the intermediate M2 (Fig. 4i) in which the Cl. . .H distance is 1.827 Å. The second dehydrochlorination can be achieved readily by overcoming a small barrier TS6 (Fig. 4m) of 1.7 kcal/mol, giving HCl:Cl2Si–OwOw(a) (Fig. 4n) with a small exothermicity of 0.6 kcal/

mol. The desorption of the physically adsorbed HCl(a) requires 8.6 kcal/mol to give the highly stable, doubly bonded Cl2Si–Ow

-Ow(a) (Fig. 4o) on the surface. Similar to the reaction of Cl3Si–

Ow(a) ? Cl3Si–hOwOw(a), the Si atom of Cl2Si–OwOw(a) can triply

bond with one more closest oxygen site, ‘Os’, forming Cl2Si–hOs

-OwOw(a) (Fig. 4q) by overcoming a high energy barrier of

28.9 kcal/mol (TS7,Fig. 4p) with an endothermicity of 25.0 kcal/ mol. The high energy barrier and the endothermicity for the forma-tion of the triply bonded adsorbate make the doubly bonded Cl2Si–

OwOw(a) adsorbate the major terminal product. Due to the surface

morphology, it is also impossible to attach to another ‘Ow’ site for

Cl2Si–OwOw(a) along the Ow-path. One of Cl atoms in Cl2Si–hOs

-OwOw(a) can in principle undergo further dehydrochlorination

through hydrogen bonding to form the third intermediate M3 (Fig. 4s) with an exothermicity of 7.2 kcal/mol by overcoming the

Fig. 3. Schematic potential energy proﬁles for the most possible reaction path of SiClxdecomposition on the hydroxylated TiO2anatase (1 0 1) surface. (For the surface sketch,

only the adsorbed sites are shown. In ﬁgure (b) and (c), one and two HCl(g) are omitted for brevity, respectively; they are included in energy calculations and are mass-balanced.)

(5)

energy barrier of 8.1 kcal/mol (TS8, Fig. 4r). The distance of the Cl. . .H hydrogen bonding in M3 (Fig. 4s) is 2.085 Å. The third dehydrochlorination can occur by overcoming TS9 (Fig. 4t) with a barrier of 10.3 kcal/mol and an exothermicity of 8.1 kcal/mol to form HCl:ClSi–OsOwOw(a) (Fig. 4u). The HCl(g) can be released

from HCl:ClSi–OsOwOw(a) with only a small endothermicity of

0.3 kcal/mol. In ClSi–OsOwOw(a), all the H atoms are too far from

the Cl atom to generate the fourth dehydrochlorination (Fig. 4v), it is left on the surface with an overall exothermicity of 7.0 kcal/ mol from the reactants, SiCl4(g) + hydroxylated anatase (1 0 1) TiO2.

In the Ow-path described above, Cl3Si–Ow(a) and Cl2Si–OwOw(a)

can be formed without thermal activation as all the transition

TS7

(p)

Cl

2

Si-O

w

O

w

(a)

(o)

HCl:Cl

2

Si-O

w

O

w

(a)

(n)

TS6

(m)

M2

(l)

TS5

(k)

Cl

3

Si-hO

w

O

w

(a)

(j)

TS4

(i)

Cl

3

Si-O

w

(a)

(h)

HCl:Cl

3

Si-O

w

(a)

(g)

TS3

(f)

M1

(e)

TS2

(d)

Cl

4

Si-hO

w

(a)

(c)

TS1

(b)

SiCl

4

:hydroxylated TiO

2

(a)

Fig. 4. Optimized geometries of adsorbed SiClxon the hydroxylated TiO2anatase (1 0 1) surface. TiO2model is only shown with one Ti layer. The bond length and angle are in

(6)

states (TS1–TS6) are lower than the initial reactants. As a result, the most likely product in the energetically downhill reactions, Cl2Si–OwOw(a), can be formed with an exothermicity of

27.0 kcal/mol relative to the reactants. The high barrier at TS8 (33 kcal/mol) for its decomposition by dehydrogenation to ClSi– OsOwOw(a) should make Cl2Si–OwOw(a) the most abundant species

produced in the reaction of SiCl4with the hydroxylated TiO2(1 0 1)

anatase surface. 3.2.3. Adsorption energies

The adsorption energies of the species SiClxon the hydroxylated

TiO2 surface are listed inTable 1. The adsorption energies were

calculated by the following equation:

Eads¼ ðEtotal Esurf EgasÞ

where Etotal, Egas, and Esurfare the electronic energies of the adsorbed

species on the surface, a gas-phase molecule, and the surface, respectively. The Esurffor Cl3Si–hOwOw(a), for example, is the

en-ergy of a fully hydroxylated TiO2 surface with one hydrogen

va-cancy for molecular adsorption.

As shown inTable 1, for the saturated Clx1Si–P1. . .P4x(a), the

average adsorption energies are 128, 170, and 307 kcal/mol for sin-gly, doubly, and triply bonded conﬁgurations, respectively. The

adsorption energies increase with the number of bondings. The same trend was noted for the over-bonded ClxSi–hP1. . .P4x(a)

with an Cl atom attached to the P1 (oxygen) site, 13, 102,

184 kcal/mol, respectively. For the same number of binding sites, the adsorption energies of the over-bonded conﬁgurations are much lower than those of bond saturated conﬁgurations. As men-tioned above, the gas phase molecules tend to adsorb on an ‘Ow’,

which is supported by the adsorption energies shown inTable 1. In the case of SiHxradical adsorptions (x = 1–3)[28], the adsorption

energies of H3Si–Ow(a) (133.2 kcal/mol), H2Si–OwOw(a) (211.2

kcal/mol), HSi–OsOwOw(a) (313.0 kcal/mol) are all larger than

those of H3Si–Os(a) (97.6 and 112.5 kcal/mol), H2Si–OsOs(a)

(148.0 kcal/mol), HSi–OwOsOs(a) (300.7 kcal/mol). A similar trend

was observed in the over-bonded cases. Finally, these high adsorp-tion energies also conﬁrm the stability of SiClxadsorption on the

hydroxylated surface.

4. Conclusions

The SiClxadsorption and reaction on the hydroxylated anatase

(1 0 1) TiO2surface have been studied by periodic DFT calculations.

The results ﬁrstly show that the hydroxylated anatase (1 0 1) TiO2

surface can be formed spontaneously by the reaction H2O(g) +

ClSi-O

s

O

w

O

w

(a)

(v)

HCl:ClSi-O

s

O

w

O

w

(a)

(u)

Fig. 4 (continued) Table 1

Calculated adsorption energies of SiClxon the hydroxylated TiO2surface (kcal/mol).

Adsorbate Site Figure Eads Adsorbate Site Figure Eads

SiCl4 hOs 4a, S3a 11.5, 8.9 SiCl3 Os S3q 112.5

hOw 4c 19.2 Ow 4h 133.2 SiCl3 hOsOs S3h 78.6 SiCl2 OsOs S3k 148.0 hOsOw S3ae 104.3 OsOw S3ai 192.1 hOwOs S3s 105.7 hOwOw 4j 118.3 OwOw 4o 211.2 SiCl2 hOsOwOw 4q 185.1 SiCl OsOwOw 4v 313.0 hOwOsOw S3ak 191.8 hOwOsOs S3m 186.3 OwOsOs S3ac 300.7 hOsOwOs S3y 173.9

(7)

anatase (1 0 1) TiO2surface. There are two kinds of oxygen

adsorp-tion sties, ‘Ow’ and ‘Os’, on the fully hydroxylated surface. The

adsorptions and reactions tend to occur more readily on the ‘Ow’

site than on the ‘Os’ site based on the predicted adsorption

ener-gies. The stepwise decomposition of SiClx(a) can be achieved by

dehydrochlorination, and the reaction paths can be divided into three types of pathways: Ow-path, Os-path and cross-path. We ﬁnd

that the Ow-path is the most favored low energy path, followed by

the cross-path and lastly the Os-path. The adsorption and

decom-position of SiCl4 following the Ow-path can take place by three

dehydrochlorination steps. Each dehydrochlorination can be achieved by three types of reactions: (1) SiClxmulti-site

adsorp-tions; (2) the intermediate Mx(x = 1–3) formation; (3) the

dehy-drochlorination giving stable ClxSi(a) adsorbates (x = 1–3). We

ﬁnd that type (1) reactions control the SiClx decomposition on

the hydroxylated anatase (1 0 1) surface. Cl3Si–Ow(a) and Cl2Si–

OwOw(a) can be formed without thermal activation; however, the

formation of ClSi–OsOwOw(a) needs to overcome a 33 kcal/mol

en-ergy barrier. Accordingly, the energetically downhill adsorption and decomposition reactions of SiCl4on the hydroxylated titania

by stepwise dehydrochlorination processes is expected to produce Cl2Si–OwOw(a) with the most abundant concentration providing a

reactive and strong molecular linker for the growth of Si thin ﬁlms and other semiconductor quantum dots for opto-electronic device and solar cell fabrication applications. Take the growth of a III–V metal nitride, for example, the Cl2Si(O)O-functionalized surface

can readily react with nitrogen and metal precursors (such as NH3and trimethyl indium) on the ﬁrst surface layer as follows:

Cl2Si(O)O(a) + NH3?2HCl + HNSi(O)O(a); HNSi(O)O(a) + (CH3)3

In ? (CH3)2InNSi(O)O(a) + CH4.

Acknowledgements

The authors are grateful to Taiwan’s National Center for High-performance Computing for the CPU’s facility and the National Science Council for the research support. MCL also wants to acknowledge Taiwan Semiconductor Manufacturing Co. for the TSMC Distinguished Professorship and Taiwan National Science Council for the Distinguished Visiting Professorship at the Center for Interdisciplinary Molecular Science, National Chiao Tung University, Hsinchu, Taiwan.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.comptc.2012 .05.035.

References

[1] K.I. Hadjiivanov, D.G. Klissurski, Surface chemistry of titania (anatase) and titania-supported catalysts, Chem. Soc. Rev. 25 (1996) 61–69.

[2] A.L. Linsebigler, G. Lu, J.T. Yates, Photocatalysis on TiO2surfaces: principles,

mechanisms, and selected results, Chem. Rev. 94 (1995) 735–758. [3] M. Gratzel, Photoelectrochemical cells, Nature 414 (2001) 338–344. [4] M.K. Nazeeruddin, A. Kay, I. Rodicio, R. Humphry-Baker, E. Muller, P. Liska, N.

Vlachopoulos, M. Gratzel, Conversion of light to electricity by cis-X2bis(2,20

-bipyridyl-4,40_{-dicarboxylate)ruthenium(II) charge-transfer sensitizers (X = Cl–,}

Br–, I–, CN–, and SCN–) on nanocrystalline titanium dioxide electrodes, J. Am. Chem. Soc. 115 (1993) 6382–6390.

[5] B. O’Regan, M. Gratzel, Solar energy conversion by dye-sensitized photovoltaic cells, Nature 353 (1991) 737–740.

[6] C.L. Huisman, A. Goossens, J. Schoonman, Aerosol synthesis of anatase titanium dioxide nanoparticles for hybrid solar cells, Chem. Mater. 15 (2003) 4617– 4624.

[7] D. Li, C. Gu, C. Guo, C. Hu, The effects of ambient gases on the surface resistance of polyoxometalate/TiO2ﬁlm, Chem. Phys. Lett. 385 (2004) 55–59.

[8] P. Yu, K. Zhu, A.G. Norman, S. Ferrere, A.J. Frank, A.J. Nozik, Nanocrystalline TiO2

solar cells sensitized with InAs quantum dots, J. Phys. Chem. B 110 (2006) 25451–25454.

[9] J.L. Blackburn, D.C. Celmarten, A.J. Nozik, Electron transfer dynamics in quantum dot/titanium dioxide composites formed by in situ chemical bath deposition, J. Phys. Chem. B 107 (2003) 14154–14157.

[10] R. Vogel, P. Hoyer, H. Weller, Quantum-sized PbS, CdS, Ag2S, Sb2S3, and Bi2S3

particles as sensitizers for various nanoporous wide-bandgap semiconductors, J. Phys. Chem. 98 (1994) 3183–3188.

[11] A. Shah, P. Torres, R. Tscharner, N. Wyrsch, H. Keppner, Photovoltaic technology: the case for thin-ﬁlm solar cells, Science 285 (1999) 692–698. [12] R.B. Bergmann, Crystalline Si thin-ﬁlm solar cells: a review, Appl. Phys. A 69

(1999) 187–194.

[13] L. Stalmans, J. Poortmans, H. Bender, M. Caymax, K. Said, E. Vazsonyi, J. Nijs, R. Mertens, Porous silicon in crystalline silicon solar cells: a review and the effect on the internal quantum efﬁciency, Prog. Photovolt. Res. Appl. 6 (1998) 233–246. [14] B. Hamilton, Porous silicon, Semicond. Sci. Technol. 10 (1995) 1187–1207. [15] R.L. Smith, S.D. Collins, Porous silicon formation mechanisms, J. Appl. Phys. 71

(1992) R1–R22.

[16] A.G. Cullis, L.T. Canham, P.D.J. Calcott, The structural and luminescence properties of porous silicon, J. Appl. Phys. 82 (1997) 909–965.

[17] P.R. McCurdy, L.J. Sturgess, S. Kohli, E.R. Fisher, Investigation of the PECVD TiO2–Si(1 0 0) interface, Appl. Surf. Sci. 233 (2004) 69–79.

[18] V.G. Erkov, S.F. Devyatova, E.L. Molodstova, T.V. Malsteva, U.A. Yanovskii, Si– TiO2 interface evolution at prolonged annealing in low vacuum or N2O

ambient, Appl. Surf. Sci. 166 (2000) 51–56.

[19] P.G. Karlesson, J.H. Richter, M.P. Andersson, J. Blomquist, H. Siegbahn, P. Uvdal, A. Sandell, UHV-MOCVD growth of TiO2on SiOx/Si(1 1 1): interfacial properties

reﬂected in the Si 2p photoemission spectra, Surf. Sci. 580 (2005) 207–217. [20] S. Takabayashi, R. Nakamura, Y. Nakato, A nano-modiﬁed Si/TiO2composite

electrode for efﬁcient solar water splitting, J. Photochem. Photobiol. A 166 (2004) 107–113.

[21] J. Abad, C. Rogero, J. Méndez, M.F. López, J.A. Martín-Gago, E. Román, Growth of subnanometer-thin Si overlayer on TiO2(1 1 0)-(1 2) surface, Appl. Surf. Sci.

234 (2004) 497–502.

[22] P. Raghunath, M.C. Lin, Adsorption conﬁgurations and reactions of boric acid on TiO2anatase (1 0 1) surface, J. Phys. Chem. C 112 (2008) 8276–8287.

[23] P. Raghunath, M.C. Lin, A computational study on the adsorption conﬁgurations and reactions of phosphorous acid on TiO2anatase (1 0 1) and

rutile (1 1 0) surfaces, J. Phys. Chem. C 113 (2009) 3751.

[24] J.-G. Chang, J.-H. Wang, M.C. Lin, Adsorption conﬁgurations and energetics of BClx(x = 0–3) on TiO2anatase (1 0 1) and rutile (1 1 0) surfaces, J. Phys. Chem. A

111 (2007) 6746–6754.

[25] W.-F. Huang, H.-T. Chen, M.C. Lin, Density-functional theory study of the adsorption and reaction of H2S on TiO2 rutile (1 1 0) and anatase (1 0 1)

surfaces, J. Phys. Chem. C 113 (2009) 20411–20420.

[26] W.-F. Huang, P. Raghunath, M.C. Lin, Computational study on the reactions of H2O2on TiO2anatase (1 0 1) and rutile (1 1 0) surfaces, J. Comput. Chem. 32 (6)

(2011) 1065–1081.

[27] C.-Y. Chang, H.-T. Chen, M.C. Lin, Adsorption conﬁgurations and reactions of nitric acid on TiO2rutile (1 1 0) and anatase (1 0 1) surfaces, J. Phys. Chem. C

113 (2009) 6140–6149.

[28] W.-F. Huang, H.-T. Chen, M.C. Lin, Density-functional calculations on the adsorption and reactions of SiHx(x = 0–4) on TiO2anatase (1 0 1) and rutile

(1 1 0) surfaces, Comput. Mater. Sci., submitted for publication.

[29] R.L. Kurtz, R. Stockbauer, T.E. Madey, E. Román, J.L.D. Segovia, Surf. Sci. 218 (1989) 178.

[30] M.B. Hugenschmidt, L. Gamble, C.T. Campbell, The interaction of H2O with a

TiO2(1 1 0) surface, Surf. Sci. 302 (1994) 329–340.

[31] P. Jones, J.A. Hockey, Infra-red studies of rutile surfaces. Part 1, Trans. Faraday Soc. 67 (1971) 2669–2678.

[32] P. Jones, J.A. Hockey, Infra-red studies of rutile surfaces. Part 2 – Hydroxylation, hydration and structure of rutile surfaces, Trans. Faraday Soc. 67 (1971) 2679–2685.

[33] L. Kavan, K. Kratochvilová, M. Gratzel, Study of nanocrystalline titanium dioxide (anatas) electrode in accumulation regime, J. Electroanal. Chem. 394 (1995) 93–102.

[34] Q.H. Powell, G.P. Fotou, T.T. Kodas, B.M. Anderson, Y. Guo, Gas-phase coating of TiO2with SiO2in a continuous ﬂow hot-wall aerosol reactor, J. Mater. Res. 12

(1997) 552–559.

[35] C.-H. Hung, J.L. Katz, Formation of mixed oxide powders in ﬂames: Part I. TiO.sub.2–SiO.sub.2, J. Mater. Res. 7 (1992) 1861–1869.

[36] P. Blochl, Projector augmented-wave method, Phys. Rev. B 17 (1994) 953–979. [37] G. Kresse, J. Furthmuller, Efﬁciency of ab initio total energy calculations for metals and semiconductors using a plane-wave basis set, J. Comput. Mater. Sci. 6 (1996) 15.

[38] G. Kresse, J. Furthmuller, Efﬁcient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54 (1996) 11169. [39] J.P. Perdew, Y. Wang, Accurate and simple analytic representation of the

electron-gas correlation energy, Phys. Rev. B 45 (1992) 13244–13249. [40] J.-H. Wang, M.C. Lin, Reactions of hydrazoic acid and trimethyl indium on TiO2

rutile (1 1 0) surface. A computational study on the formation of the ﬁrst monolayer InN, J. Phys. Chem. B 110 (2006) 2263–2270.

[41] J.-H. Wang, M.C. Lin, Y.-C. Sun, Reactions of hydrazoic acid on TiO2

nanoparticles: an experimental and computational study, J. Phys. Chem. B 109 (2005) 5133–5142.

[42] G. Mills, H. Jonsson, G.K. Schenter, Reversible work transition state theory: application to dissociative adsorption of hydrogen, Surf. Sci. 324 (1995) 305– 337.

(8)

TiO2, Proc. Natl. Acad. Sci. 99 (2002) 6476–6481.

[49] A.N. Enyashin, G. Seifert, Structure, stability and electronic properties of TiO2

nanostructures, Phys. Status Solidi B 242 (2005) 1361–1370.

[50] N.G. Park, J.V.D. Lagemaat, A.J. Frank, Comparison of dye-sensitized rutile-and anatase-based TiO2 solar cells, J. Phys. Chem. B 104 (2000) 8989–

8994.

[51] S.D. Burnside, V. Shklover, C. Barbe, Pascal Comte, F. Arendse, K. Brooks, M. Greatzel, Self-organization of TiO2nanoparticles in thin ﬁlms, Chem. Mater. 10

(1998) 2419–2425.

[52] B. O’Regan, M. Grätzel, A low-cost, high-efﬁciency solar cell based on dye-sensitized colloidal TiO2ﬁlms, Nature 353 (1991) 737–740.

[53] F. Labat, P. Baranek, C. Adamo, Structural and electronic properties of selected rutile and anatase TiO2surfaces: an ab initio investigation, J. Chem. Theory

Comput. 4 (2008) 341–352.

[54] J.K. Burdett, T. Hughbanks, G.J. Miller, J.W. Richardson Jr., J.V. Smith, Structural-electronic relationships in inorganic solids: powder neutron diffraction studies

[60] G. Mattioli, F. Filippone, A.A. Bonapasta, Reaction intermediates in the photoreduction of oxygen molecules at the (1 0 1) TiO2(anatase) surface, J.

Am. Chem. Soc. 128 (2006) 13772–13780.

[61] G.S. Herman, Z. Dohnalek, N. Ruzycki, U. Diebold, Experimental investigation of the interaction of water and methanol with anatase-TiO2(1 0 1), J. Phys.

Chem. B 107 (2003) 2788–2795.

[62] M.B. Hugenschmidt, L. Gamble, C.T. Campbell, The interaction of H2O with a

TiO2(1 1 0) surface, Surf. Sci. 302 (1994) 329–340.

[63] G. Henkelman, A. Arnaldsson, H. Jonsson, A fast and robust algorithm for Bader decomposition of charge density, Comput. Mater. 36 (2006) 354–360. [64] O. Sneh, M.L. Wise, A.W. Ott, L.A. Okada, S.M. George, Atomic layer growth of

SiO2on Si(1 0 0) using SiCl4and H2O in a binary reaction sequence, Surf. Sci.

334 (1995) 135–152.

[65] J.D. Ferguson, A.R. Yoder, A.W. Weimer, S.M. George, TiO2 atomic layer

deposition on ZrO2particles using alternating exposures of TiCl4and H2O,