Fluorous catalysis under homogeneous conditions without fluorous solvents: A "greener" catalyst recycling protocol based upon temperature-dependent solubilities and liquid/solid phase separation

Article abstract of DOI:10.1021/ja029241s

The thermomorphic fluorous phosphines P((CH₂)m(CF₂)₇CF₃)₃ (m = 2, 1a; m = 3, 1b) exhibit ca. 600-fold solubility increases in n-octane between -20 (1a = 0.104 mM) and 80 °C (63.4 mM) and 1500-fold solubility increases between -20 and 100 °C (151 mM). They catalyze conjugate additions of alcohols to methyl propiolate under homogeneous conditions in n-octane at 65 °C and can be recovered by simple cooling and precipitation and used again. This avoids the use of fluorous solvents during the reaction or workup, which are expensive and can leach in small amounts. Teflon shavings can be used to mechanically facilitate recycling, and ³¹P NMR analyses indicate >97% phosphorus recovery (85.2% la, 12.2% other). ¹⁹F NMR analyses show that 2.3% of the (CF₂)₇CF₃ moieties of 1a reach, in some form, into the n-octane (value normalized to phosphorus). 1a similarly catalyzes additions in the absence of solvent. Yield data match or exceed those of reactions conducted under fluorous/organic liquid/liquid biphase conditions. The extra methylene groups render 1b more nucleophilic than la and, thus, a more active catalyst. The temperature dependence of the solubility of la is measured in additional solvents and compared to that of the nonfluorous phosphine PPh₃.

Full text of DOI:10.1021/ja029241s

Published on Web 04/22/2003
Fluorous Catalysis under Homogeneous Conditions without
Fluorous Solvents: A “Greener” Catalyst Recycling Protocol
Based upon Temperature-Dependent Solubilities and Liquid/
Solid Phase Separation
Marc Wende and J. A. Gladysz*
Contribution from the Institut fu¨r Organische Chemie, Friedrich-Alexander-UniVersita¨t
Erlangen-Nu¨rnberg, Henkestrasse 42, 91054 Erlangen, Germany
Received November 6, 2002; E-mail: gladysz@organik.uni-erlangen.de
Abstract: The thermomorphic fluorous phosphines P((CH2)m(CF2)7CF3)3 (m ) 2, 1a; m ) 3, 1b) exhibit
ca. 600-fold solubility increases in n-octane between -20 (1a ) 0.104 mM) and 80 °C (63.4 mM) and
1500-fold solubility increases between -20 and 100 °C (151 mM). They catalyze conjugate additions of
alcohols to methyl propiolate under homogeneous conditions in n-octane at 65 °C and can be recovered
by simple cooling and precipitation and used again. This avoids the use of fluorous solvents during the
reaction or workup, which are expensive and can leach in small amounts. Teflon shavings can be used to
mechanically facilitate recycling, and 31P NMR analyses indicate >97% phosphorus recovery (85.2% 1a,
12.2% other). 19F NMR analyses show that 2.3% of the (CF2)7CF3 moieties of 1a leach, in some form, into
the n-octane (value normalized to phosphorus). 1a similarly catalyzes additions in the absence of solvent.
Yield data match or exceed those of reactions conducted under fluorous/organic liquid/liquid biphase
conditions. The extra methylene groups render 1b more nucleophilic than 1a and, thus, a more active
catalyst. The temperature dependence of the solubility of 1a is measured in additional solvents and compared
to that of the nonfluorous phosphine PPh3.
Introduction
other types of fluids are available, encompassing a broad range
of molecular weights and phase behavior.4
Over the last eight years, many new catalysts with high
affinities for fluorous solvents have been synthesized.1-3 This
has been prompted by the development of “fluorous biphase
catalysis”,2a which, as most often practiced, exploits the
markedly temperature-dependent miscibilities of organic and
fluorous solvents.4 As shown schematically in Figure 1A, most
combinations give two phases at room temperature. However,
with moderate heating, one phase is obtained. Reactions can
be catalyzed under monophasic conditions at the high-temper-
ature limit, and the products and catalyst can be separated under
biphasic conditions at the low-temperature limit. Most fluorous
solvents in current use are saturated perfluorocarbons, but many
High fluorous liquid-phase affinities can be engineered into
catalysts by attaching a number of “pony tails” of the formula
(CH2)m(CF2)n-1CF3, abbreviated (CH2)mRfn so that n represents
the number of fluorinated carbons. This often results in a low-
melting solid. In the absence of such groups, most organic
molecules of interest show marked affinities for the nonfluorous
phase (>95:<5). Typical values of m and n are 0-3 and 6-10,
respectively. The (CH2)m segment can be viewed as an electronic
“tuning element”, the length of which modulates the effect of
the highly electron-withdrawing perfluoroalkyl groups on the
reaction center.
No catalyst recovery method is without potential drawbacks.5
Accordingly, the fluorous solvent requirement in Figure 1A has
prompted various concerns, the major of which involve cost,
solvent leaching, and environmental persistence.4 For example,
the ether layer of an ether/water liquid/liquid biphase system
contains dissolved water. Small amounts of fluorous solvents
are similarly found in the organic layers of fluorous/organic
liquid/liquid biphase systems. This makes losses unavoidable
upon phase separation. Although saturated perfluorocarbons
appear to have almost infinite half-lives in the environment, it
should be emphasized that they do not contribute to ozone
(1) See: Gladysz, J. A.; Curran, D. P. Tetrahedron 2002, 58, 3823 and the
following articles in this special issue entitled “Fluorous Chemistry”.
(2) (a) Horva´th, I. T. Acc. Chem. Res. 1998, 31, 641. (b) de Wolf, E.; van
Koten, G.; Deelman, B.-J. Chem. Soc. ReV. 1999, 28, 37. (c) Cavazzini,
M.; Montanari, F.; Pozzi, G.; Quici, S. J. Fluorine Chem. 1999, 94, 183.
(d) Hope, E. G.; Stuart, A. M. J. Fluorine Chem. 1999, 100, 75. (e) Yoshida,
J.; Itami, K. Chem. ReV. 2002, 102, 3693. (f) Dobbs, A. P.; Kimberly, M.
R. J. Fluorine Chem. 2002, 118, 3.
(3) Selected full papers with extensive literature background: (a) Horva´th, I.
T.; Kiss, G.; Cook, R. A.; Bond, J. E.; Stevens, P. A.; Ra´bai, J.; Mozeleski,
E. J. J. Am. Chem. Soc. 1998, 120, 3133. (b) Juliette, J. J. J.; Rutherford,
D.; Horva´th, I. T.; Gladysz, J. A. J. Am. Chem. Soc. 1999, 121, 2696. (c)
Richter, B.; Spek, A. L.; van Koten, G.; Deelman, B.-J. J. Am. Chem. Soc.
2000, 122, 3945. (d) Zhang, Q.; Luo, Z.; Curran, D. P. J. Org. Chem. 2000,
65, 8866. (e) Colonna, S.; Gaggero, N.; Montanari, F.; Pozzi, G.; Quici, S.
Eur. J. Org. Chem. 2001, 181.
(4) Survey of practical considerations and underlying physical principles:
Barthel-Rosa, L. P.; Gladysz, J. A. Coord. Chem. ReV. 1999, 190-192,
587.
(5) For essays on the “ideal recoverable catalyst”, see: (a) Gladysz, J. A. Pure
Appl. Chem. 2001, 73, 1319. (b) Gladysz, J. A. Chem. ReV. 2002, 102,
3214.
9
10.1021/ja029241s CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 5861-5872
5861

A R T I C L E S
Wende and Gladysz
Figure 1. Possible protocols for fluorous biphase catalysis: (A) recycling by liquid/liquid phase separation; (B) recycling by liquid/solid phase separation.
depletion. Furthermore, there are no known toxicity issues or
hazardous properties, consistent with the extensive use of Teflon
in household cookware.
recovery sequences.8,9 However, to our knowledge, this strategy
has not been applied to a broad class of nonmacromolecular
catalysts.
We sought a means to eliminate the fluorous solvent
requirement for fluorous catalysis. We speculated that the same
factors that give highly temperature-dependent organic/fluorous
liquid/liquid phase miscibilities might also give highly temper-
ature-dependent organic/fluorous liquid/solid phase equilibria
(e.g., solubilities). In less conceptual terms, as we gained more
and more experience with fluorous molecules containing several
long pony tails - species that are often essentially insoluble in
organic solvents at room temperature - we began to notice
marked increases in solubilities with temperature. Of course,
above the melting point of a compound, the issue again becomes
one of liquid/liquid miscibility.
This suggested that fluorous catalysts might be utilized under
one-liquid-phase conditions involving ordinary organic solvents
as shown in Figure 1B. The system would first be warmed to
achieve monophasic reaction conditions. Subsequent cooling
would precipitate the catalyst, and recovery would involve a
simple liquid/solid phase separation. Catalysts that exhibit
strongly temperature-dependent properties - including but not
limited to solubilities - have been termed thermomorphic.6 This
term is also applied to reactions involving reversible phase
transitions such as in Figure 1. Several types of polymer-bound
catalysts with temperature-dependent solubilities have been
developed by Bergbreiter7 and exploited in clever reaction/
A test reaction to demonstrate the applicability of this concept
to fluorous catalysts was required. In the course of our extensive
synthetic and physical investigations involving fluorous aliphatic
phosphines P((CH2)mRfn)3 (1),10-12 we noticed that nearly all
were very poorly soluble in organic solvents and were solids
melting under 100 °C. This drew our attention to the many types
of phosphine-catalyzed organic reactions in the literature.13,14
We furthermore sought a transformation where the phosphine
would represent the catalyst rest state, for - as emphasized
elsewhere5 - any special property being exploited for catalyst
recovery must be associated with the rest state.
Accordingly, the conjugate addition of alcohols (2) to methyl
propiolate (3) shown in Scheme 1 was selected for study.14,15
(8) Breuzard, J. A. J.; Tommasino, M. L.; Bonnet, M. C.; Lemaire, M. J.
Organomet. Chem. 2000, 616, 37.
(9) Some relevant recent reviews: (a) Dickerson, T. J.; Reed, N. N.; Janda,
K. D. Chem. ReV. 2002, 102, 3325. (b) Bergbreiter, D. E. Chem. ReV. 2002,
102, 3345.
(10) Alvey, L. J.; Rutherford, D.; Juliette, J. J. J.; Gladysz, J. A. J. Org. Chem.
1998, 63, 6302.
(11) (a) Alvey, L. J.; Meier, R.; Soo´s, T.; Bernatis, P.; Gladysz, J. A. Eur. J.
Inorg. Chem. 2000, 1975. (b) Klose, A.; Gladysz, J. A. Tetrahedron:
Asymmetry 1999, 10, 2665. (c) See also: Soo´s, T.; Bennett, B. L.;
Rutherford, D.; Barthel-Rosa, L. P.; Gladysz, J. A. Organometallics 2001,
20, 3079.
(12) Jiao, H.; Le Stang, S.; Soo´s, T.; Meier, R.; Rademacher, P.; Kowski, K.;
Jafarpour, L.; Hamard, J.-B.; Nolan, S. P.; Gladysz, J. A. J. Am. Chem.
Soc. 2002, 124, 1516.
(13) (a) Ciganek, E. Org. React. 1997, 51, 201. (b) Buono, G.; Chiodi, O.; Wills,
M. Synlett 1999, 377. (c) Lu, X.; Zhang, C.; Xu, Z. Acc. Chem. Res. 2001,
34, 535. (d) Vedejs, E.; Daugulis, O.; MacKay, J. A.; Rozners, E. Synlett
2001, 1499. (e) Basavaiah, D.; Rao, A. J.; Satyanarayana, T. Chem. ReV.
2003, 103, 811.
(6) (a) Bergbreiter, D. E.; Osburn, P. L.; Wilson, A.; Sink, E. M. J. Am. Chem.
Soc. 2000, 122, 9058 and references therein. (b) Calhoun, C. D.; Delgado,
J. U.S. patent 5,888,650 (filed June 3, 1996); Chem. Abs. 1998, 128, 62572h.
(7) Selected lead references: (a) Bergbreiter, D. E.; Chandran, R. J. Am. Chem.
Soc. 1987, 109, 174. (b) Bergbreiter, D. E.; Zhang, L.; Mariagnanam, V.
M. J. Am. Chem. Soc. 1993, 115, 9295. (c) Bergbreiter, D. E.; Case, B. L.;
Liu, Y.-S.; Caraway, J. W. Macromolecules 1998, 31, 6053.
(14) (a) Inanaga, J.; Baba, Y.; Hanamoto, T. Chem. Lett. 1993, 241. (b) Baba,
Y. Dissertation, 1991. (c) Meier, R. Doctoral Dissertation, Universita¨t
Dortmund, 1998.
(15) The Bayliss-Hillman reaction13 would have been another obvious choice,
but an efficient fluorous catalyst has not yet been developed.
9
5862 J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003

Fluorous Catalysis without Fluorous Solvents
A R T I C L E S
Scheme 1. Phosphine-Catalyzed Addition of Alcohols (2) to
Methyl Propiolate (3)
Table 1. Catalysis by Phosphines in Non-Fluorous Solvents: Test
Reaction Parametersa
entry
R′OH
catalyst
solvent
yield (%) time (h)
1
2
3
4
5
6
7
8
9
2a PhCH2OH
P(n-Bu)3 CH2Cl2
72b
90c
83c
85b
78c
89b
86c
81c
99c
77b
86c
0.25
24d
8
0.25
48
0.25
0.25
24
8
0.25
1a
1b
CF3C6H5
CF3C6H5
2b Ph2CHOH
P(n-Bu)3 CH2Cl2
1a CF3C6H5
The mechanism is believed to involve an initial addition of the
phosphine to the CtC moiety to give a zwitterionic allenolate
(I), which then deprotonates the alcohol, yielding a vinyl
phosphonium salt (II).14a An alkoxide addition to give an enolate
(III), followed by phosphine elimination, gives the product 4
and regenerates the catalyst. Most phosphines do not indepen-
dently react with 4. Hence, as more fully documented and
analyzed below, intermediate III can be excluded as a rest state.
Under appropriate conditions, the other intermediates are also
unlikely rest states.
In this paper, we describe the successful application of the
catalyst recycling protocol in Figure 1B to the addition reaction
in Scheme 1, using the thermomorphic fluorous phosphines
P((CH2)2Rf8)3 (1a) and P((CH2)3Rf8)3 (1b).10 Catalyst solubility,
leaching, and recovery data are reported, as well as extensions
to solvent-free conditions and the use of physical supports. A
portion of this work has been communicated.16 Related efforts
from Yamamoto’s and other laboratories are summarized in the
Discussion.17-19
2c PhCH(CH3)OH P(n-Bu)3 CH2Cl2
P(n-Bu)3 CF3C6H5
1a
1b
CF3C6H5
CF3C6H5
10
11
2d CH3(CH2)7OH
P(n-Bu)3 CH2Cl2
1a CF3C6H5
48
a The mol ratio of 2:3:catalyst was 1.0:1.1:0.1. b Isolated yield (g98%
purity). c GC yield (vs internal standard); starting concentrations: 0.3 M
(2b) or 0.5 M (2a,c,d). d 95% yield after 96 h reaction time.
both P(n-Bu)3 and 1a (Table 1; entries 7, 8). After 15 min, GC
analysis of the former addition showed an 86% yield of 4 and
no remaining 2c. The latter addition was slower, with 35%, 52%,
and 81% yields of 4 after 3, 8, and 24 h. Phosphine 1a similarly
catalyzed the additions of the other alcohols 2a,b,d to 3 in
CF3C6H5 (Table 1; entries 2, 5, 11).
The electronic properties of fluorous phosphines have been
studied in detail,12 and 1a is distinctly less basic and nucleophilic
than P(n-Bu)3. This should retard the rate of the first step in
Scheme 1. However, the slower rate renders this catalyst a more
convenient test case for the protocol in Figure 1B. More reactive
fluorous phosphines are described below. Control experiments
without phosphine catalysts gave no reactions, as further detailed
below.
2. Catalysis in Fluorous Solvent Systems. Partition coef-
ficients of many fluorous compounds have been measured in
mixtures of perfluoro(methylcyclohexane) (CF3C6F11) and tolu-
ene.4 That of phosphine 1a is very high (>99.7:<0.3).10 Thus,
it should be efficiently recovered under the traditional liquid/
liquid biphase conditions in Figure 1A, and recycling protocols
involving CF3C6F11 were investigated. Experiments were first
conducted in the ternary solvent system CF3C6F11/toluene/
hexane (3:1:3 v/v/v).2a Reactions with 2a-d, 3, and 1a (1.0:
1.1:0.1 mol ratio) proceeded under one-phase conditions at 35-
40 °C. After 24-48 h, GC analyses showed the addition
products 4a-d in 88-68% yields. However, when recycling
runs were conducted under similar conditions (40-65 °C), the
phase behavior varied. More stable systems were sought.
Thus, reactions of 2a,c, 3, and 1a were conducted in the
binary solvent system CF3C6F11/n-octane (5:8 v/v). Mole ratios
of 2.0:1.0:0.1 were utilized and maintained for all recycling
experiments below. After 8 h under one-phase conditions at 65
°C, samples were cooled. Although two phases were present at
room temperature, they were separated via syringe at -30 °C.
This represents the temperature of a conveniently located freezer.
Results
1. Test Reaction Parameters. The four alcohols shown in
Table 1 - primary benzylic 2a, secondary benzylic 2b,c, and
primary aliphatic 2d - were selected for study. Experiments
were first conducted with the catalyst system originally reported,
10 mol % P(n-Bu)3 in CH2Cl2.14a Preparative reactions with 3
(1.1 equiv, room temperature, 15 min) gave the addition products
4a-d in 89-72% yields as g96:e4 E/Z mixtures, after
chromatography or distillation (Table 1, entries 1, 4, 6, 10).
Although these compounds have been described previously, only
one piece of characterization, the 1H NMR spectrum of 4a, has
been reported.14b Hence, complete analytical data are given in
the Experimental Section.
We sought to assay the effectiveness of the fluorous phos-
phine 1a as a catalyst. In this compound, two methylene groups
separate the phosphorus from the perfluoroalkyl moieties.
However, 1a is insoluble in CH2Cl2, precluding rate compari-
sons. Fortunately, the less polar solvent CF3C6H5 is normally
able to dissolve both fluorous and nonfluorous compounds.20
Thus, the reaction of 2c and 3 was repeated in CF3C6H5 with
(16) Wende, M.; Meier, R.; Gladysz, J. A. J. Am. Chem. Soc. 2001, 123, 11490.
(17) Ishihara, K.; Kondo, S.; Yamamoto, H. Synlett 2001, 1371.
(18) Xiang, J.; Orita, A.; Otera, J. AdV. Synth. Catal. 2002, 344, 84.
(19) Ishihara, K.; Hasegawa, A.; Yamamoto, H. Synlett 2002, 1299.
(20) Maul, J. J.; Ostrowski, P. J.; Ublacker, G. A.; Linclau, B.; Curran, D. P.
Top. Curr. Chem. 1999, 206, 79.
9
J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003 5863

A R T I C L E S
Wende and Gladysz
Table 2. Recycling of Fluorous Phosphine Catalyst 1a in Fluorous Solvent Systems: Reactions of 2a,c and 3 To Give 4a,ca
yield 4a or 4c (%)
reactants
entry
(2.0:1.0 mol ratio)
solvent
reaction conditions
65 °C
one liquid phase
65 °C
one liquid phase
65 °C
two liquid phases
recovery conditions
-30 °C
two liquid phases
-30 °C
two liquid phases
-30 °C
two liquid phases
cycle 1
cycle 2
cycle 3
cycle 4
cycle 5
1
2a/3
octane/CF3C6F11
(8:5)
octane/CF3C6F11
(8:5)
CF3C6F11
71
82
73
68
66
2
3
2c/3
2a/3
80
81
79
90
76
86
54
87
11
39
a 10 mol % of 1a. Additional relevant conditions are described in the text and Experimental Section.
Figure 2. Solubilities of fluorous phosphines in n-octane as a function of temperature. [ 1a, GC assay; ] 1a, 31P NMR assay; 9 1b, GC assay.
The fluorous layers were washed once with cold n-octane. GC
analyses of the n-octane layers showed 4a,c in 80% and 71%
yields (Table 2; entries 1 and 2, cycle 1). The fluorous layers
were charged with fresh 2a,c, 3, and n-octane, and the reaction
was repeated. As summarized in Table 2, good yields were
maintained for 3-5 cycles, with less deactivation in the case
of 2a. Although yield data do not provide the best measure of
catalyst recovery,5 quantitative assays and rate data are provided
for other systems below.
Because all solvent minimization strategies are of interest,
an analogous sequence involving 2a was conducted in the
absence of the n-octane organic solvent. As 2a, 3, and the
product 4a remain poorly soluble in CF3C6F11 at 65 °C, the
reaction conditions were biphasic. After reaction, n-octane was
added to efficiently separate 4a from the mixture. This was
necessary due to the small scale employed (combined mass of
2a/3/1a, ca. 0.200 g; 31 wt % 1a). However, when it is not
critical to measure the exact yield from each cycle, this can be
omitted. As shown in Table 2 (entry 3), yields were somewhat
higher than under the two-solvent reaction conditions (entry 1).
3. Temperature-Dependent Solubilities. With the viability
of catalyst recycling under organic/fluorous liquid/liquid biphase
conditions for the test reaction established, we sought to probe
whether the solubility properties of 1a were suitable for the
fluorous-solvent-free protocol in Figure 1B. We envisioned
using an aliphatic hydrocarbon solvent, with the idea this would
bring greater amounts of a nonpolar fluorous catalyst into
solution than a more polar solvent. To limit the vapor pressure,
n-octane (bp 126 °C) was selected. Thus, the solubility of 1a
was measured by GC as described in the Experimental Section
over the temperature range from -20 to 100 °C. Data are
presented in Figure 2.21
Indeed, 1a was strongly thermomorphic. Between 20 and 80
°C, its solubility in n-octane increased ca. 60-fold. Between 20
and 100 °C, the increase was 150-fold. More important were
the low absolute concentrations at 20 °C and below. Very little
1a could be detected in solution at -20 or 0 °C (0.104 and
0.308 mM). At 20 °C, millimolar concentration levels were
present (1.13 mM). Near the melting point (47 °C), solubilities
had risen ca. 20-fold (19.6 mM, 50 °C). To ensure the absence
of artifacts, all data were checked in replicate experiments. The
results were furthermore verified by 31P NMR over the
temperature range from 0 to 50 °C, as described in the
Experimental Section. These data are also incorporated into
Figure 2 (0.084, 0.967, 3.72, and 11.6 mM at 0, 20, 40, and 50
°C, respectively).
4. Fluorous Catalysis without Fluorous Solvents. With the
above data in hand, the stage was set for the title experimental
(21) The solubility data are not corrected for any change in solvent volume
with temperature.
9
5864 J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003

Fluorous Catalysis without Fluorous Solvents
A R T I C L E S
Table 3. Recycling of Fluorous Phosphine Catalysts 1a,b without
Fluorous Solvents: Reactions of 2a-d and 3 To Give 4a-d in
n-Octane at 65 °C, with Catalyst Recovery by Liquid/Solid Phase
Separation at -30 °Ca
yield 4a−d (%)
reactants
catalyst
entry (2.0:1.0 mol ratio) (10 mol %) cycle 1 cycle 2 cycle 3 cycle 4 cycle 5
1
2
3
4
5b
6
7
2a/3
2b/3
2c/3
2d/3
2a/3
2a/3
2c/3
1a
1a
1a
1a
1a
1b
1b
82
77
90
79
99
81
92
82
84
86
84
>99
89
80
71
75
66
97
82
99
81
75
95
80
77
97
a Additional relevant data are given in the text. Starting concentration
of 2, 1.25 M; cycle time, 8 h for 1a and 1 h for 1b. b n-Octane solvent
omitted.
recovered 1a are assayed in experiments below that incorporate
additional procedural enhancements.
5. Fluorous Catalysis without Solvents. In experiments
described above, either the organic or the fluorous solvent was
eliminated from the traditional recycling protocol in Figure 1A.
The logical conclusion to this line of investigation would be to
eliminate all solvents, a priority in the field of green chemistry.22
Because 2a,c,d, 3, and 4a-d are liquids at room temperature,
there are in principle no physical impediments. A solid product
could, for example, entrain the catalyst.
Accordingly, 2a, 3, and 1a were reacted as in Table 3, but in
the absence of solvent. The catalyst 1a was insoluble at room
temperature, but dissolved as the sample was warmed to 65
°C. Upon cooling, 1a precipitated as a brown solid, from which
most of the liquid phase could easily be removed via syringe.
However, on the small scales used (combined mass of 2a/3/1a,
ca. 0.200 g), it was necessary to add some n-octane to
completely extract the product 4a. In cases where it is not critical
to analyze the exact yield from each cycle, this can be omitted.
Data for a four-cycle sequence are given in Table 3, entry 5.
The yields (all g95%) are the highest realized in this study.
6. Supports for Fluorous/Organic Liquid/Solid Biphase
Recycling. The procedures in the previous two sections require
the separation of a solid catalyst from a liquid phase or product.
A potential problem, intrinsic to such methodologies, is that
the amount of catalyst manipulated can be very small. For
example, a thermomorphic fluorous palladacycle that is a highly
effective catalyst precursor for the Heck reaction is described
below.23 It can be used at 0.0001 mol % levels. On common
laboratory scales, these catalyst quantities can barely be visual-
ized.
One way to increase the mass of recycled material is to use
a support. At one extreme, the support could be inert, with only
a mechanical function. At another level, the support might
provide physical adhesion - for example, for a waxy or gumlike
catalyst. At yet another level, the support might provide
attractive interactions that would enhance recovery. Although
attractive interactions between saturated fluorocarbons are very
small,24 we sought to bring them, at least to some degree, into
play. Hence, Teflon shavings were selected for initial study.
Figure 3. Photographs of recycling sequences. Top row (Table 3, entry
1): (A) room temperature; (B) 65 °C; (C) room temperature. Bottom row
(identical, but with added Teflon shavings): (D) room temperature; (E) 65
°C; (F) room temperature.
sequence represented by Figure 1B. Compounds 2a, 3, and 1a
(2.0:1.0:0.1 mol ratio) were combined in n-octane at room
temperature. As would be expected from Figure 2, there was
no visually detectable dissolution of 1a (supernatant 65.0 mM
in 3). The sample was next warmed to 65 °C and became
homogeneous (ca. 6.5 mM in 1a). Photographs of this sequence
are given in Figure 3 (panels A and B). After 8 h, the sample
was cooled. The catalyst 1a precipitated, as shown in panel C
in Figure 3. In a few cases, the original white color was retained,
but most samples were orange and some darkened to red with
additional recycling.
The mixture was kept at -30 °C, and the supernatant was
removed via syringe. The catalyst residue was washed once with
cold n-octane. GC analysis indicated an 82% yield of 4a. As
summarized in Table 3 (entry 1), the recovered 1a was used
for four further cycles without deterioration in yield. Multiple
runs could be similarly conducted with alcohols 2b-d, affording
comparable yields of 4b-d (Table 3, entries 2-4). When the
reaction with 2c was conducted at room temperature, 4c formed
in 25% yield. No reaction occurred when 2c and 3 were similarly
kept for 8 h at 65 °C in n-octane in the absence of 1a (<1%
4c).
To better gauge the extent of catalyst recovery, the rate of
the reaction of 2a and 3 was monitored as a function of cycle.
On the first cycle, the maximum yield was reached after ca. 1
h, as illustrated in Figure 4. The second and third cycles were
similar. More importantly, when the yields are compared at
partial conversion (0.25 h), only modest decreases are noted
(67%, 63%, 55%). Thus, ca. 90% of the activity is maintained
from cycle to cycle. Both leaching and the composition of
(22) (a) Poliakoff, M.; Fitzpatrick, M.; Farren, T. R.; Anastas, P. T. Science
2002, 297, 807. (b) DeSimone, J. M. Science 2002, 297, 799.
(23) (a) Rocaboy, C.; Gladysz, J. A. Org. Lett. 2002, 4, 1993. (b) Rocaboy, C.;
Gladysz, J. A. New J. Chem. 2003, 27, 39.
(24) (a) Gross, U.; Ru¨diger, S. Houben-Weyl Methods of Organic Chemistry;
Georg Thieme: Stuttgart, Germany, 1999; Vol. E-10a, pp 18-26. (b) Kiss,
L. E.; Ko¨vesdi, I.; Ra´bai, J. J. Fluorine Chem. 2001, 108, 95.
9
J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003 5865

A R T I C L E S
Wende and Gladysz
Figure 4. Yield as a function of time for entry 1 in Table 3 (9 cycle 1, [ cycle 2, 2 cycle 3), and (inset) for entries 3 (×) and 7 (b) in Table 3 (first cycle
only).
Selected experiments from Table 3 were repeated, but in the
presence of Teflon shavings with an average x/y/z dimension
of ca. 2 mm. Yield data were comparable, and photographs are
provided in panels D-F of Figure 3. This sequence is on the
same scale as that in panels A-C. Importantly, the catalyst
residues became more compact and firmer, despite the added
mass of the Teflon, suggestive of genuine physical adhesion or
attraction. The catalyst residues were easier to manipulate, and
less leaching occurred, as described in the following section.
second showed a much clearer CF3 19F NMR signal amounting
to 2.3% of the 1a loading (normalized to the three pony tails).
Note that all compounds of the formula X(CF2)7CF3 give
essentially the same CF3 NMR chemical shift. Hence, the “total
number of leached pony tails” can be quite sensitively deter-
mined by this method.
The 1a/Teflon residue from one of the runs was treated with
a CF3C6H5 solution of a phosphorus-containing internal standard
added. A 31P NMR spectrum indicated a 85.2% yield of 1a
(-24.5 ppm) and 12.2% of new phosphorus-containing species
(97.4% phosphorus recovery). Of the latter, two were dominant
(34.4, 44.4 ppm), and a small amount of the oxide of 1a was
also present. An analogous experiment was conducted with a
CF3C6H5 solution of a fluorine-containing internal standard
added. A 19F NMR spectrum indicated a 97.9% recovery of
pony tails.
Some of the new 31P signals could represent alternative
catalyst rest states. To help probe this point, 1a and excess 4a
were combined in CF3C6H5. A 31P NMR spectrum showed no
reaction other than the formation of some oxide. However, when
1a and 3 were similarly combined, the sample turned dark black,
and 31P NMR spectra showed >90% reaction. Although the
chemical shift of the major product was sometimes close to the
one noted above, it varied (46-48 ppm), and many other peaks
were present. The multitude of species suggests that the further
reaction of intermediate I in Scheme 1 with 3 (as opposed to
alcohol 2) provides an avenue for catalyst darkening and
decomposition.26 Note that the most nucleophilic alcohol (2a)
often gives better recycling data (Table 2, entry 1 vs entry 2).
Teflon shavings also gave improved results under other types
of recycling conditions. For example, the CF3C6H5 solvent from
reactions similar to those in Table 1 can be removed under
reduced pressure, the product 4 can be extracted from the residue
with n-octane at -30 °C, and the catalyst residue can be charged
with fresh reactants and solvents. This extraction step was much
easier, and the yields were higher, when the sequence was
conducted in the presence of Teflon shavings (data for five
cycles: 90%, 99%, 95%, 97%, 94%).
7. Leaching and Catalyst Recovery. Leaching can be
analyzed from the standpoint of both the catalyst (rest state)
and the catalyst degradation products. The reactions in entries
1-4 of Table 3 and panels A-C and D-F of Figure 3 involve
the separation of an n-octane product solution from a 1a catalyst
residue at -30 °C and a subsequent n-octane extraction. From
the quantities employed, the solubility data in Figure 2 predict
catalyst leaching of ca. 0.33% per cycle (value calculated at
-20 °C). This rises to 1.0% and 3.6% if phase separations are
conducted at 0 and 20 °C.
Leaching was first assayed for reactions conducted in the
presence of Teflon shavings (Figure 3, panels D-F). The
n-octane product solutions from two runs were treated with
internal standards for 31P and 19F NMR analysis, as detailed in
the Experimental Section. The first showed a barely integratable
31P NMR signal for the oxide of 1a (41.9 ppm),10 corresponding
to ca. 0.2% leaching. No other signals could be detected.25 The
Related experiments were conducted for identical reactions
in the absence of Teflon shavings. First, 19F NMR analyses of
(25) Under the conditions of this experiment (which has no n-octane extraction),
Figure 2 predicts (for phase separation at -20 °C) ca. 0.17% leaching of
1a per cycle.
(26) All recycling experiments employ a 2.0:1.0 2/3 mol ratio. Under these
conditions, I should be a less long-lived state.
9
5866 J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003

Fluorous Catalysis without Fluorous Solvents
A R T I C L E S
Figure 5. Solubility of 1a in toluene 9, chlorobenzene O, dioxane [, and n-octane 2 as a function of temperature, as assayed by GC.
the n-octane extracts showed a higher level of leached pony
tails (first cycle, 7.1%; second cycle, 9.1% based upon 1a
present after first cycle). Second, 31P NMR analysis of the
recovered catalyst showed a 90.2:6.0:3.8 ratio of 1a, the oxide,
and other species after the first cycle, and 89.3:6.5:4.2 and 74.5:
14.0:11.5 ratios after the second and third cycles. Thus, a gradual
degradation of the phosphine 1a is evident under all recovery
conditions.
rates of these two reactions are compared in the inset in Figure
4 (1b . 1a).
9. Additional Solubility Data. Experiments were conducted
to provide further context for the above results. First, we were
curious about the temperature dependence of the solubility of
1a in other media. Results for another three solvents that are
fairly high boiling and inert to phosphines - toluene, chlo-
robenzene, and 1,4-dioxane - are given in Figure 5.21 Not
unsurprisingly, the solubilities of 1a in these more polar media
at g20 °C are lower than those in n-octane. Interestingly, the
solubilities in toluene and chlorobenzene at -20 °C are slightly
higher (0.360 and 0.205 vs 0.104 mM). In any event, the net
result is a lower concentration gradient with temperature.
8. Additional Catalysts. To test the generality of the above
phenomena and protocols, the fluorous phosphine 1b10 was
similarly studied. In this catalyst, three methylene groups
separate the phosphorus from the perfluoroalkyl moieties. The
additional “insulation” renders 1b considerably more basic and
nucleophilic than 1a, but still far below the levels of tri(n-alkyl)-
phosphines.12 It also renders the CF3C6F11/toluene partition
coefficient slightly lower than that of 1a (98.8:1.2 vs >99.7:
<0.3). However, the melting point is somewhat higher (71.5-
72.5 vs 47 °C).
However, note that from a practicality standpoint, the key
issue for the protocol in Figure 1B will usually not be achieving
a high catalyst concentration at the high-temperature limit. Any
problem in this regard can, in principle, be countered with a
more active catalyst. Rather, the critical point is to have a
catalyst that is insoluble at the low-temperature limit. Thus,
concentration gradients of the types in Figure 5 should often
suffice.
Reactions of 2a,c, 3, and 1b were conducted in CF3C6H5
under homogeneous conditions analogous to those used for 1a
and P(n-Bu)3. As shown in entries 3 and 9 of Table 1, 4a,c
formed in 83% and 99% yields over the course of 8 h at room
temperature. The results in entries 7-9, coupled with monitoring
data given above, show that 1b is a much more reactive catalyst
than 1a, but remains less reactive than P(n-Bu)3.
Another question regards the extent to which the temperature
dependence of the solubilities of fluorous phosphines differs
from those of nonfluorous phosphines. Unfortunately, there are
no readily available solid tri(n-alkyl)phosphines. As a substitute,
the solubility of PPh3 (mp 79-81 °C) in n-octane was measured.
As shown in Figure 6, a substantial temperature dependence
was also observed.21 However, note that the solubility remains
much higher than that of 1a at lower temperatures (27.9 vs 0.104
mM, -20 °C). Under the conditions of the reactions in entries
1-4 of Table 3, ca. 89% of the PPh3 would leach! Furthermore,
this solubility would markedly increase in the more polar
solvents in Figure 5. Hence, it is the miniscule low-temperature
solubilities of appropriate fluorous phosphines in nonfluorous
solvents that render them uniquely suitable for recovery via the
protocol in Figure 1B.
The solubility of 1b in n-octane as a function of temperature
was measured by GC as described for 1a, and the results are
shown graphically in Figure 2.21 The solubilities of 1a and 1b
are strikingly similar. The fact that there is little difference in
the region between the melting points (47-72 °C) indicates there
is no dramatic solubility increase associated with the solid/liquid
phase transition. As would be expected, 1b gave excellent results
in the thermomorphic, fluorous-solvent-free catalytic reactions
described above for 1a. As summarized in Table 3 (entries 6
and 7), high yields of 4a,c were obtained with much shorter
run times (1 h vs 8 h). In the case of 4c, the data were distinctly
better than those with catalyst 1a (entries 7 vs 3). The relative
9
J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003 5867

A R T I C L E S
Wende and Gladysz
Figure 6. Solubility of 1a [ and PPh3 9 in n-octane as a function of temperature, as assayed by GC.
Discussion
depicted in Figure 1B for rapid reaction. Indeed, as noted above,
1a gives a detectable background reaction of 2c and 3 in
n-octane at room temperature.
1. Merits of Methodology. The preceding data clearly
demonstrate that fluorous catalysts - here the phosphines 1a,b
- can be applied in single-phase, homogeneous reactions in
the absence of fluorous solvents and efficiently recovered via
simple liquid/solid phase separations at lower temperatures. This
protocol, which is exemplified in Figure 3 and generalized in
Figure 1B, significantly expands the scope of fluorous biphase
catalysis as originally formulated (Figure 1A).2a Although an
organic solvent (n-octane) is utilized under the standard condi-
tions, reactions and catalyst recovery can also be effected under
solvent-free conditions.
In anticipation of other applications in which catalyst loadings
might be much smaller and therefore less convenient to
manipulate, we have also established the viability of using an
auxiliary solid support. This remains insoluble during the
sequence (Figure 3, panel E) and increases the mass of the
recycled material. A support could also be used to provide
attractive fluorophilic interactions that would enhance the level
of catalyst recovery. Our choice of Teflon shavings was made
with this possibility in mind, but there are many other conceiv-
able candidates, such as fluorous reverse phase silica gel.27
Indeed, related methodologies involving fluorous silica gel have
recently been reported, as described below.28
Some limitations associated with our protocol merit note.
First, insoluble byproducts will interfere with catalyst recovery.
Although it would certainly be simple to separate a typical
inorganic salt from catalysts such as 1a,b, this adds an extra
manipulation. Second, because heating is required to achieve
homogeneity, the method is best suited for reactions normally
conducted at elevated temperatures. However, the viability of
fluorous catalysis under nonhomogeneous conditions deserves
emphasis. For example, there are many examples where
reactions of the type in Figure 1A proceed at convenient rates
at the biphasic low-temperature limit.3b,29,30 Similarly, it will
not always be necessary to achieve the homogeneous conditions
2. Solubility Considerations. Temperature-dependent solu-
bility is of course a very general phenomenon with respect to
all solvents and solutes. Figure 2 provides the first quantitative
data for fluorous compounds. The phosphines 1a,b exhibit ca.
400-, 600-, and 1500-fold solubility increases in n-octane
between -20 and 60, 80, and 100 °C, respectively.21 The
solubility increases for PPh3 in Figure 6 appear more dramatic,
but in reality that between -20 and 60 °C is only ca. 33-fold.
However, as noted above, the degree of temperature dependence
is not the critical issue. Nearly all industrially important
homogeneous catalysts operate at loadings of much less than
1%, and high solution concentrations are not necessary. Rather,
the key point is to design a catalyst with essentially zero
solubility at the low-temperature limit.
A number of fluorous compounds with multiple pony tails
containing at least eight perfluorinated carbons (gRf8) have been
found to exhibit very low solubilities at room temperature,
particularly in organic solvents but sometimes also in fluorous
solvents. This is often noticed in the course of NMR charac-
terization. Some typical examples which furthermore show
markedly enhanced solubilities at higher temperatures are
summarized in Figure 7, together with the relevant solvents.
Compounds from our group include the sulfoxides 5,31 the aryl
iodides and (diacetoxyiodo)arenes 6,32 and the N-donor and
S-donor palladacycles 7,8.23 Compounds from other groups
include the phenylboronic acid 9,17 the distannoxane 10,18 and
the 1,1-bis(sulfone) 11.19 These phenomena are clearly not
restricted to any one class of molecule or solvent.
Thus, the fine-tuning of solubility properties is of obvious
importance. Figure 2 shows that a one-carbon difference in the
methylene spacer segment, (CH2)2 versus (CH2)3, has little effect
upon solubility. However, increasing or decreasing an Rf8
(29) Rutherford, D.; Juliette, J. J. J.; Rocaboy, C.; Horva´th, I. T.; Gladysz, J.
A. Catal. Today 1998, 42, 381.
(30) Dinh, L. V.; Gladysz, J. A. Tetrahedron Lett. 1999, 40, 8995.
(31) Rocaboy, C.; Gladysz, J. A. Tetrahedron 2002, 58, 4007.
(32) Rocaboy, C.; Gladysz, J. A. Chem.-Eur. J. 2003, 9, 88.
(27) Curran, D. P. Synlett 2001, 9, 1488.
(28) (a) Tzschucke, C. C.; Markert, C.; Glatz, H.; Bannworth, W. Angew. Chem.,
Int. Ed. 2002, 41, 4500; Angew. Chem. 2002, 114, 4678. (b) Schwinn, D.;
Glatz, H.; Bannwarth, W. HelV. Chim. Acta 2003, 86, 188.
9
5868 J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003

Fluorous Catalysis without Fluorous Solvents
A R T I C L E S
Figure 7. Some fluorous compounds with very low solubilities at room temperature, but much higher solubilities at elevated temperatures.
segment by two carbons can have a profound influence.10,33
Longer perfluoroalkyl groups always diminish solubilities, and
quantitative data on the phosphines P((CH2)3Rf6)3 (a liquid) and
P((CH2)3Rf10)3 (mp 102-103 °C) will be presented in a future
report.34 Qualitatively higher solubilities have also been noted
for analogues of 11 with shorter pony tails.19 Fluorous phos-
phines that are less symmetrically substituted, such as P((C-
H2)2Rf8)((CH2)3Rf8)2 (mp 68-69 °C),11a also deserve investiga-
tion. None of the compounds in Figure 7 contain branched pony
tails, a morphology that is so far somewhat rare35 and would
be expected to affect solubilities. Additional pony tails com-
monly diminish solubilities, although with arenes substitution
patterns are often more important (1,4-disubstituted less soluble
than 1,3,5-trisubstituted).32,33
Teflon, which impart more of the solubility characteristics of
the polymer as they are lengthened. This “inverse immobiliza-
tion” strategy offers virtually infinite possibilities for engineering
the desired physical properties into the catalyst.
3. Test Reactions. Given our success with the alcohol
addition reactions in Scheme 1, appropriate fluorous phosphines
can likely serve as recoverable catalysts for other phosphine-
catalyzed reactions.13 It should be emphasized that no attempt
has been made to optimize addition rates. From the order of
catalyst activity, P(n-Bu)3 > 1b > 1a, it is obvious that still
faster rates would be obtained with the more highly “insulated”
fluorous phosphines P((CH2)4Rf8)3 (mp 44-45 °C)10 and
P((CH2)5Rf8)3 (mp 44 °C).11a Alternatively, reactions might be
conducted in more polar solvents such as chlorobenzene. The
solubilities of 1a in chlorobenzene at 60 and 70 °C (4.72 and
7.56 mM; Figure 5) are close to the concentration used for
reactions in n-octane at 65 °C (6.5 mM). The theoretical
maximum leaching of 1a would be 0.6% per cycle (calculated
for phase separation at -20 °C).
Although the catalysts 1a,b degrade somewhat as they are
recycled, no phosphine leaching can be detected by 31P NMR.
This dramatically illustrates the viability of the recovery protocol
in Figure 1B. The more sensitive 19F NMR assay shows that
other materials do leach, but this is significantly decreased in
the presence of Teflon shavings (2.7% vs 7.1%). These
substances must be either catalyst degradation products or
alternative catalyst rest states that are more soluble. In our
opinion, there are a number of ways by which the former might
be minimized. For example, an alcohol/methyl propiolate or 2/3
When molecular catalysts are covalently bound to polymeric
supports, they generally assume the solubility properties of the
host polymer. In developing thermomorphic fluorous catalysts
as described above, we like to think that we are covalently
attaching a small segment of a polymer to a molecular catalyst.
In other words, the pony tails can be viewed as short pieces of
(33) Rocaboy, C.; Rutherford, D.; Bennett, B. L.; Gladysz, J. A. J. Phys. Org.
Chem. 2000, 13, 596.
(34) Wende, M.; Gladysz, J. A., manuscript in preparation.
(35) (a) Zhang, Q.; Luo, Z.; Curran, D. P. J. Org. Chem. 2000, 65, 8866. (b)
Loiseau, J.; Fouquet, E.; Fish, R. H.; Vincent, J.-M.; Verlhac, J.-B. J.
Fluorine Chem. 2001, 108, 195. (c) Nakamura, Y.; Takeuchi, S.; Okumura,
K.; Ohgo, Y. Tetrahedron 2001, 57, 5565. (d) Ra´bai, J.; Szabo´, D.; Borba´s,
E. K.; Ko¨vesi, I.; Ko¨vesdi, I.; Csa´mpai, A.; Go¨mo¨ry, AÄ .; Pashinnik, V. E.;
Shermolovich, Y. G. J. Fluorine Chem. 2002, 114, 199. (e) El Bakkari,
M.; McClenaghan, N.; Vincent, J.-M. J. Am. Chem. Soc. 2002, 124, 12942.
(f) For silicon-based branches, see: de Wolf, E.; Richter, B.; Deelman,
B.-J.; van Koten, G. J. Org. Chem. 2000, 65, 5424.
9
J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003 5869

A R T I C L E S
Wende and Gladysz
Scheme 2. Other Examples of Fluorous Catalysis without
Fluorous Solvents by the Protocol in Figure 1B
conducted in toluene and methanol, with catalyst recoveries of
70% and 68%.
5. Other Approaches to Catalyst Recovery via Liquid/
Solid Phase Separation. Several catalyst recovery protocols
related to Figure 1B deserve to be mentioned. For example,
the solubilities of fluorous compounds in organic solvents are
markedly enhanced by the application of subsupercritical CO2
pressures (e.g., 20-70 bar).36 Thus, CO2 gas can be used as a
“solubility switch”, complementing the traditional variable of
temperature. Although this phenomenon has so far been
primarily used for crystallizations, the first application in
fluorous catalysis was recently reported.37 This featured a
fluorous analogue of Wilkinson’s catalyst, which dissolved
under CO2/H2 pressures in cyclohexane and could be recovered
after hydrogenation with no detectable rhodium leaching.
There is considerable literature on polymer-based thermo-
morphic catalysts, much of which has originated from the
Bergbreiter group.7-9 The solubilities of many types of polymers
are highly temperature dependent, and Bergbreiter has pursued
two orthogonal strategies. The first involves the functionalization
of polyethylenes with various ligands and metal fragments.7a
These normally insoluble systems can dissolve in hot toluene
and xylene and precipitate on cooling. The second involves
similar functionalizations of poly(ethylene glycols)7b and
polyacrylamides.7c However, the water solubilities of these
systems show inverse temperature dependences (less soluble at
higher temperature). Bergbreiter has used this both as a recovery
strategy and as a control element to precipitate catalysts when
the desired temperature range of a reaction has been exceeded.
While this manuscript was being reviewed, Bannwarth
described the adsorption of palladium complexes of fluorous
phosphines onto fluorous reverse phase silica gel.28 The resulting
free-flowing powders were effective catalyst precursors for
Suzuki and Sonagashira reactions, which were conducted at 80-
100 °C. They could be recovered by decantation at 0 °C with
<2% palladium loss and reused. Although it is not yet clear
whether adsorbed or dissolved palladium species are primarily
responsible for these reactions, this obviously represents another
creative and useful approach to fluorous catalysis without
fluorous solvents.
ratio greater than the 2:1 employed would decrease the lifetime
of I (Scheme 1), a species that appears to give a multitude of
products in the presence of 3 alone. However, a priority from
our standpoint was to extend Figure 1B to a broader range of
reactions. A parallel investigation of Heck reactions using the
fluorous palladacycle catalysts precursors 7,8 (Figure 7) is
detailed elsewhere.23
4. Other Thermomorphic Fluorous Catalysts. Other fluo-
rous catalysts have recently been utilized under homogeneous
conditions in nonfluorous solvents and recovered by liquid/solid
phase separations. Concurrently with our communication,
Yamamoto reported that the phenylboronic acid 9 (Figure 7)
efficiently catalyzed condensations of carboxylic acids and
amines to give amides.17 Reactions could be effected in organic/
fluorous solvent mixtures, with recovery of 9 by liquid/liquid
phase separation. Although 9 was insoluble in aromatic hydro-
carbons at room temperature, it exhibited appreciable solubilities
as higher temperatures. As shown in Scheme 2A, reactions could
be catalyzed under homogeneous conditions in refluxing toluene
or xylene, and 9 could be recovered by subsequent precipitation
at room temperature. Ten such cycles were conducted, giving
the amide product in 96% isolated yield (>99% conversion per
cycle) and 9 in 26% yield (88% recovery per cycle).
6. Additional Considerations. As emphasized above, any
property being exploited for catalyst recovery must be associated
with the rest state.5 Several special challenges in this regard
are best analyzed in retrospect. First, when a catalyst or catalyst
precursor is covalently bound to a polymer support, it can be
assumed (under idealized conditions and with an appropriate
mode of attachment) that all intermediates and the rest state
remain covalently bound. All local energy minima assume the
general physical properties of the host polymer. In contrast, for
all types of liquid/liquid biphase recovery protocols involving
molecular catalysts, all species are free to partition with different
affinities between phases. The same obviously holds for liquid/
solid biphase recovery methods.
Otera has reported that fluorous distannoxanes such as 10
(Figure 7) catalyze transesterifications in organic/fluorous
solvent mixtures.18 Although 10 was insoluble in toluene at room
temperature, it dissolved at reflux and efficiently catalyzed the
transformation in Scheme 2B, as well as others. The catalyst
precipitated upon cooling, but a fluorous solvent extraction was
utilized for recovery (100%). Most recently, Yamamoto has
utilized the bis(sulfone) 11 (Figure 7) as a super Brønsted acid
catalyst.19 This species dissolved in refluxing cyclohexane and
catalyzed the acetal formation in Scheme 2C. Cooling precipi-
tated 11, which was recovered in 96% yield. Benzoylations of
alcohols and esterifications of carboxylic acids were similarly
Note that catalysts can have a distribution of rest states and
that the dominant one during a reaction (recovery from a
continuous process) can differ from that after the consumption
(36) Jessop, P. G.; Olmstead, M. M.; Ablan, C. D.; Grabenauer, M.; Sheppard,
D.; Eckert, C. A.; Liotta, C. L. Inorg. Chem. 2002, 41, 3463.
(37) (a) Eckert, C. A.; Jessop, P. G.; Liotta, C. L. U.S. Patent 2002096550;
Chem. Abstr. 2002, 138, 26344. (b) Jessop, P. G., oral presentation, “Green
Solvents for Catalysis”, Bruchsal, Germany, October 15, 2002.
9
5870 J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003

Fluorous Catalysis without Fluorous Solvents
A R T I C L E S
used as received. The phosphines P((CH2)m(CF2)7CF3)3 (m ) 2/3, 1a/
1b) were prepared as described earlier.10 Solvents were distilled as
follows and then freeze-pump-thaw degassed (3×): n-octane (Lan-
caster), 1,4-dioxane, and toluene, from Na/benzophenone; CH2Cl2, from
CaH2; chlorobenzene, CF3C6H5, and CF3C6F11 (ABCR), from P4O10;
hexanes and ethyl acetate (for chromatography), simple distillation.
NMR spectra were recorded on Bruker or JEOL 400 MHz instruments
at ambient probe temperature and referenced to residual internal CHCl3
(1H, δ 7.27) or CDCl3 (13C, δ 77.2). Gas chromatography was conducted
on a ThermoQuest Trace GC 2000 instrument. Elemental analyses were
conducted with a Carlo Erba EA1110 instrument.
Phosphine-Catalyzed Additions: Authentic Samples. A Schlenk
flask was charged with the alcohol (2a-d, 5.00 mmol), P(n-Bu)3 (0.124
mL, 0.500 mmol), and CH2Cl2 (5 mL), and placed in a 10-15 °C bath.
Next, 3 (0.490 mL, 5.50 mmol) was added with stirring over the course
of 5 min. The mixture turned deep red-brown and then black. After 10
min, air was introduced to oxidize and facilitate separation of the P(n-
Bu)3. After 1 h, the volatiles were removed by oil pump vacuum. Next,
10:1 v/v hexanes/ethyl acetate (ca. 10 mL) was added to the black tar.
The sample was filtered through a short plug of silica, which was
washed with hexanes/ethyl acetate (150 mL). Solvents were removed
from the filtrate to give crude 4a-d as viscous yellow oils. Further
purification: 4b, silica gel column chromatography, 4:1 v/v hexanes/
ether; 4a,c,d, Kugelrohr distillation (6-8 × 10-3 Torr, fraction between
90 and 130 °C). This gave 4a-d14a,b as colorless oils, some of which
turned light tan overnight, even when stored in a refrigerator under
N2.
C6H5CH2OCHdCHCO2CH3 (4a). 0.692 g, 3.60 mmol, 72%, >99:
<1 E/Z. NMR (CDCl3): 1H (400 MHz) δ ) 7.64 (d, J ) 12.2 Hz,
dCHO), 7.37-7.24 (m, C6H5), 5.31 (d, J ) 12.7 Hz, dCHC), 4.87 (s,
CH2), 3.68 (s, CH3); 13C (125 MHz) δ ) 168.0, 162.0, 135.1, 128.7,
128.5, 127.6, 97.0, 72.8, 51.1. MS (positive FAB, 3-NBA, m/z): 193
(M+, 100%). Anal. Calcd for C11H12O3: C, 68.74; H, 6.29. Found: C,
68.74; H, 6.48.
(C6H5)2CHOCHdCHCO2CH3 (4b). 1.140 g, 4.25 mmol, 85%, 96:4
E/Z. NMR (CDCl3, E isomer unless noted): 1H (400 MHz) δ ) 7.64/
7.73 (2d, J ) 12.7/12.5 Hz, E/Z dCHO), 7.37-7.27 (m, 2C6H5), 5.95
(s, CHO), 5.34 (d, J ) 12.7 Hz, dCHC), 3.64 (s, CH3); 13C (125 MHz)
δ ) 168.1, 161.3, 139.5, 128.7, 128.3, 126.8, 98.9, 85.5, 51.1. MS
(positive FAB, 3-NBA, m/z): 267 (M+, 5%), 167 ([Ph2CH]+, 100%).
Anal. Calcd for C17H16O3: C, 76.10; H, 6.01. Found: C, 75.76; H,
6.05.
C6H5CH(CH3)OCHdCHCO2CH3 (4c). 0.918 g, 4.45 mmol, 89%,
>99:<1 E/Z. NMR (CDCl3): 1H (400 MHz) δ ) 7.51 (d, J ) 12.5
Hz, dCHO), 7.37-7.27 (m, C6H5), 5.23 (d, J ) 12.5 Hz, dCHC),
5.01 (q, J ) 6.6 Hz, CHCH3), 3.62 (s, OCH3), 1.57 (d, J ) 6.6 Hz,
CCH3); 13C (125 MHz) δ ) 168.2, 161.4, 141.2, 128.8, 128.2, 125.7,
98.0, 80.5, 51.0, 23.4. MS (positive FAB, 3-NBA, m/z): 207 (M+,
100%).
CH3(CH2)7OCHdCHCO2CH3 (4d). 0.825 g, 3.85 mmol, 77%,
>99:<1 E/Z. NMR (CDCl3): 1H (400 MHz) δ ) 7.55 (d, J ) 12.5
Hz, dCHO), 5.15 (d, J ) 12.8 Hz, dCHC), 3.79 (t, J ) 6.6 Hz, OCH2),
3.66 (s, OCH3), 1.66 (OCH2CH2), 1.39-1.21 (m, 5CH2), 0.84 (t, J )
6.8 Hz, CCH3); 13C (125 MHz) δ ) 168.3, 162.7, 95.9, 71.2, 51.0,
31.7, 29.13, 29.10, 28.8, 25.7, 22.6, 14.0. MS (positive FAB, 3-NBA,
m/z): 215 (M+, 100%), 183 ([M - OMe]+, 20%). Anal. Calcd for
C12H22O3: C, 67.25; H, 10.35. Found: C, 67.23; H, 10.53.
Catalyst Recycling. A. A 4 mL screw-top vial was charged with a
stir bar, 1a (0.0686 g, 0.050 mmol), n-undecane GC standard (0.3-
0.5 mmol added gravimetrically), 2a or 2c (1.00 mmol), 3 (0.0421 g,
0.500 mmol), n-octane (0.80 mL), and CF3C6F11 (0.50 mL). The sample
was stirred at 65 °C for 8 h (monophase conditions) and stored at -30
°C overnight. The light yellow upper organic phase was carefully
removed from the lower fluorous phase by syringe. The fluorous phase
was shaken with cold n-octane (0.8 mL, -30 °C), and the n-octane
layer was similarly separated. The organic phases were combined. An
of the limiting reagent (recovery from a batch process). Hence,
the efficient recovery of molecular catalysts under biphase
conditions requires that a number of conditions are simulta-
neously fulfilled. One simplifying advantage of the test reaction
in Scheme 1 is that a high fraction of the original phosphine
catalyst is present after the reaction has gone to completion,
representing the dominant rest state.
It is also important not to blur the distinction between a
catalyst and a catalyst precursor. In particular, many metal-
containing catalysts - including fluorous systems employed for
hydrogenations,29 hydrosilylations,30 and Heck arylations23
-
feature induction periods. An induction period usually indicates
an irreversible transformation that must occur to access the
catalytically active species. In such cases, the composition and
properties of the catalyst rest state are likely to significantly
differ from those of the catalyst precursor. Thus, in the
development of additional metal-containing fluorous catalysts
for recovery according to Figure 1B, highly relevant solubility
data will not always be available. Intuition and luck must
therefore play correspondingly larger roles.
Conclusion
To our knowledge, there have been no previous attempts to
develop a broad class of molecular catalysts that have temper-
ature-dependent solubilities, such that they can be applied under
homogeneous conditions at elevated temperatures and efficiently
recovered by simple liquid/solid phase separations at lower
temperatures.38 The above data, coupled with those of other
investigators in Scheme 2, clearly establish the applicability of
appropriately designed fluorous molecules to such recycling
protocols (Figure 1B). There is every reason to predict the
evolution of a new general method for catalyst recovery -
“fluorous catalysis without fluorous solvents” - that removes
many objections to the organic/fluorous liquid/liquid biphase
procedure in Figure 1A.
The practicality of this recovery technique is critically
dependent upon a very low catalyst solubility at the low-
temperature limit, and toward this end there is a broad palette
of structural variables that can be used for optimization and
fine-tuning (e.g., number, length, branching, and substitution
pattern of perfluoroalkyl segments). There is also the tantalizing
possibility that fluorous solid supports could be used to further
enhance catalyst recovery, as demonstrated by our first-
generation experiments with Teflon shavings and Bannwarth’s
recent data with fluorous reverse phase silica gel.28 These and
other refinements are under active investigation and will be
reported in due course.
Experimental Section
General. Reactions were conducted under inert atmospheres. Methyl
propiolate (3; Aldrich, 99%), benzyl alcohol (2a; Fluka, g99%),
1-phenylethanol (2c; Fluka, 96%), 1-octanol (2d; Fluka, g99.5%),
n-undecane, and n-tetradecane (2× Aldrich, g99%) were freeze-
pump-thaw degassed (3×). Diphenylcarbinol (2b; Aldrich, 99%), P(n-
Bu)3 (Fluka, 95%), PPh3 (ABCR, 99%), and C6F6 (Aldrich, 99%) were
(38) Because the homogeneous catalysis literature is so vast, and electronic search
methods are not capable of systematically retrieving such data, there are
likely precedents that have escaped our attention. For example, considering
the numerous amine-catalyzed reactions, many of which are conducted at
elevated temperature, it would not be surprising to find scattered examples.
However, we are confident that for whatever cases that might exist, there
would be little chance of developing a general family of catalysts
encompassing the breadth of molecules represented by 1a,b and Figure 7.
9
J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003 5871

A R T I C L E S
Wende and Gladysz
aliquot (0.2 mL) was filtered through a silica gel plug (0.5 × 1 cm)
with ethyl acetate/hexanes (10 mL, 1:10 v/v). The filtrate was analyzed
by GC (0.0010 mL autoinjection). The vial with the fluorous phase
was again charged with n-undecane, 2a or 2c, 3, and n-octane, and the
procedure was repeated. Data: Table 2, entries 1-2.
B. An analogous series of experiments was conducted without
CF3C6F11 using catalysts 1a and 1b and alcohols 2a-d. The organic
phase was separated by syringe from the solid or gumlike 1a or 1b.
The solid was washed with cold n-octane (0.8 mL, -30 °C), which
was added to the organic phase. The vial was again charged with
n-undecane, 2a-d, 3, and n-octane, and the procedure was repeated.
Data: Table 3, entries 1-4, 6, and 7.
B. A solution of C6F6 (0.0576 g, 0.310 mmol) in hexanes (3.3033
g) was prepared. An experiment identical to procedure A was conducted,
and a portion was added gravimetrically (0.0761 g, 0.00701 mmol of
C6F6) to the filtered liquid phase. A solution of C6F6 (0.0847 g, 0.455
mmol) in CF3C6H5 (4.0077 g) was prepared, and a portion was added
gravimetrically (2.0074 g, 0.223 mmol of C6F6) to the solid residue.
Both samples were analyzed by 19F NMR. The total intensities of all
(CF2)7CF3-based CF3 triplets were integrated against the C6F6 signal.
Data: see text.
C. The light yellow upper layers from vials B′ and C′ of recycling
experiment D above were, after separation from catalyst 1a, treated
with C6F6 (0.007720 (B′) and 0.008789 (C′) mmol; delivered as in
recovery experiment B). The samples were similarly analyzed by 19F
NMR (vial B′ after cycle 1, vial C′ after cycle 2). Data: see text.
D. The catalyst residues from vials A′, B′, and C′ of recycling
experiment D were dissolved in CF3C6H5. Analysis by inverse-gated
31P NMR (pulse delay ) 12 s) gave the relative amounts of phosphorus-
containing species in the recovered catalyst.39 Data: see text.
Temperature-Dependent Solubilities. A. The following is repre-
sentative. A solution of n-tetradecane (0.1414 g, 0.7127 mmol) in
n-octane (2.1841 g) was prepared. A portion of this solution (0.1184 g
delivered gravimetrically) was added to a solution of n-undecane
(0.1002 g, 0.6410 mmol) in n-octane (7.0843 g). A 2 mL round-bottom
flask was fitted with a septum and a stir bar, and charged under argon
with 1a (0.250 g, 0.182 mmol) and ca. 1.0 mL of the second solution
(0.7155 g delivered gravimetrically; 0.06283 mmol of n-undecane,
0.003556 mmol of n-tetradecane). The flask was placed in a thermo-
stated bath and stirred at -20.0 °C for 2 h. Stirring was discontinued,
and the suspension was allowed to settle. After 10 min, a 0.002 mL
aliquot was removed from the clear supernatant for GC analysis.
Samples were similarly taken after 20 min equilibration periods at
higher temperatures. Aliquots taken at or above 40 °C (0.008 mL) were
diluted with degassed CF3C6H5 (0.100 mL) to avoid gel formation (see
text). Data: Figure 2.21
C. A 4 mL screw-top vial was charged with a stir bar, 1a (0.0686
g, 0.050 mmol), n-undecane GC standard (0.3-0.5 mmol added
gravimetrically), 2a (0.1081 g, 1.000 mmol), and 3 (0.0421 g, 0.500
mmol). The sample was stirred at 65 °C for 8 h (solvent-free conditions).
n-Octane (0.8 mL) was then added, and the sample was stored at -30
°C overnight. The light yellow liquid phase was carefully removed from
the catalyst residue by syringe. The residue was treated with cold
n-octane (0.8 mL, -30 °C), and the liquid phase was similarly
separated. The liquid phases were combined. An aliquot (0.2 mL) was
filtered through a silica gel plug (0.5 × 1 cm) with ethyl acetate/hexanes
(10 mL, 1:10 v/v). The filtrate was analyzed by GC (0.0010 mL
autoinjection). The vial with the gumlike catalyst was again charged
with n-undecane, 2a, and 3, and the procedure was repeated. Data:
Table 3, entry 5.
D. A solution of 3 (0.4235 g, 5.000 mmol) and n-undecane GC
standard (0.4689 g, 3.000 mmol) in n-octane (5.6240 g) was prepared.
Three 4 mL screw top vials (A′, B′, C′) were charged with a stir bar,
1a (0.0686 g, 0.050 mmol), 2a (0.1081 g, 1.000 mmol), and a portion
of the n-octane solution (0.6516 g delivered gravimetrically; 0.500 mmol
of 3, 0.300 mmol of n-undecane). The samples were stirred at 65 °C
for 8 h, and aliquots (0.050 mL) were periodically removed from vial
A′ (0.25, 0.75, 1.5, 3.0, 5.0, 8.0 h) and diluted with CF3C6H5 (0.5 mL).
The aliquots were filtered through a silica gel plug (0.5 × 1 cm) with
ethyl acetate/hexane (10 mL, 1:10 v/v) and analyzed by GC (0.0010
mL autoinjection). The reaction mixtures were stored at -30 °C
overnight. The light yellow upper organic phase was carefully separated
by syringe from the solid 1a. The solid was washed with cold n-octane
(0.8 mL, -30 °C), which was added to the organic phase. Vials B′
and C′ were charged again with 2a (0.1081 g, 1.000 mmol) and a
portion of the n-octane solution of 3 (0.6516 g delivered gravimetrically;
0.500 mmol of 3, 0.300 mmol of n-undecane). Aliquots from vial B′
were analyzed analogously to the first cycle. A third cycle was similarly
conducted, and aliquots from vial C′ were analyzed. Data: Figure 4.
Catalyst Recovery. A. A 4 mL screw-top vial was charged with a
stir bar, 1a (0.0653 g, 0.0476 mmol), 2a (0.1081 g, 1.000 mmol), 3
(0.0421 g, 0.500 mmol), and n-octane (0.8 mL). The sample was stirred
at 65 °C for 8 h, and Teflon shavings (ca. 0.3 g, average x/y/z dimension
ca. 2 mm) were added. The sample was kept at -30 °C overnight. The
light yellow liquid phase was carefully removed from the catalyst/Teflon
residue by a syringe fitted with a filter. A solution of PPh3 (0.0619 g,
0.236 mmol) in CF3C6H5 (3.5231 g) was prepared, and a portion was
added gravimetrically (0.1287 g, 0.00874 mmol of PPh3). The solid
residue was treated with a solution of PPh3 (0.0581 g, 0.222 mmol) in
CF3C6H5 (2.0 mL). Both samples were analyzed by inverse-gated 31P
NMR (pulse delay ) 12 s), and the signals were integrated against
that of PPh3.39 Data: see text.
B. A solution of PPh3 (0.0697 g, 0.266 mmol) in n-octane (4.2416
g) was prepared. A screw-capped NMR tube was then charged under
argon with a portion of this solution (0.5530 g delivered gravimetrically;
0.0346 mmol of PPh3), 1a (0.0813 g, 0.0592 mmol), and a capillary
filled with toluene-d8. The sealed tube was immersed in a water bath
(first run: 0 °C, then 20, 40, and 50 °C) and vigorously shaken for 15
min. It was transferred to a NMR probe and spun at the corresponding
temperature for an additional 10 min. Spinning was ceased, and the
suspension/emulsion was allowed to settle. After 20 min, an inverse-
gated 31P NMR spectrum was recorded (without spinning; pulse delay
) 12 s). The concentration of 1a was calculated from the integral ratio.39
Data: Figure 2.
Acknowledgment. We thank the Deutsche Forschungsge-
meinschaft (DFG; GL 300-3/1) and the Bundesministerium fu¨r
Bildung and Forschung (BMBF) for support, Dr. Philip L.
Osburn for helpful discussions, Dr. Ralf Meier for some
exploratory experiments, and Mr. Alexander Betz for technical
assistance.
JA029241S
(39) This technique is known to give accurate 31P NMR signal integrals. A
calibration run gave a PPh3/1a response factor of 0.999.
9
5872 J. AM. CHEM. SOC. VOL. 125, NO. 19, 2003