Daylight Photocatalysis by Carbon-Modified Titanium Dioxide

Article abstract of DOI:10.1002/anie.200351577

Green titana: Carbon-doped titanium dioxide, supported onto filter paper, photocatalyzes the gas-phase degradation of the atmospheric pollutants benzene (a), acetaldehyde (b) and carbon monoxide (c) in diffuse indoor visible light (see picture).

Full text of DOI:10.1002/anie.200351577

Communications
Titanium Dioxide Photocatalysts
Daylight Photocatalysis by Carbon-Modified
Titanium Dioxide**
Shanmugasundaram Sakthivel and Horst Kisch*
In memory of Dieter Sellmann
Semiconductor photocatalysis is an efficient method for the
chemical utilization of solar energy. It is based on the surface
trapping of light-generated charges which induce interfacial
electron-transfer reactions with a great variety of substrates.
Titanium dioxide, the most promising photocatalyst is already
used in various practical applications, such as self-cleaning
[
1]
paints and window panes. However, because of its large
band gap of 3.20 eV only the small UV fraction of solar light,
about 2–3%, can be utilized. Therefore many attempts have
been made to sensitize titanium dioxide for the much larger
visible fraction.
In previous work we have shown that incorporation of a
few weight percent of a transition-metal chloride either in the
bulk or at the surface of titanium dioxide leads to visible-light
[
2]
photocatalysis. Especially the materials modified by chem-
isorbed platinum(iv) chloride were very active and sustained
photomineralization of the ubiquitous pollutant 4-chlorophe-
[
2e]
nol by visible light (l ꢀ 455 nm) for several days. Different
from modification by metal complexes, we recently found that
a coke-containing titanium dioxide was active also with visible
light. This material was prepared by a sol–gel method using
[
3]
various titanium alkoxide precursors. Since its activity,
especially at low light intensity was much smaller than the
platinum(iv) chloride modified titanium dioxide, we searched
for improvement. This was achieved during work on nitrogen-
[
4a]
doped titanium dioxide in which we hydrolyzed titanium
tetrachloride with nitrogen bases, such as tetrabutylammo-
nium hydroxide. At the subsequent calcination step we
observed that prolonged heating at 4008C afforded an
anatase material which contained carbon instead of nitrogen.
This preparation method has an excellent reproducibility and
greater applicability than the oxidation of a titanium metal
[
5,6]
sheet to a carbon-doped rutile layer in a natural gas flame.
In the degradation of 4-chlorophenol by artificial light (l ꢀ
55 nm) these new, blackish-brown powders are five times
4
more active than nitrogen-doped titanium dioxide. Herein we
report the preparation, photoelectrochemical, and photo-
[
*] Prof. Dr. H. Kisch, Dr. S. Sakthivel
Institut für Anorganische Chemie
Universität Erlangen-Nürnberg
Egerlandstrasse 1, 91058 Erlangen (Germany)
Fax: (+49)9131-852-7363
E-mail: kisch@chemie.uni-erlangen.de
[**] This work was supported by Deutsche Forschungsgemein-
schaft(SFB 583).
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
4
908 ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200351577
Angew. Chem. Int. Ed. 2003, 42, 4908 –4911

Angewandte
Chemie
catalytic properties of carbon doped TiO2 under direct
artificial and diffuse natural light.
The samples TiO -C1a, TiO -C1b, and TiO -C2 containing
2
2
2
2
.98, 0.42, and 0.03% carbon were prepared by the hydrolysis
of titanium tetrachloride with tetrabutylammonium hydrox-
ide followed by calcination at 4008C for 0.25 h (TiO -C1a)
2
and 1 h (TiO -C1b), and at 5508C for 4 h (TiO -C2).
2
2
Unmodified TiO was prepared by the same procedure but
2
replacing the nitrogen base by sodium hydroxide. Except for
TiO -C1a, which contained traces of nitrogen, the three other
2
samples were nitrogen-free. According to X-ray diffraction
(
XRD) measurements all the materials were in the anatase
modification except TiO -C2 which contained trace quantities
2
of the rutile phase.
X-ray photoelectron spectroscopy (XPS) measurements
of the C1 binding energy indicated the complete absence of
carbon in the undoped sample, whereas for TiO -C1b, peaks
2
[
7]
at 285.6, 287.5, and 288.5 eV were found. The first value
[
8,9]
arises from adventitious elemental carbon,
the latter two
[
8]
values suggest the presence of a carbonate species. This
assumption is supported by the IR spectrum of TiO -C1b
2
À1
which exhibits low-intensity peaks at 1738, 1096, and 798 cm
[
10,11]
which are indicative for the carbonate ion.
In contrast to
these results, it was proposed that in the rutile layer, prepared
by flame oxidation, carbon partially substitutes the oxide
ion.
Diffuse reflectance spectra reveal that the new absorption
Figure 1. A) Diffuse reflectance spectra of modified and pure TiO2;
[
6]
B) Plot of transformed Kubelka–Munk function versus the energy of
the light absorbed; a) TiO , b) TiO -C1a, c) TiO -C1b, d) TiO -C2.
2
2
2
2
at 400–700 nm is related to the carbon content. It is stronger
for TiO -C1a than for TiO -C1b and completely absent in
2
2
TiO -C2 (Figure 1, Table 1), which was cal-
2
cined at 5508C. Assuming the materials to be
Table 1: Carbon content, photoelectrochemical data, band-gap energies, and initial disappearance rates
ri) of 4-chlorophenol.
indirect semiconductors, like TiO , a plot of
2
(
the modified Kubelka–Munk function versus
[
0
a]
[b]
[a]
[c]
À8
À1 À1 [d]
[
12,13]
Catalyst
C [%]
pH
Ebg [eV]
U
fb (NHE) [V]
r [10 molL s ]
i
the energy of exciting light
affords band-
gap energies of 3.16, 3.02, 3.11, and 3.17 eV
for TiO , TiO -C1a, TiO -C1b, and TiO -C2,
TiO2
0.00
2.98
0.42
0.03
5.80
7.90
6.40
5.44
3.16
3.02
3.11
3.17
À0.52
À0.39
À0.48
À0.54
0.10
3.75
7.65
0.13
2
2
2
2
[
14]
TiO -C1a
2
respectively (Figure 1, Table 1). The max-
imum band-gap narrowing of 0.14 eV is
comparable with the value of 0.05 eV
recently observed for nitrogen-doped
TiO -C1b
2
TiO -C2
2
[a] Measured according to ref. [16] and calculated for pH 7. [b] Reproducibility was better than
[
4a,c]
Æ0.01 eV. [c] Reproducibility was better than Æ0.01 V. [d] Obtained by measuring the decrease of
TiO2.
absorbance at 224 nm; l ꢀ455 nm.
irr
To understand whether a shift of the
valence or conduction band edge is respon-
sible for this decrease of the band-gap
energy, the position of the flat-band potential (U ) was
fb
[
15,16]
determined by the “slurry method”
through measuring
the photovoltage as a function of the suspension pH value
Figure 2). The inflection point corresponds to the pH value
[
16]
(
0
from which the flat-band potential can be calculated
(
Table 1). The value of À0.52 V measured for TiO at pH 7
2
is in good agreement with the À0.58 reported for an anatase
[
17]
single crystal and measured by the Mott–Schottky method
and À0.47 V for anatase powder also measured by the slurry
[
18]
method. As observed for the band gap, the change in flat-
band potential is also related to the carbon content. Whereas
the potential of À0.54 V, found for the low-carbon sample
TiO -C2, does not differ significantly from that of unmodified
2
Figure 2. Variation of photovoltage with pH value for the suspension
TiO (À0.52 V) it is shifted to À0.39 and À0.48 V for TiO -
2
2
of catalyst in the presence of methyl viologen a) TiO
c) TiO -C1b, d) TiO -C2.
, b) TiO -C1a,
2 2
C1a and TiO -C1b, respectively. The anodic shifts of 130 and
2
2
2
Angew. Chem. Int. Ed. 2003, 42, 4908 –4911
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 4909

Communications
4
0 mV may be compared with the value of 50 mV found for
[
4a]
nitrogen-doped TiO2. Assuming that the distance between
flat-band potential and the conduction band edge is vanishing
for these probably highly doped n-type semiconductor
materials, the position of the valence-band edge can
be located from the corresponding band-gap energies as
+
2.63 V for all samples.
Since the carbon-modified samples start absorbing at
about 735 nm (1.70 eV; Figure 1) a variety of surface states
must exist. To determine their approximate energy, the
wavelength dependence of OH-radical formation by TiO2-
C1b was investigated in the presence of two different electron
acceptors and benzoic acid as OH scavenger.
In the presence of oxygen, salicylic acid was produced
irrespective of exciting TiO -C1b at l ꢀ 320 (3.88 eV), 455
2
(
2.73 eV), and 495 nm (2.51 eV). This can be rationalized by
À
recalling that the reduction potential of the O /O couple at
2
2
[
19]
pH 7 is À0.16 V. If light absorption at the longest wave-
length would occur from the valence band to a surface state
close to the conduction band, the resulting potential of the
trapped electron should be located at about + 0.12 V and
reduction of oxygen und subsequent OH-radical formation is
not feasible. Contrary, this process becomes possible when
absorption occurs from a surface state located near the
valence band to the conduction band, since now the trapped
electron potential is À0.39 V. In accord with this explanation,
upon replacement of oxygen by tetranitromethane as an
electron acceptor, salicylic acid formation proceeded only for
the short wavelength irradiation. Under these band-to-band
absorption conditions the OH-radical formation occurs
directly through hole-oxidation of water or surface hydroxy
groups (Figure 3). This significant difference suggests the
presence of surface states close to the valence-band edge, as
[
4c]
was also concluded from theoretical calculations.
Figure 4. A) Photomineralization of 4-chlorophenol with artificial visi-
ble light (l=455 nm; TOC and TOC =total organic carbon content
0
t
at times 0 and t); a) TiO , b) TiO -C1a, c) TiO -C1b, d) TiO -C2. B) Dif-
2
2
2
2
fuse indoor daylight degradation of 4-chlorophenol (4-CP) and remazol
red in the presence of pure TiO and TiO -C1b; a) TiO /4-CP, b) TiO /
2
2
2
2
remazol red, c) TiO -C1b/4-CP, d) TiO -C1b/remazol red. C) Gas-phase
2
2
photodegradation of acetaldehyde (a and c , 5 vol%), benzene (c ,
1
2
5
vol%), and carbon monoxide (c , 5 vol%) in air through diffuse
3
indoor light in the presence of pure TiO (a) and TiO -C1b (c).
2
2
respectively (curves b,c). In contrast, TiO -C2 and unmodi-
2
fied TiO did not significantly change the total organic carbon
2
Figure 3. Formation of salicylic acid (SA) upon irradiation of TiO -C1b
2
value (curves a,d). However, they induced a small reduction
in the concentration of 4-chlorophenol as measured by UV
absorption spectroscopy. The corresponding rates found for
TiO -C1a and TiO -C1b were about 30 and 60 times larger
in the presence of benzoic acid; a) lꢀ495 nm, under O , b,c)
2
lꢀ495 nm, under Ar and in the presence of tetranitromethane
c) lꢀ320 nm, under Ar and in the presence of tetranitromethane.
2
2
(Table 1). The twofold rate enhancement induced by TiO2-
To explore the photocatalytic activity of the new materi-
als, the degradation of some typical pollutants in water and air
by artificial and natural light was investigated. Figure 4A
illustrates the photomineralization of the ubiquitous water
pollutant 4-chlorophenol with visible light (l = 455 nm) in the
presence of the various photocatalysts. TiO -C1a and TiO -
C1b relative to TiO -C1b although the latter contains more
2
carbon may be due to the higher crystallinity of TiO -C1a as
2
suggested by XRD spectra.
The superior photocatalytic activity of the carbon-doped
materials is demonstrated by illumination experiments in
diffuse indoor daylight. In solution 4-chlorophenol and the
azo dye remazol red are efficiently mineralized only by TiO2-
2
2
C1b induced complete mineralization after 210 and 180 min,
4
910
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 4908 –4911

Angewandte
Chemie
C1b (Figure 4B, curves c,d) whereas unmodified TiO2 is
[8]E. Papirer, R. Lacroix, J.-B. Donnet, G. Nanse, P. Fioux, Carbon
995, 33, 63.
1
almost inactive (curves a,b). Furthermore TiO -C1b sup-
2
[
9]C. S. Gopinath, S. G. Hegde, A. V. Ramaswamy, S. Mahapatra,
ported on filter paper, photocatalyzes the oxidation of
gaseous acetaldehyde, benzene, and carbon monoxide (Fig-
ure 4C).
Mater. Res. Bull. 2002, 37, 1323.
10]It is noted that hydrolysis with sodium carbonate does not afford
a photocatalyst active in the visible.
11]S. Gablenz, H.-P. Abicht, E. Pippel, O. Lichtenberger, J.
Woltersdorf, J. Eur. Ceram. Soc. 2000, 20, 1053.
[
[
Experimental Section
[12]J. Tauc, R. Grigorovici, A. Vanuc, Phys. Status Solidi 1966, 15,
27.
6
Preparation of photocatalysts: A solution of tetrabutylammonium
À1
[
[
13]B. Karvaly, I. Hevesy, Z. Naturforsch. A 1971, 26, 245.
14]This corresponds to absorption onsets of 392, 411, 399, and
hydroxide (0.25 moll ) was added dropwise to 0.25m TiCl (200 mL)
4
at 08C until a pH value of 5.5 was reached. After ageing the
suspension for 24 h at room temperature, the precipitate was
collected by filtration and dried under air at 708C. The residue was
crushed into a fine powder and calcined in a muffle furnace at 4008C
for 0.25 and 1 h, and at 5508C for 4 h. Calcination temperatures were
attained at a heating rate of about 58Cmin
391 nm.
[
[
15]J. R. White, A. J. Bard, J. Phys. Chem. 1985, 89, 1947.
16]A. M. Roy, G. C. De, N. Sasmal, S. S. Bhattacharyya, Int. J.
Hydrogen Energy 1995, 20, 627. The powders (30 mg) were
À1
suspended in 50 mL of 0.1m KNO containing 6 mg of methyl-
.
3
viologen dichloride; lirr = 320 nm.
Photochemical experiments: Photomineralizations in the aque-
ous phase were carried out with catalyst powder (15 mg) suspended in
a water-cooled cylindrical 15-mL quartz cuvette; the suspension was
stirred magnetically. Appropriate cut-off filters were inserted before
irradiating with an Osram XBO 150 W xenon arc lamp. Salicylic acid
was measured through its intense fluorescence at 400–420 nm as
[17]L. Kavan, M. Grꢀtzel, S. E. Gilbert, C. Klemenz, H. J. Scheel, J.
Am. Chem. Soc. 1996, 118, 6716.
[
18]M. D. Ward, J. R. White, A. J. Bard, J. Am. Chem. Soc. 1983, 105,
7.
19]D. T. Sawyer, J. S. Valentine, Acc. Chem. Res. 1981, 14, 393.
2
[
[
2d]
described elsewhere. Daylight experiments were performed in the
laboratory by exposing the reaction system to the diffuse radiation
À2
arriving through the window (4–10 Wm at 400–1200 nm). Solution
experiments were conducted in an Erlenmeyer flask containing TiO2-
C1b (50 mg) and solutions of 4-chlorophenol (50 mL; 2.5 ”
À4
À1
À4
À1
1
0
molL ) or solutions of remazol red (50 mL; 1 ” 10 molL ).
Gas-phase experiments were conducted in a 1-L round bottom flask
containing a filter paper of 11-cm diameter onto which the catalyst
(12 mg) was supported; concentration change and CO2 formation
were measured by IR spectroscopy.
Received: April 4, 2003 [Z51577]
Publication delayed at authors request
Keywords: carbon · doping · photocatalysis · semiconductor ·
.
titanium dioxide
[
1]a) D. A. Tryk, A. Fujishima, K. Honda, Electrochim. Acta 2000,
5, 2363.
4
[
2]a) H. Kisch, L. Zang, C. Lange, W. F. Maier, C. Antonius, D.
Meissner, Angew. Chem. 1998, 110, 3201; Angew. Chem. Int. Ed.
1998, 37, 3034; b) L. Zang, C. Lange, W. F. Maier, I. Abraham, S.
Storck, H. Kisch, J. Phys. Chem. B 1998, 102, 10765; c) L. Zang,
W. Macyk, C. Lange, W. F. Maier, C. Antonius, D. Meissner, H.
Kisch, Chem. Eur. J. 2000, 6, 379; d) W. Macyk, H. Kisch, Chem.
Eur. J. 2001, 7, 1862; e) G. Burgeth, H. Kisch, Coord. Chem. Rev.
2002, 230, 41.
[
3]C. Lettmann, K. Hildenbrand, H. Kisch, W. Macyk, W. F. Maier,
Appl. Catal. B 2001, 32, 215.
[
4]a) S. Sakthivel, H. Kisch, ChemPhysChem, 2003, 4, 487; for other
preparations see: b) S. Sato, Chem. Phys. Lett. 1986, 123, 126;
c) R. Asahi, T. Morikawa, T. Ohwaki, A. Aoki, Y. Taga, Science
2001, 293, 269; d) T. Morikawa, R. Asahi, T. Ohwaki, A. Aoki, Y.
Taga, Jpn. J. Appl. Phys. 2001, 40, 561.
[
[
[
5]A. Fujishima, K. Kohayakawa, K. Honda, J. Electrochem. Soc.
1
975, 122, 1487.
6]S. U. M. Khan, M. Al-Shahry, W. B. Ingler, Jr., Science 2002, 297,
243.
7]Relevant atomic percentages are C1s 5.70, O1s 62.66, Ti2p
2.63%. We are highly indebted to Prof. P. Schmuki for these
measurements.
2
3
Angew. Chem. Int. Ed. 2003, 42, 4908 –4911
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4911