Production of hydrogen peroxide from carbon monoxide, water and oxygen over alumina-supported Ni catalysts

Article abstract of DOI:10.1016/j.molcata.2003.09.021

Novel amorphous Ni-B catalysts supported on alumina have been developed for the production of hydrogen peroxide from carbon monoxide, water and oxygen. The experimental investigation confirmed that the promoter/Ni ratio and the preparation conditions have a significant effect on the activity and lifetime of the catalyst. Among all the catalysts tested, the Ni-La-B/γ-Al ₂O₃ catalyst with a 1:15 atomic ratio of La/Ni, dried at 120°C, shows the best activity and lifetime for the production of hydrogen peroxide. The deactivation of the alumina-supported Ni-B amorphous catalyst was also studied. According to the characterizations of the fresh and used catalysts by SEM, XRD and XPS, no sintering of the active component and crystallization of the amorphous species were observed. However, it is water poisoning that leads to the deactivation of the catalyst. The catalyst characterization demonstrated that the active component had changed (i.e., amorphous NiO to amorphous Ni(OH)₂) and then salt was formed in the reaction conditions. Water promoted the deactivation because the surface transformation of the active Ni species was accelerated by forming Ni(OH) ₂ in the presence of water. The formed Ni(OH)₂ would partially change to Ni₃(PO₄)₂.

Full text of DOI:10.1016/j.molcata.2003.09.021

Journal of Molecular Catalysis A: Chemical 210 (2004) 157–163
Production of hydrogen peroxide from carbon monoxide, water and
oxygen over alumina-supported Ni catalysts
Zhong-Long Ma, Rong-Li Jia, Chang-Jun Liu∗
Key Laboratory of Green Synthesis and Transformation of Ministry of Education, School of Chemical Engineering and Technologies,
Tianjin University, Tianjin 300072, PR China
Received 23 April 2003; accepted 11 September 2003
Abstract
Novel amorphous Ni–B catalysts supported on alumina have been developed for the production of hydrogen peroxide from carbon monoxide,
water and oxygen. The experimental investigation confirmed that the promoter/Ni ratio and the preparation conditions have a significant effect
on the activity and lifetime of the catalyst. Among all the catalysts tested, the Ni–La–B/␥-Al2O3 catalyst with a 1:15 atomic ratio of La/Ni,
dried at 120 ◦C, shows the best activity and lifetime for the production of hydrogen peroxide. The deactivation of the alumina-supported Ni–B
amorphous catalyst was also studied. According to the characterizations of the fresh and used catalysts by SEM, XRD and XPS, no sintering of
the active component and crystallization of the amorphous species were observed. However, it is water poisoning that leads to the deactivation
of the catalyst. The catalyst characterization demonstrated that the active component had changed (i.e., amorphous NiO to amorphous Ni(OH)2)
and then salt was formed in the reaction conditions. Water promoted the deactivation because the surface transformation of the active Ni
species was accelerated by forming Ni(OH)2 in the presence of water. The formed Ni(OH)2 would partially change to Ni3(PO4)2.
© 2003 Elsevier B.V. All rights reserved.
Keywords: Hydrogen peroxide; Amorphous alloy; Ni–B catalyst; Carbon monoxide; Deactivation
1. Introduction
The reported TON (TON, moles of hydrogen peroxide per
mole of palladium) of the applied catalyst (palladium triph-
enylphosphine complex) was only 5. A further improvement
of the catalyst lifetime with a TON up to 87 was achieved
by using the more stable triphenylarsine as the palladium
ligand [8], but the process is still not suitable for a practi-
cal application. Recently, Bianchi et al. [9] have found that
the palladium complexes containing bidentate nitrogen lig-
ands efficiently catalyzed the reaction (1) with a productivity
comparable to that of the current commercial process.
The production of hydrogen peroxide from carbon
monoxide, water and oxygen is very promising owing to
its high safety, since it can be operated in a wider range
of non-ignitable gas phase composition. However, a draw-
back in the investigation of H2O2 synthesis from CO, water
and O2 is that palladium or other noble metal catalysts
are required [7–11]. In addition, most of these processes
employed a complex homogeneous catalytic system that
requires extra separation processes or acid additives for the
reaction. Only one reported work was related to the use
of heterogeneous catalytic synthesis over Pd/CaCO3 and
Ru/graphite catalysts [10].
Hydrogen peroxide is a chemical that is becoming more
and more popular as an environmentally friendly oxidant,
due to its unique feature of producing water as the only
by-product of the oxidation [1]. The utilization of hydrogen
peroxide for the novel green chemical synthesis has also
made great progress [2–5]. The traditional production of
hydrogen peroxide is not an economic and simple process.
Commercially, hydrogen peroxide is primarily produced by
the alternative oxidation and reduction of alkylantraquinone
derivatives [6]. Since this method is quite complex (a stoi-
chiometric amount of quinone must be used and recycled),
a great effort has been made worldwide towards a develop-
ment of a simple but efficient production of hydrogen per-
oxide. Ermakov and co-workers [7] reported a synthesis of
H2O2 from carbon monoxide, oxygen and water in 1979:
O2 + CO + H2O → H2O2 + CO2
(1)
∗
Corresponding author. Fax: +86-22-27890078.
E-mail address: changliu@public.tpt.tj.cn (C.-J. Liu).
1381-1169/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2003.09.021

158
Z.-L. Ma et al. / Journal of Molecular Catalysis A: Chemical 210 (2004) 157–163
Table 1
Since the first report of the study on the amorphous al-
Amorphous catalysts used in this study
loys as catalysts in 1980 [12], the amorphous alloys have
received much attentions over the past years. Their unique
isotropic structure and high concentration of coordinatively
unsaturated sites [13] lead to a better catalytic activity and
selectivity compared to their crystalline counterparts. Vari-
ous amorphous alloys have been prepared and employed in
catalysis including electrolysis [14], hydrogenation [15,16],
oxidation [17], isomerization [18], and methanation of car-
bon dioxide [19]. The amorphous alloy catalysts prepared
by chemical reduction seem to be suitable for catalysis too.
All the proposed reactions concerning the preparation of the
amorphous catalyst can be described as follows [20]:
Number Composition Atomic ratio
Preparation conditions
of La/Ni
Temperature (◦C)/
drying time (h)
a
b
1
2
3
4
5
6
7
8
Ni–B/␥-Al2O3
Ni–La–B/␥-Al2O3
Ni–B/␥-Al2O3
Ni–La–B/␥-Al2O3
Ni–La–B/␥-Al2O3
Ni–La–B/␥-Al2O3
Ni–La–B/␥-Al2O3
Ni–La–B/␥-Al2O3
–
–
–
b
1:15
a
–
120/10
120/10
120/10
120/10
60/10
1:20
1:15
1:10
1:15
1:15
300/10
a
No La in the catalyst.
No further thermal treatment with the fresh catalyst.
b
BH4− + 2H2O = BO2− + 4H2
(2)
BH4− + 2M2+ + 2H2O = 2M + BO2− + 4H+ + 2H2 (3)
support were completely reduced. During the reduction,
the temperature was kept at 0 ◦C in an ice-bath and the
mixture was stirred slowly. The resulted Ni–B/␥-Al2O3
(or Ni–La–B/␥-Al2O3) catalysts were then washed several
times with the oxygen-free pure water. The samples were
further washed with alcohol, and then heated at different
temperatures (60, 120, and 300 ◦C) in air for 10 h.
BH4− + H2O = B + OH− + 2.5H2
(4)
Supported amorphous alloy catalysts have been proved
to be promising in industrial applications owing to the re-
markable improvement in the active surface area and ther-
mal stability of catalysts caused by the high dispersion of
the amorphous alloy particles on the support and the strong
metal-support interaction [21]. Among various amorphous
alloy catalysts, the Ni-based amorphous alloys have been
studied thoroughly [22]. These amorphous alloy catalysts ex-
hibit higher catalytic activity, better selectivity and stronger
sulfur resistance in many hydrogenation reactions.
In this work, we report an amorphous Ni oxide-supported
catalyst, which shows the capability in the synthesis of H2O2
from CO, water and O2. However, the deactivation of the
catalyst was observed after reaction for several hours. The
aim of this paper is also to examine the possible factors
responsible for the deactivation.
2.2. Catalyst characterization
The amorphous structures of catalysts were determined
by X-ray diffraction (XRD) using a Rigaku C/max-2500
diffractometer with a Cu K␣ radiation. The surface mor-
phology of the catalyst was studied by a scanning electron
microscopy (SEM) using a Philips XL-30 ESEM system.
The surface composition and the surface electronic states of
the catalysts were determined by X-ray photoelectron spec-
trometry (XPS) using a Perkin-Elmer PHI1600 ESCA sys-
tem with a Al K␣ radiation (1486.6 eV).
2.3. Activity test
2. Experimental
The reaction was conducted in an autoclave with 350 ml
water containing phosphorous acid (the concentration is
about 0.01 M) and 5 g catalyst. Before the reaction, the
pressure was increased to 3000 kPa with a partial pressure
of CO (99.9%, Danpu Gas, Beijing) of 300 kPa and that of
O2 (99.9%, Huabei Gas, Tianjin, China) of 2700 kPa. The
reactions were carried out at 20 ◦C. The aqueous samples
taken at intervals were analyzed by titration using the iodo-
metric method. The used catalyst was kept to determine if
any sintering, crystallization, or surface poisoning occurred
after deactivation.
2.1. Catalyst preparation
Table 1 presents the catalysts applied in this work.
Each catalyst precursor was prepared by the chemi-
cal reduction method. The supported Ni–B/␥-Al2O3 (or
Ni–La–B/␥-Al2O3) catalysts were prepared with the follow-
ing procedures. The dried ␥-Al2O3 carriers (40–60 mesh,
197 m2/g) were first impregnated with a desired amount
of an aqueous solution of nickel nitrate (Wenda Chemical,
Tianjin, China; >99%) (or with a common solution of lan-
thanum nitrate (Wenda Chemical, Tianjin, China; >99%)
and nickel nitrate) over night. The total amount of metal
Ni loaded was 15 wt.%. The obtained samples were dried
at 110 ◦C for 6 h, and then the dried supported catalyst
precursors were reduced by the addition of an aqueous so-
lution of 0.5 M potassium borohydride (Wenda Chemical,
Tianjin, China; >99%). The initial atomic ratio of B/Ni
was adjusted to 3/1 to ensure that all the Ni2+ ions on the
3. Results and discussion
3.1. Characterization of catalysts
The amorphous structure of catalysts was identified by
XRD. Typical results of amorphous Ni alloys over alumina
are shown in Fig. 1. One broad peak around 2θ = 46◦ was

Z.-L. Ma et al. / Journal of Molecular Catalysis A: Chemical 210 (2004) 157–163
159
observed from each sample (except catalyst 8), which indi-
cates the catalysts (1–7) are the amorphous structure [23],
and the introduction of the alumina support and La does not
change the amorphous structure of Ni–B alloys. The catalyst
8, which was treated at 300 ◦C, exhibited a sharp diffrac-
tion peak around 2θ = 43.3◦ on the XRD patterns, showing
the occurrence of the slight crystallization since Ni–La–B
amorphous alloys are thermodynamically metastable.
SEM determines the surface morphology of catalysts. One
can see that the catalysts 3 and 5, shown in the Fig. 2,
exhibit a sponge-like morphology comprised of thousands
of small particles with the average size of less than 150 nm,
consistent with those of other Ni-based amorphous alloy
samples. It is also observed that, although with similar nickel
loading, the alloy particles in Ni–La–B/␥-Al2O3 were more
homogeneously dispersed than those in Ni–B/␥-Al2O3.
Table 2 shows the surface atomic composition of the
fresh amorphous catalysts (3–6) and the corresponding
Fig. 1. XRD patterns of the catalyst samples: (a) Amorphous NiO and
(b) crystal NiO.
Fig. 2. SEM photos of (a) catalyst 3 and (b) catalyst 5.

160
Z.-L. Ma et al. / Journal of Molecular Catalysis A: Chemical 210 (2004) 157–163
Table 2
Comparison of the surface atomic composition of the catalysts
XPS level Atomic composition (%)
Catalyst 3
Catalyst 4
Catalyst 5
Catalyst 6
Fresh Used Fresh Used Fresh Used Fresh Used
Ni2p
B1s
La3d
7.3
3.9
–
4.0
3.9
–
8.5
3.3
0.2
4.0
3.8
0.3
9.7
3.9
0.4
5.1
4.0
0.5
10.3
6.0
0.5
4.5
6.0
0.5
a
a
a
No La in the catalyst.
deactivated ones after reaction. From them, we can see a
significant decrease in the Ni content after reaction. Because
a 40–60% decrease of the surface Ni atoms was observed,
the Ni of active sites of the catalyst surface must have been
reduced significantly after reaction.
Fig. 3. B1s spectra of the catalyst samples and the deactivated catalysts:
(a) catalyst 3; (b) deactivated catalyst 3; (c) catalyst 5; and (d) deactivated
catalyst 5.
In order to evaluate the electronic state of the catalysts,
the binding energies of Ni and B were determined by XPS
under present conditions. The reference binding energies are
listed in Table 3 [24]. The peaks at 853.0 eV, at 853.9 eV,
and at 855.5 eV in Ni2p3/2 spectra are ascribed to elemental
nickel, nickel oxide, and Ni(OH)2, respectively, while the
peaks at 187.0 eV, at 193.9, and 193.5 eV are assigned to
elemental boron, boron oxide, and H3BO3, respectively.
Fig. 3 shows B1s spectra of the fresh and deactivated cata-
lysts. In comparison with the binding energy of the reference
substances in Table 3, the binding energy of B2O3 shown
in Fig. 3(a) and in Fig. 3(c) is found in the XPS spectra
of the catalysts 3 and 5, indicating that boron exists mainly
as B2O3. However, the XPS spectra of the deactivated cat-
alysts exhibit the peaks at about 193.5 eV, which point to
that boron is changed from B2O3 to H3BO3 after the cat-
alysts deactivated completely. The formation of H3BO3 is
from the presence of phosphorous acid and H2O.
The XPS spectra shown in Fig. 4(b) and (d) are the Ni2p3/2
spectra of the deactivated catalyst, which is obviously differ-
ent from the fresh catalysts (see Fig. 4(a) and (c)). Before re-
action, most of nickel are in oxide (B.E. = 854.0 eV), while
on the deactivated catalyst, Ni(OH)2 (B.E. = 855.5 eV) are
the predominant species. Little amount of Ni(OH)2 exists on
the catalyst samples, which indicates that during the prepara-
tion of catalysts, Ni(OH)2 has been produced. When prepar-
ing the catalysts, the presence of water from the aqueous
solution of potassium borohydride contributed to the forma-
tion of Ni(OH)2.
3.2. Activity of catalysts
As shown in Fig. 5, the fresh amorphous Ni–B/␥-Al2O3
catalyst (1) without further thermal treatment can catalyze
the synthesis of hydrogen peroxide from carbon monox-
ide, water and oxygen. The catalyst, however, deactivated
quickly, and H2O2 concentration in the clave reached the
maximum (0.6 mmol/l) when the reaction time was 2 h.
The Ni–La–B/␥-Al2O3 catalyst (2), which contains a small
amount of La, exhibited a longer lifetime than the catalyst
1 did. This indicates that the addition of La to Ni–B amor-
phous alloys can improve the stability of the catalyst and
lifetime reached 3 h. But the activity of the catalyst 2 was
slightly lower than that of catalyst 1 before deactivation,
Table 3
Binding energies (eV) of reference substances (all of these substances are
commercially available)
Standard substance
XPS level
Binding energies (eV)
Pure Ni
NiO
Ni (OH)2·H2O
Amorphous B
H3BO3
B2O3
Na2B4O7
KBH4
Ni2p3/2
Ni2p3/2
Ni2p3/2
B1s
B1s
B1s
853.0
853.9
855.5
187.0
193.5
193.9
192.0
187.3
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
Fig. 4. Ni2p3/2 spectra of the catalyst samples and the deactivated catalysts:
(a) catalyst 3; (b) deactivated catalyst 3; (c) catalyst 5; and (d) deactivated
catalyst 5.
B1s
B1s

Z.-L. Ma et al. / Journal of Molecular Catalysis A: Chemical 210 (2004) 157–163
161
Fig. 5. Comparison of activities between the dried catalysts and the
catalysts without drying treatment.
Fig. 6. Comparison of catalytic activities with different La/Ni ratios.
because La species cover part of the surface active area of
Ni–B alloys.
After being dried at low temperature, the activities of the
catalyst 3 and 5 increase obviously. The activity of the cata-
lyst 5, dried for 10 h at 120 ◦C, is almost three times higher
than that of the catalyst 2. The treatment at 120 ◦C induced
a significant increase in the maximal H2O2 concentration
(from 0.6 to 2.0 mmol/l). The catalysts were treated at low
temperature, which was evidently lower than the crystalliza-
tion temperature [23]. As exhibited in Fig. 1, the catalysts
still kept amorphous structure.
The peak of elemental nickel (B.E. = 853.0 eV) was not
detected in the XPS spectra (Fig. 4) of the catalysts dried
at 120 ◦C for 10 h. Together with the fact that the activities
of the dried catalysts (3–6) were obviously higher than that
of the catalysts without drying treatment (catalysts 1 and 2)
(see Fig. 5), we conclude that NiO has much better activity
than elemental nickel for this reaction.
Fig. 7. Effect of drying temperature on the catalytic activities.
conditions, the activity and lifetime of Ni–La–B/␥-Al2O3
(4–6) were increased and prolonged, compared to that of
catalyst 3. Fig. 6 also exhibits that the stability of catalyst
was improved with increase of atomic ratio of La/Ni. The
lifetime of the catalyst 6 reached 6 h. La has been applied
Fig. 6 compares results of activities of catalysts with dif-
ferent La/Ni ratios. Obviously, under the same preparative
Fig. 8. SEM surface morphology of the catalyst 8.

162
Z.-L. Ma et al. / Journal of Molecular Catalysis A: Chemical 210 (2004) 157–163
here to keep the stability of the Ni-supported catalyst. It has
been considered that La species may take up part of active
sites (Ni) of Ni–B alloys. Upon the XPS analysis shown in
Fig. 4, the addition of La can significantly slow down the
oxidation of Ni under the oxidative conditions. The lifetime
of the catalyst is thereby prolonged. However, a further in-
crease in La content leads to a reduced activity. The maximal
H2O2 concentration was obtained with the catalyst 5.
Fig. 7 presents the effect of drying temperature on cat-
alytic activities of catalysts with the same La/Ni ratio. From
Fig. 7, we can see that, when drying temperature is either
60 ◦C (catalyst 7) or 120 ◦C (catalyst 5), the catalyst shows
a sufficiently high activity. The activity of the catalyst 7 is
just a little lower than that of the catalyst 5. However, upon
the further increase in drying temperature to 300 ◦C (catalyst
8), the activity reduces quickly. From the XRD characteri-
zation shown in Fig. 1, when the drying temperature reaches
300 ◦C, the amorphous Ni–La–B alloy started to crystallize
partially that induces a decrease in the catalytic activity. In
Fig. 9. XRD patterns of the deactivated catalysts: (a) catalyst 3; (b)
catalyst 4; (c) catalyst 5; and (d) catalyst 6.
Fig. 10. SEM surface morphology of (a) deactivated catalyst 3 and (b) deactivated catalyst 5.

Z.-L. Ma et al. / Journal of Molecular Catalysis A: Chemical 210 (2004) 157–163
163
addition, a SEM characterization (Fig. 8) suggested that a
sintering occured after catalyst 8 was dried at 300 ◦C.
water poisoning and the loss of the surface active Ni species.
Water plays an important role in the deactivation process
of the amorphous Ni–B/␥-Al2O3 or Ni–La–B/␥-Al2O3 cat-
alyst, and the active Ni species to the inactive Ni speeds up
the transformation to the salt.
3.3. Analysis of deactivation
In general, the deactivation of supported nickel catalysts
may result from (a) sintering, (b) poisoning, and (c) loss of
active species [25]. SEM characterization exhibits that dur-
ing the present preparation all the catalysts did not sinter
except for catalyst 8. Since the reaction temperature was
comparatively low (20 ◦C), sintering can be excluded. Fig. 9
shows the XRD patterns of the deactivated catalysts (3–6),
which is the typical XRD pattern of the alumina support.
The fact that we did not observe the crystallization of cat-
alysts indicates that the main reason of deactivation is not
crystallization, thermal deactivation can be also excluded.
On the other hand, water has a negative effect on the sup-
ported nickel catalyst [26]. During the reaction, the active
Ni species (NiO) reacted with H2O in the media to form cat-
alytically inactive Ni(OH)2. But Ni(OH)2 was not a ultimate
deactivated substance of active Ni. Phosphorous acid in the
reaction media, which played an important role in the sta-
bility of H2O2, can react with Ni(OH)2 to form Ni3(PO4)2.
With the progress of reaction, the formed surface Ni salt can
partially dissolve in the media. So a significant decrease of
the surface Ni atoms occurred. The result of XPS character-
ization also showed that 1.5–2.0% P existed on the surface
of the deactivated catalysts.
Acknowledgements
The support from the 985 Project of Tianjin University
is very appreciated. The assistance from Ms. Fei He, Ms.
Haiyan Du, and Mr. Sen Han is also very appreciated.
References
[1] G. Strukul, in: G. Strukul (Ed.), Catalytic Oxidations with Hydrogen
Peroxide as Oxidant, Kluwer Academic Publishers, Dordrecht, 1992,
p. 1.
[2] A. Corma, L.T. Nemeth, M. Renz, S. Valencia, Nature 412 (2001)
423.
[3] G. Strukul, Nature 412 (2001) 388.
[4] H.R. Kalish, L. Latos-Grayski, A.L. Balch, J. Am. Chem. Soc. 122
(2000) 12478.
[5] G. Goor, in: G. Strukul (Ed.), Catalytic Oxidations with Hydrogen
Peroxide as Oxidant, Kluwer Academic Publishers, Dordrecht, 1992,
p. 13.
[6] F.E. Jacobson, Chem. Abstr. 108 (1988) 40590x.
[7] V.N. Zudin, V.A. Likholobov, Yu.A. Ermakov, Kinet. Katal. 20 (1979)
1599.
Fig. 10 shows SEM images of the surface morphology
of the deactivated catalysts. We can see that the deactivated
catalysts are made of even layers, and the characteristics of
the amorphous structure disappeared completely.
[8] S.E. Jacobson, US Patent 4 711 772 (1987).
[9] D. Bianchi, R. Bortolo, R. D’Aloisio, J. Mol. Catal. A 150 (1999) 87.
[10] W.F. Brill, US Patent 4 462 978 (1984).
[11] D. Bianchi, US Patent 5 783 164 (1998).
[12] G.V. Smith, W.E. Brower, M.S. Matyjaszczyk, in: Proceedings of
the Seventh International Congress on Catalysis, 1980, p. 355.
[13] A. Molnar, G.V. Smith, M. Bartok, Adv. Catal. 36 (1989) 329.
[14] M.M. Jaksic, J. Mol. Catal. 38 (1986) 161.
[15] H. Yamashita, M. Yoshikawa, T. Funabiki, S. Yoshida, J. Chem.
Soc., Faraday Trans. I 82 (1986) 1771.
[16] J.F. Deng, H.Y. Chen, X.H. Bao, M. Muhler, Appl. Surf. Sci. 81
(1994) 341.
[17] K. Machida, M. Enyo, I. Toyoshima, Y. Toda, T. Masumoto, Surf.
Coat. Technol. 27 (1986) 359.
4. Conclusions
Brill reported that the catalytic activity of the 5%
Pd/CaCO3 and 1% Ru/graphite catalysts for H2O2 pro-
duction reaches 0.1 mmol/(g h). In this work, the highest
activity of the catalyst 5 is 0.07 mmol/g h which is compa-
rable to the activity reported with noble metal catalysts. La
incorporation has a strong effect on Ni amorphous catalyst,
and the stability of the catalyst will thereby be improved.
Suitable drying temperature can raise the activity of the
catalyst. The Ni–La–B/␥-Al2O3 catalyst with a 1:15 atomic
ratio of La/Ni, dried at 120 ◦C, shows the best result.
In the present work, we also studied the deactivation
behavior of the amorphous Ni–B/␥-Al2O3 or Ni–La–B/␥-
Al2O3 catalyst for the production of hydrogen peroxide from
carbon monoxide, water, and oxygen. It was found that, un-
der the reaction conditions, the cause of the deactivation is
[18] W.E. Beower, M.S. Matyjaszczyk, T.L. Pettit, G.V. Smith, Nature
(Lond.) 301 (1983) 497.
[19] K. Hashimoto, Mater. Sci. Eng. A 226 (1997) 891.
[20] J.Y. Shen, Z.Y. Li, Q.J. Yan, Y. Chen, J. Phys. Chem. 97 (1993) 8564.
[21] H. Li, H.X. Li, J.F. Deng, Appl. Catal. 193 (2000) 9.
[22] S. Linderoth, S. Morup, J. Appl. Phys. 69 (1991) 5256.
[23] W.J. Wang, M.H. Qiao, S.H. Xie, J.F. Deng, Appl. Catal. A 163
(1997) 101.
[24] H. Li, H.X. Li, W.L. Dai, W.J. Wang, Zh.G. Fang, J.F. Deng, Appl.
Surf. Sci. 152 (1999) 25.
[25] S.H. Xie, H.X. Li, H. Li, J.F. Deng, Appl. Catal. A 189 (1999) 45.
[26] W.J. Wang, M.H. Qiao, H.X. Li, W.L. Dai, J.F. Deng, Appl. Catal.
A 168 (1998) 151.