Selective half-hydrogenation of adiponitrile to aminocapronitrile on Ni-based catalysts elaborated from lamellar double hydroxide precursors

Article abstract of DOI:10.1016/S0021-9517(02)93766-8

Layered double hydroxides with a hydrotalcite-like structure and containing Ni²⁺/Mg²⁺/Al³⁺ cations in different amounts were prepared and activated under various conditions. These catalysts were tested in the liquid-phase hydrogenation of adiponitrile with the aim of producing aminocapronitrile (ACN). The reaction was carried out in a batch reactor at 323-353 K and 2.5 MPa H₂ pressure with catalysts reduced at 823 K. The products were ACN, hexamethylenediamine (HMDA), azacycloheptane (ACH), and C₁₂ compounds. The ACH and C₁₂ byproducts are formed by condensation between "imine-" and "amine-like" adsorbed species on metal and acid sites (bifunctional mechanism), and on the metal sites as well. The tuned addition of Mg (Mg/(Mg + Ni)=0.20) allows us to reach the highest selectivity and yield in ACN (66% selectivity at 70% conversion, 50% yield at 85% conversion). The IR spectroscopy of adsorbed CO provided evidence of the presence of smaller Ni⁰ ensembles on the sample with Mg/(Mg + Ni)=0.20, as well as larger back-donation from Ni⁰ sites to the 2π* orbitals of CO. On that account, it is proposed that the lower formation of HMDA, ACH, and C₁₂ byproducts may be mainly due to (i) a faster desorption of ACN from the Ni⁰ surface before deeper hydrogenation and (ii) the decrease of transimination reactions which need large Ni⁰ ensembles to proceed. The correlation between acidity and condensation reactions is not obvious, since upon Mg substitution for Ni, the number of acid sites increases but their strength decreases concurrently, as shown by temperature-programmed desorption of NH₃.

Full text of DOI:10.1016/S0021-9517(02)93766-8

Journal of Catalysis 211, 511–520 (2002)
doi:10.1006/jcat.2002.3766
Selective Half-Hydrogenation of Adiponitrile to Aminocapronitrile
on Ni-based Catalysts Elaborated from Lamellar Double
Hydroxide Precursors
∗
∗
∗
∗,1
Didier Tichit, Robert Durand, Alice Rolland, Bernard Coq, Joseph Lopez,† and Philippe Marion†
∗
Laboratoire de Mat e´ riaux Catalytiques et Catalyse en Chimie Organique, UMR CNRS-ENSCM 5618, ENSCM, 8 rue d l’Ecole Normale, 34296
Montpellier Cedex 5, France; and †Rhodia, Centre de Recherche de Lyon, 85, avenue des Fr e` res Perret, BP 62, 69192 Saint-Fons Cedex, France
Received May 1, 2002; revised July 5, 2002; accepted July 5, 2002
diaminohexane (hexamethylenediamine, HMDA). A few
Layered double hydroxides with a hydrotalcite-like structure and catalytic materials have been described as selective in this
containing Ni2 /Mg /Al cations in different amounts were pre- transformation. A first patent which dates back to 1952
+
2+
3+
pared and activated under various conditions. These catalysts were
tested in the liquid-phase hydrogenation of adiponitrile with the
aim of producing aminocapronitrile (ACN). The reaction was car-
ried out in a batch reactor at 323–353 K and 2.5 MPa H2 pressure
with catalysts reduced at 823 K. The products were ACN, hexam-
Ni/Fe2O3 (3), and Ni/α-Al2O3 (4) catalysts with a reactants
ethylenediamine (HMDA), azacycloheptane (ACH), and C12 com-
pounds. The ACH and C12 byproducts are formed by condensation
between “imine-” and “amine-like” adsorbed species on metal and
acid sites (bifunctional mechanism), and on the metal sites as well.
claims 92% ACN yield with the use of Pd/silica-gel in a
liquid-phase process at 30 MPa, 383 K, and a large ex-
cess of NH3 (2). In a gas-phase process, 100% ACN se-
lectivity at 52% ADN conversion was claimed on K-doped
molar ratio H2/ADN ≈ 300. High ACN selectivities have
also been obtained in the liquid-phase hydrogenation of
ADN by using catalysts containing Rh supported on high-
The tuned addition of Mg (Mg/(Mg + Ni) = 0.20) allows us to reach surface MgO, and performing the reaction in the presence
the highest selectivity and yield in ACN (66% selectivity at 70% of a large excess of NH3 (5). Yields of ca. 64% were thus
conversion, 50% yield at 85% conversion). The IR spectroscopy of claimed in the hydrogenation at 373 K of ADN on 5 wt%
0
adsorbed CO provided evidence of the presence of smaller Ni en-
Rh/Al2O3 in a batch reactor and of 3 MPa H2 pressure with
a large excess of tetrahydrofuran and NH3 as solvents (6).
sembles on the sample with Mg/(Mg + Ni) = 0.20, as well as larger
0
∗
back-donation from Ni sites to the 2π orbitals of CO. On that
account, it is proposed that the lower formation of HMDA, ACH,
and C12 byproducts may be mainly due to (i) a faster desorption
A 35% Ni/MgO promoted with Fe yielded 71.3% ACN in
the hydrogenation of ADN in a batch reactor under 7 MPa
H2 pressure with a large excess of NH3 or methanol (10/1,
vol/vol) with respect to ADN (7). With lower NH3 excess,
a powdered promoted Fe-based catalyst gave 56.8% yield
in a continuous process operating at 25 MPa (8). Raney Ni
of ACN from the Ni0 surface before deeper hydrogenation and
0
(
ii) the decrease of transimination reactions which need large Ni
ensembles to proceed. The correlation between acidity and con-
densation reactions is not obvious, since upon Mg substitution for
Ni, the number of acid sites increases but their strength decreases catalysts exhibited good ACN selectivities when the reac-
concurrently, as shown by temperature-programmed desorption of tion was carried out in excess of NH3, or methanol with
NH3.
ꢀc 2002 Elsevier Science (USA)
an inorganic base (9). The hydrogenation at 323 K and
MPa of concentrated HMDA solution of ADN over a
promoted Raney Ni catalyst lead to an ACN yield of ca.
6% at 84% ADN conversion with a small addition of H2O
and KOH (10). Depending on the processes, the main sec-
ondary products are HMDA from deep hydrogenation of
ACN, azacycloheptane (ACH), and condensed C12 amines
2
INTRODUCTION
-Amino-5-cyanopentane (aminocapronitrile, ACN) is
the key intermediate in a new route for synthesis of
caprolactam for the fabrication of nylon-6 (1). An ele-
gant catalytic route could be considered through the partial
hydrogenation of 1,4-dicyanobutane (adiponitrile, ADN)
over heterogeneous catalysts. The challenge is to stop
the reaction halfway to avoid deep hydrogenation to 1,6-
5
1
(
BHT); the latter products are formed by inter- and in-
tratransimination between imine and amine intermediates
9, 11, 12). From these selected examples it comes out that
(
a high dilution of ADN, especially by methanol or NH3,
favors the ACN selectivity. However, this is detrimental to
productivity. On the other hand, Ni-based materials are ex-
cellent candidates for preparing active and selective cata-
lysts.
1
To whom correspondence should be addressed. E-mail: coq@cit.
enscm.fr.
511
0
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.
ꢀ

5
12
TICHIT ET AL.
We have recently shown that lamellar Ni-containing obtained by coprecipitation at constant pH of 11 ± 0.2 of
double hydroxides (LDH) are efficient precursors for an aqueous solution containing in appropriate amounts
catalysts that are very selective in the hydrogenation of Ni(NO3)2 · 6H2O, Mg(NO3)2 · 6H2O, and Al(NO3)3 · 9H2O
nitriles (13, 14). In particular, a catalyst obtained from with a solution of NaOH (1 M). The dropwise addition was
Ni/Mg/Al (0.42/1.43/1) LDH yielded 91.7% ethylamine in performed at 293 K under vigorous stirring. The precipi-
the gas-phase hydrogenation of acetonitrile at 393 K and tated gels were heated at 353 K during 15 h, centrifugated,
atmospheric pressure (13). There is an optimal Mg content and washed several times with demineralized water until
resulting from a compromise between the reducibility of neutral pH was obtained. The solids were finally dried in
Ni and the acidity of the material, both decreasing with an oven at 353 K for 20 h. The chemical analyses of the solids
the introduction of Mg. An increase in basicity indeed fa- were carried out at the Service Central d’Analyse du CNRS
vors the desorption of amine and decreases to some extent (Solaize, France) by ICP-MS. Compositions of the samples
2
+
3+
the transimination reactions between amines and imines. In are given in Table 1. The ratio (ꢀM /Al ) was fixed at 3
their pioneering work on the liquid-phase hydrogenation of in the starting synthesis solution for samples AR39, AR44,
nitriles, von Braun et al. (15) claimed that these condensa- AR5, and AR6, and at 2 for sample AR7. The chemical
tion steps occur in the homogeneous phase, but Dallons composition of the solids is generally close to that of the
et al. (16) concluded that the transimination reactions take synthesis solution. The presence of carbonates comes from
place at the catalyst surface. On supported metal catalysts CO2 in the surrounding atmosphere. We have assigned car-
2
3
−
two pathways can exist for the formation of byproducts at bonate species to CO alone, but part of these species are
−
the catalyst surface:
1) A condensation reaction between adsorbed imine
and amine proceeding at the metal surface (17, 18).
2) On supported metal catalysts, the condensation be-
tween amine and protonated imine can also occur on acidic
sites of the carrier to yield a protonated aminal, which in
turn loses NH3 giving a secondary imine, further hydro-
genated to amine at metal sites; this is a bifunctional mech-
anism (13, 19, 20).
HCO (22), which explains the excess of negative charge
3
found in some samples. The remaining Na content from pre-
cipitation procedure never exceeded 100 ppm. The samples
for catalysis were reduced for 2 h at 823 or 893 K in H2/N2
20/80, vol/vol) flow (150 cm min , ramp: 3 K min ).
ThematerialswerecharacterizedbyX-raydiffraction, N2
sorption at 77 K, temperature-programmed reduction by
H2 (TPR), H2 chemisorption, IR spectroscopy of adsorbed
CO, and temperature-programmed desorption of NH3.
XRD patterns were recorded on a CGR Theta 60 in-
strument using monochromatized Cu Kα1 radiation at
λ = 1.5405 A (40 kV and 50 mA). N2 sorption was per-
formed at 77 K with a Micromeritics ASAP 2000 set-up.
Before N2 sorption, the sample was outgassed at 523 K at
(
(
3
−1
−1
(
Moreover, due to the presence of both Al and Mg, such
materials can be easily shaped, and thereby constitute at-
tractive candidates as catalytic materials for the partial hy-
drogenation of ADN to ACN. It is worth noting that LDH-
like materials of widely varying compositions have recently
been claimed as active in the deep hydrogenation of dini-
triles (21). This is the aim of this study to evaluate Ni-based
LDH as precursors for the selective hydrogenation of ADN
to ACN in the liquid phase.
˚
−
4
10
Pa. The surface areas were determined using the BET
equation.
3
−1
TPRbyH2 (H2/Ar:3/97, flow = 30cm min , m = 30mg,
−
1
ramp = 10 K min ) was carried out with a Micromerit-
ics AutoChem 2910 apparatus using TCD detection. The
H2 consumption was determined after H2O was trapped
at ca. 200 K. TPR experiments were run with the aim to
determine the degree of reduction of Ni before chemisorp-
tion. The TPR were carried out after in situ reactivation
of the sample under the same conditions as those for H2
chemisorption (vide infra).
EXPERIMENTAL
Preparation and Characterization of the Materials
LDH materials containing Ni2+/Mg /Al cations with
2+
3+
a large range of compositions were prepared. They were
TABLE 1
Some Characteristics of the Ni/Mg/Al LDHs
Chemical formulaa
(Mg + Ni)/Al Mg/(Mg + Ni)
2−
−
3+
Sample
ꢀ(2CO3 + NO )/Al
3
−
−
AR39
AR44
AR5
AR7
AR6
[Ni0.74Al0.25(OH)2](CO2 )0.05(NO )0.18 · mH2O
2.92
2.96
3.00
1.76
2.89
0.0
1.02
1.04
1.29
1.03
1.36
3
3
2−
−
[Ni0.67Mg0.07Al0.25(OH)2](CO3 )0.04(NO )0.18 · mH2O
0.095
0.20
0.22
0.39
3
[Ni0.60Mg0.15Al0.25(OH)2](CO2 )0.09(NO )0.18 · mH2O
−
−
3
3
[Ni0.49Mg0.14Al0.36(OH)2](CO2 )0.06(NO )0.25 · mH2O
−
−
3
3
[Ni0.45Mg0.29Al0.26(OH)2](CO2 )0.09(NO )0.18 · mH2O
−
−
3
3
a
After drying at 353 K; trace amounts of Na (<100 ppm) were identified in some samples.

LAMELLAR DOUBLE HYDROXIDE PRECURSORS
513
0
TheaccessibilitytoNi phasewasestimatedbyvolumetry
of H2 chemisorption with a Micromeritics ASAP 2010 and a
dedicated attachment. The sample was reactivated at 773 K
in H2 for 2 h, then outgassed at the same temperature for 1 h
For condensed C12 products (BHT),
moli
mol0 − molt )ADN
2
Si = 100
.
(
−4
at 10 Pa. The Ni surface was then saturated by step with
H2 at 373 K until 40–50 kPa was reached. The H2 uptake
was estimated by the extrapolation to zero pressure of the
linear part of the isotherm.
In situ DRIFT experiments of adsorbed CO were con-
ducted on a Bruker Equinox 55 spectrometer with a special
cell (Graseby Specac) allowing a gas flow and high temper-
ature operation. A sample of 80 mg of powder was reacti-
vated in flowing H2 overnight at 723 K, then outflowed with
Ar at the same temperature for 2 h. After cooling to room
temperature in Ar, CO/He (1/99) was fed to the reactor
cell. After evacuation of gaseous CO with Ar, the collec-
tion of IR spectra (100 scans, resolution 4 cm ) was started
from RT to 523 K by step of 20 to 30 K.
NH3-TPD experiments were carried out using a Mi-
cromeritics Autochem 2910 with TCD detection. NH3 was
adsorbed at 373 K on the reduced and outgassed catalysts.
The sum of product selectivities should be 100%. Actu-
ally, in most of the experiments it was lower (92–95%). The
imbalance should be accounted for by condensed imines
(
C12 imines), which are quite unstable during the GC anal-
ysis. Someparallelmeasurements ofimineconcentrationby
apolarographicmethodorNMRhaveshownthatthedeficit
of identified products in the reaction can be accounted for
by these heavier imines.
The properties of these materials were compared with
those of a commercial Raney-type Ni catalyst (STREM).
RESULTS AND DISCUSSION
−1
Characterization of the Materials
The XRD patterns of precursor materials are those
of well-crystallized LDH. No excess phase was identified
2
+
2+
−
1
showing that Ni was isomorphously substituted for Mg
Heating was then started up to 873 K at 10 K min in He
in the layers. The XRD patterns show the appearance of
flow.
0
Ni phase upon reduction at 823 K.
Catalytic Experiments
The textural properties of the materials strongly de-
pend on the thermal treatments. The uncalcined lamel-
The liquid-phase hydrogenation of ADN was carried out
in a tank reactor (100 cm ) under strirring at 323–353 K and
2
−1
lar structure exhibits low surface area of 15–35 m g
,
3
which increases upon thermal treatment at high temper-
ature. The SBET values are given in Table 2 for every sam-
ple after reduction at 823 K of the materials. The high-
est surface area was found for AR5 (Mg/Mg + Ni) = 0.20,
2
.5 MPa H2 pressure. An aliquot of the passivated catalyst
(
400–500 mg, grain size: 63–125 µm) was first reactivated
in situ in HMDA (24 g) with various amounts of water and
KOH. The reactivation was carried out for 2 h at 423 K and
2
−1
SBET = 290 m g ). The surface area of the calcined and/or
reduced samples depends on their chemical composition,
which determined the relative amounts of mixed oxide and
spinel-like phases. The former leads to higher surface area
than the latter, which is favored at high Ni content (23).
The best compromise between these phases is obtained for
3
.5 MPa H2 pressure. After cooling to the reaction temper-
ature (generally 353 K), the solution was purged with N2
flow. The reactant ADN (m = 25 g) was then pushed into
the reactor by H2 flow. The H2 pressure was then main-
tained constant into the reactor at 2.5 MPa by H2 feed
regulation. The consumption of H2 was monitored by a
pressure transducer in an upstream reservoir. The initial
reaction rate was determined from H2 consumption, and
samples of reaction mixture were withdrawn from the reac-
tor from 50 to 90% ADN conversion. The sample composi-
tions were analyzed offline by gas chromatography (Varian
TABLE 2
BET Surface Area, H2/Ni Molar Consumption Ratio in the
Various TPR Experiments, and Accessibility to Ni Sites
SBETa
−1
Ni accessible
0
3
0
300)onaRestekRTX-2aminecapillarycolumn(30m;i.d.,
.25 mm).
2
H2/Nib
H2/Nic
H/Nid
−1
sites (mmol g )
Sample
(m g
)
The concentration of reactant and products in the with-
AR39
AR44
AR5
AR7
AR6
166
176
293
220
172
0.94
0.90
0.99
1.12
0.86
0.07
—
0.30
0.32
0.38
0.05
0.08
0.10
0.065
0.05
0.52
0.82
0.95
0.63
0.35
drawn samples was determined with respect to the internal
standard diphenylether. Thereafter, the selectivity to the
product i has been expressed as the number of moles of
ADN which has reacted to i divided by the total number of
moles of transformed ADN:
a
After drying at 393 K and reduction at 823 K.
b
For uncondensed C6 products (ACN, HMDA, ACH),
From TPR experiments of the dried samples at 393 K (H2 uptake
at T > 580 K).
c
From TPR experiments of the reduced samples then in situ reactivated
at 773 K in H2.
moli
Si = 100
.
d
(
mol0 − molt )ADN
After drying at 393 K and reduction at 823 K.

5
14
TICHIT ET AL.
AR5. If the reduction of AR5 is carried out at 893 K, the
2
−1
SBET decreases to 165 m g . All the samples exhibit a
T4-type histeresis loop for N2 desorption. The curves pro-
vide evidence for the absence of microporosity, and a mean
mesopore size of 5–7 nm.
The TPR profiles of the LDH samples are given in
Fig. 1A. Two peaks of H2 consumption are observed with
maxima at 550 K for the first and ranging from 700 to 900 K
for the second. The maximum of the second broad peak
is regularly shifted toward higher temperatures when the
Mg content increases. In a detailed previous study on the
reduction of Ni-containing LDH (23), we clearly demon-
strated with mass spectrometry coupling that the first H2
consumption peak at 550 K comes from the reduction of
nitrate anions mainly to NO. In full agreement with this
fact, the H2 consumption represented by the broad peak at
high temperature is very close to that corresponding to the
2+
0
reduction of Ni into Ni in the present samples. In these
TPR experiments (dynamic conditions, H2/Ar = 3/97), it
thus appears that almost full Ni reduction is achieved at
1
000 K for the less reducible sample AR6. Since the ma-
terials are reduced at 823 K in pure H2 overnight to yield
the catalysts, it is of utmost importance to know the de-
gree of Ni reduction in the catalysts after this treatment.
That is why we carried out TPR experiments after reac-
tivation of some reduced catalysts at 823 K in flowing H2
overnight (Fig. 1B). The remaining H2 consumption shows
that the degree of Ni reduction in the catalyst ranges from
6
2% in AR6 (high Mg content) to 93% in the Mg-free
AR39.
0
The accessibility to the Ni phase was estimated from H2
chemisorption. The data are collected in Table 2 for the un-
calcined samples directly reduced at 823 K. The accessibility
to Ni goes through a maximum value ca. 10% for AR5. It is
0
worth noting the good accessibility to the Ni phase in spite
of the very high Ni loading of 68%. In comparison, a H/Ni
ratio of ca. 0.18 was reported for a 21%Ni/MgO reduced at
6
73 K (24).
All these observations are in perfect agreement with the
previous detailed study on the activation of Ni- and Mg-
containing LDH (23).
Catalytic Experiments
The possible occurrence of interparticle mass transfer
limitations was checked on the most active sample: Raney
Ni. The rotation speed was varied from 100 to 900 rpm;
the initial rate reached a plateau value between 700 and
8
00 rpm, then decreased after appearance of a vortex. The
value of 730 rpm was then chosen for any further experi-
ments. Intraparticle mass transfer was checked with AR5 by
sieving two fractions, <63 µm and 63–125 µm; the catalytic
data were the same.
In the course of ADN selective hydrogenation over
LDH-originating catalysts, the target product is ACN, with tion at 823 K; conditions: H2/Ar (3/97), ramp: 10 K min
FIG. 1. (A) H2-TPR profiles of Ni/Mg/Al LDH precursors; (a) AR39,
(
1
b) AR44, (c) AR5, (d) AR7, (e) AR6; conditions: H2/Ar (3/97), ramp:
−1
0 K min . (B) H2-TPR profiles of AR39 and AR6 catalysts after reduc-
−
1
.

LAMELLAR DOUBLE HYDROXIDE PRECURSORS
515
FIG. 2. Formal reaction scheme for adiponitrile hydrogenation.
HMDA, ACH, and C12 compounds as byproducts. This is Ni-basedcatalystsfromLDHprecursors, thesamebehavior
obviously in line with the general scheme of α, ω-dinitrile occurs as shown by the data reported in Table 3. The trace
hydrogenation (5, 9, 11, 12). It is worth noting that in amounts of Na (<100 ppm) found in some samples cannot
spite of the presence of KOH, the well-known Thorpe– interfere with the effect of KOH since the ratio Na/K is ca.
Ziegler intramolecular condensation of ADN to 2- 1/50 mol/mol. In the absence of KOH, the loss of selec-
aminomethylcyclopentylamine (C–C bond formation) was tivity occurs by transimination with the formation of ACH
not occurring (11, 12). A simplified representation of this mainly. Ziemecki (9) interpreted the promoter effect of in-
scheme with the products only detected in our experiments organic base addition by the stabilization of an intramolec-
is displayed in Fig. 2. There are several types of transimina- ular interaction of ACN to an heterocyclic form contain-
tion between imino and amino groups. One leads to dimeric ing imino form of the CN group. Such cyclic conformation
compounds when transimination occurs between the imino would impede readsorption of the amino-nitrile, and its fur-
intermediate compounds and an amine. The second type ther hydrogenation to 1-amino-6-imino-hexane. However,
is the intramolecular transimination of 1-amino-6-imino- KOH addition should neutralize some acid sites responsible
hexane intermediate yielding ACH.
for the intramolecular condensation of 1-amino-6-imino-
Before discussing the main point in this study, which is hexane to ACH. Neutralization of acid sites was indeed
the influence of catalyst composition on the reaction, we postulated to interprete the better selectivity to butylamine
shall first present the effect of operating conditions on the in the hydrogenation of butyronitrile over Raney Ni in the
catalyticproperties, i.e., thepresenceofKOHandwater, the presence of NaOH (25). It is obvious that the doping effect
temperature of in situ reactivation, and the reconstruction of KOH addition is determined by its adsorption isotherm;
of the layered structure of the material by water.
the latter being likely different depending on H2O content
and the nature of the catalyst. Fundamental studies are fur-
Parametric study. The addition in the reaction medium ther needed to better understand that point. However, it is
of a certain amount of an inorganic base is recommended to worth noting that the same KOH addition was found opti-
improve the selectivity to aminonitrile in the hydrogenation mum in order to promote the selectivity of Raney Ni in the
of dinitriles on Raney Ni (9, 21). In the case of the present present work.

5
16
TICHIT ET AL.
TABLE 3
Catalytic Properties of AR5 Catalysts for the Hydrogenation
of Adiponitrile at 70% Conversion
Selectivities (mol%)
Initial rate
−
1
−1
)
(
mol g
s
ACN
HMDA
ACH
BHT
KOH addition in mol/molNi (other conditions: 9.6 vol% H2O, in-situ reactivation at 423 K)
−
−
−
6
6
6
0
0
0
.0
.01
.014
7.8 × 10
9.6 × 10
11.6 × 10
47.4
66.0
59.4
12.8
14.7
12.8
22.0
9.5
11.3
4.0
1.0
3.6
In-situ reactivation in H2 pressure (other conditions: 9.6 vol% H2O, 0.01 molKOH/molNi)
−
−
−
−
6
6
6
6
No reactivation
7.0 × 10
11.7 × 10
9.6 × 10
9.4 × 10
41.4
49.5
66.0
64.5
30.6
14.3
14.7
7.1
10.3
11.9
9.5
22.6
20.4
1.0
Reactivation at 393 K
Reactivation at 423 K
Reactivation at 438 K
9.9
10
Water addition in vol% (other conditions: 0.01 molKOH/molNi, in-situ reactivation at 423 K)
−
−
−
−
−
6
6
6
6
6
0
8
9
%
%
.6%
5.2 × 10
7.5 × 10
9.6 × 10
7.3 × 10
8.1 × 10
49.3
59.1
66.0
64.1
56.5
17.0
13.0
14.7
15.5
11.5
11.9
14.3
9.5
12.6
11.8
28.6
4.0
1.0
6.0
10.2
1
1
0.0%
2.8%
Notes. Conditions: (ADN)0 = 0.222 mol, (HMDA)0 = 0.207 mol, H2 pressure = 2.5 MPa,
TR = 353 K.
Another very important point is the possible modifi- ity to ACN is higher after activation at 423 K. Attempts to
cation of the catalyst structure in the course of reactiva- reactivate the material at temperatures higher than 423 K
tion and reaction steps. As mentioned above, after reduc- lead to catalysts of slightly lower selectivity to ACN.
tion, the Ni-containing materials exhibit Ni metal and a
One could expect that the extent of reconstruction to the
Mg(Al)O periclase-type phase with various extents of Ni LDH structure will depend on the hydration degree of the
substitution for Mg depending on the sample composition. medium. The latter was thus varied and data from Table 3
It has been shown that the LDH structure can be recon- show that an optimum water content of ca. 9.6 vol% exists
structed by rehydroxylation of the mixed oxide phase in an for which the selectivity to ACN is the highest. However,
aqueous solution, using the so-called memory effect (26). XRD patterns did not provide a clear distinction in the
This process is more difficult for the Ni-containing mixed extent of the LDH reconstruction, with the exception of
oxides than the Mg-containing oxides (26–28). Elements the experiment carried out in the water-free medium where
about the occurrence of this phenomenon is of utmost im- no reconstruction occurred. Moreover, this optimum value
portance since textural and acid–base properties are modi- of water content is very close to that found with Raney Ni
fied, with the appearance of strong Brønsted basic sites (29). as catalyst (10), in which case any reconstruction cannot
The extent of reconstruction depends on the medium and obviously be invoked. On that account, one can anticipate
the temperature. Moreover it is clear that in the case of the that the promotion of ACN selectivity by H2O is coming
present catalysts, which have been previously reduced, the from both (i) the reconstruction of LDH with an increase
0
Ni phase will be hardly reconstructed. This aspect has not in surface basicity and (ii) the formation of an hydrophilic
been previously described, and its impact on ADN hydro- interface at the metal surface, which makes the desorption
genation has been investigated for the AR5 catalyst. The of ACN easier, to the benefit of ADN adsorption.
catalytic properties of AR5 are reported without any reacti-
The parametric study thus evidenced the most promis-
vation step (direct hydrogenation at 353 K) and after in situ ing operating conditions to reach high ACN yield:
reactivation at various temperatures and 3.5 MPa H2 pres- 0.01 molKOH/molNi, 9.6 vol% H2O, and in situ reactivation
sure (Table 3). The selectivity to ACN is better with in situ in H2 (3.5 MPa) of the catalyst at 423 K. Under these con-
reactivated material, and XRD patterns provide evidence ditions, the influence of the material composition on the
of LDH structure reconstruction to some extent, very likely catalytic properties was approached.
x+
−
of the [Mg1−x Alx (OH)2] (OH ) type. It is worth noting
that the LDH is reconstructed during the reactivation step.
Influence of catalyst composition. Table 4 presents the
An in situ reactivation at 393 K does not significantly mod- catalytic properties of the various samples for the hydro-
ify the XRD pattern (the presence of the LDH peak at genation of ADN under the standard conditions defined
◦
2
θ = 11.5 with similar amplitude). However, the selectiv- above. The initial rate on Raney Ni is twentyfold faster

LAMELLAR DOUBLE HYDROXIDE PRECURSORS
517
TABLE 4
and ACN (30). In this model, the ratio of adsorption con-
stant λACN/λADN is important and likely dominated by the
electronic state of Ni d-band. It was found equal to 2.4 in
the three-phase hydrogenation of ADN on Raney Ni (31).
On the other hand, the transimination reactions leading
to ACH and BHT can occur on both Ni surface and acidic
sites of the support (vide supra). On similar materials, the
adsorption of ethylamine followed by calorimetry provided
Catalytic Properties of Ni/Mg/Al/O LDH-Originated Catalysts
for the Hydrogenation of Adiponitrile
Selectivities (mol%)
ADN
Initial rate TOF
−
1
−1 −1
s ) (h ) ACN HMDA ACH BHT
Sample conversion (mol g
−
−
−
−
−
6
6
6
6
6
6
AR39
AR44
AR5
AR7
AR6
70
70
70
70
70
70
5.6 × 10
12.1 × 10
9.6 × 10
6.4 × 10
3.0 × 10
53
70
42
41
20
—
54.0
54.1
66.0
64.6
57.0
46.8
—
6.3
14.7
7.5
14.0
32.5
20.6 23
17.4 14.8
9.5
11.1
8.4 16.6
11.0 0.6
−1
evidence of a decrease by 40 kJ mol of the heat of adorp-
1.0
9.2
tion upon introduction of Mg (Mg/(Mg + Ni) = 0.23) (13).
This is reflective of a surface of lower acid strength, less
prone for catalyzing intra- and intermolecular transimina-
tion reactions.
To correlate any change of the acid–base properties of
the materials, as well as the electronic state of Ni, the TPD
of NH3, and the IR spectroscopy of adsorbed CO were
carried out.
Raney Ni
210 × 10−
Notes. Conditions: (ADN)0 = 0.222 mol, (HMDA)0 = 0.207 mol, H2
pressure = 2.5 MPa, 9.6 vol% H2O, 0.01 molKOH/molNi, TR = 353 K, in-situ
reactivation at 423 K.
than that on AR5. However, the rate slows down rapidly
on Raney Ni, and 80% conversion was reached after 54 min
on Raney Ni but 100 min on AR5. This might reflect dif-
ferent kinetic dependences or poisoning for these two ma-
terials. The experimental set-up did not allow us to carry
out successive batch reactions in order to evaluate the ex-
tent of catalyst deactivation with accuracy. Sampling of the
reaction medium was carried out between 50 and 90% con-
version, where the ACN yield goes through a maximum
value. For the sake of comparison, the catalytic properties
were reported at 70% conversion.
It first appears that the initial rate (t = 0), expressed on
a per gram basis, decreases when the Mg content increases.
However, when expressed as TOF, the number of ADN
molecules hydrogenated per accessible Ni site, the intrin-
sic activity only changes by a factor of 3 to 4 as a function
of catalyst composition. The rate of ADN hydrogenation
probably depends on the respective coverage of H2 and
ADN. Calorimetric studies of H2 and acetonitrile adsorp-
tion on similar materials have shown that the heat of ad-
sorption of both compounds is only slightly affected by the
introduction of Mg (13). One may thus postulate that the
moderate change of TOF in ADN hydrogenation is reflec-
tive of small changes in H2 and ADN coverages.
Figure 4 shows the IR spectra of adsorbed CO in the re-
gion 2200–1750 cm , characteristic of mono- and multicar-
−1
bonyl species. It should be mentioned that bands of carbon-
−1
ates appeared in the spectral region 1700–1200 cm ; these
carbonate species were formed by interaction of CO with
basic sites (32). The high frequency broadband (A band)
−1
between 2100 and 2000 cm can be assigned to monocarb-
onyl species on metal sites. On supported Ni catalysts, two
bands of linearly adsorbed CO were observed at ca. 2040
−1
and 2070 cm (33). The former was due to CO adsorbed
0
on unperturbated Ni of facets. The latter was assigned to
CO bonded on Ni atoms in close interaction with an oxide
0
phase (unreduced Ni or other oxide phase), or CO on low
coordination sites. It should be mentioned that the pres-
2+
−1
ence of Ni could give rise to a band at ca. 2195 cm (33),
which was never observed in our experiments.
Regarding the effect of catalyst composition on product
selectivity, when the Mg content increases the tendency is
for the HMDA selectivity to increase and the ACH selectiv-
ity to decrease, whereas the BHT selectivity goes through
a minimum value. On that account, the selectivity (Table 4)
and the yield (Fig. 3) to ACN are the highest for the AR5
sample.
The reaction scheme shown in Fig. 2 is composed of con-
secutive and parallel reactions. The selectivity to the target
compounds is thus determined by (i) the consecutive hy-
drogenations ADN → ACN → HMDA and (ii) the consec-
utiveandparalleltransiminationtoACHandBHT. Thefor-
mer hydrogenations obey Langmuir–Hinshelwood kinetics
FIG. 3. Aminocapronitrile yield as a function of adiponitrile conver-
sion in the hydrogenation of adiponitrile on AR39 (ꢀ), AR44 (ꢁ), AR5
with competition of adsorption between hydrogen, ADN, (ꢂ), AR6 (ꢃ), and Raney Ni (ꢄ) catalysts; standard reaction conditions.

5
18
TICHIT ET AL.
FIG. 4. (A) DRIFT spectra of adsorbed CO on in situ reduced AR39 catalysts: (a) after CO admission at 298 K, and after desorption in Ar at
98 K (b), 373 K (c), 443 K (d), 523 K (e). (B) DRIFT spectra of adsorbed CO on in situ reduced AR5 catalysts: (a) after CO admission at 298 K, and
2
after desorption in Ar at 298 K (b), 373 K (c), 443 K (d), 523 K (e). (C) DRIFT spectra of adsorbed CO on in situ reduced AR6 catalysts: (a) after CO
admission at 298 K, and after desorption in Ar at 298 K (b), 373 K (c).

LAMELLAR DOUBLE HYDROXIDE PRECURSORS
519
The low frequency bands between 2000 and 1800 cm−1 (B
bands) are generally assigned to multibonded CO species
commonly found on Ni-based catalysts (33, 34), with νCO
−
1
at ca. 1935 cm for Ni2CO bridged species, and below
1
−
1
900 cm for Nix CO multicentered species (x = 3, 4), at
full CO coverage. The evolution of CO IR spectra at in-
creasing desorption temperatures call forth several com-
ments:
(
1) The intensity of the A band is the highest for the AR5
sample, and more specifically that corresponding to CO on
low coordination sites.
(
2) The integrated intensity of the CO bands is the low-
est for AR6. The decrease can be roughly estimated of one
order of magnitude. However, the much lower amount of
adsorbed CO might be thus accounted for by a dilution of
the Ni surface by some migrating oxide species, in agree-
ment with the lowest intensity of the B band in this sample.
(
3) CO is more weakly bonded in AR6, as compared to
AR39 and AR5. This is evidenced by the decrease by 80%
of the integrated intensity of the A band after desorption
at 373 K; band A disappeared after desorption at 413 K. A
similar decrease of the A band is observed for a desorption
temperature of ca. 523 K for AR39 and AR5.
FIG. 5. NH3-TPD profiles of AR39, AR5, and AR6; ramp: 10 K min 1
−
.
(
4) At low CO coverage, the νCO goes through a mini-
−1
−1
mumvalueof2000cm forAR5, ascomparedto2005cm
for AR39 and 2020 cm for AR6. This is reflective of a
−1
Figure 5 shows the NH3-TPD plots of some samples. The
amounts of NH3 desorbing during these experiments are of
.35, 0.77, 0.85, and 0.60 mmolNH3 g for samples AR39,
larger electron back donation from Ni d-band toward the
∗
2
π antibonding orbital of CO. This enhanced back dona-
−1
0
tion results from a higher electron density at the Ni sites.
AR44, AR5, and AR6, respectively. It is shown that the
amount of NH3 adsorbed increases with the Mg content,
but concurrently the proportion of NH3 strongly bonded
decreases with the appearance of a broad desorption peak
between 373 and 673 K. One can postulate the appearance
of weaker acid sites upon Mg substitution for Ni, proba-
bly associated to Mg belonging to acid–base pairs (22).
Weaker acid sites upon Mg introduction have been indeed
identified previously by calorimetry of ethylamine adsorp-
tion on similar materials (13). The antagonistic evolution
of number and strength of acid sites as a function of the
Mg content in the material makes them difficult to pro-
pose a correlation between acidity and condensed products
formed through transimination reactions on the acid sites.
As a concluding comment to these observations, it is
shown that AR5 sample focuses attention by (i) a higher
proportion of linear Ni–CO species, in particular on low
coordination sites, and (ii) a stronger back donation of elec-
trons to the antibonding CO bond. In a previous study on
the gas-phase hydrogenation of acetonitrile on a catalyst
from Co/Ni/Mg/Al LDH, we proposed that the transimi-
nation reactions at the Ni sites occur through a Langmuir–
Hinshelwood mechanism between two vicinal multibonded
adsorbed species (35). The introduction of Co dilutes the
Ni phase in small ensembles less prone to accommodate
neighboring multibonded species. One can propose that the
presence of monocarbonyl Ni–CO species in higher propor-
tion in AR5 reflects the presence of smaller Ni ensembles.
Moreover, an increase of the electron density at Ni sites
might decrease the adsorption strength of the “more basic”
compound ACN with comparison to ADN. From these de-
2
+
CONCLUSIONS
Ni/Mg/Al LDHs constitute good candidates as precur-
ductions one can understand that among the various sam- sors of catalysts for the hydrogenation of ADN into ACN
ples, AR5 would be the less prone to favor the formation with good yields. The catalyst obtained from Ni/Mg/Al
of byproducts through transimination at the metal sites.
(0.60/0.15/0.25, AR5), calcined at 393 K and reduced at
However, as mentioned above the transimination reac- 823 K, exhibits the highest selectivity and yield to ACN
tion can also take place through bifunctional catalysis in- (66% selectivity at 70% conversion, 50% yield at 85%
volving metal and acid sites. The latter sites are responsi- conversion). We suggest that this behavior comes from (i)
ble for the rate determining step of condensation between a faster desorption of ACN thus decreasing the consecu-
amine and protonated imine.
tive hydrogenation to HMDA and (ii) the lowering of the

5
20
TICHIT ET AL.
undesired consecutive transimination reactions which can 13. Medina Cabello, F., Tichit, D., Coq, B., Vaccari, A., and Dung, N. T.,
J. Catal. 167, 142 (1997).
occur on both the acid and the metal sites between
imines- and amines-like adsorbed species. The transimina-
tion on the metal sites could occur through a Langmuir–
1
4. Tichit, D., Medina Cabello, F., Durand, R., Mateo, C., Coq, B., Sueiras,
J. E., and Salagre, P., in “Heterogeneous Catalysis and Fine Chemical
IV” (H. U. Blaser, A. Baiker, and R. Prins, Eds.), p. 297. Elsevier,
Amsterdam, 1997.
Hinshelwood mechanism between multibonded adsorbed
0
species. This reaction is not favored on small Ni ensembles, 15. von Braun, J., Blessing, G., and Zobel, F., Chem. Ber. 36, 1988 (1923).
1
1
1
1
2
6. Dallons, J. L., Van Gysel, A., and Jannes, G., in “Catalysis of Organic
Reactions” (W. E. Pascoe, Ed.), p. 93. Dekker, New York, 1992.
7. Pavlenko, N. V., Tripol’skii, A. I., Prokhorenko, E. V., and Golodets,
G. I., Kinet. Katal. 28, 1382 (1987).
8. Hochard, F., Jobic, H., Massardier, J., and Renouprez, A. J., J. Mol.
Catal. 95, 165 (1995).
which are found on the AR5 sample. It was impossible to
conclude on a positive effect of Mg substitution for Ni with
respect to the bifunctional-catalyzed transimination since
upon introduction of Mg the number of acid sites increases
but their strength decreases. Finally, the ACN desorption
would be the fastest on the AR5 sample thanks to a higher
electron density at the Ni sites, as evidenced by IR spec-
troscopy of adsorbed CO.
9. Freidlin, L. Kh., Sladkova, T. A., and Englina, F. F., Kinet. Katal. 3,
4
17 (1962).
0. Verhaak, M. J. F. M., Van Dillen, A. J., and Geus, J. W., Catal. Lett. 26,
7 (1994).
3
2
2
1. Patent WO9710052 (1996) to Rh oˆ ne Poulenc.
2. Prinetto, F., Ghiotti, G., Durand, R., and Tichit, D., J. Phys. Chem. B
ACKNOWLEDGMENT
1
04, 11,117 (2000).
3. Tichit, D., Medina, F., Coq, B., and Dutartre, R., Appl. Catal., A 159,
41 (1997).
2
2
2
A.R. thanks the Rhodia company for a scholarship.
2
4. Turlier, P., Praliaud, H., Moral, P., Martin, G. A., and Dalmon, J. A.,
Appl. Catal. 19, 287 (1986).
5. Thomas-Pryor, S. N., Manz, T. A., Liu, Z., Koch, T. A., Sangupta, S. A.,
and Delgass, W. N., in “Catalysis of Organic Reactions” (F. Herkes,
Ed.), p. 195. Dekker, New York, 1998.
REFERENCES
1
2
3
. McCoy, M., Chem. Eng. News 78(40), 32 (2000).
. German Patent 848 654 (1952) to BASF.
. Medina, F., Salagre, P., Sueiras, J. E., and Fierro, J. L. G., Appl. Catal.,
A 92, 131 (1992).
26. Sato, K., Wakabayasi, T., and Shimada, M., Ind. Eng. Chem. Prod. Res.
Dev. 25, 89 (1986).
4
. Medina, F., Salagre, P., Sueiras, J. E., and Fierro, J. L. G., J. Chem. Soc.
Faraday Trans. 89, 3981 (1993).
27. Sato, K., Fujita, H., Endo, T., Shimada, M., and Tnunashima, A., React.
Solids 5, 219 (1988).
5
. Mares, F., Galle, J. E., Diamond, S. E., and Regina, F. J., J. Catal. 112,
28. Rebours, B., d’Espinose de la Caillerie, J. B., and Clause, O., J. Am.
Chem. Soc. 116, 1707 (1994).
1
45 (1988).
6
7
8
9
. U.S. Patent 4,389,348 (1983) to Allied Signal Corporation.
. Patent WO9947492 (1999) to Dupont de Nemours.
. Patent WO9944984 (1999) to BASF.
29. Prinetto, F., Tichit, D., Teissier, R., and Coq, B., Catal. Today 55, 103
(2000).
30. Mathieu, C., Dietrich, E., Delmas, H., and Jenck, J., Chem. Eng. Sci.
47, 2289 (1992).
. Ziemecki, S. B., in “Heterogeneous Catalysis and Fine Chemical III”
(
M. Guisnet, J. Barbier, J. Barrault, C. Bouchoule, D. Duprez, G. P e´ rot,
31. Joly-Vuillemin, C., Gavroy, D., Cordier, G., de Bellefon, C., and
Delmas, H., Chem. Eng. Sci. 49, 4839 (1994).
and C. Montassier, Eds.), p. 283. Elsevier, Amsterdam, 1993.
1
1
0. Patent WO9618603 (1995) to Rh oˆ ne Poulenc.
32. Lavalley, J. C., Catal. Today 27, 377 (1996).
1. Marion, Ph., Grenouillet, P., Jenck, J., Joucla, M., in “Hetero-
geneous Catalysis and Fine Chemical II” (M. Guisnet, J. Barrault,
C. Bouchoule, D. Duprez, G. P e´ rot, R. Maurel, and C. Montassier,
Eds.), p. 329. Elsevier, Amsterdam, 1991.
33. Primet, M., Dalmon, J. A., and Martin, G. A., J. Catal. 46, 25 (1977).
34. Eischens, R. P., and Pliskin, W. A., in “Advances in Catalysis” (D. D.
Eley, W. G. Frankenburg, V. I. Komarevski, and P. B. Weisz, Eds.),
Vol. 10, p. 1. Academic Press, San Diego, 1958.
1
2. de Bellefon, C., and Fouilloux, P., Catal. Rev.-Sci. Eng. 36, 459 (1994).
35. Coq, B., Tichit, D., and Ribet, S., J. Catal. 189, 117 (2000).