Mixed oxides supported low-nickel formulations for the direct amination of aliphatic alcohols with ammonia

Journal of Catalysis 356 (2017) 133–146

Contents lists available at ScienceDirect

Journal of Catalysis

journal homepage: www.elsevier.com/locate/jcat

Mixed oxides supported low-nickel formulations for the direct

amination of aliphatic alcohols with ammonia

A. Tomer ^a^,^b, Z. Yan , A. Ponchel , M. Pera-Titus ^b^,

b

a,

⇑

a

Univ. Artois, CNRS, Centrale Lille, ENSCL, Univ. Lille, UMR 8181, Unité de Catalyse et de Chimie du Solide (UCCS), F-62300 Lens, France

Eco-Efﬁcient Products and Processes Laboratory (E2P2L), UMI 3464 CNRS – Solvay, 3966 Jin Du Road, Xin Zhuang Ind. Zone, 201108 Shanghai, China

b

a r t i c l e i n f o

a b s t r a c t

Article history:

The present study focuses on the selective synthesis of primary amines from aliphatic alcohols

and ammonia using alumina-ceria supported nickel formulations based on very low nickel loading

Received 14 June 2017

Revised 27 July 2017

Accepted 15 August 2017

(ꢀ2 wt%) and without any additive or external H

2

supply. The effect of the catalyst preparation methods

and modes of nickel impregnation were studied in detail and comprehensively characterized. The best

formulation afforded 80% n-octanol conversion with 78% selectivity to n-octylamine at optimized reac-

tion conditions, which were far better than control catalysts and benchmark Ni-alumina formulations

relying on high Ni loadings. The enhanced activities of the alumina-ceria supported nickel catalysts

was attributed to three combined effects: (1) a higher reducibility of surface nickel oxide species, (2)

the genesis of very small and homogeneously distributed nickel nanoparticles (2–3 nm), and (3) a strong

decline in the formation of nickel aluminates. Furthermore, unlike benchmark Ni catalysts, these formu-

lations afforded a higher resistance to leaching.

Keywords:

n-Octanol

Ammonia

Direct amination

CeO

2 2 3

-Al O

Nickel

1

. Introduction

2

hazardous reagents, consume H stoichiometrically, and/or pro-

duce signiﬁcant amount of waste (e.g., salts). As an alternative,

the alkylation of amines with (bio)alcohols appears as an eco-

efﬁcient method for amine production as water is generated as

the main byproduct [12].

The most extended catalysts for the direct synthesis of amines

from alcohols rely on homogeneous Ru and Ir complexes operating

Synthetic amines are extensively used as intermediates for the

production of pharmaceuticals, agrochemicals, detergents, poly-

mers, varnishes and dyes [1,2]. Alkylamines constitute a particular

family of amines derived from fatty acids, oleﬁns or alcohols issued

either from natural sources (fats, oils) or from petrochemical raw

materials. The global fatty amines market was valued at US$ 1,7

Mn in 2014 and is estimated to grow to US$ 2,2 Mn by 2020 mainly

driven by the Asia-Paciﬁc region excluding Japan (APEJ), account-

ing for more than 25% of the revenue share in 2014 [3]. The world

market of alkylamines is segmented in 6 sectors: (1) water treat-

ment, (2) agrochemistry, (3) oilﬁeld, (4) asphalt additives, (5)

anti-caking, and (6) others (personal care, mining, fabric softening,

paints and coatings) [4,5]. Water treatment constitutes the largest

market with more than 29% of the global revenue (2014), and it is

anticipated to grow by 4.7% until 2020 [3].

via the borrowing H

external H supply [13–18]. In parallel, heterogeneous catalysts

based on Raney Ni [19,20], and Ni [21–25], Cu [26–29], NiCu

[30–32], NiCuFeO [33] and NiCuZn [34] nanoparticles supported

2 2

or H auto-transfer mechanism without

2

x

over alkaline or amphoteric oxides have shown high versatility

for the additive-free alkylation of amines and ammonia with aro-

matic and aliphatic alcohols. In particular, Shimizu and co-

2 3

workers recently showed that 10 wt%Ni/h-Al O could afford high

yields (70–96%) in the reaction of primary and secondary alcohols

with excess ammonia at 160 °C for 13–72 h [23]. The major short-

coming of non-noble based catalysts is often ascribed to the use of

large metal contents (most often >15 wt%), as well as to their

heterogeneous particle size distributions and low metal dispersion,

which impacts their activity for amination. High-Ni loaded formu-

lations may result in the formation of a NiO crystalline phase on

the support, resulting into large Ni particles and accordingly low

Ni dispersion during the reduction process. Furthermore, high-Ni

loaded catalysts are prone to Ni leaching upon exposure to

Various methods are available for synthesizing alkylamines,

including alkyl halide amination, reductive amination of ketones

or aldehydes, hydroamination of oleﬁns, hydroaminomethylation,

and nitrile hydrogenation [6–11]. However, most of these methods

suffer from poor selectivity to the desired amines, use toxic or

⇑

Corresponding authors.

E-mail addresses: anne.ponchel@univ-artois.fr (A. Ponchel), marc.pera-titus-

ext@solvay.com (M. Pera-Titus).

https://doi.org/10.1016/j.jcat.2017.08.015

0

1

34

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

ammonia and polar solvents, thereby affecting not only the catalytic

activity, but also favoring the contamination of the amine product by

Ni. In this context, the development of eco-efﬁcient processes with

stable and selective heterogeneous catalysts minimizing the Ni con-

tent for the direct amination of alcohols is hour of need.

aluminum mixed oxides (Ce-Al) (used as supports) using two dif-

ferent methods, whereas the second step encompassed Ni impreg-

nation (2 wt%) over the mixed oxides either in the dried or calcined

forms.

Ceria is a well-known promoter that can modify the structural

or electronic properties of catalysts in many industrial reactions

such as automotive exhaust gas conversion (i.e. 3-way catalysts)

2.2.1. Ce-Al mixed oxides prepared by wet impregnation (WI)

2 3

In a typical preparation, 5 g of c-Al O suspended in 70 mL of

deionized water in a two-neck round bottom ﬂask were thermally

treated at 145 °C for 1 h. Subsequently, 20 mL of an aqueous solu-

tion of cerium nitrate (2.65 g) were added for 20 min at a constant

stirring speed of 600 rpm using a dropping funnel. The mixture

was aged for 2 h, cooled down to room temperature and the excess

water was slowly removed at 60 °C using a rotary evaporator. The

ﬁnal solid (yellowish) was dried in an oven overnight at 120 °C and

further calcined at 500 °C for 6 h using a heating rate of 3 °C min

in a mufﬂe furnace under static air. The solid prepared by wet

impregnation is hereinafter denoted as Ce-Al_WI.

[

2

35–37], methane reforming with CO and/or with steam, and

water-gas shift reactions [38–42]. In particular, ceria is well known

to enhance the thermal stability of alumina [43,44], reduce the Ni

particle size and mitigate coke formation and metal sintering in Ni/

2 3

c-Al O catalysts, especially for CO methanation and steam-

reforming reactions [39,40,42,45,46]. In parallel, several patents

have reported the amination of aliphatic and ethoxylated alcohols

relying mostly on Ni, Cu or Co, either in bulk form or supported

over single and mixed oxides containing rare-earth oxides [47–

ꢂ1

5

2]. In particular, a process patented by Imperial Chemical Indus-

tries (UK) describes the amination of ethanol in the presence of a

catalyst comprising NiO-CoO-Al -CeO prepared by co-

2.2.2. Ce-Al mixed oxides by the co-precipitation (PPT)

2

O

3

2

A series of Ce-Al oxides were also synthesized by a co-

precipitation method as described elsewhere [59]. In a typical

preparation, 5 g of c-Al O suspended in 50 mL of deionized water

2 3

precipitation in the presence of sodium carbonate [48]. The cata-

lyst afforded 98% ethanol conversion to ethylamine, diethylamine

and triethylamine with 43%, 44% and 13% yields, respectively, with

the simultaneous formation of high boiling byproducts. A patent by

BASF SE disclosed the use of a catalyst based on Cu supported over

La O -Al O support for the liquid-phase synthesis of secondary

2 3 2 3

amines from primary and secondary alcohols such as ethanol, iso-

propanol and cyclohexanol and primary amines [52].

in a two-neck round bottom ﬂask were thermally treated at 80 °C

for 1 h. A 20-mL solution of cerium nitrate was added to the above

solution and stirred for 5 min. Subsequently, a 25-mL solution of

NaOH and citric acid (NaOH/citric acid molar ratio = 1) was added

during 20 min using a dropping funnel and the solution was fur-

ther aged for 2.5 h. Finally, the solid mixture was cooled down to

room temperature, vacuum ﬁltered and thoroughly washed until

neutral pH using ca. 2 L deionized water. The as-obtained solid

(yellowish) was dried overnight at 120 °C and further calcined at

To the best of our knowledge, no report is available in neither

the open nor the patent literature on low-Ni formulations (ꢀ2 wt

%

Ni) for the direct amination of alcohols, especially over alumina.

ꢂ1

This is probably due to the fact that low-Ni formulations are poorly

active for amination due to the formation of inactive surface Ni

aluminates even when subjected to low calcination temperatures

500 °C for 6 h in a mufﬂe furnace using a heating rate of 3 °C min

under static air. The solid prepared by co-precipitation is here-

inafter denoted as Ce-Al_PPT.

[

53,54]. Ni-aluminates are also favored at higher Ni loadings (2–

5

wt% Ni) at a calcination temperature >550 °C by promoting

2.2.3. Ni impregnation over mixed oxides by the IWI method

2+

solid-state diffusion of Ni cations into the alumina lattice [55–

8]. Here we report for the ﬁrst time a cooperative effect between

alumina and ceria in 2 wt.%Ni/CeO -Al catalysts affording high

Nickel (2 wt% nominal loading) was impregnated by Incipient

Wetness Impregnation (IWI) over the prepared mixed oxides (Ce-

Al_WI and Ce-Al_PPT). Before impregnating the nickel nitrate solu-

tion, the dried mixed oxides were divided into two halves. On the

one hand, one half of the dried mixed oxide was directly impreg-

nated with the nickel nitrate solution and then calcined at 500 °C

for 6 h in a mufﬂe furnace. On the other hand, the second half of

the dried mixed oxide was subjected to a ﬁrst calcination step at

5

2

2 3

O

activity and selectivity to primary amines in the liquid-phase

direct amination reaction of n-octanol with ammonia.

2

. Experimental

2.1. Materials

5

00 °C for 6 h before impregnation and then calcined at similar

conditions. Two routes for Ni impregnation were considered,

namely after (A) and before (B) calcination. The ﬁnal samples dis-

played different characteristic colors regardless of the support

2

c

-Al

HSA5, 250 m /g), were used as supports for catalyst synthesis. Cer-

ium nitrate hexahydrate (Ce(NO O, >99 wt%) and nickel

ꢁ6H

nitrate hexahydrate (Ni(NO O, >99 wt%), both supplied by

ꢁ6H

2 3 2

O (Puralox Sasol Scca-5/170, 154 m /g), CeO (Solvay

2

3

)

2

(

Fig. S1).

Along with the above-mentioned catalysts, control monometal-

lic catalysts were also prepared, namely 2Ni/Al , 8Ni/Al and

Ni/CeO -HS_300 (CeO _HS pre-calcined at 300 °C before Ni

3

)

2

Sigma-Aldrich, were used as precursors for the synthesis of mixed

oxides and for Ni impregnation, respectively. NaOH (>98 wt%) and

citric acid monohydrate (>99 wt%) were procured from J&K. Ani-

line, benzyl alcohol, n-octanol and ammonia, all supplied by J&K

2

O

3

2 3

O

2

impregnation). All the catalysts were prepared by the IWI method

using the nickel nitrate solution and then calcined at 400 °C for 2 h

in a mufﬂe furnace.

(

99.5% purity), were used in the catalytic tests. O-xylene (J&K, pur-

ity 99.5%) was used as solvent in the n-octanol/ammonia amina-

tion tests. N-benzylideneaniline, N-benzylaniline, N,N-dibenzyl-

aniline, benzonitrile, toluene, benzene, n-octylamine, di-n-

octylamine, trioctylamine and octanenitrile standards for GC cali-

bration were all purchased from J&K (purity 99.5%). All the reac-

tants were used as received without further puriﬁcation.

2

.3. Catalyst characterization

The bulk metal composition of the calcined catalysts was mea-

sured using a Varian Inductively Coupled Plasma-Optical Emission

Spectrometry (ICP-OES) available at the REALCAT platform at UCCS

Lille. Before the analyses, the dried and ground sample (10 mg) was

2.2. Catalyst synthesis

dissolved in 1.5 mL of concentrated aqua regia and 250 lL of HF

solution. The solutions were heated to 50 °C and stirred for 24 h.

The preparation of Ni-supported mixed oxides was divided into

The speciﬁc surface area and pore volume of the different cata-

two steps. The ﬁrst step focused on the synthesis of cerium-

2

lysts was measured from N adsorption/desorption isotherms at

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

135

ꢂ1

ꢂ196 °C using a Micromeritics ASAP 2010 Surface Area Analyzer.

a ﬂow rate of 75 mL(STP) min . Approximately, 10 mg of sample

was heated in an open Pt crucible.

The surface areas were calculated by the Brunauer-Emmett-Teller

(

BET) method in the relative pressure range 0.05 < P/P

0

< 0.25,

The morphology and local composition of the different catalysts

was characterized ﬁrst by SEM/EDS on a Zeiss EVO-18 microscope.

Subsequently, the catalysts were inspected by STEM/EDS using a

200 kV Tecnai F20 microscope equipped with a FEG electron gun,

a STEM unit and an EDAX Optima T60 SDD-EDS spectrometer.

The images were analyzed using EDAX Team microanalysis soft-

ware. Before analysis, the solid powder was directly dispersed over

the holey carbon Cu 400 mesh grid (Agar, Ref S147-4).

The surface composition of the different catalysts was analyzed

by X-ray photoelectron spectroscopy (XPS) using a Kratos Axis

Ultra DLD apparatus equipped with a hemispherical analyzer and

a delay line detector. The spectra were recorded using an Al

monochromated X-ray source (10 kV, 15 mA) with a pass energy

of 40 eV (0.1 eV/step) for high resolution spectra, and a pass energy

of 160 eV (1 eV/step) for survey spectrum in hybrid mode and slot

lens mode, respectively. The C1 s binding energy (285.0 eV) was

used as an internal reference. Prior to the measurements, the sam-

while the pore volumes were measured at P/P = 0.99. The

0

Barrer-Joyner-Halenda (BJH) method was used for measuring the

interparticle pore size distributions. Prior to the measurements,

the catalysts were degassed overnight at 100 °C.

The phases present in the different catalysts were analyzed by

powder X-ray diffraction (PXRD). The PXRD patterns were recorded

on a Bruker D8 Advance diffractometer in Bragg-Brentano geome-

try equipped with a copper anode (k = 1.5418 Å) and a 1D PSD

Lynxeye detector. The spectra were collected in the 2h range 8–

8

0° with a step size of 0.02°. The patterns were indexed using

the Joint Committee on Powder Diffraction (JCPDS) database and

interpreted using MDI JADE 5.0 software. The Scherrer equation

was used to estimate the average size of the oxide particles from

the XRD line broadening.

The Diffuse Reﬂectance UV–Vis (DRUV–Vis) spectra of the cal-

cined catalysts were recorded in diffuse reﬂectance mode on a

Lambda 650 Perkin-Elmer spectrophotometer equipped with

ples were reduced in a pre-treatment chamber at 580 °C for 30 min

ꢂ1

6

0 mm integrating sphere. BaSO

4

was used as standard. The

under a 10% H

2

ﬂow (50 mL(STP) min ). Despite the intrinsic com-

Schuster-Kubelka-Munk (SKM) absorption function, expressed by

plexity of estimating the reduction level of ceria from the different

components of the XPS Ce 3d core-level spectrum [60], we used the

following commonly accepted expression [61]

2

F(R) = (1 ꢂ R

where R is the reﬂectance of a thick solid layer, and K and S indi-

cate the absorption and diffusion coefﬁcients, respectively.

The temperature-programmed reduction (H -TPR) proﬁles were

measured on Micromeritics AutoChem 2920 instrument

equipped with a thermal conductivity detector (TCD) and a cold

trap before the detector. The H -TPR proﬁles were recorded by

1 1

) /2 R = K/S, was applied for data interpretation,

1

0

v þ u

3

þ

Ce ð%Þ ¼

ð1Þ

2

R

ðv þ uÞ

a

where

v and u are the spin-orbit components of the 3p₅_/₂and 3 p₃_/₂

2

states of the Ce 3d core level.

reducing the the catalysts (55 mg) under 10%H

2

-Ar ﬂow

ꢂ1

(

40 mL(STP) min ) in the temperature range 30–1000 °C using a

2.4. Catalytic activity measurements

ꢂ1

heating rate of 10 °C min

.

The CO-TPD measurements were carried out using

Micromeritics AutoChem 2920 instrument. Brieﬂy, a known quan-

a

The catalytic activity of the different Ni catalysts was assessed

ﬁrst in the model amination reaction of benzyl alcohol with aniline

(Scheme 1). The reaction was performed using 2 mmol of aniline,

6 mmol of benzyl alcohol and 65 mg of pre-reduced catalyst in

Biotage microwave vials under atmospheric Ar at 160 °C for 3 h

and stirring speed of 600 rpm. In a second reaction, the amination

of n-octanol with ammonia was conducted in a 30-mL stainless

steel autoclave equipped with a pressure gauge and a safety valve

(Scheme 2). In a given experiment, the reactor was charged with n-

octanol (1.3 mmol) and 65 mg of pre-reduced catalyst. The reactor

was sealed and evacuated by applying vacuum followed by charg-

ing ammonia (7 bar). The reactor was then placed on a hot plate

equipped with a magnetic stirrer for different reaction time and

temperatures.

tity of calcined catalyst (200 mg) was reduced under a H

2

ﬂow

ꢂ1

(

1

40 mL(STP) min ) at 580 °C for 30 min using a heating rate of

ꢂ1

0 °C min and cooled down to 40 °C. Subsequently, a 3% CO-He

ꢂ1

ﬂow (40 mL(STP) min ) was passed through the sample for

2

0 min. After this period, the CO excess was removed from the

ꢂ1

sample under a He ﬂow (40 mL(STP) min ) and the CO-TPD proﬁle

was measured by monitoring the CO desorption from 40 °C to

6

00 °C under He using a TCD detector. The CO uptake allowed

the measurement of the metal dispersion (%D), the metal surface

area per gram of catalyst and per gram of metal (A ), and the aver-

age particle size of metal (S ).

m

The thermogravimetric analyses (TGA) were conducted using a

TA Instruments SDT 2960 apparatus equipped with a ﬂow gas sys-

The reactants (aniline, benzyl alcohol and n-octanol) and the

expected products for both reactions, i.e. N-benzylideneaniline

(BZIA), N-benzylaniline (BZA), N,N-dibenzylaniline (DBZA),

tem. The catalyst was treated from room temperature to 800 °C

ꢂ1

using a heating rate of 5 °C min under a 20% O

2

/He mixture with

Scheme 1. Potential amination products in the reaction between benzyl alcohol with aniline.

Scheme 2. Potential amination products in the reaction between n-octanol with ammonia.

1

36

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

n-octylamine (OA), di-n-octylamine (DOA), trioctylamine (TOA),

octanenitrile (ON) and n-octanol were analyzed and quantiﬁed

using an Agilent 7890 GC equipped with a HP-5 capillary column

with 5 wt% phenyl groups using biphenyl as internal standard.

Mass balances were accurate to within 5% for all the catalytic tests.

The conversion of the limiting reactant (LR: aniline and

n-octanol) and the selectivity and yield to the N-containing prod-

ucts were deﬁned as follows:

in more uniform and narrower pore size distributions, whereas

the catalysts synthesized by the precipitation method, i.e. Ce-

Al_PPT (Fig. S3, proﬁle d), display broader distributions and slightly

higher surface areas and pore volumes. Ce-Al_WI shows a lower

speciﬁc surface, pore volume and mean pore size compared to

the parent

suggesting a partial pore blockage by ceria particles grown over

-alumina. Opposing this observation, Ce-Al_PPT shows compara-

ble textural properties as in the case of -alumina, which can be

explained by a higher dispersion of ceria.

2 3 2

c-Al O and CeO -HS_300 (Table 1, entries 1, 3 and 5),

c

n

LR

Conversion ð%Þ ¼ 1 ꢂ

Selectivity ð%Þ ¼

ð2Þ

ð3Þ

ð4Þ

0

LR

The Ni impregnation method (i.e. before or after calcination)

also impacts the textural properties of the ﬁnal catalysts. A slight

reduction of the speciﬁc surface area is observed for 2Ni/Ce-

n

i

0

n

ꢂ nLR

2

LR

Al_WI(A) compared to the parent support (127 m /g vs.

2

1

34 m /g), whereas it keeps almost unchanged for 2Ni/Ce-Al_WI

n

i

Yield

i

ð%Þ ¼ _n

(B) and for 2Ni/Ce-Al_PPT(A,B). In contrast, all the catalysts exhibit

a remarkable decline of the total pore volume and mean pore size

after Ni impregnation (except 2Ni/Ce-Al_WI(A)) compared to the

0

LR

Moles of formed products þ reactants

2 3 2

parent supports, 2Ni/Al O and 2Ni/CeO -HS_300 (Table 1, entry

Carbon balance ¼

ð5Þ

Moles of reactants

7

). This observation can be explained by a partial pore blockage

where n ⁰L R and nLR refer to the initial and ﬁnal mole number of the

limiting reactant, respectively, whereas n corresponds to the mole

by NiO nanoparticles grown over

impregnation.

c-alumina/ceria after

i

number of N-containing products formed. In addition to the above

stated main products, toluene was formed as side product by dehy-

droxylation of benzyl alcohol.

Finally, the turnover number (TON) at a given time was com-

puted by dividing the number of moles of OA formed by the num-

ber of moles of surface Ni

3

.1.2. XRD patterns

Fig. 1 plots the XRD patterns of the calcined catalysts at 500 °C.

In all cases, neat reﬂections belonging to

0

angles of 28.4° (11 1), 33.0° (200), 47.4° (220), 56.2° (311),

59.0° (222) and 69.3° (400) can be clearly visualized in the parent

supports (Ce-Al_WI and Ce-Al_PPT) and after Ni impregnation for

c

4

-alumina (JCPDS 10-

(JCPDS 65-3102/10-

425), CeO

2

(JCPDS 34-0394) and NiAl

2

O

339) can be identiﬁed. In particular, the ceria reﬂections at 2h

ꢀ

ꢁ

mmol OA

mmol Nisurf

n

OA

TON

^¼n

ð6Þ

Ni_s_u_r_f

2

Ni/Ce-Al_WI(A,B). In contrast, 2Ni/Ce-Al_PPT(A,B) show less

intense reﬂections, suggesting either a lower crystallinity of the

ceria phase, or a better dispersion of ceria over -alumina probably

3

. Results and discussion

c

3

.1. Characterization

due to a partial dissolution during Ni impregnation. This latter

hypothesis is supported by the higher SBET and pore volume

observed for 2Ni/Ce-Al_PPT(A,B) as well by the broader pore size

distribution after Ni impregnation.

3.1.1. Textural properties

Table 1 (entries 6–10) lists the textural properties of the differ-

ent calcined Ni/Ce-Al catalysts prepared in this study, whereas

Figs. S2 and S3 plot the corresponding N adsorption/desorption

A negative shift is observed for the Ce(111) reﬂection in

Ce-Al_WI (ꢂ0.22%), which becomes more moderate but detectable

for 2Ni/Ce-Al_WI(A) and 2Ni/Ce-Al_WI(B) (ꢂ0.10% and ꢂ0.06%,

respectively). This observation can be explained by a possible

2

isotherms at ꢂ196 °C and pore size distributions. All the catalysts

exhibit Type IV isotherms with a H2-type hysteresis loop according

to the IUPAC classiﬁcation that are reminiscent of mesoporous

materials with a porous network consisting of interparticle pores

with an ill-deﬁned shape and a wide pore size distribution [62].

The pore size distribution of the different catalysts is affected to

a certain extent by the preparation method. The catalysts synthe-

sized by wet impregnation, i.e. Ce-Al_WI (Fig. S3, proﬁle a), result

2

+

2+

incorporation of Ni with smaller ion radius (rNi = 0.72 Å) into

4

+

the surface or subsurface of ceria (rCe = 0.92 Å), but not into the

ceria lattice, to form substituted Ni-CeO [46,63–66]. The charac-

x

teristic reﬂections at 43.5° and 63.0° belonging to the NiO phase

are not observed, suggesting a high NiO dispersion over ceria-

alumina within the limits of the experimental error. Besides, the

Table 1

Main properties of the different catalysts prepared in this study.

BET (m²

g

ꢂ1

)

V

(cm³

g

ꢂ1

)^a

Pore size (A)^b

CeO

2

particle size (nm)^c

Entry

Catalyst

-Al

2Ni/Al

CeO -HS_300

2Ni/CeO2-HS_300

Ce-Al_WI

2Ni/Ce-Al_WI(B)

2Ni/Ce-Al_WI(A)

Ce-Al_PPT

2Ni/Ce-Al_PPT(B)

2Ni/Ce-Al_PPT(A)

S

g

1

2

3

4

5

6

7

8

9

c

2

O

3

150

148

236

–

134

135

127

146

148

146

0.528

0.504

0.201

–

0.407

0.374

0.382

0.505

0.461

0.463

98.8

96.5

32.3

–

93.5

90.2

94.2

100.1

99.0

97.4

–

10

–

6.5

6.4

7.0

9.0

3.0

3.6

2

O

3

d

2

1

0

a

b

c

From BJH desorption branch.

From adsorption branch.

From XRD at 2h = 28.5° (111).

d

2

CeO -HS pre-calcined at 300 °C.

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

137

γ-Al O

NiO

CeO₂

NiAl O₄

_O2- _C_e4+

2

3

2

CeO adsorpꢀon edges

2

O Ni

h

O²^-Ce³⁺

O2p Ce4f

T Ni

h

j

i

j

i

h

g

f

h

g

e

d

c

b

a

f

e

d

c

Fig. 1. XRD patterns of Ni/Ce-Al catalysts: (a)

Al_WI, (d) Ce-Al_PPT, (e) 2Ni/Ce-Al_WI(B), (f) 2Ni/Ce-Al_WI(A), (g) 2Ni/Ce-Al_PPT

2 3 2

c-Al O , (b) CeO -HS_300, (c) Ce-

(

B), (h) 2Ni/Ce-Al_PPT(A), (i) 2Ni/Al

2

O

3

and (j) 2Ni/CeO

2

-HS_300.

reﬂections belonging to NiAl

Al due to overlapping.

Table 1 lists the CeO (111) crystallite sizes estimated from the

Scherrer equation at 2h = 28.4°. The CeO crystallite size is about

.5 nm and 9.0 nm for Ce-Al_WI and Ce-Al_PPT, respectively. The

2

4

O are hardly distinguishable from c-

b

a

2 3

O

2

6

2

CeO crystallite size keeps almost unchanged after Ni impregna-

tion for 2Ni/Ce-Al_WI(B) (i.e. before calcination) with an average

size about 6.4 nm, whereas a slight increase is observed for

Fig. 2. DRUV–Vis spectra of different catalysts: (a)

Ni/Al , (d) 2Ni/CeO -HS_300, (e) Ce-Al_WI, (f) Ce-Al_PPT, (g) 2Ni/Ce-Al_WI(B),

h) 2Ni/Ce-Al_WI(A), (i) 2Ni/Ce-Al_PPT(B) and (j) 2Ni/Ce-Al_PPT(A).

2 3 2

c-Al O , (b) CeO -HS_300, (c)

2

Ni/Ce-Al_WI(A) with an average particle size about 7 nm. This

2

(

2

O

3

2

increase is probably related to partial sintering of ceria nanoparti-

cles when contacting the support with the Ni precursor solution

followed by calcination. Opposing this observation and irrespective

of the Ni impregnation mode, the Ni/Ce-Al_PPT(A,B) catalysts exhi-

bit a remarkable decrease of the ceria crystallite size down to 3 nm,

suggesting a higher dispersion of ceria upon Ni impregnation.

and higher wavelength regions (proﬁle c). The band at 380 nm

can be assigned to a d-d transition for octahedrally (O ) coordi-

h

nated Ni in the NiO lattice, while the band appearing at higher

3.1.3. DRUV-Vis

wavelength (600–650 nm) can be assigned to a d-d transition

The chemical states of the Ni-supported catalysts and the par-

for tetrahedrally coordinated (T

2Ni/CeO -HS_300 (proﬁle d) exhibits a charge transfer band at ca.

320 nm and a broader band in the range 350–450 nm that can be

assigned to a d-d transition for O -coordinated Ni and to the

h

) Ni in the Ni-aluminate spinel.

ent supports (i.e.

Al_PPT) were assessed by DRUV-Vis. Fig. 2 plots the corresponding

spectra. The -alumina support exhibits a distinct band at 280 nm

proﬁle a), while CeO -HS_300 exhibits two bands (proﬁle b): (1) a

c-Al

2

O

3

,

CeO

2

-HS_300, Ce-Al_WI and Ce-

2

c

h

(

2

inter-band transition of ceria, respectively. The adsorption edge

is observed at about 500 nm, falling into the visible range. This cat-

alyst also shows a weak broad band in the UV region (<230 nm),

2

ꢂ

4+

band centered at 320 nm corresponding to O ? Ce charge

transfer, and (2) a band near 350 nm that can be assigned to

inter-band transitions [67,68]. The position of the adsorption edge

is observed at about 400–450 nm. No adsorption band at 255 nm

2ꢂ

3+

which can be attributed to O ? Ce charge transfer. This band

is also observed for 2Ni/Ce-Al_WI(B) (proﬁle g), but with weaker

intensity and a promoted adsorption edge at 635 nm. In contrast,

the spectrum of 2Ni/Ce-Al_WI(A) (proﬁle h) is very similar to that

of the bare support (proﬁle e) with an adsorption end at about

380 nm.

2

ꢂ

3+

corresponding to

Ce-Al_WI and Ce-Al_PPT exhibit similar spectra compared to

CeO -HS_300 (proﬁles e and f, respectively), but with a blue shift

of the adsorption edge from 400 nm to 360 nm as compared to

CeO -HS_300. However, this shift is more pronounced for the for-

O

? Ce

charge transfer is observed.

2

The spectra of 2Ni/Ce-Al_WI(A,B) and 2Ni/Ce-Al_PPT(A,B) (pro-

ﬁles g, h and i, j respectively) differ from those of the parent oxides

and are strongly affected by the Ni impregnation mode. For exam-

ple, 2Ni/Ce-Al_WI(B) exhibits three bands: (1) a weak band at

mer support (Ce-Al_WI), also showing a decrease in intensity,

which can be most likely attributed to a smaller crystallite size

of ceria for Ce-Al_WI (6.5 nm) compared to Ce-Al_PPT (9.0 nm)

2

ꢂ

3+

(

Table 1). This hypothesis is supported by the study reported by

Bensalem et al. [60,69], where an absorbance occurring at

375 nm may imply the presence of ﬁner ceria crystallites and

the shift can be explained by prevailing presence of

220 nm corresponding to O ? Ce charge transfer, (2) a band

at 320 nm being attributed to O ? Ce⁴⁺charge transfer, and (3)

2ꢂ

ꢀ

a broad d-d transition band in the range 400–550 nm being indica-

tive of O -coordinated Ni in NiO or due to O 2p ? Ce 4f charge

h

transfer transitions. The broadening of this band could also be

indicative of a higher density of surface defects for this catalyst

[70,71]. Notably, the latter band is only observed for 2Ni/Ce-

Al_WI(B), but vanishes from the Ni/Ce-Al_WI(B) catalysts prepared

a

4

+

4+

Ce -oxygen charge transfer occurring on lower-coordinated Ce

ions.

Concentrating our attention now to the control catalysts, the

Ni/Al O catalyst exhibits distinct absorption bands in the lower

2 3

2

1

38

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

with lower (0.5 wt%) and higher (8 wt%) Ni loadings (Fig. S4). In the

case of 2Ni/Ce-Al_WI(A), the DRUV-Vis spectrum does not show

any band in the lower wavelength region (<300 nm) and in the

range 400–450 nm, suggesting a weaker Ni-Ce interaction. Finally,

none of the Ni-impregnated catalysts (Fig. 2, proﬁles g-j) show a

band in the 600–650 nm region ascribed to Ni aluminates as in

to the reduction of two different NiO species, (2) a third band

appearing in the range 300–450 °C that can be ascribed to the

reduction of surface ceria, and (3) a fourth band centered at

T > 700 °C corresponding to the reduction of bulk ceria.

When the mixed oxides are prepared by the WI method, Ce-

Al_WI (Fig. 3, proﬁle d) exhibits a relatively different reduction

proﬁle compared to Ce-Al_PPT. Ce-Al_WI shows the presence of

two main bands centered at 450 °C and a smaller band at a temper-

ature >850 °C that can be assigned to the reduction of surface ceria

2 3

the case of 2Ni/Al O , suggesting a potential role of ceria as barrier

between Ni and alumina.

4

+

3+

[

3

72], and to Ce ? Ce reduction into CeAlO [44,80,81]. No evi-

3

.1.4. Reducibility of the catalysts

Fig. 3 plots the H -TPR proﬁle of 2Ni/Ce-Al_WI and 2Ni/Ce-

Al_PPT until 1000 °C, as well as the corresponding proﬁles for the

Ce-Al_WI and Ce-Al_PPT supports, and the benchmark 2Ni/CeO

HS_300 and 2Ni/Al . Ce-Al_PPT (proﬁle a) displays two bands

dence of bulk ceria reduction is observed. Upon Ni impregnation,

the 2Ni/Ce-Al_WI catalysts (proﬁles e and f) display ﬁve reduction

bands: (1) a small band centered at 275 °C that can be ascribed to

2

-

2 3

Ni O reduction or to the reduction of oxygen species generated

2 3

O

from a Ni-O-Ce solid solution [82–84]; (2) a band at ca. 450 °C cor-

responding to the reduction of surface ceria; (3) a broad band in

the range 550–700 °C being ascribed to the reduction of surface

with different intensities centered at 470 °C and 648 °C that can

be ascribed to the reduction of surface and bulk ceria, respectively

[

72,73]. Upon Ni impregnation, Ni/Ce-Al_PPT shows three major

bands (proﬁles b and c): (1) a ﬁrst band centered at 160–200 °C

corresponding to the reduction of Ni and to a possible reduction

Ni species (b

range 730–850 °C corresponding to the reduction of Ni aluminate

spinel species (NiAl ); and (5) a band appearing at very high

temperature (>850 °C) that can be attributed to Ce ? Ce reduc-

tion encompassing CeAlO formation. Note that the reduction of

surface Ni species occurs at lower temperature for 2Ni/Ce-Al_WI

B) than for 2Ni/Ce-Al_WI(A) (645 °C instead of 665 °C). Altogether,

1 2 2 3

and b ) interacting with c-Al O ; (4) a band in the

2 3

O

2 4

O

2+

of oxygen species generated upon dissolution of Ni into the ceria

4

+

3+

surface or subsurface [74–77]; (2) a second band at 550–650 °C

3

ascribed to the reduction of surface NiO species (b

acting with -Al [55–57,78,79]; and (3) a third band at

T > 700 °C corresponding to the reduction of bulk NiAl spinel

1 2

and b ) inter-

c

2 3

O

(

2 4

O

the results conﬁrm that the reducibility of surface Ni species

depends to an important extent on the Ni impregnation mode.

The appearance of a reduction band ascribed to CeAlO irrespec-

3

tive of the presence of Ni may inform about the particle size of

in strong interaction with surface alumina [40,45,55–57,79]. The

latter band is also present at a much higher intensity for the 2Ni/

Al

ascribed to CeAlO

Opposing these observations, 2Ni/CeO

2

O

3

control catalyst (proﬁle g). Finally, no band at T > 850 °C

species is observed for Ni/Ce-Al_PPT catalysts.

-HS_300 (proﬁle h) shows

3

CeO2ꢂx crystallites in the catalysts. As outlined by Kaufherr et al.

2

[

3

85], the formation of CeAlO depends on the reduction tempera-

a different reduction proﬁle being characterized by four major

bands: (1) two bands centered at 165 °C and 208 °C corresponding

ture, which is governed by the average particle size of ceria. As a

rule, a higher reduction temperature is required for larger ceria

particles to favor Ce diffusion into alumina to generate a solid solu-

3

tion. In our case, the reduction temperature for CeAlO formation

Ni

2

O

3

S. CeO

α-NiO

2

B. CeO

β-NiO

2

γ-NiO CeAlO

3

decreases from 870 °C for Ce-Al_WI to 865 °C for 2Ni/Ce-Al_WI

(B), while it increases to 880 °C for 2Ni/Ce-Al_WI(A). This observa-

tion can be interpreted by the presence of smaller dispersed ceria

particles in 2Ni/Ce-Al_WI(B), favoring its reducibility and accord-

ingly the formation of surface Ni species (b type).

h

357

762

165

208

2

The H -TPR study was further extended to Ni/Ce-Al_WI cata-

g

f

lysts with variable Ni loading (0.5–8 wt%) (Fig. S5). The reduction

proﬁle of a very low Ni-loaded catalyst (0.5Ni/Ce-Al_WI) is very

similar to that observed for the parent Ce-Al_WI (Fig. 3, proﬁle

d). In contrast, the catalyst displaying the highest Ni loading

3

15

480

805

2

75

(

8Ni/Ce-Al_WI) exhibits a major band in the temperature range

00–700 °C corresponding to the reduction of surface Ni (b and

type), as well as to minor bands at 747 °C and 850 °C attributed

to NiAl and CeAlO , respectively. Unlike 2Ni/Ce-Al_WI, the

reduction bands corresponding to surface Ni species and NiAl

6

65

800

880

5

1

e

b

2

O

4

3

2

645

765

2 4

O

865

d

c

shift to lower temperature at higher Ni loading, indicating weaker

Ni-alumina interactions. Interestingly, the intensity of the band at

874 °C is lower for 8Ni/Ce-Al_WI(B), which could be probably due

to a higher surface occupancy by Ni instead of ceria, and thus to a

lower degree of interaction between alumina and ceria.

450

870

230

615

7

60

b

a

3

.2. Catalytic results

2

80

620

780

.2.1. Amination of benzyl alcohol with aniline

The Ni/Ce-Al catalysts were ﬁrst tested in the model amination

648

reaction of benzyl alcohol with aniline at 160 °C for 3 h using

5 mg of catalyst pre-reduced at 580 °C for 30 min. Fig. 4 plots

the results obtained for the different catalysts, as well for the

benchmark 2Ni/Al and 2Ni/CeO _300, while a complete list is

6

2

O

3

2

provided in Table S1 including control catalysts. The parent sup-

ports (Ce-Al_WI and Ce-Al_PPT) display very low activity (ꢃ10%

aniline conversion) with formation of BZIA as major product.

Fig. 3. H

method: (a) Ce-Al_PPT, (b) 2Ni/Ce-Al_PPT(B), (c) 2Ni/Ce-Al_PPT(A), (d) Ce-Al_WI,

e) 2Ni/Ce-Al_WI(B), (f) 2Ni/Ce-Al_WI(A), (g) 2Ni/Al , and (h) 2Ni/CeO -HS_300.

2

-TPR proﬁles of Ni/Ce-Al catalysts synthesized by the wet impregnation

(

2

O

3

2

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

139

Among the Ni-impregnated catalysts, 2Ni/CeO

2

-HS_300 exhibit the

compiles the results obtained. The parent supports without metal

do not display any activity, as previously observed in the model

reaction. Among the Ni-impregnated catalysts, 2Ni/CeO -HS_300

2

shows activity with 32% n-octanol conversion and 79% of OA selec-

tivity. The Ni/Ce-Al catalysts are very active for n-octanol amina-

2 3

tion compared to 2Ni/Al O . Among the different catalysts, those

synthesized by the wet impregnation method and displaying a

high density of b-type Ni species (i.e. 2Ni/Ce-Al_WI) exhibit higher

activities. The Ni impregnation mode impacts the catalytic proper-

ties with 2Ni/Ce-Al_WI(B) impregnated just before calcination

showing a higher activity (27% n-octanol conversion, 78% OA selec-

tivity). This trend opposes to that observed for the catalysts pre-

pared by co-precipitation: while 2Ni/Ce-Al_PPT(B) displays no

activity, 2Ni/Ce-Al_PPT(A) exhibits 23% n-octanol conversion and

77% OA selectivity.

highest activity with 28% aniline conversion and 68% BZA selectiv-

ity. Among the different 2Ni/Ce-Al catalysts, 2Ni/Ce-Al_WI(B)

shows the highest activity with ꢃ70% aniline conversion and 88%

BZA selectivity. In contrast, 2Ni/Ce-Al_WI(A) displays a lower

activity (23% aniline conversion and 43% BZA selectivity) with BZIA

as major product with 56% selectivity. The opposite trend (i.e. a

higher activity) is observed for 2Ni/Ce-Al_PPT(A) synthesized by

the precipitation method and for which Ni was impregnated after

calcination.

To survey the effect of the reduction temperature on the cat-

alytic activity for amination, the best performing catalyst, i.e.

2

Ni/Ce-Al_WI(B), was reduced at different temperatures in the

range 400–800 °C and the activity for benzyl alcohol amination

was further measured (Table S1). The reduction at lower tempera-

ture (400 °C) results into a lower activity with an aniline conver-

sion of ꢃ10% towards BZIA. A similar effect is observed at very

high reduction temperatures (800 °C) with only 18% aniline con-

version and 44% BZA selectivity. Opposing these results, a reduc-

tion of temperature 580 °C affords a high aniline conversion

3.2.2.2. Effect of the catalyst loading and reaction time. The catalytic

performance of 2Ni/Ce-Al_WI (B) was further optimized by tuning

the loading and the reaction time (Fig. 6). An increase of the cata-

lyst weight from 65 mg to 235 mg (5 mol%) while keeping the reac-

tion time constant (4 h) results in an enhancement of the n-octanol

conversion to 64% with an OA selectivity of 80% (46% OA yield).

Furthermore, at a constant catalyst loading of 65 mg, an increase

of the reaction time from 4 h to 24 h also results in a remarkable

increase of the n-octanol conversion and OA selectivity to 55%

and 94% after 24 h (43% OA yield), respectively. An increase of

the catalyst loading from 65 mg to 110 mg at a reaction time of

(

70%) and BZA selectivity (88%).

The above stated activity trends as a function of the reduction

temperature can be explained on the basis of the H

of 2Ni/Ce-Al_WI(B) (Fig. 3, proﬁle e). When the catalyst is reduced

at 400 °C, only Ni species are expected to be reduced to Ni,

2

-TPR proﬁle

2 3

O

resulting in a very low activity. In contrast, when the catalyst is

reduced by 600 °C, other Ni species are expected to come into

act, encompassing mainly bulk Ni species (200–300 °C), reduced

surface ceria species (400–500 °C) and surface Ni species (500–

2

4 h results in an increase of the n-octanol conversion and OA

selectivity to 94% and 78% (60% OA yield), respectively.

6

00 °C). The presence of surface Ni species after reduction at

00 °C, probably interacting with surface ceria and alumina,

Finally, a series of Ni/Ce-Al_WI(B) catalysts with variable Ni

loading were prepared and tested for n-octanol amination with

ammonia at 24 h using 65 mg of catalyst. No activity is observed

for the catalyst with the lowest Ni loading i.e. 0.5Ni/Ce-Al_WI(B),

while 8Ni/Ce-Al_WI(B) affords almost full n-octanol conversion in

favor of DOI (47% selectivity) with an OA yield of only 20%. This

body of results clearly point out the beneﬁts of 2Ni/Ce-Al_WI(B)

over the other formulations prepared in this study for enhancing

the catalytic activity towards OA.

appears to be a main driver for enhancing the catalytic activity.

Finally, the catalyst reduced at 800 °C shows the preferential

reduction of bulk ceria and Ni aluminate species, which are not

active for catalysis. Hence, it can be concluded that a reduction

temperature about 600 °C is compulsory for favoring the formation

of surface Ni species, which can help achieving a better reducibility

and controlled Ni-Ce or Ni-Al interactions. Accordingly, this reduc-

tion temperature was used in the remainder of this study to assess

the performance of the different 2Ni/Ce-Al catalysts for n-octanol

amination with ammonia.

3

.2.2.3. Catalytic properties of 2Ni/Ce-Al_WI(B) compared to 8Ni/Al

a further step of our study, we compared the catalytic properties of

Ni/Ce-Al_WI(B) against the benchmark 8Ni/Al also prepared

2 3

O . In

2

2 3

O

3

.2.2. Amination of n-octanol with ammonia

.2.2.1. Catalytic properties of Ni/Ce-Al catalysts. Fig. 5 plots the per-

by the IWI method in the amination of n-octanol reaction with

ammonia. Interestingly, when representing the OA selectivity vs.

n-octanol conversion, an almost constant trend is observed for

2Ni/Ce-Al_WI(B), whereas a decreasing trend is observed for 8Ni/

formance of the different 2Ni/Ce-Al catalysts for n-octanol amina-

tion with ammonia at 180 °C for 4 h after pre-reduction at 580 °C,

as well as for different benchmark catalysts, whereas Table S2

2 3 2 3

Al O (Fig. 7). The best performance for 8Ni/Al O towards OA

Fig. 4. Performance of 2Ni/Ce-Al catalysts in the amination reaction of benzyl alcohol with aniline. Reaction conditions: Aniline- 2 mmol, Benzyl alcohol- 6 mmol, T- 160 °C,

Time- 3 h, Cat- 65 mg, no solvent, rpm- 600, Ar atm.

1

40

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

Fig. 5. Performance of 2Ni/Ce-Al catalysts in the amination reaction of n-octanol with ammonia. The numbers in parentheses indicate the carbon balance. Reaction

conditions: Oct- 1.3 mmol, NH - 7 bar, T- 180 °C, Time- 4 h, Cat- 65 mg, Solvent- 3 mL o-xylene, rpm- 600.

3

Fig. 6. Performance of Ni/Ce-Al_WI(B) catalysts in the amination reaction of n-octanol with ammonia. The numbers in parentheses indicate the carbon balance. Other

reaction conditions as in Fig. 5.

encompasses 83% n-octanol conversion and 73% OA selectivity

(

54% OA yield) after 4 h. An increase of the reaction time results

in a decrease of the OA selectivity in favor of the secondary amine

DOA). Moreover, when comparing the turnover number (TON) for

(

2

Ni/Ce-Al_WI(B) and the corresponding TON for 8Ni/Al

4 h, the former catalyst outstands the benchmark catalyst

2 3

O after

(

Table 2). Also noteworthy, the TON achieved for 2Ni/Ce-Al_WI

B) after 24 h compares well with the TON reported by Shimizu

et al. [23] over 10Ni/h-Al

3 h), even if the Ni loading used here is seven times smaller

1.4 mol%Ni vs. 5 mol%Ni). In other words, 2Ni/Ce-Al_WI(B) affords

a save about ꢃ75% of Ni metal for achieving a comparable OA yield

60%) due to a higher Ni dispersion in the form of b-species. In

2

O

3

catalyst (TON = ꢃ85 at 160 °C for

1

(

addition to these beneﬁts, 2Ni/Ce-Al_WI(B) affords a much better

carbon balance (95% vs. 85%).

3.3. Understanding the structure and catalytic behavior of 2Ni/Ce-

Al_WI

Fig. 7. Evolution of the OA selectivity vs. n-octanol conversion for 2Ni/Ce-Al_WI(B)

and 8Ni/Al

Fig. 5.

2 3

O as a function of the reaction time. Other reaction conditions as in

The catalytic results for the model and n-octanol amination

reactions presented above reveal an enhanced activity for the cat-

alysts prepared by wet impregnation (2Ni/Ce-Al_WI) instead of the

precipitation method (2Ni/Ce-Al_PPT). In light of these results, we

explored in more detail the structure of 2Ni/Ce-Al_WI before and

after reaction along with control catalysts.

The elemental composition of 2Ni/Ce-Al_WI(A,B) was con-

ﬁrmed by ICP-OES. The Ni and Ce loading was found to be very near

to the nominal values. As a matter of example, the Ni and Ce load-

ing estimated for both catalysts is 1.7 wt% and 15 wt%, respec-

tively, for 2Ni/Ce-Al_WI(B), and 1.8 wt% and 15 wt%, respectively,

for 2Ni/Ce-Al_WI(A). The morphology of 2Ni/Ce-Al_WI(B) was

inspected by SEM combined with EDS analysis (Figs. S6-S8,

Table S3). The SEM micrographs show the presence of spherical

2 3

c-Al O particles with a size in the range of 40–90 mm. The Ce

and Ni loadings show values at ca. 15 wt% and 1.8–2.3 wt% match-

ing the theoretical loadings measured from ICP-OES analysis. A

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

141

very homogeneous Ni distribution is observed for 2Ni/Ce-Al_WI(B)

Fig. S8).

Fig. 8 plots the XPS spectra for the Ce 3 d, Ni 2 p and O 1s core

levels for 2Ni/Ce-Al_WI(B) and 2Ni/Ce-Al_WI(A) and the corre-

sponding deconvolution after pre-reduction, whereas Table 3 sum-

marizes the surface concentrations measured for Ce, Al and Ni in

the calcined and reduced states. The deconvoluted Ce 3 d core-

level spectra could be assigned to eight spin-orbit coupling levels

(Fig. 8b and e). Moreover, 2Ni/Ce-Al_WI(A) displays an additional

small band centered at 852.0 eV that can be ascribed to Ni⁰(5%),

along with a band at ꢃ861 eV that can be assigned to a shake-up

(

2 3

satellite band of NiO or Ni O [88,89]. The absence of band corre-

sponding to reduced Ni species in 2Ni/Ce-Al_WI(B) might be

ascribed to the formation of a Ni-O-Ce solid solution on the ceria

surface or subsurface, even after pre-reduction. This hypothesis

can be supported by the low Ni/Ce molar ratio approaching 0.5

for this sample (Table 3, column 6), favoring Ni-Ce interaction

[63,90].

The XPS spectra for the O 1s core level of 2Ni/Ce-Al_WI(B)

and 2Ni/Ce-Al_WI(A) (Fig. 8c and f) show the presence of a

major band centered at 531.5 eV that agrees well with the BE

of lattice O in the alumina phase. In addition, a shoulder band

centered at 533 eV can be observed, matching the band com-

monly observed for pure ceria accounting for weakly bounded

oxygen (e.g., hydroxyl groups, vacancies) (Fig. S9b). No band is

observed at about 529–530 eV belonging to lattice O in pure

ceria, suggesting an increase by 2 eV of the BE compared to pure

ceria. Such an increase has already been described in highly dis-

corresponding to the 3d5/2

The components show binding energies (BE) at 882.7 eV (

84.1 eV (

nents show BEs at 901.3 eV (u), 903.4 eV (u ), 907.3 eV (u ) and

(

v

) and 3d3/2 (u) states (Fig. 8a and d).

),

), while the u compo-

v

0

00

000

8

v

), 888.5 eV (

v

) and 898.3 eV (

v

0

00

0

00

9

17.0 eV (u ) [61,86,87]. Noteworthy, the lack of increase of the

BE for the components in the presence of Ni compared to the

v

value commonly accepted for ceria suggests very low Ni incorpora-

tion in the ceria surface or subsurface [63–66]. Furthermore, the

0

presence of v and u components combined with the presence of

0

00

000

large bands for v and u components indicates the presence of

Ce in the surface of ceria, but without generating CeAlO [81].

3

The proportion of Ce can be estimated at ca. 40% for both 2Ni/

3

+

3

+

Ce-Al_WI(B) and 2Ni/Ce-Al_WI(A), as inferred from Eq. (1) imple-

persed ceria in CeO

to a change in the coordination environment of ceria, favoring its

interaction with Al by the formation of Ce-O-Al bonds. Fur-

thermore, the formation of O-ion vacancies in the subsurface

layer of ceria due to partial incorporation of Ni cannot be ruled

out [91–93]. Finally, when we compare the relative areas for the

2 2

-SiO composites [61], and can be attributed

mented with the integrated areas for the

v and u components in

the Ce 3d spectra (Table 3, column 9). Besides, the Ce/Al molar ratio

is higher for 2Ni/Ce-Al_WI(B) than for 2Ni/Ce-Al_WI(A) both in the

calcined (0.059 vs. 0.027) and pre-reduced catalysts (0.039 vs.

2 3

O

2

+

0

.028) (Table 3, column 5 and 8), suggesting a higher proportion

of ceria on the catalyst surface for 2Ni/Ce-Al_WI(B).

O1s bands in 2Ni/CeO

2Ni/Ce-Al_WI (A) (Fig. S9e), the area for 2Ni/ Ce-Al_WI(A) is very

similar to that of 2Ni/Al . Also, the area for the O1s band in

2 2 3

-HS_300, 2Ni/Al O , 2Ni/Ce-Al_WI(B) and

The Ni 2p3/2 spectra for the pre-reduced 2Ni/Ce-Al_WI(B) and

2

Ni/Ce-Al_WI(A) exhibit two main bands centered at 856.5 eV

2 3

O

2

+

and 856.8 eV that are indicative of the presence of Ni

2

2Ni/Ce-Al_WI(B) is closer to that of 2Ni/CeO -HS_300. These

Ce 3d

Ni 2P 3/2

(c)

O 1s

(a)

(b)

u

u’”

v’”

v

Ni (II)

u’

v’

Satellites

u”

Ni

v”

Ni (0)

2

Ni/Ce-Al_WI(B)

(f)

(d)

(e)

u

u’”

v

Ni (II)

v’”

u’

v’

u”

Satellites

Ni

v”

Ni (0)

2

Ni/Ce-Al_WI(A)

Binding energy (eV)

2

Fig. 8. XP spectra of Ni/Ce-Al_WI(B) and Ni/Ce-Al_WI(A) for the Ce 3 d (a), Ni 2p (b) and O 1s (c) core levels. The samples were pre-reduced at 580 °C for 30 min under H ﬂow

before the XPS analyses.

1

42

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

Table 2

Comparison between the performance of 2Ni/Ce-Al_WI(B) and Al

2 3

O

-supported Ni catalysts for the direct amination of n-octanol with ammonia.^a

Entry

Catalyst

%D^b

Loading (mol%)

%Octanol Conv^c

%Selectivity^d

%Yield

OA

TON^e

Carbon balance

ON

OA

DOA

TOA

DOI

1

2

3

4

6

2Ni/Ce-Al_WI(B)

35

15

16

1.4

2.3

6

5

55

82

83

97

90

2.4

7.4

10.1

5.4

–

94

78

73

28

91

2.4

11

16

66

9.3

0.6

1.6

0.7

–

0.6

2.0

–

43

60

54

25

70

85

69

54

25

83

95

96

95

96

85

2Ni/Ce-Al_WI(B)

f

8Ni/Al

2

O

3

2

3

g

10Ni/h-Al

2

O

3

–

a

b

c

d

e

f

3

Reaction conditions: Oct- 1.3 mmol, NH - 7 bar, T- 180 °C, Time- 24 h, Solvent- 3 mL o-xylene, rpm- 600.

Dispersion calculated from CO-TPD considering Ni/CO, S.F. = 2.

Estimated from GC.

Normalized selectivity.

TON: moles of OA formed per moles of surface Ni at 24 h.

Reaction time: 4 h.

3

Reaction conditions: Oct- 3 mmol, NH - 4 bar, T- 160 °C, Time- 13 h, Solvent -4 mL o-xylene (Ref. [23]).

g

Table 3

Composition of 2Ni/Ce-Al_WI catalysts measured by XPS.

Entry

Catalyst

Calcined catalysts

Ni/Ce

Reduced catalysts^a

Ni/Ce

Ni/Al

Ce/Al

Ni/Al

Ce/Al

Ce³⁺

%

1

2

3

2Ni/Al

2Ni/Ce-Al_WI(B)

2Ni/Ce-Al_WI(A)

2

O

3

–

0.046

0.045

0.044

–

0.026

0.022

0.032

–

40

0.767

1.648

0.059

0.027

0.547

1.142

0.039

0.028

a

2

Catalysts pre-reduced under at 580 °C for 30 min under H ﬂow.

observations suggest a higher oxygen contribution from CeO

the Al lattice in 2Ni/Ce-Al_WI(B).

2

to

pointing out that only a fraction of Ni is available for interacting

with CO in 2Ni/Al after ceria-doping. A higher Ni dispersion

2

O

3

2 3

O

Fig. 9 plots the CO-TPD proﬁles of 2Ni/Ce-Al_WI(B) and 2Ni/Ce-

Al_WI(A), as well as the corresponding proﬁles for the control cat-

alysts, whereas Table S4 lists the corresponding Ni dispersion,

average particle size and total amount of CO desorbed from the

support and the Ni phase. The bands in the lower temperature

region (40–150 °C) are generally attributed to CO desorption from

metallic Ni, whereas the bands in the mid-temperature region

(35%) is observed for 2Ni/Ce-Al_WI(B) with a average particle size

for Ni of ꢃ2.5 nm, whereas 2Ni/Ce-Al_WI(A) shows a Ni dispersion

of 21% with a average particle size of Ni ꢃ4 nm for Ni (Table S4,

entries 4 and 5). In addition to a higher Ni dispersion, the genesis

of very small Ni nanoparticles over 2Ni/Ce-Al_WI(B) favoring the

exposition of edges and corners can also be regarded as a driver

for the enhanced catalytic activity on this catalyst for n-octanol

amination.

A closer inspection to 2Ni/Ce-Al_WI(B) by STEM/EDS (Figs. 10

and 11 and Fig. S10) indicates a distribution of ceria nanoparticles

with a size <10 nm supported over alumina arranged in small

agglomerates. Ni appears as homogenously dispersed over alumina

(

200–400 °C) are ascribed to CO desorption from ceria. The remain-

ing high-temperature bands (T > 400 °C) are attributed to CO des-

orption as CO from Al -CeO . Noteworthy, unlike 2Ni/Ce-

Al_WI(B) and 2Ni/Ce-Al_WI(A), no band in the low-temperature

2

O

3

2

region (ꢃ80 °C) is observed for neither

2 3

c-alumina nor 2Ni/Al O ,

(Fig. 12), matching the observations pointed out above using SEM/

EDS, but being preferentially located near the ceria nanoparticles.

The Ni nanoparticles show non-uniform shape and average particle

size in range 2–3 nm, which is in fairly good agreement with the

average particle size measured by CO-TPD. This morphology con-

ﬁrms a major role of ceria in assisting Ni dispersion over alumina

during impregnation, hampering in turn the formation of

aluminates.

From the above characterization and catalytic results, we can

conclude that the enhanced catalytic activity of 2Ni/Ce-Al_WI(B)

for the production of primary amines from the direct amination

reaction of n-octanol with ammonia can be attributed to the gene-

sis of small Ni nanoparticles over alumina promoted by small ceria

nanoparticles assembled in agglomerates. Furthermore, the Ni-

ceria interaction favors the generation of readily reducible surface

Ni species, as well as the suppression of inactive Ni aluminates. A

plausible structure of 2Ni/Ce-Al_WI(B) is proposed in Fig. 13.

a

b

c

d

e

3.4. Recyclability and post-catalytic characterization of 2Ni/Ce-Al_WI

(

B) catalyst

The recyclability of the catalyst showing best performance (2Ni/

Fig. 9. CO-TPD proﬁles for Ni/Ce-Al and control catalysts synthesized by the wet

Ce-Al_WI(B)) was surveyed in two consecutive cycles. Brieﬂy, after

impregnation method: (a)

and (e) 2Ni/Ce-Al_WI(A).

2 3 2 3

c-Al O , (b) Ce-Al_WI, (c) 2Ni/Al O , (d) 2Ni/Ce-Al_WI(B)

st

the 1 run, the catalyst was separated by centrifugation, washed

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

143

a

b

c

2.4 nm

d

1

.9 nm

2.7 nm

2

.3 nm

2

.6 nm

2

.1 nm

Fig. 10. STEM micrographs of 2Ni/Ce-Al_WI(B). The sample consists of non-uniform Ni nanoparticles with a particle size in the range 2–3 nm: (a) 100 nm, (b) 100 nm, (c)

0 nm and (d) 20 nm.

2

with ethanol to remove the organic content, dried in an oven and

calcined at 500 °C for 6 h. The calcined catalyst was used in a 2

selectivity to 68% after the 2ⁿ^drun. This drop in the catalytic activ-

ity suggests a modiﬁcation of the active surface Ni species during

the reaction. However, in line with reported studies on low-Ni cat-

nd

reaction run after reduction under H

2

at 580 °C for 30 min.

st

nd

Table S5 lists the results obtained for n-octanol amination with

alysts [94], no Ni leaching during the 1 run and 2 run has been

observed in our study as conﬁrmed by ICP analysis. In the case of

cerium content, a slight decrease is observed from an initial value

ammonia at 180 °C for 24 h. The n-octanol conversion decreases

st

nd

from 55% after the 1 run to 7% after the 2 run with only a slight

decrease of the OA selectivity from 94% to 82%. A further increase

st

of 14.8 wt% to 13.8 wt% after the ﬁrst 1 run.

st

of the reduction temperature to 750 °C for 30 min after the 1 run

To rationalize the catalyst evolution during reaction, a part of

st

enhances the n-octanol conversion to 13% with decrease in the OA

the spent 2Ni/Ce-Al_WI(B) catalyst after the 1 run was calcined

a

b

e

c

1

00 nm

50 nm

d

f

50 nm

Fig. 11. EDS maps of 2Ni/Ce-Al_WI(B) during STEM analysis: (a) area of EDS map acquisition marked by white rectangle, (b) mixture, (c) O K, (d) Al K, (e) Ce L, and (f) Ni K.

1

44

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

ceria is also relevant for boosting the catalytic activity. The decon-

volution of the H -TPR proﬁles of the fresh and spent catalysts

2

shows a clear demarcation between ceria and Ni in the spent cat-

alyst (Fig. S11). The integrated areas under the individual bands are

listed in Table S6. The results show that relative areas of the bands

keep almost unchanged. However, the band centered at 441 °C

decreases to 396 °C, whereas the band centered at 596 °C shifts

to a higher temperature (655 °C).

100 nm

Fig. 14-right plots the DRUV–Vis spectra of the fresh and spent

2

Ni/Ce-Al_WI(B) catalyst calcined at 500 °C under air ﬂow. The

2ꢂ

4+

original band centered at 320 nm due to O ? Ce charge trans-

Al K

Ce L

fer is still visible in the spent catalyst, showing a small shoulder at

ꢃ380 nm being indicative of the presence of O

the NiO lattice. Nevertheless, the broad band in the range 450–

50 nm present in the fresh catalyst vanishes after reaction,

h

-coordinated Ni in

5

reﬂecting a change in the coordination environment of Ni and also

probably in the density of O vacancies in ceria. Partial incorpora-

tion of Ni in the ceria surface or subsurface during reaction is not

excluded. The small hump at 640 nm can be ascribed to the pres-

h 2 4

ence of T -coordinated Ni in NiAl O .

Interestingly, the DRUV-Vis spectrum of the spent

Ni/Ce-Al_WI(B) (Fig. 14-right, proﬁle b) is very similar to that

2

obtained on the fresh 2Ni/Ce-Al_WI(A) for which Ni-impregnation

was performed after calcination (Fig. 14-right, proﬁle c). This

observation appears to indicate that the Ni-Ce interaction might

disturb the environment of Ni during reaction, or that the spent

catalyst might be mainly constituted of only surface Ni as for the

20 nm

Ni K

Ni K Ce L

fresh 2Ni/Ce-Al_WI(A). Furthermore, when comparing the H

proﬁles of the spent 2Ni/Ce-Al_WI(B) (Fig. 14-left, b) and the fresh

Ni/Ce-Al_WI(A) (Fig. 14-left, c), very similar reduction patterns

2

-TPR

2

are observed, suggesting a similar Ni speciation. Adding the obser-

vation that the catalytic activity of the spent 2Ni/Ce-Al_WI(B) and

the fresh 2Ni/Ce-Al_WI(A) are very similar, we can conclude that,

after the reaction, 2Ni/Ce-Al_WI(B) has been restructured and

behaves in a similar way as the fresh 2Ni/Ce-Al(A).

At last, we investigated coke deposition on 2Ni/Ce-Al_WI(B)

st

after the 1 run using TG analysis. The TG proﬁle of the fresh

2

1

Ni/Ce-Al_WI(B) (Fig. S12, proﬁle a) shows a small weight loss near

00 °C due to water removal. In contrast, the TG proﬁle of the spent

2

0 nm

20 nm

catalyst (Fig. S12, proﬁle b) shows two weight loss regions at lower

and higher temperatures. The weight loss of 2% near 100 °C can be

attributed to physically adsorbed water, whereas the larger weight

loss (9%) in the region 300–400 °C can be attributed to the combus-

Fig. 12. EDS images of selected area of 2Ni/Ce-Al_WI(B) during STEM analysis

showing highly dispersed Ni.

a

tion of amorphous carbon (C ) deposits [95]. This observation indi-

cates that coke is not formed in large amount during n-octanol

amination with ammonia and that coke can be completely

removed from the catalyst after calcination at 400 °C.

Al O

2

3

Ce(IV)

Ce(III)

Ni

4

. Conclusions

A series of low-Ni catalytic formulations (2 wt% Ni) supported

Fig. 13. Proposed structure for 2Ni/Ce-Al_WI(B).

over Ce-Al mixed oxides achieved an unprecedented activity and

selectivity to primary amines in the direct amination reaction of

n-octanol with ammonia. The addition of ceria as promoter to

2

2 3

Ni/Al O improved not only the reducibility by generating active

at 500 °C under air ﬂow and subjected to H

2

-TPR analysis to assess

Ni surface species, but also promoted the formation of very small

nickel nanoparticles (2–3 nm) in detriment of Ni aluminates. How-

ever, the properties of the catalysts were strongly dependent on

the method of preparation. The best results were achieved on a

formulation based on Ce-Al supports prepared by wet impregna-

tion of ceria over alumina followed by incipient wetness impregna-

tion of the Ni precursor solution before calcination. After

optimization, this catalyst afforded a comparable activity to the

benchmark catalysts, but saving up to 75% Ni. The present study

its reducibility compared to the fresh catalyst (Fig. 14-left). The

spent 2Ni/Ce-Al_WI(B) catalyst shows a similar reduction pattern

to that of the fresh catalyst, being characterized by ﬁve reduction

bands. Nonetheless, the spent catalyst displays a shift to higher

1 2

temperatures for surface Ni species (b and b type), suggesting

the presence of a stronger interaction between Ni and alumina

st

after the 1 run. This observation suggests that, besides the pres-

ence of active surface Ni species, the interaction between Ni and

A. Tomer et al. / Journal of Catalysis 356 (2017) 133–146

145

c

b

a

c

b

a

Fig. 14. H

2

-TPR proﬁles (left) and DRUV–Vis spectra (right) of 2Ni/Ce-Al_WI(B) before (a) and after reaction (b), 2Ni/Ce-Al_WI(A) before reaction (c). Reaction conditions: Oct-

- 7 bar, T- 180 °C, Time- 24 h, Cat- 65 mg, Solvent- 3 mL o-xylene, rpm- 600.

1

.3 mmol, NH

3

opens an avenue for further investigations on low-Ni formulations

for various catalytic applications.

[20] A. Mehta, A. Thaker, V. Londhe, S.R. Nandan, Appl. Catal. A: Gen. 478 (2014)

41–251 .

2

[

21] G.A. Vedage, L.A. Emig, H-X. Li, J.N. Armor, US 5,917,092, 1998.

22] G.A. Vedage, K.S. Hayes, M. Leeaphon, J.N. Armor, US 5,932,769, 1999.