Tuning the recovery of an Rh-containing catalyst with silica-based (poly) amine ion exchangers

Article abstract of DOI:10.1039/b108457g

Highly efficient Rh-recovery from different adsorption media has been effected with silica-based (chelating) ion exchangers containing (poly) amine functionalities; recoveries have been found to correlate well with the stability of the metal-to-ligand com

Full text of DOI:10.1039/b108457g

View Article Online / Journal Homepage / Table of Contents for this issue
Tuning the recovery of an Rh-containing catalyst with silica-based
(poly) amine ion exchangers
Jurjen Kramer, Ana Ríos García, Willem L. Driessen* and Jan Reedijk
Leiden Institute of Chemistry, Leiden University, P.O. Box 9502 2300 RA. Leiden, The Netherlands.
E-mail: driessen@chem.leidenuniv.nl
Received (in Cambridge, UK) 19th September 2001, Accepted 12th October 2001
First published as an Advance Article on the web 29th October 2001
Highly efficient Rh-recovery from different adsorption
media has been effected with silica-based (chelating) ion
exchangers containing (poly) amine functionalities; recov-
eries have been found to correlate well with the stability of
the metal-to-ligand complexes.
latter two ion exchangers proved to be some 3–10 times higher
over the entire tested pH range (Fig. 1).
An explanation for the observed behaviour lies in the ability
of 2 and 3 to form highly stable chelate rings with Rh, whereas
1 can only coordinate in a monodentate fashion. Another
striking feature is the observed log D difference between 2 and
4. The experiment clearly demonstrates, that the complex
formation of ligands that contain five-membered chelate rings
with Rh metal-ions is favoured over those forming six-
membered rings. This is in agreement with previous findings on
metal ion selectivity, that already indicated the preference of
relatively large metal ions to coordinate with chelating ligands
via five-membered rings whereas relatively small transition
metals prefer larger, six-membered rings.13
Rh-containing homogeneous catalysts play an important role in
the improvement of many important industrial processes, such
as hydroformylation and hydrogenation reactions.1 As these
catalysts are widely used and since Rh is a very scarce element,
the efficient separation of the used metal-containing catalyst
from product streams is of paramount importance. However, the
complete recovery of the often very low concentration of
catalyst is a difficult task. A highly efficient technique for the
recovery of low concentrations of metal ions is ion exchange,
due to the ease of separation of the adsorbed metal ions from the
product stream. Earlier findings indicated that silica is a very
suitable backbone, due to its high chemical and thermodynam-
ical stability and fast recovery kinetics.2
One of the difficulties is to find optimal binding. This means,
the ligandsA binding capacity should be sufficiently strong. On
the other hand, if the coordination is too strong, the desorption
becomes very difficult, as previous studies on the recovery of
two Rh-containing model catalyst with ion exchangers that
possessed ligands with both N and S donor atoms already
pointed out.3
Amine ligands are known to coordinate well to platinum
group metals and a number of studies have focussed on their
application, mainly as anionic extractants for hydrometallur-
gical purposes.4,5 The efficient recovery of Rh almost always
proved to be the bottleneck, though. The Rh extraction is
complicated due to its chemical behaviour in chloride media,
and desorption of the loaded ion exchangers is often incomplete
or only possible under very harsh elution conditions.6,7
This communication describes our first investigations on the
tuning of the recovery of RhCl3, which is both used as a catalyst
precursor and as a real catalyst,8–11 with silica-based ion
exchangers containing (poly) amine functionalities.
At low pH ( < 2), the amines are protonated14 and binding
will probably proceed via an ion association way, with
positively charged amine ligands binding in an ionic fashion to
a
negatively charged Rh complex of the type
[RhCl6 2 n(H2O)n](3 2 n)2 (with n generally 0–2).6 The some-
what lower log D values at pH 1–2, are caused by the presence
of Cl2 ions from the HCl/NaCl buffers which compete with the
metal ions for binding to the ion exchangers. At higher pH,
HOAc/NaOAc buffers have been used, and these coordinate
less strongly to the Rh metal-ions. Still, the acetate ions, present
in increasing amounts with increasing pH, also compete when
the ligand to metal complexes are weaker (Fig. 1, 1 and 4).
For the recovery of Rh, ion exchangers 1–4 were added to a
fresh solution of RhCl3·3H2O‡ in water, 1 M HCl or EtOH
(absolute), to study the effect of the adsorption medium on the
stripping possibility. The desorption was done in 2 M HNO3, as
it proved to be the most effective stripping agent. The results are
depicted in Fig. 2. Since the ligand concentrations vary
significantly between the different ion exchangers (see Table 1)
it is more correct to not only look at the uptake capacity (C,
mmol g21) but to compare the ligand occupation (L, %) as well.
This value is defined in the following way: L (%) = [uptake
capacity (C) for Mn+ (mmol g21)]/[ligand concentration (mmol/
g silica)].
The ion exchangers used for this investigation contain a
monoamine (1), ethylenediamine (2), diethylenetriamine (3)12
and propylenediamine (4) functionalities (Table 1).†
In order to gain insight into the stability of the binding of the
different ligands with the Rh metal-ions, distribution coeffi-
cients have been measured at different pH values (Fig. 1).3
Clearly, the log D values of ion exchangers 1 and 4 are much
lower than of 2 and 3. The stability of Rh-complexes with the
The maximum ligand capacities have been found to increase
with increasing ligand concentration. A maximum of over 0.5
mmol g21 ion exchanger could be extracted ( > 50 mg Rh/g ion
exchanger). The extraction percentage is clearly very dependent
on the adsorption medium; the Rh adsorption in water and
ethanol proved to be much more efficient (about 40–70%) than
Table 1 Ion exchanger characteristics
Ion
exchanger
Ligand conc./
mmol g21a
Grafted group
1
2
3
4
–(CH2)3NH2
1.14
1.02
0.63
0.68
–(CH2)3NHCH2CH2NH2
–(CH2)3NH(CH2CH2NH)2H
–(CH2)3NH2CH2CH2CH2NH2
Fig. 1 The distribution coefficients for Rh3+ uptake using ion exchangers
1–4. D = (mmol Mn+/g of dry ion exchanger)/(mmol Mn+/ml of solution).
The ligand to metal ratio is 5+1.
a Based on the N content.
2420
Chem. Commun., 2001, 2420–2421
This journal is © The Royal Society of Chemistry 2001
DOI: 10.1039/b108457g

View Article Online
Fig. 2 Recovery experiments of Rh3+-loaded ion exchangers 1 (top left), 2 (top right), 3 (bottom left) and 4 (bottom right).§ Black bars indicate the loaded
ion exchangers after adsorption of Rh3+ in H2O, 1 M HCl or absolute EtOH for 48 h. White bars indicate the amount of Rh3+ left on the ion exchanger after
desorption in 2 M HNO3 for 48 h. Reagents and conditions: adsorption: RhCl3·3H2O (an amount, corresponding to a ligand to metal ratio of 1+1) in 50 ml
of solvent (H2O, 1 M HCl or absolute EtOH), 200 mg ion exchanger, agitation for 2 days. Desorption: 100 mg loaded ion exchanger, 50 ml 2 M HNO3,
agitation for 2 days.
in 1 M HCl (maximum 25%). The difference can be explained
by a different extraction mechanism under the adsorption
conditions, as discussed above. The large amounts of chloride
ions strongly compete with the metal ions for the positively
charged binding sites. The possible coordination of more than
one ligand to a single Rh metal-ion might be an explanation for
the inability of any of the ion exchangers to attain a 100% ligand
occupancy.
stimulating discussions. Mr J. J. M. van Brussel is thanked for
his technical assistance.
Notes and references
† A mixture of 3-chloropropyltrimethoxysilane (83 mmol) and 1,3-diami-
nopropane (250 mmol) was heated under argon at 150 °C for 6 h. After
cooling, the white solid phase was filtered off and the liquid phase was
purified by vacuum distillation (bp 112–114 °C/1 mm Hg). Yield 5.18 g
(22%).1H NMR (300 MHz, CDCl3): d 0.61 (t, 2H, SiCH2) 1.00 (br s, 2H,
NH2), 1.21 (t, 9H, OCH2CH3), 1.55–1.65 (m, 4H, SiCH2CH2,
NHCH2CH2CH2NH2), 2.59–2.64 (2 3 t, 4H, CH2NHCH2), 2.74 (t, 2H,
CH2NH2), 3.80 (q, 6H, OCH2CH3). Subsequent immobilisation was
performed similarly to ion exchangers 1–3.12
‡ As RhCl3 shows extensive aquation (the formation of kinetically rather
inert aquochloro complexes) with time, freshly prepared RhCl3-solutions
( < 10 min old) were always used. Furthermore, hardly any variation in the
pH has been observed.
The desorption possibility (Fig. 2, right bars), after Rh
adsorption in water, increased in the following way: 3 í 2 í
4 = 1
The stripping percentages range from 45% when the
diethylenetriamine ion exchanger 3 is used to > 90% for the
monoamine ion exchanger 1. Evidently, the formation of five-
membered chelate rings has a strongly negative effect on the
desorption. This effect already becomes clear with ethylenedi-
amine but is more strongly displayed when multiple chelate
rings are formed. Apparently, the six-membered chelate rings,
that ion exchanger 4 forms with the metal ions, are stable
enough for good adsorption (64%), but not too stable, to prevent
easy desorption.
§ The determinations of the uptake capacities and desorption, are based
upon the metal content on the ion exchangers, established by concentrated
acid digestion as previously described.3
1 B. Cornils and W. A. Herrmann, Applied Homogeneous Catalysis with
Organometallic Compounds, ed. B. Cornils and W. A. Herrmann, VCH
Publishers, NY, 1996, vol. 29.
2 H. J. Hoorn, P. deJoode, D. J. Dijkstra, W. L. Driessen, H. Kooijman, N.
Veldman, A. L. Spek and J. Reedijk, J. Mater. Chem., 1997, 7, 1747.
3 J. Kramer, A. Scholten, W. L. Driessen and J. Reedijk, Inorg. Chim.
Acta, 2001, 315, 183.
These desorption differences are absent when the adsorption
was performed in 1 M HCl, demonstrating the absence of
binding in a chelate fashion under these conditions. Although
the adsorptions were much lower, a good desorption was
obtained with all the ion exchangers ( > 80%).
The desorption, after adsorption in ethanol, proved to be very
difficult for all the ion exchangers used. This has most probably
to do with partial reduction of the Rh metal ions in this solvent,
as reported by Hernan et al.15 and Kriek et al.16
4 G. P. Demopoulos, Hydrometallurgy, 1989, 82, 165.
5 I. A. Kovalev, L. V. Bogacheva, G. I. Tsysin, A. A. Formanovsky and
Y. A. Zolotov, Talanta, 2000, 52, 39.
6 E. Benguerel, G. P. Demopoulos and G. B. Harris, Hydrometallurgy,
1996, 40, 135.
7 M. S. Alam, K. Inoue and K. Yoshizuka, Hydrometallurgy, 1998, 49,
213.
8 H. J. Borg, L. C. A. Vandenoetelaar and J. W. Niemantsverdriet, Catal.
In conclusion, it is shown that very efficient recovery of Rh3+
has been achieved with amine-containing ion exchangers that
form complexes with log D values lying roughly between 2.0
and 2.5. Both the ligand capacity and the elution percentage of
the metal ion have proven to be very dependent on the
adsorption conditions.
Lett., 1993, 17, 81.
9 J. Blum, Russ. Chem. Bull., 1993, 42, 1619.
10 M. Lin and A. Sen, Nature, 1994, 368, 613.
11 M. M. Mdleleni, R. G. Rinker and P. C. Ford, Inorg. Chim. Acta, 1998,
270, 345.
The work is currently being extended to the possible
application in a continuous system. This implies that important
factors such as adsorption and desorption kinetics, successive
recovery and the use of other Rh-containing catalysts will be
investigated.
Financial support comes from the IOP (Innovation Oriented
Research Programme, The Netherlands) Environmental Tech-
nology/Heavy Metals cluster Separation, project number
IZW97411. Dr W. Buijs (DSM-Research) is acknowledged for
12 J. Kramer, W. L. Driessen, K. R. Koch and J. Reedijk, Hydrometallurgy,
to be submitted.
13 P. Comba, Coord. Chem. Rev., 1999, 185–186, 81.
14 B. K. Tait and D. P. Shillington, S. Afr. J. Chem., 1992, 45, 17.
15 P. Hernan, C. del Pino and E. Ruiz-Hitzky, Chem. Mater., 1992, 4,
49.
16 R. J. Kriek, W. J. Engelbrecht and J. J. Cruywagen, J. S. Afr. Inst. Min.
Metall., 1995, 75.
Chem. Commun., 2001, 2420–2421
2421