Catalyst recycling via specific non-covalent adsorption on modified silicas

Article abstract of DOI:10.1039/c2dt32047a

This article describes a new strategy for the recycling of a homogeneous hydroformylation catalyst, by selective adsorption of the catalyst to tailor-made supports after a batchwise reaction. The separation of the catalyst from the product mixture is based on selective non-covalent supramolecular interactions between a ligand and the support. Changing the solvent releases the active catalyst back into the reactor and allows a subsequent batch reaction with the recycled active catalyst. For this purpose, the bidentate NixantPhos ligand has been equipped with a pyridine group. The corresponding rhodium pre-catalyst [Rh(Nix-py)(acac)] (acac = acetylacetonate) forms a very selective, active and highly stable catalyst, and able to reach a turnover number (TON) of 170:000 in a single run (reaction performed in nearly neat 1-octene, S/C ratio of 200:000, at 140 °C, 20 bars syngas pressure). Various commercially available supports have been explored in binding studies and recycling experiments. The end-capped silica-alumina performs the best so far with respect to ligand-adsorbing properties for the current purpose. Although this system has not been fully optimized, four recycling runs could be performed successfully.

Full text of DOI:10.1039/c2dt32047a

Dalton
Transactions
View Article Online
View Journal | View Issue
PAPER
Catalyst recycling via specific non-covalent adsorption
on modified silicas†‡
Cite this: Dalton Trans., 2013, 42, 3609
Alexander M. Kluwer,*a,b Chretien Simons,b Quinten Knijnenburg,b
Jarl Ivar van der Vlugt,c Bas de Bruinc and Joost N. H. Reek*a,c
This article describes a new strategy for the recycling of a homogeneous hydroformylation catalyst, by
selective adsorption of the catalyst to tailor-made supports after a batchwise reaction. The separation of
the catalyst from the product mixture is based on selective non-covalent supramolecular interactions
between a ligand and the support. Changing the solvent releases the active catalyst back into the reactor
and allows a subsequent batch reaction with the recycled active catalyst. For this purpose, the bidentate
NixantPhos ligand has been equipped with a pyridine group. The corresponding rhodium pre-catalyst
[Rh(Nix-py)(acac)] (acac = acetylacetonate) forms a very selective, active and highly stable catalyst, and
able to reach a turnover number (TON) of 170 000 in a single run (reaction performed in nearly neat
1-octene, S/C ratio of 200 000, at 140 °C, 20 bars syngas pressure). Various commercially available sup-
ports have been explored in binding studies and recycling experiments. The end-capped silica-alumina
performs the best so far with respect to ligand-adsorbing properties for the current purpose. Although
this system has not been fully optimized, four recycling runs could be performed successfully.
Received 4th September 2012,
Accepted 5th December 2012
DOI: 10.1039/c2dt32047a
www.rsc.org/dalton
So far, several strategies for advanced catalyst recycling have
been explored, which can be broadly grouped into two types.
Introduction
The efficient separation and subsequent recycling of homo- The first group links the active catalyst covalently to some kind
geneous transition metal catalysts remains a topic that not of soluble or insoluble support and separation is accom-
only constitutes a scientific challenge but also has significant plished by a (nano)filtration technique.2,3 Various scaffolds
commercial relevance. Homogeneous catalysis facilitates an have been used such as dendritic,4 polymeric organic, in-
atom-economic synthetic approach in which all atoms of the organic, or hybrid supports.5,6 The other type of recycling
reagents end up in the product. In addition, a high degree of methodology is the selective partitioning of the catalyst from the
tunability of the catalyst activity, chemoselectivity, regio- reactants and products that can be achieved by employing two
selectivity and enantioselectivity can be achieved.1 However, the immiscible solvents in the separation step. An example of suc-
lack of efficient recycling methodologies that can be applied cessful industrial implementation is the biphasic Ruhrchemie/
on industrial scale currently restricts the use of homogeneous Rhône-Poulenc process for the hydroformylation of propene.7
catalysis to the production of valuable fine-chemicals.2 Stan- Other biphasic systems such as supported aqueous phase cata-
dard industrial purification operations such as distillation, lysis,8 fluorous biphasic catalysis,9 and the use of (supported)
extraction, and filtration are often incompatible or at least lim- ionic liquids10 and supercritical fluids11 have been reported.
iting when employing temperature-, hydrolysis- or oxidation-
sensitive molecular catalyst systems.
Previously, we reported the use of a non-covalent, supra-
molecular host–guest anchoring strategy for the immobilization
of homogeneous catalysts on silica supports with well-defined
binding sites.12 The supramolecular interaction between the
binding motif tethered to the transition-metal catalyst and the
binding site located on the support was sufficiently strong to
enable catalyst immobilization and efficient catalyst recycling.
The host–guest binding of the silica-supported systems was
found to correspond well with the analogous systems in solu-
tion. Importantly, it was shown that the binding is fully revers-
ible, and just changing the solvent allows the reuse of the
support. The supramolecular host–guest immobilization was
further developed for a Reverse-Flow Adsorption (RFA) reactor
aInCatT B.V. Science Park 904, 1098 XH Amsterdam, The Netherlands.
E-mail: joost.reek@incatt.nl, sander.kluwer@incatt.nl; Fax: +31 20 5256422;
Tel: +31 20 5256437
bCatfix B.V. Science Park 904, 1098 XH Amsterdam, The Netherlands
cVan’t Hoff Institute for Molecular Sciences, University of Amsterdam, Science Park
904, 1098 XH Amsterdam, The Netherlands. E-mail: j.n.h.reek@uva.nl,
A.M.Kluwer@uva.nl; Fax: +31 20 5256422; Tel: +31 20 5256437
†This contribution is dedicated to Professor David J. Cole-Hamilton on the
occasion of his 65th birthday.
‡Electronic Supplementary Information (ESI) available: Ligand synthesis, experi-
mental details and additional NMR spectra. See DOI: 10.1039/c2dt32047a
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 3609–3616 | 3609

View Article Online
Paper
Dalton Transactions
Scheme 1 Rhodium catalyzed hydroformylation of 1-octene leading to alde-
hyde products.
Fig. 1 Catalyst recycling via a reversible catalyst adsorption strategy using non-
covalent interactions.
that integrates selective recovery and recycling of the homo-
geneous catalysts from the reactor effluent.13 For this purpose
the binding motif to adsorb the catalyst was adjusted such
that the binding was weaker (Kads = 103), and a complementary
ligand-binding motif was covalently attached to the silica
support. This enables a one-step adsorption of both the tran-
sition metal complex and the excess ligand. Importantly, the
metal complex remained intact during the adsorption and
reaction conditions, which is crucial for efficient catalyst recy-
cling. A continuous process-integrated recycling of a rhodium-
catalyst for asymmetric hydrogenation of methyl acetamido-
acrylate and asymmetric hydrosilylation of acetophenone has
been demonstrated.13 The system is still far from perfect as the
adsorption process is limited by adsorption kinetics. As a
result, only a few percent of the binding sites on silica could
be effectively used. Also, in this non-optimized system, a large
part of the catalyst resides on the adsorption bed and is there-
fore not used. In addition, implementation of this concept
also requires a significant investment. Based on these results
and arguments we wondered if we could apply similar adsorp-
tion strategies for batchwise reactions.
We decided to investigate this catalyst recycling strategy for
the industrially important Rh-catalyzed hydroformylation. The
concept, displayed in Fig. 1, implies that after a batch reaction
the catalyst is selectively scavenged by the tailor-made support
using non-covalent binding. After separation of the catalyst
from the product mixture, the adsorption bed is washed with a
different (more polar) solvent to disrupt the supramolecular
interactions, thus releasing the catalyst from the bed. Reintro-
duction of the catalyst into the reactor and addition of new
reactants (1-octene and syngas) allows the subsequent catalytic
run with the recycled active catalyst. Here we report the use of
homogeneous Rh-catalysts tethered with a support-specific
binding motif for the hydroformylation of 1-octene under
homogeneous conditions (Scheme 1). Various commercially
Scheme 2 Functionalization of NixantPhos (1) with a pyridyl function for
reversible binding to Lewis and Brønsted acidic supports.
available supports with Lewis and Brønsted acidic sites have
been investigated in combination with a pyridyl-functionalized
NixantPhos (2) ligand. Two supports were selected for further
application in the hydroformylation reaction of 1-octene and
used in several consecutive reactions.
Results
XantPhos-type ligands are excellent ligands for Rh-catalyzed
hydroformylation.14 The combination of the large bite-angle
and the rigidity of the backbone allows α-olefins to be hydro-
formylated with high linear/branched ratios. Within the Xant-
Phos-family, NixantPhos (1) is most readily modified and
functionalized at the N–H and here we performed a hydroami-
nation reaction using 4-vinyl pyridine under basic conditions,
introducing a pyridine moiety in nearly quantitative yield (2,
Nix-py, see Scheme 2).
Catalyst Rh(2) was first evaluated in the hydroformylation
of 1-octene under batch conditions and compared to the per-
formance of the Rh/XantPhos catalyst (see Table 1). Both
systems display similar activity and selectivity, which implies
that the presence of the pyridyl moiety has no adverse effect
on the hydroformylation reaction. Clearly, the reaction can
also be run at more elevated temperatures (140 °C) with a high
S/C ratio (200 000), confirming that Rh(2) is sufficiently stable
to reach high turn-over numbers (TON of 170 000, see entry 3,
Table 1). Based on these experiments the catalyst recycling
experiments were designed to run in nearly neat 1-octene. The
3610 | Dalton Trans., 2013, 42, 3609–3616
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
Table 1 Hydroformylation of 1-octenea
Entry Ligand
S/Ce ratio Additive
Time (h) Conv. (%) Linear ald (%) Branched ald (%) Isomers (%) l : b
TOFc (h−1
)
1
XantPhos
10 000
10 000
200 000
10 000
—
—
23
23
1.8
21
92.2
89.6
87.1
22.0
93.8
93.6
91.5
89.0
2.7
2.4
2.3
2.7
3.5
4.0
6.1
8.3
35.0
38.9
39.0 33 137
32.8 103
406
386
2
Nix-py
Nix-py
Nix-py
3b
4
5 mL THF
SiO2/Al2O3
d
a 80 °C, 20 bars (H2/CO 1 : 1), 50 mL 1-octene, 7.8 mg Rh(acac)(CO)2, L/Rh = 5. b 140 °C, 20 bars (H2/CO 1 : 1), 45 mL 1-octene, 0.32 mg Rh(acac)-
(CO)2, L/Rh = 5. c Turn-over frequency defined as molaldehyde/molcatalyst × hour. d 1 g uncapped silica-alumina was added to the reaction mixture.
e S/C ratio = substrate to catalyst ratio.
Table 2 Adsorption constant (Kads) and number of binding sites for the
adsorption of 2 to different solid supports
Number of
Kads
Kads (M−1
nonanal
)
binding sites
(M−1
THF
)
Entry
Support
(mmol g−1
)
1
2
(SiO2)n–CH2CH2COOH
(SiO2)n–CH2CH2CH2–
NHC(O)NH(p-chloro)-
phenyl
7.0 × 102
<1
0.1
—
n.d.
n.d.
3
4
γ-Alumina
<1
—
0.3
n.d.
<5
Silica-alumina
capped
2 × 103
5
6
Silica-alumina
Silica-alumina
4 × 105
0.23
0.3
<5
n.d.
8.5 × 103
(5% THF)
7
Silica-alumina
6.4 × 103
0.3
n.d.
(15%THF)
Fig. 2 Flow experiments using 2.5 mM
2 and non-capped silica-alumina
(200 mg) (black), end-capped silica-alumina (200 mg) (red) and carboxylic acid
silica (300 mg) (green). After passing 5 mL nonanal containing 2 (0.5 mM) over
the column, the solvent was changed to pure THF and the flow was reversed.
Samples of 0.5 mL were collected. Amounts of 2 have been scaled by dividing
the cumulative amount measured (Csample × Vsample) by the initial concentration
(C(0) × Vtotal).
recycling experiments have been performed at lower reaction
temperatures (80 °C) and substrate/catalyst ratio (S/C = 10 000).
Furthermore, THF and nonanal, the principal product
obtained from hydroformylation, were used as solvents in the
binding studies.
The sorption characteristics of 2 were studied by determin-
ing the adsorption constants (Kads) of 2 to different solid sup- reported by De Haan et al. for adsorption/desorption processes
ports that contain either (1) only Brønsted acid sites ((SiO2)n– in RFA.17
CH2CH2COOH), (2) Lewis acid sites (neutral γ-alumina) or (3)
To study the adsorption/desorption in more detail, flow
both (silica-alumina).15 The results are collected in Table 2. experiments were conducted wherein a stock solution of
The adsorption constant (Kads) of 2 has been determined by 2.5 mM 2 in nonanal was passed over a packed column. The
fitting the data, obtained by direct titration, with a (modified) concentration of the ligand after the bed was determined
Langmuir-isotherm.16 Interestingly, there appears to be no using UV-Vis spectroscopy. After passing 5 mL of the stock
affinity of 2 for neutral γ-alumina while a very high binding solution over the column the flow was reversed and THF was
constant for the untreated silica-alumina was found (compare used as the desorption solvent for the ligand and the efficiency
entries 3 and 5). Capping the silica-alumina support with was determined by measuring the concentration of 2. Fig. 2
dimethyldimethoxysilane to convert the silanol groups into displays the flow experiments employing the different sup-
unreactive methyl end-groups reduces the binding constant by ports. The retention of 2 on the support is in line with the
a factor of 100 while the accessible number of binding sites adsorption constants in Table 2, with the ligand 2 strongly
remains the same. A similar drop in adsorption constant was adsorbed on silica-alumina and fortunately also being efficien-
obtained for the untreated silica-alumina if 5% or 15% THF tly desorbed. As silica-alumina showed the best retention-and-
was added to the nonanal solution during the adsorption release performance for 2, we studied the use of this combi-
experiment. This shows that limited amounts of desorption- nation in more detail.
solvent THF affect the binding to a limited extent, but under
Increasing the amount of solid support resulted in a better
these conditions the adsorption is still sufficient for selective retention of 2 (Fig. 3). Leaching of 2 (measured at 4.5 mL of
catalyst removal. The adsorption strength of 2 to capped silica- stock solution passed) was reduced from 14% for 200 mg to
alumina corresponds well to the optimum binding strengths 4% for 400 mg silica-alumina. In addition, desorption with
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 3609–3616 | 3611

View Article Online
Paper
Dalton Transactions
Fig. 5 Reactor setup and separation column. The red arrows indicate the flow-
direction of the reaction mixture (adsorption) and the blue arrows indicate the
flow of the THF (desorption).
Fig. 3 Flow experiments using 2.5 mM 2 with various amounts of silica-
alumina at a constant flow-rate (0.25 mL min−1). After passing 5 mL nonanal
containing 2 (0.5 mM) over the column, the solvent was changed to pure THF
and the flow was reversed. Samples of 0.5 mL were collected. Amounts of 2
have been scaled by dividing the cumulative amount measured (Csample
sample) by the initial concentration (C(0) × Vtotal).
×
subsequent runs. To probe a potential effect of the silica-
alumina support on the hydroformylation results, the reaction
was performed with the silica-alumina present in the reaction
mixture. Entry 3 of Table 2 shows that the selectivity of the
hydroformylation reaction is not affected, while the activity is
significantly lower, probably due to (partial) physisorption or
entrapment of the catalyst rather than actual deterioration of
the system although we have no conclusive proof to exclude
catalyst decomposition at this point.
V
Recycling experiments have been performed using the set-
up as shown in Fig. 5 under very similar reaction conditions as
in the batch reactions (Table 2). At all times during the re-
cycling process the syngas pressure was kept at 20 bars and the
conversion was kept below 90%. From preliminary experi-
ments it was found that the retention of the catalyst to the
adsorption-bed was improved when the reaction mixture was
cooled to room temperature before pumping the reaction
mixture over the absorber-bed. Unmodified silica-alumina
(entry 5, Table 2) with high binding affinity was initially
Fig. 4 Flow experiments using 2.5 mM 2 at different flow-rates with a constant selected for further studies. The flow-rate during the
amount of silica-alumina (400 mg). After passing 5 mL nonanal containing 2
adsorption and desorption process was kept constant at
(0.5 mM) over the column, the solvent was changed to pure THF and the flow
0.5 mL min−1
.
was reversed. Samples of 0.5 mL were collected. Concentrations have been
scaled by dividing by the start concentration (C(0)). Amounts of 2 have been
scaled by dividing the cumulative amount measured (Csample × Vsample) by the
start concentration (C(0) × Vtotal).
The hydroformylation results using the unmodified silica-
alumina are presented in Table 3. The first run shows
the typical results expected from this rhodium complex, com-
parable with a Rh-catalyst bearing the non-functionalized
NixantPhos ligand, but recycling experiments lead to partial
THF was not greatly affected and a high recovery of 2 was erosion of catalyst performance in the three subsequent runs.
observed in all cases (87–93%). Changing the flow-rate of the The selectivity for the linear product remains high and sub-
solution revealed more insight into the kinetics of the sorption strate isomerisation only becomes significant in the last reac-
processes. From Fig. 4 it is clear that a lower flow rate and tion. The activity drops substantially after each recycling
thus increased contact time greatly improves the retention of 2 procedure. Analysis of the collected effluent showed that the
to the support and reduces the leaching at 4.5 mL to 4% (com- rhodium leaching is low, between 0.5 and 1.3%. These data
pared to 18% for a flow rate of 2.0 mL min−1), while also the strongly indicate that the Rh(2) catalyst is scavenged very well
desorption profile improves with a recovery of 2 of 87%.
from the reaction mixture by the solid support, but that the
For optimal recycling of the catalyst, the rhodium complex active catalyst is either not efficiently desorbed or is deacti-
should remain in its active form to be efficiently used in the vated by the support. Possibly, the catalyst resting state
3612 | Dalton Trans., 2013, 42, 3609–3616
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
Table 3 Recycle of the hydroformylation catalyst using silica-aluminaa
Run
Time (h)
Conv. (%)
Linear ald (%)
Branched ald (%)
Isomers (%)
l : b
TOFb (h−1
)
[Rh]c mg L−1
[P]c mg L−1
1
2
3
4
18
18
19
19
92.3
64.5
62.5
25.6
94
93
91
81
2.2
2.2
2.4
2.6
4.1
4.6
6.3
43.0
41.8
38.6
31.8
1134
797
732
269
0.16 (0.7%)
0.13 (0.5%)
0.32 (1.3%)
0.16 (0.7%)
11.3 (11%)
13.1 (13%)
14.3 (14%)
6.15 (8%)
15.8
a 90 °C, 20 bars (H2/CO 1 : 1), 45 mL 1-octene, 5 mL THF, 3.1 mg Rh(acac)(CO)2, S/Rh = 22 500, L/Rh = 5, 2.6 g SiO2/Al2O3. b Turn-over frequency
defined as molaldehyde/(molcatalyst × hour). c Concentration of rhodium and phosphorus in the collected product. Numbers in brackets are the
percentage of initial concentration.
Table 4 Recycle of the hydroformylation catalyst using end-capped silica-aluminaa
Run
Time (h)
Conv. (%)
Linear ald (%)
Branched ald (%)
Isomers (%)
l : b
TOFb (h-1)
[Rh]c mg L−1
[P]c mg L−1
1
2
3
4
5
6
7
8
18
18
19
19
25
17
18
20
90
73
75
60
63
77
61
68
92
93
56
43
17
84
92
71
2.1
2.2
4.1
5.7
4.4
2.4
2.3
3.1
5.7
4.5
39.5
50.9
78.6
14.0
6.1
43.3
42.3
13.9
7.6
1116
924
557
385
129
920
768
612
0.13 (0.6%)
0.20 (0.8%)
1.8 (7.4%)
0.8 (3.5%)
0.3 (1.3%)
0.5 (2.1%)
0.3 (1.2%)
n.d.
14 (20%)
11 (15%)
13 (18%)
8 (11%)
8 (11%)
25 (—)
3.9
34.8
40.4
22.7
11 (—)
n.d.
25.5
a 90 °C, 20 bars (H2/CO 1 : 1), 45 mL 1-octene, 5 mL THF, 3.1 mg Rh(acac)(CO)2, S/Rh = 22 500, L/Rh = 5, 2.6 g Aldrich Catalyst carrier SiO2/Al2O3.
b Turn-over frequency defined as molaldehyde/(molcatalyst × hour). c Concentration of rhodium and phosphorus in the collected product. Numbers
in brackets are the percentage of initial concentration.
Table 5 Recycle of the hydroformylation catalyst using end-capped silica-alumina with one equivalent ligand added in every runa
Run
Time (h)
Conv. (%)
Linear ald (%)
Branched ald (%)
Isomers (%)
l : b
TOFb (h−1
)
[Rh]c mg L−1
1
2
3
4
5
18
18
18
18
18
87
58
56
69
37
94
92
91
90
45
2.2
2.2
2.2
2.5
4.7
3.7
6.0
6.6
8.0
50
43.2
41.4
41.0
36.2
9.6
1107
721
688
832
18
<3 (<0.2%)
10 (0.6%)
13 (0.7%)
96 (5.3%)
27 (1.5%)
a 90 °C, 20 bars (H2/CO 1 : 1), 45 mL 1-octene, 5 mL THF, 3.1 mg Rh(acac)(CO)2, S/Rh = 22 500, L/Rh = 5, 2.6 g Aldrich Catalyst carrier SiO2/Al2Os.
b Turn-over frequency defined as molaldehyde/(molcatalyst × hour). c Concentration of rhodium in the collected product. Numbers in brackets are the
percentage of initial concentration.
[Rh(2)H(CO)2] reacts with Brønsted acidic sites of the support, restored to initial levels in the seventh run. Although the reten-
producing cationic rhodium complexes.6b To circumvent these tion of the ligand was not ideal, these results show the proof-
problems the end-capped silica-alumina, lacking reactive of-principle of a reversible supramolecular catalyst scavenging
(acidic) OH-groups, was subsequently investigated in the technology.
catalyst recycling.
With the knowledge that part of the ligand is lost during
The hydroformylation results using the end-capped silica- the recycling, the recycling experiments were performed with
alumina are presented in Table 4. Again, the first reaction an additional equivalent ligand in each batch reaction. As can
showed the typical results expected from a Rh(NixantPhos) be seen from Table 5, four successful recycling experiments
complex, with high activity and selectivity for the linear alde- were performed where the catalyst was taken out of the reactor,
hyde obtained. Upon recycling, the catalyst activity gradually passed over an adsorption bed and reintroduced into the
drops from a TOF of 1116 to 557 in the first four runs. Also the reactor. The activity dropped slightly in the second run but
selectivity towards linear aldehyde drops significantly in the remained stable until the fifth run. Importantly, the selectivity
third run. The linear/branched ratio in the fifth run is close to can be retained in the first four reactions under these con-
the ligand-free hydroformylation system, suggesting that the ditions, but is almost completely lost in the fifth run.
ligand leached from the reactor. This is also confirmed by the Rhodium analysis of the effluent (see Table 5) and the
phosphorus leaching data in Table 4. Extra ligand was added rhodium levels in the autoclave (27% of the initial amount)
to the reaction mixture in the sixth run and indeed the hydro- show that a significant amount of rhodium still resides on the
formylation activity and selectivity were almost completely catalyst scavenger bed. Although the properties of the catalyst
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 3609–3616 | 3613

View Article Online
Paper
Dalton Transactions
are lost in the fifth run, these results clearly show that the Carboxylic Acid) was purchased from Silicycle Inc., γ-alumina
combination of batchwise reactions and catalyst scavenging (Merck KGaA, 70–230 mesh) was purchased from VWR Inter-
using a selective adsorbent is very promising. Further research national. Capping of silica alumina was performed as
should focus on optimization of selective adsorbent materials.
described earlier.18
Synthesis of 4,6-bis(diphenylphosphino)-10-(2-(pyridin-4-yl)-
ethyl)-10H-phenoxazine (2). To
a mixture of NixantPhos
(10.3 g, 18.7 mmol) and tetrabutylammonium bromide (1.3 g,
3.9 mmol) in THF (100 mL) was added vinyl pyridine (20 mL,
Conclusions
In this contribution we have demonstrated a new strategy for 185 mmol) and 10% NaOH solution (30 mL). The mixture was
the recycling of a homogeneous hydroformylation catalyst, by heated to gentle reflux for 24 hours. After cooling, the mixture
selective adsorption of the catalyst to tailor-made supports was evaporated till dryness and water (500 mL) was added. The
after a batchwise reaction. The binding of the catalyst is based yellow solid was filtered off and washed 3 times with water
on selective non-covalent interactions between ligand and (300 mL). The solid was dissolved in dichloromethane and
support. For this purpose, the NixantPhos ligand has been subsequently washed with 200 mL water. The organic phase
functionalized with pyridine groups. The corresponding Rh was dried over Na2SO4 and the solvent was evaporated. The
catalyst is very selective, active and highly stable, reaching a remaining solid was recrystallized from a mixture of acetone
TON of 170 000. Various commercially available supports were (400 mL) and methanol (600 mL). Off-white 4,6-bis(diphenyl-
explored, with the end-capped silica-alumina showing the best phosphino)-10-(2-(pyridin-4-yl)ethyl)-10H-phenoxazine was col-
results of this study with respect to ligand-adsorbing proper- lected (11.70 g, 96%).
3
ties. The related unfunctionalized analogue binds better, but
1H NMR (400 MHz): δ (ppm) 8.56 (d, J = 6.0 Hz, 2H), 7.22
this material accelerates catalyst deactivation during the recy- (m, 12H), 6.72 (pseudo-triplet, 3J = 7.8 Hz), 6.52 (d, 3J = 7.2 Hz,
cling experiment. Under optimized conditions, four successful 2H), 6.05 (d, 3J = 9.2 Hz, 2H), 3.8 (m, 2H), 3.0 (m, 2H).
recycling experiments were performed (at 90 °C, 20 bars
13C NMR (101 MHz): δ (ppm) 150.01 (C–H), 147.13 (Cq),
syngas pressure, substrate/catalyst ratio 22 500). The activity 136.65 (t, 1J(P,C) = 6.0 Hz, Cq), 133.68 (t, 2J(P,C) = 11.1 Hz,
dropped slightly in the second run but remained stable C–H). 132.41 (Cq), 128.13 (C–H), 128.00 (C–H), 125.54 (C–H),
until the fifth run after which the activity and selectivity 123.83 (C–H), 123.59 (C–H), 111.56 (C–H), 44.75 (CH2), 30.48
were lost. Further optimization is required before such recyc- (CH2).
ling strategies can be efficiently applied for commercial
processes, but proof-of-concept for this supramolecular
31P NMR (162 MHz): δ (ppm) −19.02 (s).
FAB+-MS: m/z calcd for C43H35ON2P2 ([MH]+): 657.2225,
scavenging strategy is demonstrated by the experiments out- obsd 657.2232.
lined in this paper.
Determination of adsorption constant and number of binding
sites
Experimental
General procedures
For the determination of the adsorption constant and the
number of binding sites, ten portions of silica were accurately
weighed (ranging from 0–80 mg) and transferred to sealable
vials. 2 mL portions of a 1.5 mM stock solution of 2 were
added to each silica containing vial. The mixture was shaken
at 21 °C for 17 hours. The equilibrated solutions were
measured in a 1 mm light path cuvette. For optimum response
250 μL of the solution and 2000 μL of the corresponding
solvent were added (9 times dilution). The cuvette was
Unless stated otherwise, reactions were carried out under an
inert atmosphere of nitrogen or argon using standard Schlenk
techniques. THF was distilled from sodium benzophenone
ketyl, methanol was distilled from CaH2, and toluene was dis-
tilled from sodium, all under a nitrogen atmosphere. NMR
spectra were measured on a Bruker AV 400 (1H: 400 MHz, 31P-
{1H}: 162 MHz and 13C{1H}: 101 MHz) at room temperature.
Chemical shifts are reported in ppm and are given relative to
tetramethylsilane (1H and 13C{1H}) or 85% H3PO4 (31P{1H}) as
external standards. High resolution mass spectra (HRMS) were
recorded using FAB+ ionization on a JEOL JMS SX/SX 102A
four-sector mass spectrometer with 3-nitrobenzyl alcohol as
the matrix. Rhodium and phosphorus analyses were per-
formed on an ICP-OES, PerkinElmer Optima 3000XL with a
detection limit of 1.4 μg l−1 for rhodium (1.4 ppb) and a deter-
measured
3 times. After base-line correction the three
measurements were averaged and analyzed. For the analysis
the absorption at the absorption maximum 343 nm (ε = 9140
M−1 cm−1) was determined and the results were plotted as
[guest]equilibrium versus the amount of silica (in mg) and fitted
to a (modified) Langmuir isotherm.16
Recycling experiments
mination limit of 4.2 μg l−1 for rhodium (4.2 ppb). All commer- A stainless steel HPLC column (25 cm × 4.6 mm ID × 1/4″ OD)
cially available reagents were used without further was charged with 2.7 g methylated aluminosilicate, after which
purification, unless indicated otherwise. Silica alumina (silica- it was mounted in the reactor setup as depicted in Fig. 5. The
alumina catalyst support, grade 135) was purchased from lines, including the column, were washed with syngas satu-
Sigma Aldrich, (SiO2)n–CH2CH2COOH silica (SiliaBond® rated toluene prior to the hydroformylation experiments.
3614 | Dalton Trans., 2013, 42, 3609–3616
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
In a typical experiment, 1-octene (45 mL, 287 mmol) and
THF (5 mL) were added to NixPy (39.3 mg, 0.06 mmol) and Rh-
(acac)(CO)2 (3.12 mg, 0.012 mmol) under an Ar atmosphere. At
a temperature of 90 °C the reaction vessel was pressurized to
20 bars H2/CO (1 : 1) by four purge cycles. After the desired
time the reaction mixture was allowed to cool down to approxi-
mately 30 °C, at which temperature it was pumped out at
0.5 mL min−1. During the entire procedure the system
remained under 20 bars of syngas. When the reaction mixture
was almost completely removed the flow direction was reversed
and the column was flushed by 5 mL THF under 20 bars
syngas at 0.15 mL min−1. In the same way 45 mL 1-octene was
transferred into the reaction vessel and the second run is
initiated by heating the mixture to 90 °C. This entire procedure
is repeated for the additional runs.
and F. R. Hartley, Chem. Rev., 2002, 102, 3215, thematic
issue; Y. R. de Miguel, E. Brulé and R. G. J. Margue,
J. Chem. Soc., Perkin Trans. 1, 2001, 3085; S. Kobayashi and
R. Akiyama, Chem. Commun., 2003, 449; S. Bräse,
F. Lauterwasser and R. E. Ziegert, Adv. Synth. Catal., 2003,
345, 869; Z.-L. Lu, E. Lindner and H. A. Mayer, Chem. Rev.,
2002, 102, 3589.
6 Other interesting contributions: (a) A. Ajjou and H. Alper,
J. Am. Chem. Soc., 1998, 120, 1466; (b) A. J. Sandee, L. van
der Veen, J. N. H. Reek, P. C. J. Kamer, M. Lutz, A. L. Spek
and P. W. N. M. van Leeuwen, Angew. Chem., Int. Ed., 1999,
38, 3231; (c) A. J. Sandee, J. N. H. Reek, P. C. J. Kamer and
P. W. N. M. van Leeuwen, J. Am. Chem. Soc., 2001, 123,
8468; (d) M. R. Buchmeiser and K. Wurst, J. Am. Chem. Soc.,
1999, 121, 11101; (e) D. E. Bergbreiter, Y.-S. Liu and
P. L. Osburn, J. Am. Chem. Soc., 1998, 120, 4250;
(f) M. K. Richmond, S. L. Scott and H. Alper, J. Am. Chem.
Soc., 2001, 123, 10521; (g) A. J. Sandee, D. G. I. Petra,
J. N. H. Reek, P. C. J. Kamer and P. W. N. M. van Leeuwen,
Chem.–Eur. J., 2001, 7, 1202; (h) D. E. Bergbreiter,
P. L. Osburn and Y.-S. Liu, J. Am. Chem. Soc., 1999, 121,
9531; (i) K. Nozaki, Y. Itoi, F. Shibahara, E. Shirakawa,
T. Ohta, H. Takaya and T. Hiyama, J. Am. Chem. Soc., 1998,
120, 4051; ( j) M. O. Farrell, C. H. van Dyke, L. J. Boucher
and S. J. Metlin, J. Organomet. Chem., 1979, 172, 367;
(k) S. C. Bourque and H. Alper, J. Am. Chem. Soc., 2000, 122,
956; (l) M. Janssen, J. Wilting, C. Müller and D. Vogt,
Angew. Chem., Int. Ed., 2010, 49, 7738; (m) R. Sablong,
J. I. van der Vlugt, R. Thomann, S. Mecking and D. Vogt,
Adv. Synth. Catal., 2004, 347, 633.
7 Chapter 6 in Aqueous-Phase Organometallic Catalysis: Con-
cepts and Applications, ed. B. Cornils and W. A. Herrmann,
Wiley-VCH Verlag, Weinheim, 1998, p. 271; J. Herwig
and R. Fischer, in Rhodium Catalyzed Hydroformylation,
ed. P. W. N. M. van Leeuwen and C. Claver, Kluwer
Academic Publishers, Dordrecht, The Netherlands,
2000.
8 J. P. Arhancet, M. E. Davis, J. S. Merola and B. E. Hanson,
Nature, 1989, 339, 454; M. E. Davis, in Aqueous-Phase
Organometallic Catalysis: Concepts and Applications,
ed. B. Cornils and W. A. Herrmann, Wiley-VCH Verlag,
Weinheim, 1998, pp. 241; A. J. Sandee, V. F. Slagt,
J. N. H. Reek, P. C. J. Kamer and P. W. N. M. van Leeuwen,
Chem. Commun., 1999, 1633.
Notes and references
1 B. M. Trost, Science, 1991, 254, 1471.
2 R. T. Baker and W. Tumas, Science, 1999, 284, 1477;
D. J. Cole-Hamilton, Science, 2003, 299, 1702; Catalyst Sep-
aration, Recovery and Recycling: Chemistry and Process
Design, ed. D. J. Cole-Hamilton and R. P. Tooze, Springer,
Dordrecht, The Netherlands, 2006.
3 M. Janssen, C. Müller and D. Vogt, Green Chem., 2011, 13,
2247.
4 R. van Heerbeek, P. C. J. Kamer, P. W. N. M. van Leeuwen
and J. N. H. Reek, Chem. Rev., 2002, 102, 3717; R. Kreiter,
A. W. Kleij, R. J. M. Klein Gebbink and G. van Koten, in
Dendrimers IV: Metal Coordination Self-Assembly, Catalysis,
ed. F. Vögle and C. A. Schalley, Springer-Verlag, Berlin,
2001, pp. 163; G. E. Oosterom, J. N. H. Reek, P. C. J. Kamer
and P. W. N. M. van Leeuwen, Angew. Chem., Int. Ed., 2001,
40, 1828; R. M. Crooks, M. Zhao, L. Sun, V. Chechik and
L. K. Yueng, Acc. Chem. Res., 2001, 34, 181; J. N. H. Reek,
D. de Groot, G. E. Oosterom, P. C. J. Kamer and
P. W. N. W. van Leeuwen, Rev. Mol. Biotechnol., 2002, 90,
159; D. Astruc and F. Chardac, Chem. Rev., 2001, 101,
2991.
5 Extensive reviews on polymer-immobilized catalysts
include: F. R. Hartley, Supported Metal Complexes. A New
Generation of Catalysts, Reidel, Dordrecht, The Netherlands,
1985; Y. Iwasawa, in Tailored Metal Catalysts Metal Cata-
lysts, ed. R. Ugo and B. R. James, Reidel, Dordrecht, The
Netherlands, 1986; W. Keim and B. Driessen-Hölscher, in
Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözin-
ger and J. Weitkamp, Wiley-VCH, Weinheim, Germany,
9 I. T. Horváth and J. Rábai, Science, 1994, 266, 72;
I. T. Horváth, Acc. Chem. Res., 1998, 31, 641; E. de Wolf,
G. van Koten and B.-J. Deelman, Chem. Soc. Rev., 1999, 28,
37; R. H. Fish, Chem.–Eur. J., 1999, 5, 1677.
1997, vol. 1, pp. 231; E. Lindner, T. Schneller, F. Auer and 10 P. Wasserscheid and T. Welton, in Ionic Liquids in Synthesis,
H. A. Mayer, Angew. Chem., Int. Ed., 1999, 38, 2154;
J. M. Thomas, Angew. Chem., Int. Ed., 1999, 38, 3588;
J. M. Thomas, E. Lindner, T. Schneller, F. Auer,
H. A. Mayer, W. Keim, B. Driessen-Hölscher, Y. Iwasawa
and F. R. Hartley, J. Mol. Catal. A, 2002, 177, 1, thematic
issue; J. M. Thomas, E. Lindner, T. Schneller, F. Auer,
H. A. Mayer, W. Keim, B. Driessen-Hölscher, Y. Iwasawa
Wiley-VCH, Weinheim, Germany, 2003; T. Welton,
Chem. Rev., 1999, 99, 2071; R. P. J. Bronger, S. M. Silva,
P. C. J. Kamer and P. W. N. M. van Leeuwen, Chem.
Commun., 2002, 3044; W. Keim, D. Vogt, H. Waffenschmidt
and P. Wasserscheid, J. Catal., 1999, 186, 481; X. Jin,
D. Yang, X. Xu and Z. Yang, Chem Commun., 2012, 48, 9017
Supported ionic liquids: S. Shylesh, D. Hanna, S. Werner
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 3609–3616 | 3615

View Article Online
Paper
Dalton Transactions
and A. T. Bell, ACS Catal., 2012, 2, 487; C. P. Mehnert, 14 P. W. N. M. van Leeuwen and Z. Freixa, in Modern Carb-
R. A. Cook, N. C. Dispenziere and M. Afeworki, J. Am.
Chem. Soc., 2002, 124, 12932; C. P. Mehnert, Chem.–Eur. J.,
2005, 11, 50.
onylation Methods, ed. L. Kollár, WILEY-VCH Verlag,
Weinheim, 2008.
15 D. Ballivet, D. Barthomeuf and P. Pichat, J. Chem. Soc.,
Faraday Trans. 1, 1972, 68, 1712.
11 M. F. Sellin, P. B. Webb and D. J. Cole-Hamilton, Chem.
Commun., 2001, 781; P. B. Webb, M. F. Sellin, T. E. Kunene, 16 S. Sohn and D. Kim, Chemosphere, 2005, 58, 115;
S. Williamson, A. M. Z. Slawin and D. J. Cole-Hamilton,
A. W. Adamson, Physical Chemistry of Surfaces, John Wiley &
J. Am. Chem. Soc., 2003, 125, 15577.
Sons, 5th edn, 1990, p. 591.
12 R. Chen, R. P. J. Bronger, P. C. J. Kamer, P. W. N. M. van 17 T. Dekic Zivkovic and A. B. de Haan, AIChE J., 2008, 54,
Leeuwen and J. N. H. Reek, J. Am. Chem. Soc., 2004, 126,
14557.
13 F. Marras, P. W. N. M. van Leeuwen and J. N. H. Reek,
Chem.–Eur. J., 2011, 17, 7460.
2392.
18 F. Marras, A. M. Kluwer, J. R. Siekierzycka, A. Vozza,
A. M. Brouwer and J. N. H. Reek, Angew. Chem., Int. Ed.,
2010, 49, 5480.
3616 | Dalton Trans., 2013, 42, 3609–3616
This journal is © The Royal Society of Chemistry 2013