1
94
A. Peeters et al. / Applied Catalysis A: General 469 (2014) 191–197
of the material was examined with EDX after sputtering with car-
bon. This analysis showed that no chlorine could be detected with
EDX, even for the 5 wt% Ru/HAP sample. This is in accordance with
literature. Chlorine was found to be absent in samples with a low
ruthenium content and only at higher loadings (>15 wt%) it was
detected [13,22,27]. The nature of the Ru sites in Ru/HAPs has been
debated in literature; most likely a simple ion exchange is not the
only mechanism determining the speciation in metal loaded HAPs
[
18]. EXAFS studies showed evidence for hydroxy-ligands on the Ru
sites [16], and later investigations by the same group using DRIFT
spectroscopy of adsorbed CO and ATR-IR spectroscopy revealed the
3+
presence of two different species: isolated Ru ions and a hydrated
3+
3+
Ru -oxide like phase (Ru hydroxide) with high redox activity
14,18].
[
3.2. Amination of benzyl alcohol with N-methylbenzylamine
The reaction of N-methylbenzylamine with benzyl alcohol was
chosen as a test reaction to optimize the reaction conditions with a
wt% Ru/HAP catalyst. With this secondary amine, only the mono-
2
Fig. 3. Hot filtration test: yield (Y%) of N,N-dibenzylmethylamine before and after
filtration of reaction mixture after 3 h of reaction time (Reaction conditions: 0.5 mmol
of N-methylbenzylamine, 2 mmol of benzyl alcohol, 1 mL mesitylene, 2 wt% Ru cata-
lyst (3 mol% Ru with respect to the understoichiometric reactant; 15 mol Ru; 75 mg
catalyst), under Ar atmosphere, 130 C. GC yield based on amine, calibrated with
tetradecane internal standard (0.25 mmol)).
N-alkylation product can be formed, i.e. N,N-dibenzylmethylamine
(3a, reaction outline, Table 2). Hardly any benzaldehyde was
observed, indicating that the first step in the sequence, leading
to benzaldehyde, is the slowest one. All subsequent steps in the
sequence, i.e. imine formation and reduction must be fast for the
model reaction used in Table 2. Primary amines can be alkylated
twice with an alcohol generating both secondary and tertiary amine
products. The imine formed in the second step of the borrowing
hydrogen mechanism is relatively stable when working with a pri-
mary amine and is often detected in large quantities in the reaction
mixture. This is avoided in the test reaction.
◦
3
. Results and discussion
3.1. Catalyst characterization
Hydroxyapatite [Ca10(PO ) (OH) ] is a naturally occurring min-
4
6
2
eral that has gained interest in heterogeneous catalysis during the
2+
past decade as basic support material. The Ca -ions in the lat-
tice can be partially replaced by other, catalytically active metal
ions, such as ruthenium. High catalytic activity of Ru/HAP for oxida-
tions has been ascribed to the combination of redox active Ru sites
with basic sites in the lattice [14]. The amination of alcohols via
the borrowing hydrogen mechanism has similar catalyst require-
ments, since a redox site is needed to dehydrogenate the alcohol
and hydrogenate the imine, and base additives are known for their
co-catalytic effect on e.g. the dehydrogenation.
When varying the ratio of amine and alcohol reactants, it became
clear that an excess of alcohol has a beneficial effect on the yield of
3a. With an excess of amine, or a stoichiometric amount of alcohol
and amine, the yield did not exceed 24% after 24 h of reaction at
◦
130 C (Table 2, entries 1 and 2). A 1:2 or 1:4 alcohol excess results
in high yields of the desired product (Table 2, entries 3 and 4). When
benzyl alcohol is used as solvent (1 mL) the reaction is improved
even further. Working in a polar aprotic solvent such as DMF had
a detrimental effect on the reaction, most likely due to a stronger
coordination of this solvent on the active sites of the catalyst. In the
nonpolar solvents mesitylene and nonane the reaction proceeded
with ease. For nonane however some mesitylene co-solvent had
to be utilized to dissolve all products due to poor solubility of the
alcohol and amine in this solvent.
Since the basicity of the support presumably contributes to the
overall activity of the Ru/HAP catalyst, it was studied by acid titra-
tion with bromothymol blue (pKa = 7.2) as pH indicator [23,24]. The
titrated number of basic sites was determined to be 0.14 mmol/g.
The HAP support material and a material with 2 wt% Ru were
investigated with XRD (Fig. 1). Both before and after ion exchange
HAP is the only structure that could be detected. The patterns are
in good agreement with literature data, with strong reflections at
Blank reactions were performed to assess the need for an effi-
cient catalyst. When working without any additives, no product
◦
was detected even after 24 h of reaction at 130 C (Table 3, entry 1).
◦
◦
◦
◦
◦
◦
2
5.9 , 31.8 , 32.2 , 32.9 , 46.7 and 49.5 2Â which are assigned to
When the HAP support was added to the reaction mixture, still no
Miller indices (0 0 2), (2 1 1), (1 1 2), (3 0 0), (2 2 2) and (2 1 3) of the
hexagonal crystal system of HAP [25,26].
product was formed (Table 3, entry 2). The Ru precursor RuCl ·nH O
3
2
was used as a homogeneous catalyst under the same reaction con-
ditions, and while initially good yields of 3a were obtained with
the homogeneous analogue, the yields obtained after long reaction
times are clearly inferior to those for the heterogeneous Ru/HAP,
which is due to inactivation of the homogeneous catalyst; the Ru
salt formed a black precipitate in the reaction mixture after 24 h.
RuCl ·nH O was supported an a high-surface alumina support;
The elemental compositions of the HAP support and of the Ru-
exchanged materials were investigated with ICP and AAS. The Ca/P
ratio of the carrier HAP was 1.65, which is slightly below the stoi-
chiometric amount of 1.67 for Ca10(PO ) (OH) . The ruthenium
4
6
2
amount was determined and is shown in Table 1.
Nitrogen adsorption and desorption isotherms were measured
for a 2 wt% Ru/HAP sample and they are shown in Fig. 2. It can
be concluded that hardly any microporosity is present in the HAP.
3
2
however, reaction yields were moderate in comparison to the
Ru/HAP (Table 3, entry 5). On HAP the simultaneous presence of
2
3+
The BET surface area was calculated to be 61 m /g, while the HAP
the Ru redox site and intrinsic basicity is an excellent combina-
2
support itself was found to have a surface area of 59 m /g. This
tion to obtain high conversions for this N-alkylation. Yamaguchi
et al. obtained a functional catalyst by treating the suspension of
RuCl ·nH O and alumina or titania with 1 M of NaOH until the sus-
indicates that the ion exchange procedure had no significant effect
on the surface area.
The morphology of the samples was investigated with SEM and
this showed no alterations in morphology of the powder after Ru
exchange (see Supporting Information). The surface composition
3
2
pension reached a pH of 13.2 [11]. With HAP as support this harsh
base treatment of the catalyst is not necessary, since the support
has intrinsic basicity.