Immobilized bisdiazaphospholane catalysts for asymmetric hydroformylation

Article abstract of DOI:10.1021/ja501568k

Condensation reactions of enantiopure bis-3,4-diazaphospholanes (BDPs) that are functionalized with carboxylic acids enable covalent attachment to bead and silica supports. Exposure of tethered BDPs to the hydroformylation catalyst precursor, Rh(acac)(CO)₂, yields catalysts for immobilized asymmetric hydroformylation (iAHF) of prochiral alkenes. Compared with homogeneous catalysts, catalysts immobilized on Tentagel resins exhibit similarly high regioselectivity and enantioselectivity. When corrected for apparent catalyst loading, the activity of the immobilized catalysts approaches that of the homogeneous analogues. Excellent recyclability with trace levels of rhodium leaching are observed in batch and flow reactor conditions. Silica-bound catalysts exhibit poorer enantioselectivities.

Full text of DOI:10.1021/ja501568k

Article
pubs.acs.org/JACS
Immobilized Bisdiazaphospholane Catalysts for Asymmetric
Hydroformylation
Tyler T. Adint and Clark R. Landis*
Department of Chemistry, University of Wisconsin-Madison, 1101 University Avenue, Madison Wisconsin 53706, United States
S
* Supporting Information
ABSTRACT: Condensation reactions of enantiopure bis-3,4-
diazaphospholanes (BDPs) that are functionalized with
carboxylic acids enable covalent attachment to bead and silica
supports. Exposure of tethered BDPs to the hydroformylation
catalyst precursor, Rh(acac)(CO)2, yields catalysts for
immobilized asymmetric hydroformylation (iAHF) of prochi-
ral alkenes. Compared with homogeneous catalysts, catalysts
immobilized on Tentagel resins exhibit similarly high
regioselectivity and enantioselectivity. When corrected for
apparent catalyst loading, the activity of the immobilized
catalysts approaches that of the homogeneous analogues. Excellent recyclability with trace levels of rhodium leaching are
observed in batch and flow reactor conditions. Silica-bound catalysts exhibit poorer enantioselectivities.
synthesis of high-performing BDP ligands involves the coupling
of the intermediate tetracarboxylic acid bisdiazaphospholane
((S,S)-tetraacid) (Scheme 1) with amines to form tetracarbox-
INTRODUCTION
■
Immobilization is an obvious, but not always successful,
approach to retaining the high selectivity and activity of
homogeneous catalysts while facilitating catalyst/product
separation and catalyst recycling.1,2 Immobilization of catalysts
for linear selective alkene hydroformylation, one of the largest
industrial processes that uses homogeneous, precious metals as
catalysts, has been achieved, though apparently not imple-
mented on an industrial scale.3−5 Fewer reports detail the
immobilization of asymmetric hydroformylation (AHF) cata-
lysts, even though the generally higher costs of enantiopure
catalysts and the need for high-purity pharmaceutical
intermediates provide strong driving forces for the development
of recyclable, heterogeneous AHF catalysts. To date, the report
by Nozaki and co-workers of BINAPHOS-derivatives that are
cross-linked with polystyrene constitute the most successful
development of immobilized AHF catalysts.6−9 While immo-
bilized BINAPHOS ligands exhibit good selectivity relative to
the homogeneous BINAPHOS catalysts, recycling studies using
this catalyst are limited to relatively low TON (∼300) and
require high syn gas pressures in flow (88−120 atm).9
Application of immobilized AHF at both bench and production
scales requires thorough understanding of catalyst activity,
along with selectivity and catalyst leaching/recyclability, under
practical conditions.
Scheme 1. Synthesis of (S,S,S)-BisDiazphos (previous work)
amide ligands. Generally speaking, the carboxamide resulting
from coupling (S,S)-tetraacid with (S)-methylbenzylamine
gives the most selective catalyst for a broad range of olefins.17
Carboxamide formation provides a simple route to BDP
immobilization. Previously, we demonstrated immobilization of
monodiazaphospholanes by coupling the carboxylic acid to an
amine-functionalized polystyrene resin. Such immobilized
ligands provided high enantioselectivity in Pd-catalyzed allylic
alkylation reactions.18 In this work we report the synthesis,
characterization, and application of supported AHF catalysts
based on immobilized BDP ligands. The overall plan of this
paper begins with the synthesis of a small collection of resin-
bound and silica-supported catalysts with different linkers and
supports. Subsequent screening and recyclability study results
are reported for shaker-style reactors. The section on resin-
based catalysts concludes with NMR and rhodium-loading
Bis-3,4-diazaphospholane (BDP) ligands reported by our
group exhibit high selectivity and activity for a broad range of
substrates, thus enabling efficient syntheses of chiral building
blocks and complex structures.10−14 Notable examples include
the Burke group’s synthesis of (+)-Patulolide C, for which the
key step was the AHF of a Z-enolacetate, and Leighton’s
application of Rh(BDP)-catalyzed AHF for synthesis of two of
the three fragments of Dictyostatin.15,16 A key step in the
Received: February 13, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 2. Synthesis of Resin-Bound Bisdiazaphospholanes (L1−L3) and Rh Catalysts (C1−C3)
a
Table 1. Results for Simultaneous AHF of Three Substrates with Supported and Homogeneous Bisdiazaphospholane Ligands
allyloxy tert-
butyldimethylsilane
styrene
vinyl acetate
b
c
b
c
b
c
entry
catalyst
b:l ratio
% ee
b:l ratio
% ee
b:l ratio
% ee
1
2
3
4
C1
C2
C3
2.3:1
4.7:1
6.3:1
20:1
61
77
73
91
13:1
48:1
50:1
40:1
89
92
92
97
1.7:1
1.2:1
1.7:1
1.7:1
85
88
87
(S,S,S)-BisDiazphos-Rh(acac)(CO)2 (1.2:1)
>95
a
Conditions: 24 h, 60 °C, 150 psig H2/CO (1:1), 0.2 M concentration of each substrate,1 mL total volume, 600:1 sub:cat, catalyst concentration
b
c
1
estimated for C1−C3 from data in Chart 2, complete conversion of olefin to aldehyde. Determined by H NMR spectroscopy. Determined by
chiral GC analysis.
studies to better understand the effective active site count of the
immobilized ligands. The paper concludes with application of
the resin-bound catalysts in flow reactors.
the first linkage to the resin support. Thus, one expects two
diastereomers of the final ligand to result from the
immobilization procedure. Ligand synthesis is completed by
capping with acetic anhydride any resin-based primary amine
sites that were not functionalized with tetraacyl fluoride BDP.
Free amine sites on the resin are undesirable because they
could promote racemization of chiral aldehydes and provide a
site for coordination to rhodium catalyst precursors. AHF
catalysts (C1−C3) were formed by the gentle agitation of the
catalyst precursor, Rh(acac)(CO)2, with a suspension of BDP-
containing beads for 2 h followed by several washings.
The immobilized catalysts C1−C3 were screened for
simultaneous rhodium-catalyzed iAHF of three substrates:
styrene, vinyl acetate, and alloxy-tert-butyldimethylsilane (Table
1). Previously we have used this protocol to screen steric effects
for homogeneous bisdiazaphospholane catalysts.21−23
RESULTS AND DISCUSSION
■
Tetraacid BDP was immobilized on three different amine resins
using a three-step procedure to yield ligands L1−L3 (Scheme
2). The first step involves conversion of the (S,S)-tetraacid
BDP to the activated tetraacyl fluoride BDP as previously
reported by our group.17 Resins with different length PEG
linkers were examined; previous studies have shown that PEG
linkers yield more solution-like behavior for immobilized
catalysts.19,20 Overnight reaction of tetraacyl fluoride BDP
with a substoichiometric amount of amine-terminated beads (in
order to maximize the probability that just one carboxamide-
linkage is formed between the ligand and the bead) presumably
yields a ligand that is immobilized by one carboxamide
attachment, with the three acyl fluoride groups remaining.
This covalently linked BDP ligand is then reacted with (S)-
methylbenzylamine (MBA) to generate asymmetrical tetracar-
boxamides. Note that the tetraacyl fluoride BDP comprises two
pairs of diastereotopic acyl groups either of which may make
The simultaneous screen results establish that the immobi-
lized catalysts are active and selective. The activity of the
catalyst will be discussed later in this manuscript. The data
indicate that, although the regioselectivities and enantioselec-
tivities obtained with the immobilized catalyst are respectably
high, they are slightly lower than those obtained with the
B
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Because previous work by our group showed that both the
regio- and enantioselectivities of styrene hydroformylation
increase as the CO partial pressure is raised from 50 to 500
psig, we decided to examine pressure effects on styrene AHF
using the Tentagel-S-NH2-based catalyst, C3 (Chart 1).12
Interestingly, increased syn gas pressure dramatically increases
the regioselectivity of styrene (>25:1) but has little influence on
the enantiomeric ratio (∼7:1). Such differences between
immobilized and homogeneous catalysts suggest intrinsic
selectivity differences between the catalysts rather than simply
slower mass transport of gases to the immobilized catalyst
active sites.
Chart 1. AHF of Styrene at Increasing Syn Gas Pressure
One motivation for catalyst immobilization is the recycling of
expensive precious metals and chiral ligands. Using the protocol
for simultaneous hydroformylation of three substrates in a
shaker reactor, we examined the recyclability of Tentagel-linked
C3. A single batch of immobilized catalyst was used for nine
AHF cycles albeit with some variation of conditions between
cycles 5−9 to examine different reaction conditions on the
immobilized catalysts (Figure 1). After each cycle the reaction
mixture was filtered, and beads were washed three times with
clean solvent. Rhodium leaching was determined by ICP-OES
analysis of the solution after each cycle following removal of the
beads by filtration. At the end of the recycling experiments the
total Rh content of the beads was determined by acid digestion
of the beads followed by ICP-OES of the digest. Generally
speaking, the regio- and enantioselectivities for each substrate
were consistently high and similar to those of the initial run.
a
Conditions: 23 h, 50 °C, 1.2 M styrene; 1600:1 sub:cat, changes in
selectivity are assumed to be affected by pCO and unaffected by pH2
based off previous studies.12
homogeneous catalyst. The observation of high enantioselec-
tivity requires that the bulk, but not all, of product formation
proceeds through BDP-ligated catalysts. Many factors could
yield lower selectivity in the immobilized catalysts: diastereo-
meric environments created by immobilization of BDPs,
variations in the amount of catalyst ligated by chiral phosphine,
or mass transport limitations that yield different steady-state
concentrations of reactants.
Figure 1. Recycling studies using 20 mg of C3. All reactions @ 60 °C, 1 mL total volume, 23 h. Beads washed three times with reaction solvent
between cycles. All reactions went to complete conversion of olefin. Cycles 1−5, 0.4 M in each substrate = 1600:1 sub:cat. Cycles 6−9, 0.8 M in each
substrate = 3200:1 sub:cat. Rh leaching determined by ICP analysis of Rh in product mixtures and decomposition of beads.
C
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 3. Synthesis of Tentagel-proline-bisdiazaphospholanes L4(R,R,S) and L5(S,S,S) and Rh catalysts C4 and C5
a
Furthermore, the Rh leaching was negligible (i.e., at the limits
of detection) for most runs.24 Two exceptions to these general
features are seen: (1) higher styrene regioselectivity results
when the CO pressure was increased (cycles 5, 7, and 9); (2)
high Rh leaching occurs after the first run and after the seventh
cycle when the solvent was changed from toluene to THF.
Overall, such results demonstrate that immobilized rhodium
AHF catalysts are robust and are capable of achieving
reasonable catalyst recyclability and simple separation of
catalyst from product. Using 20 mg of C3 containing the
equivalent of 1.3 mg of (S,S,S)-BisDiazphos, 2.1 g of combined
aldehydes were produced over nine cycles, resulting in a TON
of 20800.
Recently we examined a library of homogeneous bisdiaza-
phospholane catalysts that varied in the carboxamide
substituent.17 During this work we found that in the AHF of
2,3- and 2,5-dihydrofuran (DHF), bisdiazaphospholanes
coupled with methyl prolinate instead of methylbenzylamine
give higher selectivity and activity. We hypothesized that
attachment of proline to a support followed by coupling the
proline amine to the tetraacyl fluoride BDP would create an
immobilized catalyst environment that was maximally similar to
the analogous homogeneous catalyst (because all four
carboxamides have prolines attached) while creating a practical
catalyst for chiral furan building blocks that are useful in organic
synthesis.25 Proline-based immobilized bisdiazaphospholanes
L4 and L5 were synthesized using the procedure outlined in
Scheme 3.
Hydroformylation of the two substrates, 2,3- and 2,5-DHF
were performed with catalysts containing immobilized ligands
C3, C4, and C5 (see Table 2). For the substrate, 2,3-DHF, the
preferred regioisomer is the α-carboxaldehyde (2-formyl
tetrahydrofuran). The proline-functionalized catalysts C4 and
C5 yield significantly higher regioselectivity and activity than
C3, with similar enantioselectivity (see entries 1−3).
Hydroformylation of 2,5-DHF is complicated due to
competitive olefin isomerization to 2,3-DHF. Because AHF of
2,3-DHF is less selective and prefers the opposite antipode of
the β-product (3-formyl tetrahydrofuran), isomerization of 2,5-
DHF could result in substantially lower regio- and
enantioselectivity of the major product.17 However, 2,5-DHF
commonly hydroformylates faster than 2,3-DHF, such that
isomerization does not degrade selectivity at incomplete
conversion.17,26−28 Results for C4 and C5 were surprising
because similar conversions were obtained starting from either
2,3- or 2,5-DHF. Also surprising was the finding that high
selectivity is maintained during the AHF of 2,5-DHF, even
though some isomerization to 2,3-DHF occurs and the
apparent hydroformylation rates of 2,5- and 2,3-DHF seem
Table 2. Immobilized AHF of Dihydrofurans
α:β
ratio
% CHO,
% 2,3-DHF
b
b,
c
c
c
entry catalyst substrate
% ee α % ee β
1
2
3
4
5
6
7
C3
C4
C5
C3
C4
C5
C4
2,3-DHF
2,3-DHF
2,3-DHF
2,5-DHF
2,5-DHF
2,5-DHF
1.3:1
3.7:1
4.0:1
32, 68
73, 27
61, 39
58, 12
69, 15
70, 13
10, 48
82(S)
86(S)
80(S)
n.d.
42(S)
54(S)
40(S)
84(R)
91(R)
88(R)
88(R)
1:100
1:50
1:50
1:14
n.d.
n.d.
d
2,3/2,5
-DHF
77(S)
a
Conditions: 0.8 M substrate, 1090:1 substrate:catalyst, 45 °C, 150
b
c
psig CO/H2 (1:1), 21 h. Determined by 1H NMR spectroscopy. 2,3-
DHF in entries 4−6 from isomerization of 2,5-DHF during reaction,
remaining material is 2,5-DHF. Determined by chiral GC analysis.
1.6 M in 2,3-DHF and 1.6 M in 2,5-DHF, 4000:1 substrate:catalyst,
c
d
40 °C, 21 h.
similar. Starting with a mixture containing equal concentrations
of both 2,5- and 2,3-DHF and running to low overall
conversion (10%), nearly all of the consumed olefin is 2,5-
DHF (Table 2 entry 7).29 Thus, when both substrate isomers
are present with immobilized catalysts, 2,5-DHF apparently
outcompetes 2,3-DHF. Under competitive conditions, most of
the 2,3-isomer formed by isomerization accumulates until the
2,5-isomer is consumed.
Determination of Effective Catalyst Loading. Active
site counting and characterization is a ubiquitous problem in
catalysis that is made more complicated with immobilized
catalysts. For the immobilized catalysts used in this study, there
are no guarantees that all amine sites in the support precursor
are functionalized with ligand, that all ligand sites are bound to
rhodium, that all immobilized rhodium is bound to chiral
ligand, or that all rhodium sites are accessible to substrates.
As previously mentioned for the recycling studies, ICP-OES
measurements provide a basis for quantitating the amount of
rhodium that is retained in C3 after synthesis and washing.
Partial digestion of the beads by heating in aqua-regia greatly
reduces the size of the beads, presumably by oxidation and
cleavage of the PEG linker. After filtering to remove particulate
PS, analysis of the filtrate solution by ICP-OES C3 leads to an
estimated loading of 36 μmol of rhodium per gram.30 For
comparison of Rh retention in nonphosphinated beads, excess
Rh(acac)(CO)2 (relative to amine linkers) was added directly
to Tentagel-NH2 beads, as well as Tentagel-NH2 beads treated
D
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
with acetic anhydride (Tentagel-NHAcyl). After washing both
beads and acidic workup, the Tentagel-NH2 beads retained 46
μmol Rh/g, while the Tentagel-NHAcyl beads retained just 6
μmol Rh/g. Both C3 and Tentagel-NHAcyl beads show a
negative ninhydrin test for free amines, suggesting that all the
amine sites have been functionalized. The majority of the
retained Rh on C3 is, therefore, likely bound to a BDP since the
synthesis of C3 was followed by acylation of any remaining
amine groups. It should be noted that the addition of BDP, Rh,
and acetyl groups to C3 corresponds to less than 10% of the
original mass of the beads. Therefore, it is reasonable to
normalize all values for rhodium loading to the mass of the
bead.
Although we have measured the amount of Rh added to the
beads and the amount of Rh recovered by acid workup and ICP
analysis of the bead, these quantities do not indicate the
number of phosphine sites that are attached to Rh and
catalytically active. Comparison of reaction rates of homoge-
neous and immobilized catalysts gives a rough indication, only,
of the active site count because of heterogeneity of the
immobilized catalyst site activities arising from different steric
environments and accessibility.
In order to assess the effective activity of immobilized
catalysts while minimizing the consumption of precious
enantiopure BDP, the ligand L6, a racemic analogue of L3,
was prepared by first coupling racemic tetra acyl fluoride BDP
with Tentagel-S-NH2 followed by capping the remaining acyl
fluorides with benzyl amine.
Equal masses of L6 were mixed overnight with different
amounts of a standardized Rh(acac)(CO)2 solution and then
were rinsed and filtered (Scheme 4). Beads exposed to different
Chart 2. Hydroformylation Activity versus Rhodium Added
a
Conversion of 4.1 mmol of vinyl acetate vs the amount of
Rh(acac)(CO)2 added to 20 mg of L6 followed by washing three
times with 1 mL toluene. Conditions: 5 h, 4 M vinyl acetate in toluene,
150 psig CO/H2, 44 °C.
state 31P MAS NMR spectra of immobilized ligands were
unsuccessful due to low signal-to-noise ratio. However, solution
31P NMR of bead suspensions give spectra useful for
characterization of the immobilized ligand. Tentagel beads
were loaded into a Shigemi tube after reacting with a slight
excess of tetraacyl fluoride BDP and washing three times with
dichloromethane. Shigemi tubes concentrate the less dense
beads in the detection region of the probe, allowing for
integration versus an internal standard in solution. Although
immobilized ligands yield broader spectra than the free BDP
ligands, two clear doublets with a coupling constant of 185 Hz
are seen in the 31P NMR spectrum. This coupling value is
3
consistent with JPP coupling of diastereotopic P atoms, the
result of desymmetrization by attachment of the bisdiazaphos-
pholane to the support through one amide bond (Figure 2).
Scheme 4. Racemic Tetrabenzyl Carboxamide BDP L6 and
Rh Catalyst C6 for Activity Determination
Figure 2. Solution-phase 31P NMR of tetraacyl fluoride BDP reacted
with Tentagel-S-NH2, followed by washing with DCM.
quantities of Rh were then used for the hydroformylation of
vinyl acetate to determine if reactivity scales linearly with Rh
loading, as seen for the homogeneous catalysts (Chart 2). If all
BDP sites of L6 have equal accessibility to Rh, one expects a
plot of rate vs added rhodium to be linear until the all BDP
sites are saturated with rhodium at which point there will be a
breakpoint.
As shown in Chart 2, breakpoint behavior is observed with
maximal rate observed at ∼50 μmol Rh/g bead. This implies
that exposure of L6 to more than 50 μmol of rhodium per gram
of bead does not lead to more active sites. While this value
(∼50 μmol/g) is slightly larger than the amount (∼36 μmol/g)
recovered by ICP analysis of the beads exposed to excess
Rh(acac)(CO)2, it is reasonable, considering possible error in
weighing these materials and variability between different
batches of immobilized phosphines. So far the data is consistent
with the supposition that all of immobilized rhodium in the
resin is effective for AHF.
The other major species at 5 ppm likely is noncovalently
attached tetraacyl fluoride BDP, based on the chemical shift and
lack of splitting. Fewer solvent washes were performed in the
preparation of the samples for NMR analysis compared to the
standard procedure in order to limit exposure to oxygen.31
Multiple peaks corresponding to acyl fluorides are present in
the 19F NMR spectrum but cannot be assigned.
Unfortunately, addition of benzyl amine to form L6 results in
much broader NMR peaks that are less revealing (Figure 3).
Crude integration of the broad L6 peaks versus an internal
standard (triphenylphosphine oxide) indicates loading of
approximately 37 ( 18) μmol bisdiazaphospholane/g bead.
Finally, the addition of a solution of Rh(acac)(CO)2 to the L6-
functionalized beads in the NMR (50 μmol Rh/g bead) gives a
ratio of metalated-to-free ligand around 3:1. Further addition of
Rh(acac)(CO)2 gives a negligible change to this ratio.
A more direct method of characterizing immobilized ligands
and catalysts is NMR spectroscopy. Attempts to obtain solid-
E
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
under optimized reaction conditions in well-stirred pressure
bottles,21 but provide a direct comparison of the homogeneous
and immobilized catalysts under a single set of conditions.
More detailed analysis of the rates of the immobilized catalysts
is complicated by the inability to observe catalyst preactivation
of the BDP−Rh(acac) complex with CO/H2 to form the
BDP−RhH(CO)2 complex, which is the resting state for the
AHF catalyst.
Overall the Tentagel-immobilized bisdiazaphospholane cata-
lysts C3−C6 perform AHF with high selectivity, exhibit 3.5-
fold lower normalized activities than the homogeneous catalyst,
and can be recycled. While developing these catalysts, we also
pursued similar methods for immobilization on silica-based
supports, due to their robustness and prevalence in the
literature.3,32−34
Silica-Immobilized Catalysts. Rh−BDP complexes can be
supported on silica either by covalent attachment of a metalated
BDP complex to the support or by first attaching the BDP
ligand to the silica followed by metalation. The former method
involves coupling 3-aminopropyltriethoxysilane to the tetraacyl
fluoride BDP precursor to give a tetracarboxamide BDP ligand
with triethoxysilane groups (L7). Addition of a substoichio-
metric amount of Rh(acac)(CO)2 followed by condensation on
high-purity silica gel results in the immobilized catalyst C8
(Scheme 5; middle). The latter method treats mesoporous
silica SBA-15 with 3-aminopropyltriethoxysilane. Subsequent
addition of tetraacyl fluoride BDP presumably forms at least
one carboxamide linkage to the functionalized silica, followed
by addition of methylbenzyl amine as in the synthesis of L1−
L3. Washing and addition of Rh(acac)(CO)2 gives the SBA-15
immobilized Rh-BDP catalyst C9 (Scheme 5; bottom).
The silica-immobilized Rh−BDP catalysts C8 and C9, along
with homogeneous catalyst C7, were screened using the one-
pot/multiolefin screen (styrene/vinyl acetate/allyloxy tert-
butyldimethylsilane) used for bead-based catalysts. Unfortu-
nately, these catalysts give decreased regio- and enantioselec-
Figure 3. 31P NMR of L6 (top), addition of 2.44 μmol Rh(acac)-
(CO)2 to 49 mg L6 (50 μmol Rh/g bead) (middle), and addition of
4.88 μmol Rh(acac)(CO)2 to 49 mg L6 (100 μmol Rh/g bead)
(bottom).Taken together, these data suggest that just 40−50 μmol/g
of the nominal 290 μmol amine sites per gram Tentagel loading
become functionalized with BDP ligand. Apparently, most of the BDP
sites can be metalated with rhodium precursors. All of these numbers
are somewhat crude, but it is clear that the nominal amine content of
the bead far exceeds the active catalyst site count.
Now that we have a better understanding of the active site
count of the immobilized catalysts, activity comparisons with
the homogeneous catalysts can be made. Crude rate values
taken between 10 and 30% total conversion show the
immobilized catalyst to be 3−4 times slower than the
corresponding homogeneous catalyst when normalized using
the approximate active site counts. Using vinyl acetate, rates of
80 TOF h−1 were calculated for C6, compared to 280 TOF h−1
for the equivalent homogeneous catalyst (44 °C in shaker, 150
psig (1:1) CO/H2, 4.1 M vinyl acetate, 4100:1 sub:cat). These
rates are slower than those seen for homogeneous catalysts
Scheme 5. Synthesis of Silica-Bound Bisdiazaphospholanes
F
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Table 3. Results of One-Pot AHF Screening of Silica-Immobilized Bisdiazaphospholanes and Homogeneous Precursor
allyloxy tert-
butyldimethylsilane
styrene
vinyl acetate
b
c
b
c
b
c
entry
catalyst
pressure conv. (%) b:l ratio % ee conv. (%) b:l ratio
% ee conv. (%) b:l ratio % ee
1
2
3
4
C8
C9
C8
C9
C7
150
150
300
300
150
150
15
43
6:1
1.9:1
4.6:1
3.7:1
25:1
20:1
63
56
62
53
89
91
25
74
30:1
40:1
30:1
24:1
30:1
40:1
78
75
81
83
89
97
34
80
1.2:1
0.8:1
1.3:1
1.1:1
1.7:1
1.7:1
82
67
70
49
89
57
79
88
25
47
49
d
5
90
68
100
100
e
6
(S,S,S)BisDiazphos−Rh(acac)(CO)2 (1.2:1)
100
100
>95
1
a
b
Conditions: 24 h, 50 °C, 0.6 M total substrate (equal concentrations of each ∼200:1 sub:cat (C9), ∼1000:1 sub:cat (C8). Determined by H
NMR spectroscopy. Determined by chiral GC analysis. 16 h, 4.3 M total substrate (equal concentrations of each), 1600:1 total substrate:cat. 60
c
d
e
°C, 600:1 sub:cat. Catalyst loadings for C8 and C9 are estimates based on loading of the materials assuming all Rh added forms a BDP catalyst.
tivities for all substrates examined compared to those of the
homogeneous catalysts or Tentagel-supported catalyst C3.
While previous experiments showed that increasing the CO
pressure for C3 improved the selectivity of styrene AHF,
negligible changes in selectivity are observed for C8−C9.
However, increasing syn gas pressure to 300 psig does improve
the activity of C8 (this effect has not been observed for the
other immobilized or homogeneous BDP catalysts). This result
may indicate that mass transport is limiting activity at 150 psig
for C8. After filtering the solid catalyst away from aldehyde
products, the product solution contains less than 1% of the
rhodium loading as determined by ICP-OES.
Control experiments using C7 under homogeneous reaction
conditions show high regio- and enantioselectivities (Table 3,
entry 5). Therefore, it is unlikely that the low selectivities for
the silica-supported catalysts C8 and C9 relative to the bead-
supported catalysts C3-C6 results from the different carbox-
amide groups used for support attachment. Reasonable
enantioinduction is still observed for the AHF of vinyl acetate
and allyloxy t-butyldimethylsilane and is the first example of an
iAHF on silica supports with greater than 60% ee to the best of
our knowledge. Unfortunately, the AHF product of styrene
racemizes upon extended reaction times with the silica-based
catalysts, while no change in the enantiomeric excess of the
other two chiral aldehydes was observed after 24 h.
AHF in Flow. Recently the application of asymmetric
catalysis in flow reactors has received considerable attention,
but examples of AHF in flow reactors are uncommon.35,36
Tentagel and other PS−PEG cografted beads have been used
for immobilization of homogeneous catalysts and in flow
reactors for hydrogenations, allylations, and other reac-
tions.2,37−39
We set out to demonstrate the application of the more
selective Tentagel-immobilized AHF catalysts in a plug flow
reactor. Particularly for the bench-scale chemist, a packed
column of recyclable catalyst would be easier to rinse, store, and
set up as needed. Our goals for this setup were to demonstrate
that these immobilized catalysts could be easily applied to a
flow reactor and determine if syn gas concentration in solution
could be maintained while passing through the catalyst bed.
Our primitive flow reactor consists of a glass pressure bottle
and manifold, an HPLC pump with a max flow rate of 10 mL/
min, and a 1/4 in. × 5.9 in. HPLC column packed with C6.
Substrate solution is stirred under syn gas inside the pressure
bottle, which acts as a reservoir while the gas-saturated solution
is pumped through the column. Solution from the column then
drips back into the pressure bottle, thus effecting recycle and
gas-saturation of the substrate/product solution.
One limitation with this design is that dissolved syn gas is
consumed as solution passes through the catalyst bed. This may
lead to different syn gas concentrations at the beginning and
end of the column. Previous work has demonstrated that the
AHF selectivity for aryl alkenes such as styrene is particularly
sensitive to the effective concentration of CO in solution.12 To
probe the effective gas concentration in the plug flow reactor
we examined the effect of flow rate on the selectivity of styrene
hydroformylation using the racemic, supported catalyst C6.
One anticipates increased regioselectivity with increased flow
rate if the reactor zone is significantly depleted in gas. This
behavior is observed (Table 4). Increased flow rate yields
a
Table 4. Hydroformylation of styrene in plug flow reactor
flow rate (mL/min)
conv. (%)
b:la
1
2
5
9
19
11
25
24
1.3:1
2.2:1
9.0.:1
13:1
a
Conditions: 3 h, 150 psig, 60 °C, 1.2 M styrene; 1000:1 sub:cat. 0.28g
C6 (≡ 14 μmol Rh based off batch reactor experiments).
increased regioselectivity with a plateau b:l ratio of about 13:1
reached at the highest flow rate examined (10 mL/min). This
plateau corresponds to the regioselectivity observed in shaker-
style batch reactors with the structurally similar catalyst C3 at
the same temperature and gas pressure (Chart 2). We conclude
that the plug flow reactor using immobilized catalyst and
sufficiently high flow rates mimics the behavior of immobilized
catalysts in batch reactors.
Recycling studies were performed in which the C6 catalyst
effected the hydroformylation of eight separate batches of vinyl
acetate in the plug flow reactor (Table 5). The Rh content of
the product solutions was measured by ICP-OES after each
batch. Catalyst was loaded onto the beads by placing
phosphinated beads into the plug flow reactor and then
circulating a solution of Rh(acac)(CO)2 through the system
G
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
per gram of bead. This is significantly lower than the nominal
amine site count of 290 μmol per gram of native bead,
demonstrating the importance of measuring, rather than
assuming, active site counts in making comparisons between
homogeneous and heterogeneous catalysts. On the basis of
catalyst active site counts, the immobilized catalyst exhibits
about 30% of the activity of the homogeneous catalyst. Like the
homogeneous Rh(BDP) catalysts, the Tentagel immobilized
catalysts demonstrate higher selectivity in styrene hydro-
formylation as the synthesis gas pressure is increased and
faster conversion of 2,5-DHF than 2,3-DHF.
Table 5. Hydroformylation of Vinyl Acetate Using a Single
Catalyst Loading in Plug Flow Reactor
a
b
c
d
run
[substrate], time
conv. (%)
b:l
Rh leached (%)
1
2
3
4
5
6
7
8
1 M, 15 h
1 M, 4 h
1 M, 4 h
1 M, 4 h
1 M, 4 h
1 M, 4 h
4.2 M, 4 h
4.2 M, 4 h
100
100
100
100
100
100
48
58:1
50:1
54:1
55:1
56:1
60:1
46:1
51:1
6.1
2.2
0.7
0.7
0.7
0.3
0.4
0.5
51
Tentagel-immobilized BDP ligands provide heterogeneous
AHF catalysts that can be recycled with no loss of selectivity
over nine cycles and applied in shaker-style batch reactors or
plug flow reactors. Using only 20 mg of resin containing 1.3 mg
of active catalyst, grams of aldehyde can be produced in high
selectivity containing only trace amounts of precious metals.
Flow reactors are a safe, scalable method of implementing
reactions using dangerous gases such as CO and H2 in general
use facilities such as those in many pharmaceutical and fine-
chemical companies. The work reported herein allows practical
access to immobilized BDP ligands that are recyclable, active,
and selective catalysts for AHF.
a
Conditions. 50 °C, 150 psig 1:1 CO/H2, 4 mL/min flow rate. 0.38 g
C6, 19 μmol Rh based on Chart 2, 485:1 substrate:catalyst for 1 M
runs. Conversion percentages are estimates due to a leak in the
b
HPLC pump, ∼20% (first run) to ∼50% (final run). Leaked solution
was combined with reaction solution for analysis, but some vinyl
c
acetate was lost due to its volatility. Determined by 1H NMR
d
spectroscopy. Determined by ICP analysis of rhodium in the product
solution compared to beads following acid workup.
under N2. Clean solvent was circulated through the system
multiple times to remove any unbound Rh from the reactor.
As seen with beads in shaker reactors, immobilized catalyst in
a plug flow reactor exhibits low catalyst leaching and stable
activity and selectivity metrics. Recycling studies using the plug
flow reactor show 6% of the total Rh loading leaches out in the
first cycle, 2% in the second cycle, and ∼0.5% in the third and
successive cycles. Initial turnover frequencies for the plug flow
reactor are in the range of 100−200 TOF h−1. This is
comparable to the rates observed in batch reactor experiments.
Over eight additions of fresh substrate solutions, nearly 5000
turnovers of the catalyst were achieved with no significant
changes in apparent rates or selectivity.
EXPERIMENTAL SECTION
■
General Considerations. All phosphines were prepared
under N2 using standard Schlenk line techniques. Rh(acac)-
(CO)2 was recrystallized (green needles) from toluene/hexanes
prior to use. THF and toluene were distilled over Na/
benzophenone under a nitrogen atmosphere. THF and toluene
used for asymmetric hydroformylation reactions were further
deoxygenated by at least three freeze−thaw cycles prior to use.
Dichloromethane was distilled under nitrogen over P2O5.
Routine NMR experiments (1H, 13C, 19F, 31P) were carried out
on a Bruker AC-300, Varian Mercury-300, or a Varian Unity-
500. Proton (1H) and carbon (13C) NMR spectra were
referenced to TMS (0.00 ppm) and CDCl3 (77.0 ppm)
respectively. The fluorine (19F) spectra were referenced to
Immobilized AHF reactions achieve high rates and selectivity
in plug flow reactors. For substrates that are sensitive to gas
partial pressures, the plug flow setup should be arranged such
that equilibrium concentrations of dissolved gas in the catalytic
reaction zone are maintained.
1
TMS in the H spectra, using the Unified Scale. Phosphorus
(31P) chemical shifts were referenced to an external 85%
phosphoric acid (H3PO4) sample. The percent conversion and
regioisomer ratios were determined by 1H NMR analysis of the
crude reaction mixture. Gas chromatography (GC) was
performed on a Varian 2010 using a β-DEX 225 column (30
m × 0.25 mm ID). Deoxo-Fluor, diisopropylethylamine, benzyl
amine, (S)-methylbenzylamine, anhydrous grade dimethylfor-
mamide, diisopropylcarbodiimide, hydroxybenzotriazole hy-
drate, 3-aminopropyltriethoxysilane, and high-purity silica gel
(Merck grade 7734, pore size 60 Å, 70−230 mesh) were
purchased from Sigma-Aldrich and used without further
purification. (R)- and (S)-Proline methyl esters were
synthesized from (R)- and (S)-proline respectively, and
recrystallized. Vinyl acetate, styrene, TBSO allyl ether, 2,5-
dihydrofuran, and 2,3-dihydrofuran were all sparged with N2
before use. Tentagel-S-NH2 was purchased from Anaspec (0.29
mmol/g, 90 μm particle size). PS-MBHA resin was purchased
from Polymer laboratories (1% cross-linking, loading 1.6
mmol/g). Synthesis of the EBES linker and attachment to
the PS-MBHA resin was carried out following literature
procedure.18 SBA-15 treated with Ph2SiCl2 and 3-amino-
propyltriethoxysilane to give an amine loading of 5.7 mmol/g
was prepared by Daniel Resasco at University of Oklahoma.
Synthesis of the tetraacid bisdiazaphospholane and tetraacyl
CONCLUSION
■
Ultimately, larger-scale application of catalysts containing
precious metals and ligands requires recycling of the catalyst.
We report a complete analysis of immobilized AHF that
includes synthesis on both silica and resin supports,
investigation of catalyst selectivity and activity, characterization
of the often elusive “active site count”, determination of catalyst
leaching rates, and application to gram-scale synthesis in flow
reactors. Ligand immobilization takes advantage of the versatile
ligand precursor, tetraacyl fluoride BDP. Reaction of the acyl
fluoride groups with amines enables covalent attachment of the
ligand to supports and further elaboration of the BDP
environment. Of the supports examined, immobilization on
Tentagel-S-NH2 beads yields the best combination of regio-
and enantioselectivity in AHF reactions. In fact, the regio- and
enantioselectivities obtained with immobilized catalysts ap-
proach the high bench-marks established by the homogeneous
BDP catalyst.
Comparison of catalyst activities for immobilized and
homogeneous catalysts depends critically on accurate determi-
nation of the active site counts. For the BDP-functionalized
Tentagel-S-NH2 beads, the active site counts are determined by
breakpoint titration methods to be approximately 40−50 μmol
H
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
fluoride bisdiazaphospholane followed literature procedure.10,17
The C2 enantiomers of tetraacid bisdiazaphospholane are
resolved by SFC using a Chiralpak IC-H column, (60% MeOH
+ 0.2% diethylamine)/CO2, 2.4 mL/min, 100 bar, 40 °C.
Reported substrate:catalyst ratios for C3−C6 are based on the
data for C6 in Chart 2, corresponding to an estimated 50 μmol
Rh/g bead. Substrate:catalyst ratios for the silica-based catalysts
C8/C9 are estimates based on the amount of Rh(acac)(CO)2
added to the support.
Synthesis of Resin-Supported Bisdiazaphospholanes
C4−C5. One hundred and fifty mg of Tentagel-S-NH2 resin
was loaded into a 6 mL SPE tube with a PP frit. The tube was
sealed with a septum and a two-way valve and needle on the
bottom. The needle was then placed in a 250 mL Schlenk flask
connected to Schlenk line to allow for draining reactant
solution while maintaining a nitrogen atmosphere. The flask
and SPE tube were pump-backfilled with N2 three times. The
beads were washed sequentially with 3 mL DMF, 1 mL DIEA, 3
mL DMF, and 3 mL DCM to remove any water inside the
beads. The beads were then swollen in 1 mL DMF while being
agitated by inserting a 22 gauge needle in the septum, allowing
N2 to slowly purge out. In 2 mL of DMF were dissolved 0.17
mmol of Boc-L-proline and 0.17 mmol of HOBT hydrate and
added to beads, agitated for 10 min, followed by addition of
0.17 mmol of DIC. Beads were agitated overnight; solution was
drained, and then washed with 2 × 2 mL DMF, 2 × 2 mL
MeOH, and 2 × 2 mL DCM. Boc groups were cleaved by
adding a mixture of 3 mL DCM, 1 mL TFA, and 0.05 mL
anisole and agitating for 45 min. The solution was then drained
and washed with 2 × 2 mL DMF, 2 × 2 mL MeOH, and 2 × 2
mL DCM. In 3 mL of DCM was dissolved 0.9 mmol of
tetraacyl fluoride bisdiazaphospholane ((R,R)-L4/(S,S)-L5)
and added to the bead by syringe. Added by syringe was 0.44
mmol of DIEA. The solution was mixed overnight, adding
DMF as needed due to solvent evaporation. The solution was
drained, followed by washing with 2 × 2 mL DMF, 2 × 2 mL
MeOH, and 2 × 2 mL DCM. The remaining acyl fluorides on
the immobilized bisdiazaphospholanes were then functionalized
by suspending the beads in 3 mL DMF, followed by addition of
0.9 mmol of L-Pro-OMe and 0.9 mmol of DIEA. The beads
were agitated overnight; the solution was drained, and then
washed with 2 × 2 mL DMF, 2 × 2 mL MeOH, and 2 × 2 mL
DCM. A solution of 0.3 mL acetic anhydride, 0.5 mL 2,6-
lutidine, a grain of DMAP, and 2 mL THF was prepared and
added to the beads; the beads were agitated for 20 min,
drained, followed by washing of the beads with 2 × 2 mL DMF,
2 × 2 mL MeOH, and 2 × 2 mL DCM. Two mL of toluene was
added to the beads, followed by a solution of 0.07 mmol of
Rh(acac)(CO)2 dissolved in 2 mL toluene. The beads were
agitated for 1 h, adding toluene as needed. The solution was
drained, followed by washing with 5 × 2 mL toluene; beads
were then dried in vacuo.
Synthesis of Resin-Supported Bisdiazaphospholanes
L6. Two hundred mg of Tentagel-S-NH2 resin was loaded into
a 6 mL SPE tube with a PP frit. The tube was sealed with a
septum with a two-way valve and needle on the bottom. The
needle was then placed in a 250 mL Schlenk flask connected to
Schlenk line to allow for draining reactant solution, while
maintaining a nitrogen atmosphere. The flask and SPE tube
were pump-backfilled with N2 three times. The beads were
washed sequentially with 3 mL DMF, 1 mL DIEA, 3 mL DMF,
and 3 mL DCM to remove any water inside the beads. The
beads were then swollen in 1 mL DMF while being agitated by
inserting a 22 gauge needle in the septum, allowing N2 to
slowly purge out. Dissolved in 3 mL DCM and added to the
beads by syringe was 0.12 mmol of tetraacyl fluoride
bisdiazaphospholane (RAC). Added by syringe was 0.6 mmol
of DIEA. The solution was mixed overnight, adding DMF as
needed as solvent evaporates. The solution was drained,
followed by washing with 2 × 2 mL DMF, 2 × 2 mL MeOH,
and 2 × 2 mL DCM. The beads were suspended in 3 mL DMF,
followed by addition of 0.6 mmol of benzylamine and 0.6 mmol
General Asymmetric Hydroformylation Procedure.
Reactions were performed in a multiwell Cat24 Parr reactor
that was dried in the oven overnight. The supported catalysts
were loaded into oven-dried glass tubes in a dinitrogen-filled
glovebox; substrate and solvent were then added by 200 and
1000 μL Eppendorf pipettes. The reactor was sealed, removed
from the glovebox, placed in a fume hood, purged three times
with 150 psig of (1:1) synthesis gas to remove dinitrogen and
filled to the appropriate synthesis gas pressure. The shaker
speed used for all reactions was 300 rpm. Upon completion of
the reaction, the reactor was allowed to cool to room
temperature and was vented inside the fume hood down to a
slight overpressure. The reactor was then brought back into a
dinitrogen glovebox and disassembled. For recycling experi-
ments, the supported catalyst was filtered and washed three
times inside a glovebox with 1 mL of reaction solvent, followed
by repeating the reaction setup. An aliquot of the reaction
1
mixture was dissolved in d8-toluene for H NMR to determine
percent conversion and regioselectivity. The enantiomeric
excess of the branched hydroformylation product was
determined by chiral GC.
Synthesis of Resin-Supported Bisdiazaphospholanes
C1−C3. Two hundred mg of amine resin (MBHA/MBHA-
EBES/Tentagel-S-NH2) was loaded into a 6 mL SPE tube with
a polypropylene frit. The tube was sealed with a septum on top
and a two-way valve and needle on the bottom. The needle was
then placed in a 250 mL Schlenk flask connected to Schlenk
line to allow for draining reactant solution while maintaining a
nitrogen atmosphere. The flask and SPE tube were pump-
backfilled with N2 three times. The beads were washed
sequentially with 3 mL DMF, 1 mL DIEA, 3 mL DMF, and
3 mL DCM to remove any water inside the beads. The beads
were then swollen in 1 mL DMF while being agitated by
inserting a 22 gauge needle in the septum, allowing N2 to
slowly purge out. Two equivalents (relative to amine loading on
the bead) of tetraacyl fluoride bisdiazaphospholane (S,S) was
dissolved in 3 mL DCM and was added to the beads by syringe,
followed by 10 equiv of DIEA. The solution was then mixed
overnight, adding DMF as needed as solvent evaporates. The
solution was then drained, followed by washing with 2 × 2 mL
DMF, 2 × 2 mL MeOH, and 2 × 2 mL DCM. The beads were
then suspended in 3 mL DMF, followed by addition of 10 equiv
of (S)-methylbenzylamine and 10 equiv of DIEA. The beads
were agitated overnight; solution was drained, followed by
washing with 2 × 2 mL DMF, 2 × 2 mL MeOH, and 2 × 2 mL
DCM. A solution of 0.3 mL acetic anhydride, 0.5 mL 2,6-
lutidine, a grain of DMAP, and 2 mL THF was prepared and
the solution was added to the beads, agitated for 20 min,
drained, followed by washing of the beads with 2 × 2 mL DMF,
2 × 2 mL MeOH, and 2 × 2 mL DCM. 20 mg Rh(acac)(CO)2
was dissolved in 4 mL toluene and added to the beads. The
beads were agitated for 2 h, adding toluene as needed. The
solution was drained followed by washing with 5 × 2 mL
toluene; the beads were then dried in vacuo.
I
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. Diagram and photo of plug flow reactor.
of DIEA. The beads were agitated overnight; drained, and then
washed with 2 × 2 mL DMF, 2 × 2 mL MeOH, and 2 × 2 mL
DCM. A solution of 0.3 mL acetic anhydride, 0.5 mL 2,6-
lutidine, a grain of DMAP, and 2 mL THF was prepared and
added to the beads, agitated for 20 min, drained, followed by
washing of the beads with 2 × 2 mL DMF, 2 × 2 mL MeOH,
and 2 × 2 mL DCM. The beads were then dried in vacuo.
Synthesis of L7. In 100 mL DCM under N2 was dissolved
0.18 mmol of tetraacyl fluoride bisdiazaphospholane (S,S). By
syringe was added 1 mmol of DIEA, followed by 0.2 mL (1
mmol) 3-aminopropyltriethoxysilane. The reaction was stirred
at room temperature overnight followed by aqueous extraction
with sat. NaHCO3, 1 M HCL, and brine. The organic layer was
with 2 × 2 mL DMF, 2 × 2 mL MeOH, and 2 × 2 mL DCM.
Silica was suspended in 3 mL DMF, followed by addition of 5.7
mmol of (S)-methylbenzylamine and 5.7 mmol of DIEA. Silica
was agitated overnight; the solution drained, followed by
washing with 2 × 2 mL DMF, 2 × 2 mL MeOH, and 2 × 2 mL
DCM. A solution of 0.3 mL acetic anhydride, 0.5 mL 2,6-
lutidine, a grain of DMAP, and 2 mL THF was prepared and
added to the silica, agitated for 20 min, drained, followed by
washing with 2 × 2 mL DMF, 2 × 2 mL MeOH, and 2 × 2 mL
DCM. Silica was suspended in 2 mL toluene, followed by
addition of 0.8 mmol of Rh(acac)(CO)2 dissolved in 2 mL
toluene. Silica was agitated for 1 h, adding toluene as needed.
Solution was drained followed by washing with 5 × 2 mL
toluene; silica was then dried in vacuo.
1
concentrated by rotovap to give L7 in quantitative yield. H
Hydroformylation Procedure for a Plug Flow Reactor.
A portion of 0.28 g of L6 was loaded in a 1/4 in. × 5.9 in.
HPLC column inside a glovebox, then connected to an HPLC
pump and 200 mL glass pressure bottle (Figure 4). The
pressure bottle was connected to a nitrogen source via quick-
connect, purging the pressure bottle and pump by opening a
three-way valve attached to the column, keeping the column
closed off. Twelve mL of substrate solution was added to the
pressure bottle via syringe, and the valves were then positioned
to allow solution to circulate. Circulation and heating were
maintained until a constant temperature was reached (∼30
min) before pressurizing/venting the pressure bottle with
synthesis gas three times to 150 psig. At the end of the desired
reaction time the solution was drained through a three-way ball
valve attached to the column for analysis. The apparatus was
then rinsed with 3 × 10 mL toluene between recycles.
NMR (CDCl3, δ, ppm); 8.46 (t, J = 5.6 Hz, 2H) 7.63 (d, J = 7.5
Hz, 4H) 7.50 (m, 4H) 7.21 (m, 6H) 6.98 (m, 4H) 6.68 (t, J =
7.6 Hz, 2H) 6.23 (s, 2H) 6.16 (d, J = 8 Hz, 2H) 3.84 (m, 24H)
3.6−3.2 (m, 8H) 2.6−24 (m, 8H) 1.8−1.4 (m, 8H) 1.25 (m,
36H) 0.9−0.4 (m, 8H). 31P{1H} NMR (CDCl3, δ, ppm) 7.25
(s).
Synthesis of Silica-Supported Bisdiazaphospholane
C8. Three hundred mg (0.18 mmol) of L7 was dissolved in 50
mL toluene, and added to a solution of 0.9 mmol of
Rh(acac)(CO)2 in 3 mL toluene, giving a 2:1 ligand:Rh ratio.
The solution was stirred at room temperature for 1 h, and then
added to 2 g of Merck 60 Å silica suspended in 50 mL toluene.
The solution was stirred overnight at room temperature,
followed by filtering the solution through Schlenk flask with a
frit side arm. C8 was then washed with 4 × 50 mL of toluene
and dried in vacuo.
Synthesis of Silica-Supported Bisdiazaphospholane
C9. One hundred mg of SBA-15 with an amine loading of 5.7
mmol/g was loaded into a 6 mL SPE tube with a PP frit. The
tube was sealed with a septum and a two-way valve and needle
on the bottom. The needle was then placed in a 250 mL
Schlenk flask connected to Schlenk line to allow for draining
reactant solution while maintaining a nitrogen atmosphere. The
flask and SPE tube were pump-backfilled with N2 three times.
The silica was washed twice with DCM. Silica was suspended in
1 mL DMF while being agitated by inserting a 22 gauge needle
in the septum, allowing N2 to slowly purge out. Dissolved in 3
mL DCM and added by syringe was 1.2 mmol of tetraacyl
fluoride BDP (S,S). Added by syringe was 5.7 mmol of DIEA.
Suspension was mixed overnight, adding DMF as needed as
solvent evaporates. Solution was drained, followed by washing
ASSOCIATED CONTENT
■
S
* Supporting Information
Preparation of L6 for rhodium titration, method for preparing
ICP solutions, description of plug flow reactor part,
homogeneous AHF data for C7, and GC conditions for chiral
aldehydes. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
landis@chem.wisc.edu
Notes
The authors declare no competing financial interest.
J
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(30) Average of two runs that were within 1 μmol Rh/g.
ACKNOWLEDGMENTS
■
(31) Standard procedure used is 2 × DMF, 2 × MeOH, and 2 ×
DCM washes to remove excess reagents, beads in Shigemi tube were
only washed 2 × DCM.
(32) Merckle, C.; Blumel, J. Top. Catal. 2005, 34, 5.
(33) Lu, Z.-l.; Lindner, E.; Mayer, H. A. Chem. Rev. 2002, 102, 3543.
(34) Han, D.; Li, X.; Zhang, H.; Liu, Z.; Li, J.; Li, C. J. Catal. 2006,
243, 318.
(35) Tsubogo, T.; Ishiwata, T.; Kobayashi, S. Angew. Chem., Int. Ed.
2013, 52, 6590.
(36) Webb, D.; Jamison, T. F. Chem. Sci. 2010, 1, 675.
(37) Bayston, D. J.; Travers, C. B.; Polywka, M. E. C. Tetrahedron:
Asymmetry 1998, 9, 2015.
The assistance of Rob McClain for ICP instrumentation and a
HPLC pump is greatly appreciated. We thank Daniel Resasco
(University of Oklahoma) for a sample of SBA-15 treated with
3-aminopropyltriethoxysilane, the Dow Chemical Company for
their generous donation of Rh(acac)(CO)2, and Merck & Co.
(Dr. Neil Strotman) for resolution of tetracarboxylic acid
bisdiazaphospholane. We also acknowledge the NSF (CHE-
9208463 and CHE-0342998) for funding the NMR facilities
and some of the reactor equipment (CHE-0946901) and the
NSF (CHE-1152989).
(38) Jee, J.-E.; Cheong, J. L.; Lim, J.; Chen, C.; Hong, S. H.; Lee, S. S.
J. Org. Chem. 2013, 78, 3048.
(39) Nakano, H.; Takahashi, K.; Suzuki, Y.; Fujita, R. Tetrahedron:
Asymmetry 2005, 16, 609.
REFERENCES
■
(1) Baleizao, C.; Garcia, H. Chem. Rev. 2006, 106, 3987.
(2) Zhao, G.; Chai, Z. In Recoverable and Recyclable Catalysts; John
Wiley & Sons, Ltd: New York, 2009; p 49.
(3) Marras, F.; Wang, J.; Coppens, M.-O.; Reek, J. N. H. Chem.
Commun. 2010, 46, 6587.
(4) Lu, S.-M.; Alper, H. J. Am. Chem. Soc. 2003, 125, 13126.
(5) van Leeuwen, P. W. N. M.; Sandee, A. J.; Reek, J. N. H.; Kamer,
P. C. J. J. Mol. Catal. A: Chem. 2002, 182−183, 107.
(6) Shibahara, F.; Nozaki, K.; Hiyama, T. J. Am. Chem. Soc. 2003,
125, 8555.
(7) Shibahara, F.; Nozaki, K.; Matsuo, T.; Hiyama, T. Bioorg. Med.
Chem. Lett. 2002, 12, 1825.
(8) Nozaki, K.; Shibahara, F.; Hiyama, T. Chem. Lett. 2000, 29, 694.
(9) Nozaki, K.; Itoi, Y.; Shibahara, F.; Shirakawa, E.; Ohta, T.;
Takaya, H.; Hiyama, T. J. Am. Chem. Soc. 1998, 120, 4051.
(10) Clark, T. P.; Landis, C. R.; Freed, S. L.; Klosin, J.; Abboud, K. A.
J. Am. Chem. Soc. 2005, 127, 5040.
(11) McDonald, R. I.; Wong, G. W.; Neupane, R. P.; Stahl, S. S.;
Landis, C. R. J. Am. Chem. Soc. 2010, 132, 14027.
(12) Watkins, A. L.; Landis, C. R. J. Am. Chem. Soc. 2010, 132, 10306.
(13) Watkins, A. L.; Hashiguchi, B. G.; Landis, C. R. Org. Lett. 2008,
10, 4553.
(14) Wong, G. W.; Adint, T. T.; Landis, C. R. Org. Synth. 2012, 89,
243.
(15) Risi, R. M.; Burke, S. D. Org. Lett. 2012, 14, 1180.
(16) Ho, S.; Bucher, C.; Leighton, J. L. Angew. Chem., Int. Ed. 2013,
52, 6757.
(17) Adint, T. T.; Wong, G. W.; Landis, C. R. J. Org. Chem. 2013, 78,
4231.
(18) Landis, C. R.; Clark, T. P.; Halpern, J. Proc. Natl. Acad. Sci.
U.S.A. 2004, 101, 5428.
(19) Bayer, E.; Albert, K.; Willisch, H.; Rapp, W.; Hemmasi, B.
Macromolecules 1990, 23, 1937.
(20) Bayer, E. Angew. Chem., Int. Ed. 1991, 30, 113.
(21) Klosin, J.; Landis, C. R. Acc. Chem. Res. 2007, 40, 1251.
(22) Cobley, C. J.; Gardner, K.; Klosin, J.; Praquin, C.; Hill, C.;
Whiteker, G. T.; Zanotti-Gerosa, A. J. Org. Chem. 2004, 69, 4031.
(23) Cobley, C. J.; Klosin, J.; Qin, C.; Whiteker, G. T. Org. Lett.
2004, 6, 3277.
(24) Cycles 3−5 are within error of 0 Rh g/mL.
(25) Stepan, A. F.; Karki, K.; McDonald, W. S.; Dorff, P. H.; Dutra, J.
K.; DiRico, K. J.; Won, A.; Subramanyam, C.; Efremov, I. V.;
O’Donnell, C. J.; Nolan, C. E.; Becker, S. L.; Pustilnik, L. R.; Sneed, B.;
Sun, H.; Lu, Y.; Robshaw, A. E.; Riddell, D.; O’Sullivan, T. J.; Sibley,
E.; Capetta, S.; Atchison, K.; Hallgren, A. J.; Miller, E.; Wood, A.;
Obach, R. S. J. Med. Chem. 2011, 54, 7772.
(26) Chikkali, S. H.; Bellini, R.; de Bruin, B.; van, d. V.; Reek, J. N. H.
J. Am. Chem. Soc. 2012, 134, 6607.
(27) Chikkali, S. H.; Bellini, R.; Berthon-Gelloz, G.; van der Vlugt, J.
I.; de Bruin, B.; Reek, J. N. H. Chem. Commun. 2010, 46, 1244.
(28) Fernan
Lett. 2013, 15, 3634.
́ ́
dez-Perez, H.; Benet-Buchholz, J.; Vidal-Ferran, A. Org.
(29) Calculated from enantiomeric excess and branch to linear ratios
of combined run to individual runs.
K
dx.doi.org/10.1021/ja501568k | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX