Highly active Co-B amorphous alloy catalyst with uniform nanoparticles prepared in oil-in-water microemulsion

Article abstract of DOI:10.1016/j.jcat.2008.07.015

Uniform Co-B nanoparticles were synthesized for the first time by chemical reduction of cobalt ion with borohydride in an oil-in-water microemulsion system comprising cyclohexane, polyethylene glycol, and water. The particle size was controlled by modulating the cyclohexane content. With the characterization of X-ray diffraction, selective area electronic diffraction, X-ray photoelectron spectroscopy, and transmission electron microscopy, the resulting Co-B nanoparticles were identified to be amorphous alloys ranging in size from 6 to 20 nm. During liquid-phase cinnamaldehyde hydrogenation, the as-synthesized Co-B catalyst was extremely active and more selective than the regular Co-B prepared in aqueous solution. Furthermore, this catalyst also was found to be more durable during the hydrogenation process.

Full text of DOI:10.1016/j.jcat.2008.07.015

Journal of Catalysis 259 (2008) 104–110
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Highly active Co–B amorphous alloy catalyst with uniform nanoparticles prepared
in oil-in-water microemulsion
Hui Li a, Jun Liu a, Songhai Xie b, Minghua Qiao b, Weilin Dai b, Hexing Li a,∗
a
Department of Chemistry, Shanghai Normal University, Shanghai 200234, PR China
Department of Chemistry, Fudan University, Shanghai 200433, PR China
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Uniform Co–B nanoparticles were synthesized for the first time by chemical reduction of cobalt
ion with borohydride in an oil-in-water microemulsion system comprising cyclohexane, polyethylene
glycol, and water. The particle size was controlled by modulating the cyclohexane content. With
the characterization of X-ray diffraction, selective area electronic diffraction, X-ray photoelectron
spectroscopy, and transmission electron microscopy, the resulting Co–B nanoparticles were identified
Received 29 March 2008
Revised 28 July 2008
Accepted 29 July 2008
Available online 23 August 2008
to be amorphous alloys ranging in size from
6 to 20 nm. During liquid-phase cinnamaldehyde
Keywords:
hydrogenation, the as-synthesized Co–B catalyst was extremely active and more selective than the regular
Co–B prepared in aqueous solution. Furthermore, this catalyst also was found to be more durable during
the hydrogenation process.
Co–B amorphous alloy catalyst
Uniform diameter
Microemulsion
Cinnamaldehyde
Hydrogenation
© 2008 Elsevier Inc. All rights reserved.
Cinnamyl alcohol
1. Introduction
kinds of M–B amorphous alloy catalysts [12,13]. But vigorous and
−
exothermic reactions between metallic ions and borohydride (BH4
)
usually induce particle aggregation, thereby reducing catalytic ac-
tivity because of the low surface area in the catalysts. Meanwhile,
the nonuniform nanoparticles also are responsible for the ready
crystallization of amorphous alloys [14]. Although monodispersed
nanostructured metal colloids have been reported [15], few at-
tempts have been made to prepare M–B amorphous alloy in uni-
form nanoparticles, possibly because the reduction of metallic ions
The hydrogenation of cinnamaldehyde (CMA) to cinnamyl al-
cohol (CMO) is widely applied in the production of perfumes,
flavorings, pharmaceuticals, and other fine chemicals [1]. Up to
now, only the Ir-, Os-, and Pt-based catalysts have been used to
achieve high selectivity to CMO, because hydrogenation of C=C
is always easier than hydrogenation of C=O [2–6]. However, the
high cost and nonavailability of these catalysts pose a problem.
Co-based catalysts are more selective toward CMO compared with
other nonnoble metal catalysts [7], but their practical applica-
tions remain limited due to their very low activity. Amorphous
alloys (metallic glasses), metastable materials with long-range dis-
ordered but short-range ordered structure, have attracted growing
attention from both academia and industry due to their supe-
rior corrosion resistance, high mechanical toughness, and excellent
magnetic, electronic, and catalytic properties compared with their
crystalline counterparts [8]. As heterogeneous catalysts, amorphous
alloys have been widely studied because of their excellent cat-
alytic performances [8–11]. To form and stabilize the amorphous
structure, some metalloids (e.g., B) should be incorporated in
amorphous alloys, significantly affecting their physical and chem-
ical properties [12,13]. The synthesis of ultrafine particles through
chemical reduction has been most often used to prepare these
−
by BH4 is uncontrollable and strongly exothermic in preparation.
Previously, we reported a Co–B amorphous alloy catalyst with uni-
form particle size prepared by ultrasound-assisted reduction of
−
Co(NH3)2+ with BH4 in aqueous solution [14]. But the slow reduc-
6
tion process resulted in relatively large (>50 nm) Co–B particles
unfavorable for catalytic activity.
Recently, size-and-morphology control synthesis of nanoparti-
cles has been achieved through reverse microemulsion [16–18].
A series of Ni–B amorphous alloy catalysts were successfully syn-
thesized in a water-in-oil microemulsion system [19]. But using
large amounts of organic phase in microemulsions may cause sev-
eral problems, including low concentration of products, difficult
separation, and even environmental pollution [20]. Thus, there is a
great need to design simple methods that can produce large quan-
tities of uniform nanoparticles in a desired size range.
In this paper, we report the synthesis of uniform Co–B amor-
phous alloy nanoparticles by conducting the chemical reduction
−
of Co2+ with BH4 in an oil-in-water microemulsion system. The
Corresponding author. Fax: +86 21 64322272.
*
E-mail address: hexing-li@shnu.edu.cn (H. Li).
nanoparticles could be easily collected and the particle size freely
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.07.015

H. Li et al. / Journal of Catalysis 259 (2008) 104–110
105
adjusted by changing the amount of oil phase in the microemul-
sion. During liquid-phase hydrogenation of CMA to CMO, the as-
prepared Co–B exhibited better catalytic performance (activity, se-
lectivity, and durability) than the Co–B prepared by direct reduc-
using the ideal gas equation. The RSH should represent the intrinsic
activity, because the effect of metal dispersion has been ruled out.
During the reaction, samples were withdrawn from the reaction
mixture every 30 min for product analysis using a gas chromato-
graph (GC 102) with a flame ionization detector. The stationary
phase was 15% Apiezon (L)/Gas Chrom (red) at 523 K with N2 as
carrier gas. All results have been reproduced, and the errors were
limited within 5%.
−
tion of Co2+ with BH4 or even by ultrasound-assisted reduction of
−
Co(NH3)2+ with BH4 . The correlation of the catalytic performances
to the st6ructural properties has been tentatively established.
2. Experimental
3. Results and discussion
2.1. Catalyst preparation
3.1. Structural and electronic characteristics
In a typical run of catalyst synthesis, CoCl2·6H2O (5.0 mmol,
1.2 g) and polyethylene glycol (Mw = 20,000, 5.0 g) were first dis-
solved in deionized water (40 ml), then an appropriate amount
of cyclohexane was added to this solution. The mixture thus ob-
tained was treated under ultrasonic radiation at 293 K for 30 min,
with an oil-in-water microemulsion formed. Then a 2.0 M KBH4
aqueous solution (10 ml) was added dropwise under vigorous stir-
ring at 293 K. After reaction was complete, the black precipitate
The TEM images (Fig. 1a) show that Co–B-0 displayed irreg-
ular, broadly dispersed particles, apparently due to the particle
−
agglomeration, because the reaction between Co2+ and BH4 is
strongly exothermic [12]. In contrast, the Co–B nanoparticles syn-
thesized in an oil-in-water microemulsion were spherical and rela-
tively uniform, and the particle size decreased from 20 to 6 nm
as the cyclohexane/water volume ratio was increased from 0.1
to 0.25 (Figs. 1b–1d), which was obviously much smaller than the
Co–B-50-30 synthesized through ultrasound-assisted reduction of
−
+
was washed free from Cl or K ions with deionized water until
a pH ∼7 was achieved, followed by three washings with abso-
lute alcohol (EtOH). Finally, the mixture was stored in EtOH until
use. The as-prepared Co–B samples were designated Co–B-x, with
x representing the volume ratio of cyclohexane to water. The di-
rect reduction of CoCl2 with KBH4 in pure aqueous solution re-
sulted in the regular Co–B sample [21] and was designated Co–B-0.
The Co–B-50-30 was prepared by ultrasound-assisted reduction of
−
Co(NH3)2+ with BH4 [14]. The relatively narrow distribution for
6
the as-prepared nanoparticles can be attributed mainly to the dis-
persing effect of oil droplets in microemulsion, which can restrict
particle growth and even prevent aggregation of the Co–B nanopar-
ticles. An increase in the cyclohexane content may enhance the
dispersing effect, leading to decreased particle size (Figs. 1b–1d);
however, a further increase in the amount of cyclohexane could in-
duce significant agglomeration due to the extremely high surface
energy in small Co–B nanoparticles (Fig. 1e).
−
Co(NH3)2+ with BH4 in aqueous solution [14], where 50 and 30
6
refer to the ultrasound power (W) and ultrasonication time (min),
respectively.
ICP analysis revealed that the bulk compositions in all Co–B-x
samples were very close (Table 1). The higher B content in Co–
B-x compared with in Co–B-0 may be due to the inhibition of
2.2. Characterization
−
BH4 hydrolysis in the microemulsion, which would enhance the
The bulk composition was analyzed by inductively coupled
plasma optical emission (ICP; Varian VISTA-MPX). The amor-
phous structure was determined by both X-ray diffraction (XRD;
Rigaku D/Max-RB with CuKα radiation) and elective area elec-
tronic diffraction (SAED; JEOL JEM-2010). Surface morphology and
particle size were observed by transmission electron microscopy
(TEM; JEOL JEM2010). The size distribution was evaluated from
about 300 randomly selected particles. Thermal stability was deter-
mined by differential scanning calorimetry (DSC; Shimadzu DSC-
60) under N2 atmosphere, at a heating rate of 10 K min−1. The
active surface area (SCo) was measured by hydrogen chemisorption
acidity of the reaction solution and the formation of B-enriched
amorphous alloys [24]. Table 1 also shows that the SCo of Co–
B-x samples initially increased and then decreased with increas-
ing cyclohexane content in the microemulsion. The maximum SCo
was obtained for Co–B-0.25, obviously due to the smallest par-
ticle size (see Fig. 1). Co–B-50-30 displayed a higher SCo than
Co–B-0, due to the ultrasound effect, which can clean the cata-
lyst surface and also inhibit the particle agglomeration. Co–B-0.25
showed even higher SCo than Co–B-50-30, obviously due to its
smaller particle size. The XRD patterns presented in Fig. 2 show
that, similar to the Co–B-0 and Co–B-50-30 [14], all Co–B samples
synthesized in oil-in-water microemulsion (e.g. Co–B-0.25) exhib-
on a Quantachrome CHEMBET 3000 system, assuming H/Co(s) = 1
−20
and a surface area of 6.5 × 10
m2 per Co atom [22]. The hy-
◦
ited a single broad peak around 2θ = 45 indicative of amorphous
drogen temperature-programmed desorption (H2-TPD) curves were
obtained on the same instrument in argon flow at a ramping rate
of 20 K min−1. The surface electronic states were investigated by
X-ray photoelectron spectroscopy (XPS; Perkin-Elmer PHI 5000C
ESCA using AlKα radiation), and all Co–B samples were dried in
pure Ar atmosphere before measurements. The binding energy (BE)
values were calibrated using C 1s = 284.6 eV as a reference.
structure [25], which can be further confirmed by a successive
diffraction halo in an attached SAED image [26]. Treatment of the
fresh Co–B-0.25 sample (at 673 K in N2 for 2 h) resulted in the
appearance of various diffraction peaks corresponding to metallic
Co and crystalline Co–B alloy. The appearance of Co–B crystalline
phases during the crystallization process verified the formation of
an alloy between Co and B for the as-prepared Co–B. In addition,
our previous work also demonstrated the presence of Co–B bond
through EXAFS analysis [21]. The presence of a shoulder peak at
around 1.7 Å at the radial distribution functions obtained from the
Co χ(k)k3 edge by fast Fourier transformation suggests the forma-
tion of a Co–B bond. Furthermore, calculations based on the EXAFS
reveal that the Co–B bond length was much shorter than the sum
of Co radius and B radius, which also suggests the formation of
Co–B alloy. We infer that except for the crystallization, partial de-
composition of a Co–B amorphous alloy may occur in this process.
The DSC analysis (Fig. 3) reveals that the crystallization of the Co–
B-0 sample involved two steps: rearrangement of Co–B amorphous
2.3. Activity test
Liquid-phase hydrogenation of CMA was carried out in
a
200-ml stainless autoclave containing 0.30 g of catalyst, 4 ml of
CMA, 40 ml of EtOH, and 1.0 MPa of H2 at 373 K. The reaction
system was stirred vigorously (1000 rpm) to eliminate the diffu-
sion effects [23]. According to the PH drop with the reaction time,
2
both the specific activity [i.e., H2 uptake rate per gram of cobalt,
RmH (mmol h
g
−1)] and the areal activity [i.e., H2 uptake rate per
−1
Co
−1
m2 of the active surface area, RSH (mmol h
m
−2)] were calculated
Co

106
H. Li et al. / Journal of Catalysis 259 (2008) 104–110
Fig. 1. TEM morphologies and their corresponding size distribution histograms (inserts) of the fresh Co–B samples. (a) Co–B-0, (b) Co–B-0.1, (c) Co–B-0.2, (d) Co–B-0.25, and
(e) Co–B-0.3.

H. Li et al. / Journal of Catalysis 259 (2008) 104–110
107
Table 1
Structural properties and catalytic performances of the as-prepared catalystsa
Catalyst
Composition
(at%)
SCo
RHm
RHS
Timeb
(h)
Yieldb
(%)
−1
)
(m2
g
−1 −1
−1 −2
(mmol h
g
)
(mmol h
m
)
Co
Co
Co–B-0
Co75.4B24.6
Co60.2B39.8
Co62.5B37.5
Co61.9B38.2
Co59.7B40.3
15.8
13.6
36.9
42.6
32.2
19.5
20.4
57.0
69.9
48.3
1.2
1.5
1.5
1.6
1.5
8.0
9.0
4.0
3.5
4.0
80.3
80.0
81.6
85.6
74.3
Co–B-0.1
Co–B-0.2
Co–B-0.25
Co–B-0.3
Crystallized
Co61.9B38.2
Co77.7B22.3
12.6
25.5
11.5
0.9
1.4
10.0
4.5
68.5
80.0
Co–B-0.25c
Co–B-50-30d
34.7
a
Reaction conditions: catalyst containing 0.3 g Co, 4 ml of CMA, 40 ml of EtOH, T = 373 K, PH = 1.0 MPa, stirring rate = 1000 rpm.
2
b
c
Values corresponding to the maximum CMO yield.
Obtained by treating Co–B-0.25 amorphous alloy at 673 K for 2 h in N2.
−
Prepared via reduction of Co(NH3)2+ with BH4 with ultrasound power of 50 W and 30 min [14].
d
6
Fig. 2. XRD patterns of (a) the fresh Co–B-0, (b) the fresh Co–B-0.25, and (c) the
Co–B-0.25 sample after being treated at 673 K for 2 h in N2 flow. The insert is the
SAED image of the fresh Co–B-0.25 sample.
Fig. 4. XPS spectra of (a) Co–B-0, and (b) Co–B-0.25 samples.
Fig. 3. DSC curves of (a) Co–B-0, and (b) Co–B-0.25 samples.
increased B content in Co–B alloys [24]. On the other hand, the
uniform particle size may inhibit the particle migration and ag-
glomeration, a key step in the crystallization process [27]. The XPS
spectra (Fig. 4) revealed that for all Co species in either Co–B-0
or Co–B-0.25, the core level of Co 2p3/2 was at 778.2 eV, indicat-
ing that all Co atoms were present in metallic state [28]. But the
B species were present in both the elemental B and the oxidized B,
with B1s BE values of around 188.2 and 192.5 eV. The B1s BE of
alloy structure and the crystallization of Co–B amorphous alloy
with exothermic peaks at around 414 and 580 K [21]. But Co–B-
0.25 displayed only one exothermic peak at 658 K, higher than that
for Co–B-0 by 78 K. Obviously, the as-prepared Co–B-x samples ex-
hibited much stronger thermal stability against crystallization than
Co–B-0. The enhanced thermal stability is obviously related to the

108
H. Li et al. / Journal of Catalysis 259 (2008) 104–110
Scheme 1. Possible reaction pathways for CMA hydrogenation.
the elemental B in Co–B-0.25 exceeded that of pure B (187.1 eV) by
1.1 eV [29], further indicating that the elemental B is alloyed with
the metallic Co. In alloys, partial electrons may be transferred from
B to Co, as has been detected in a Co–B-0 sample [21] and demon-
strated by a theoretic calculation based on the ComB2 (m = 1–4)
cluster models using density functional theory [30]. This also was
confirmed by the finding that the metal in the M–B amorphous al-
loy had a much stronger antioxidation ability than the pure metal,
due to the presence of electron-deficient B that could effectively
protect the metal from oxidation [31]. The failure in observing the
BE shift of the metallic Co can be understood by considering its
relatively greater atomic weight compared with the B atom [29].
Thus, considering the results from XRD, SAED, DSC, and XPS, we
can conclude that Co–B amorphous alloy was formed through the
present method. According to the calculation based on XPS peak
areas, the surface molar ratio of the alloying B to the metallic Co
in Co–B-0.25 was 38/62, higher than that in either Co–B-0 (28/72)
or even in Co–B-50-30 (32/68) [14], implying that Co–B-0.25 was
surface-enriched with the alloying B [24].
(a)
3.2. Catalytic performance
Fig. 5 shows the changes of solution composition with reaction
time during CMA hydrogenation over the Co–B-0 and Co–B-0.25
samples. Along with the production of CMO, dihydrocinnamalde-
hyde (HCMA) and 3-phenylpropanol (HCMO) were identified as
two byproducts. The CMA hydrogenation process is illustrated in
Scheme 1 [1]. The main product CMO is produced from the se-
lective hydrogenation of the C=O bond in CMA. The competitive
hydrogenation of C=C against C=O in CMA results in the for-
mation of HCMA. HCMO is produced by further hydrogenation of
either CMO or HCMA.
(b)
Fig. 5. Reaction profiles of CMA hydrogenation over (a) Co–B-0, and (b) Co–B-0.25
samples. Reaction conditions are given in Table 1.
We compared the catalytic efficiencies of Co–B-0.25 and Co–B-0
amorphous catalysts and several conventionally synthesized Co-
based catalysts, including Co/SiO2 [32], Co/Al2O3 [33], and Raney
Co [34]. As shown in Fig. 6, the as-prepared Co–B amorphous
alloys exhibited much greater CMO yields compared with the Co-
based catalysts reported previously. The superior catalytic proper-
ties of the as-prepared catalysts can be attributed to their unique
short-range ordering, whereas a long-range disordering amorphous
structure has proven favorable for hydrogenation processes [8–13].
This is supported by the fact that fresh Co–B-0.25 exhibited much
higher activity and selectivity to CMO than the corresponding crys-
tallized catalyst obtained by treating the fresh catalyst at 673 K for
2 h in N2 flow (see Table 1). The superior performance of the Co–B
amorphous alloy catalyst over its corresponding crystalline catalyst
can be interpreted in terms of both structural and the electronic
effects.
Fig. 6. CMO yield obtained on Co–B-0.25, Co–B-0, Co/SiO2, Co/Al2O3, and Raney Co
catalysts.
Co active sites, and the lower N corresponded to more highly un-
saturated Co active sites, which may promote the adsorption of
reactant molecules and the surface hydrogenation [35]. Meanwhile,
the higher σ implies a more homogeneous distribution of Co ac-
tive sites, and the uniform Co active sites are beneficial for the
selectivity to CMO [32].
3.2.1. Structural effect
According to our previous EXAFS analysis [23], the Co–B amor-
phous alloy displayed a shorter Co–Co bonding length (RCo–Co),
a lower coordination number (N) of Co–Co first neighboring shell,
and a higher DW factor (σ) than the crystalline catalysts (Ta-
ble 2). The RCo–Co represented stronger synergistic effect between

H. Li et al. / Journal of Catalysis 259 (2008) 104–110
109
Table 2
Structural parameters of Co-based samples calculated from EXAFS analysis
Sample
RCo–Co
Coordination
DW factor
(Å)
number (N)
σ (nm)
Co–B amorphous alloy
Crystallized Co–B
Raney Co
2.42
2.55
2.51
8.7
10.6
9.9
0.0248
0.0212
0.0226
Scheme 2. A plausible model of CMA adsorption and hydrogenation on Co–B amor-
phous alloy.
Fig. 7. Hydrogen TPD curves of (a) Co–B-0, and (b) Co–B-0.25 samples.
Table 1 also revealed that Co–B-0.25 exhibited much higher RHm
and shorter time to obtain optimum CMO yield than Co–B-0. This
may be attributed to both the larger metallic area (SCo) and the
enhanced intrinsic activity (RSH). As demonstrated by the XPS re-
sults, Co–B-0.25 has a higher B content than Co–B-0, which can
account for the higher RSH from Co–B-0.25 compared with Co–B-0.
On one hand, the Co in Co–B-0.25 was more electron-enriched
than Co–B-0, due to the electron back-donation from B [29], which
resulted in enhanced catalytic activity as discussed above. Co–B-
0.25’s greater CMO yield made it more selective than Co–B-0, due
to the uniform Co–B nanoparticles [14]. The H2-TPD curves (Fig. 7)
reveal four desorption peaks at around 493, 593, 640, and 822 K
obtained for Co–B-0, but only a single intense peak at 593 K and
a small shoulder peak at 443 K for Co–B-0.25. Those profiles sug-
gest that Co–B-0.25 contained uniform Co active sites and that the
hydrogen adsorption was stronger on Co–B-0.25 than on Co–B-0.
The hydrogen atoms strongly adsorbed on Co active sites were
favorable for the competitive hydrogenation of C=O against C=C
coexisting in CMA molecule [41]. Based on the RmH and reaction
time data presented in Table 1, Co–B-0.25 obviously was more ac-
tive than Co–B-50-30. This also can be attributed to the higher SCo
and the enhanced intrinsic activity (see RSH values). Meanwhile,
Co–B-0.25 was slightly more selective to CMO than Co–B-50-30
(see the CMO yields in Table 1), which also can be attributed to
the higher B content in Co–B-0.25.
Table 1 shows the effect of cyclohexane content on the cat-
alytic performance of Co–B samples. The maximum catalytic ac-
tivity was measured from Co–B-0.25, apparently arising from the
SCo, because the RSH remained almost unchanged regardless of
the cyclohexane content. In accordance with the maximum CMO
yields listed in Table 1, Co–B-0.25 was more selective toward CMO
than Co–B-0.1 and Co–B-0.2, possibly due to the greater B-content
and greater dispersion of the Co–B nanoparticles in the Co–B-0.25
sample. Abrupt decreases in both activity and the selectivity were
observed for the Co–B-0.3 sample, likely due to the significant ag-
glomeration of Co–B nanoparticles.
3.2.2. Electronic effect
Blyholder et al. [36] confirmed that a metal–oxygen bond,
rather than a metal–carbon bond, was responsible for the bond-
ing between the carbonyl group and Co metal. Noller et al. [37]
also confirmed that the C=O hydrogenation started with the attack
−
of H on the C atom. The aforementioned XPS spectra demon-
strated the strong electronic interaction between Co and B in the
Co–B amorphous alloy, making Co electron-enriched but making
B electron-deficient. On one hand, the higher electron density on
−
Co active sites could facilitate the formation of H species [37],
possibly promoting CMA hydrogenation. On the other hand, the
electron-enriched Co active sites were favorable for activating the
adsorbed C=O group through an electron back-donation from
dx2–y2 orbital of Co to the πC∗=O antibonding orbital of the C=O
group, leading to the enhanced hydrogenation activity [38]. Mean-
while, the electron-enriched Co active sites could effectively inhibit
the adsorption of C=C bond due to the presence of a benzene ring,
which also is electron-enriched, which could retard the hydrogena-
tion of C=C bond, leading to the enhanced selectivity to CMO.
Furthermore, the electron-deficient B could strengthen the adsorp-
tion for C=O group through a side-bond interaction (Scheme 2),
activating the C=O bond toward hydrogenation [39]. But no elec-
tron donation-acceptance was present in the Co-based crystalline
catalysts, which could account for the lower activity and selectiv-
ity to CMO compared with the corresponding Co–B amorphous
alloy catalysts [40]. The XPS spectra revealed that all Co species
were present in the metallic state. Nevertheless, both the elemen-
tal B and the oxidized B were detected in all of the as-prepared
Co–B samples. Thus, the presence of oxygen is related to boron ox-
ide.
To investigate the effect of boron oxide on the catalytic prop-
erties of Co active sites, we prepared pure Co and Co/B2O3 cata-
lysts by chemical reduction of Co2+ and Co2+/B2O3 precursor with
NH2NH2 alkaline solution at 363 K and conducted CMA hydro-
genation over these catalysts under the same reaction conditions
as specified in Table 1. The control experiments reveal that pure
Co and Co/B2O3 catalysts had similar hydrogenation processes and
catalytic properties. After reaction for 12 h, the maximum CMO
yield was obtained as 25.6% and 27.3%, respectively. Apparently, all
of the Co–B amorphous alloy catalysts had much better catalytic
properties than the pure Co and Co/B2O3 catalysts. Therefore, the
influence of boron oxide on the catalytic properties of Co–B amor-
phous was insignificant compared with the B alloying with Co.
Fig. 8 shows the recycling tests using the Co–B-0 and Co–B-
0.25 samples. Significant losses in activity (14%) and CMO selectiv-
ity (11%) were observed for Co–B-0 after four recyclings, whereas
Co–B-0.25 exhibited a longer lifetime, with activity and selectiv-
ity decreasing by only 12% and 3% after seven recyclings. The FTIR
spectra of the Co–B-0.25 sample show that the absorbance peaks
indicative of polyethylene glycol molecular disappeared completely
after being washed, as described in Section 2. Thus, the possible
degradation of polyethylene glycol also can be ruled out. Moreover,

110
H. Li et al. / Journal of Catalysis 259 (2008) 104–110
amorphous alloys with controllable size. The nanoparticles were
prepared mainly through chemical reduction of metallic ions with
−
BH4 in the confined voids formed by oil droplets in an oil-in-water
microemulsion system. The Co–B sample thus produced demon-
strated superior catalytic performance to the regular Co–B during
liquid-phase CMA hydrogenation. Our findings demonstrate the ad-
vantages of this approach.
Acknowledgment
This work was supported by the National Natural Science Foun-
dation of China (20603023), the 863 Project (2007AA03Z339), the
Preliminary 973 Project (2005CCA01100), and Shanghai Govern-
ment (05QMX1442, T0402, 06JC14093, 07dz22303).
References
Fig. 8. CMA conversions and CMO selectivity on Co–B-0 and Co–B-0.25 samples as
a function of recycling runs. Reaction conditions are given in Table 1.
[1] T.B.L.W. Marinelli, S. Nabuurs, V. Ponec, J. Catal. 151 (1995) 431.
[2] F. Delbecq, P. Sautet, J. Catal. 220 (2003) 115.
[3] Á. Zsigmond, I. Balatoni, K. Bogár, F. Notheisz, F. Joó, J. Catal. 227 (2004)
428.
[4] S.I. Fujita, Y. Sano, B.M. Bhanage, M. Arai, J. Catal. 225 (2004) 95.
[5] A. Ghosh, R. Kumar, Microporous Mesoporous Mater. 87 (2005) 33.
[6] X.Q. Wang, R.Y. Saleh, U.S. Ozkan, J. Catal. 231 (2005) 20.
[7] P. Gallezot, D. Richard, Catal. Rev.-Sci. Eng. 40 (1998) 81.
[8] K. Klement, R.H. Willens, P. Duwez, Nature 187 (1960) 869.
[9] J. Van Wonterghem, S. Mørup, C.J.W. Koch, S.W. Charles, S. Well, Nature 322
(1986) 622.
[10] A. Molnar, G.V. Smith, M. Bartok, Adv. Catal. 36 (1989) 329.
[11] A. Baiker, Faraday Discuss. Chem. Soc. 87 (1989) 239.
[12] Y. Chen, Catal. Today 44 (1998) 3.
[13] J.F. Deng, H.X. Li, W.J. Wang, Catal. Today 51 (1999) 113.
[14] H.X. Li, H. Li, J. Zhang, W.L. Dai, M.H. Qiao, J. Catal. 246 (2007) 301.
[15] H. Bonnemann, R.M. Richards, Eur. J. Inorg. Chem. (2001) 2455.
[16] M.P. Pileni, J. Phys. Chem. 97 (1993) 6961.
[17] M.J. Schwuger, K. Stickdorn, R. Schomäcker, Chem. Rev. 95 (1995) 849.
[18] S. Eriksson, U. Nylén, S. Rojas, M. Boutonnet, Appl. Catal. A 265 (2004) 207.
[19] S.J. Chiang, B.J. Liaw, Y.Z. Chen, Appl. Catal. A 319 (2007) 144.
[20] W.Z. Zhang, X.L. Qiao, J.G. Chen, Colloids Surf. A 299 (2007) 22.
[21] H.X. Li, Y.D. Wu, H.S. Luo, M.H. Wang, Y.P. Xu, J. Catal. 214 (2003) 15.
[22] J.J.F. Scholten, A.P. Pijers, A.M.L. Hustings, Catal. Rev.-Sci. Eng. 27 (1985) 151.
[23] H.X. Li, X.F. Chen, Y.P. Xu, Appl. Catal. A 225 (2002) 117.
[24] H.X. Li, H. Li, W.L. Dai, M.H. Qiao, Appl. Catal. A 238 (2003) 119.
[25] A. Yokoyama, H. Komiyama, H. Inoue, T. Masumoto, H.M. Kimura, J. Catal. 68
(1981) 355.
[26] K.S. Martens, J.A. Parton, R. Vercruysse, K. Jacobs, P.A. Maier, Catal. Lett. 38
(1996) 209.
[27] H.X. Li, W.J. Wang, H. Li, J.F. Deng, J. Catal. 194 (2000) 211.
[28] N.S. McIntyre, M.G. Cook, Anal. Chem. 47 (1975) 2208.
[29] H. Li, H.X. Li, W.L. Dai, W.J. Wang, Z.G. Fang, J.F. Deng, Appl. Surf. Sci. 152 (1999)
25.
[30] Z.G. Fang, B.R. Shen, K.N. Fan, Chinese J. Chem. Phys. 15 (2002) 17.
[31] W.L. Dai, H.X. Li, Y. Cao, M.H. Qiao, K.N. Fan, J.F. Deng, Langmuir 18 (2002)
9605.
[32] Y. Nitta, Y. Hiramatsu, T. Imanaka, J. Catal. 126 (1990) 235.
[33] C. Ando, H. Kurokawa, H. Miura, Appl. Catal. A 185 (1999) L181.
[34] B. Liu, L. Lu, T. Cai, K. Iwatani, Appl. Catal. A 180 (1999) 105.
[35] G.D. Angel, J.L. Benitez, J. Mol. Catal. 94 (1994) 409.
[36] G. Blyholder, D. Shihabi, J. Catal. 46 (1977) 91.
Fig. 9. XRD patterns of (a) Co–B-0 after recycling for 4 times, and (b) Co–B-0.25
after running for 7 times.
ICP analysis reveals that the leaching of Co species was <1.0 ppm,
which can rule out the loss of active phases and the advanced re-
duction of the catalysts. Apparently, partial crystallization occurred
for these spent catalysts (Fig. 9), being responsible for the deacti-
vation. On one hand, the crystallization resulted in the decrease of
SCo due to agglomeration of particles [27]; on the other hand, it
resulted in transformation from the amorphous structure to the
crystalline structure, which was unfavorable for the hydrogena-
tion activity. Meanwhile, the electron density in those electron-
enriched Co active sites may have decreased gradually during the
crystallization process due to the decomposition of Co–B alloys,
leading to the lower CMO selectivity. Co–B-0.25 exhibited better
durability than Co–B-0 because of its higher thermal stability, as
confirmed by the aforementioned DSC analysis (Fig. 4).
[37] H. Noller, W.M. Lin, J. Catal. 85 (1984) 25.
[38] E. Boellaard, R.J. Vreeburg, O.L.J. Gijzeman, J.W. Geus, J. Mol. Catal. 92 (1994)
299.
[39] Y.Z. Chen, S.W. Wei, K.J. Wu, Appl. Catal. A 99 (1993) 85.
[40] Y. Nitta, Y. Hiramatsu, T. Imanaka, Chem. Express 4 (1989) 281.
[41] H.R. Hu, M.H. Qiao, S. Wang, K.N. Fan, H.X. Li, B.N. Zong, X.X. Zhang,
J. Catal. 221 (2004) 612.
4. Conclusion
In summary, the strategy reported herein provides a facile,
one-step, and potentially general procedure for the synthesis of