close

Вход

Забыли?

вход по аккаунту

?

Vibrational Spectroscopy in Supercritical Fluids From Analysis and Hydrogen Bonding to Polymers and Synthesis.

код для вставкиСкачать
REVIEWS
Vibrational Spectroscopy in Supercritical Fluids : From Analysis and Hydrogen
Bonding to Polymers and Synthesis**
Martyn Poliakoff," Steven M. Howdle, and Sergei G. Kazarian
r
Supercritical fluids are beginning to be
used widely in cnemistry. Applications
range from extraction and chromatography in analytical chemistry to solvents
for reaction chemistry and preparation
of new materials. Spectroscopic monitoring is important in much of supercritical chemistry. and vibrational spectroscopy is particularly useful in this
context because the vibrational spectrum o f a given molecule is usually quite
sensitive to the environment of that molecule. Thus. vibrational spectra are excellent probes of conditions within the
fluid. In this review, we describe a vari-
ety of techniques and cells for IR and
Raman spectroscopy in supercritical fluids and illustrate the breadth of applications in supercritical fluids. The examples include: the use of supercritical
Xe as a spectroscopically transparent
solvent for chemistry and for supercritical fluid chromatography with FTIR
detection of analytes; Raman spectroscopy as a monitor for gases dissolved in supercritical CO,; the effect of
solvent density on hydrogen bonding in
supercritical fluids and the formation of
reverse micelles; IR as a monitor for the
supercritical impregnation/extraction of
1. Introduction
Supercritical fluids are currently the subject of major research
projects in both academic and industrial chemistry."] The fluids
have already had a significant impact on traditional analytical
chemistrylz] and may well transform wide areas of chemistry.
The systematic study of supercritical fluids is therefore a matter
of both urgency and interest.
In its broadest sense, the term supercritical fluid can be applied to any gas that is compressed to a pressure greater than its
critical pressure, p , , and heated to a temperature above its critical temperature, T, (Table 1 ) . In practice, the definition is generally restricted either to gases close to their critical temperatures or. if the temperature is farBbove T,, to those gases
compressed to densities close to their critical densities, p,. The
behavior of such gases is so Par from ideal that supercritical
fluids have been a source of continuing scientific interest since
I*]Prof.
M . Poli;ikoft Dr. S. M. Howdle
Dqxirtiiienl 01' Chemistry. University of Nottingham
Kuttinghklm. NC7 2RD ( U K )
Tclcfm I n ( code (115) 951 3563
e-mail: M:II tyn. Poliakofffo Nottingham.ac.uk
+
1-1
Presmt addrcs\
Dr. S. G . KaLurian
Georgia Institute of Technology
Atlantii. G A 30332-0100 (USA)
Thioughout thia review' the abbreviation sc is used to indicate the supercritical
state. for chaniple scC0,.
polymers and the reactions of organometallic compounds impregnated into
polymers; reactions of organometallic
compounds in supercritical fluids; and
finally, the use of miniature flow reactors for laboratory-scale preparative
chemistry. Overall, our aim is to provide
a starting point from which individual
readers can judge whether such measurements might usefully be applied to
their own particular problems.
Keywords: analytical chemistry
IR
spectroscopy . photochemistry - Raman
spectroscopy . supercritical fluids
the middle of the nineteenth century. In recent years, this interest has been significantly increased by the possibility of using
supercritical fluids for a wide range of extraction processes and
chemical reactions, applications which exploit the almost counter-intuitive power of supercritical fluids to dissolve solid substances.[']The result has been a general shift in focus of research
Fable 1. Critical data for selected substances [a]
Substance
T,iK
p,:MPa
p.:gmL-'
Ar
CH,
Kr
C,H,
Xe
CiFh
CHF,
CCIE;
COz
CLHh
N2O [bl
SF,
H,S
C2H,(NH,),
H,O
150.9
190.5
209.4
282.3
289.7
293.0
299.3
302.0
304.1
305.4
309.6
318.7
373.2
593
647.1
4.898
4.604
5.50
5.041
5.84
3.06
4.858
3.870
7.375
4.884
7.255
3.77
8.94
6.27
22.06
0.536
0.162
0.919
0.214
1.110
0 622
0.528
0.579
0 468
0.203
0 452
0.735
0.346
0.29
0 322
[a] For critical data of other substances, see the excellent compilation by Reid et a1
[112]. [b] Safety warning: N,O has similar critical parameters and rather better
solvent properties than scC0, but there have been reports [I131 of potentially
disastrous explosions when scN,O has been used with even small amounts of organic compounds. Therefore. scN,O should only be used with extreme caution.
REVIEWS
from pure fluids to the behavior of more complex supercritical
solutions. Traditionally, most investigations of supercritical fluids have involved macroscopic phenomena (phase changes,
thermodynamics, etc) but the complexity of these mixtures has
led to the need for more detailed information on a molecular
level. It is the purpose of this review to show how vibrational
spectroscopy, both infrared and Raman, is now providing
molecular information relevant to a wide range of processes in
supercritical fluids.
The application of vibrational spectroscopy to the investigation of supercritical fluids is relatively new. In this review, we
outline the scope of the technique and highlight how it has been
applied to problems of chemical interest. We also discuss the
main technical problems and indicate how they have been overcome. Mostly, we consider fluids whose critical temperatures lie
in the range close to room temperature (i.e. from scC,H, with
T, = 9°C to scSF, with T, = 46'C) since these have had the
widest impact on both synthetic and analytical chemistry.", 21
scH,O is only briefly discussed, not because scH,O is less interesting (it has considerable potential for the oxidation of toxic
M. Poliakoff et al.
materials and environmental contaminant^[^]) but because the
high temperature and pressure ( T , = 374 "C and p , = 218 atm)
render most of the technical problems quite different. Our review complements Buback's excellent summary of near-IR spectroscopy in supercritical fluids,[4] and the reader's attention is
drawn to a number of recent multiauthored books[', 51 which,
together, constitute an excellent introduction to the current
state of supercritical science and its applications to chemistry.
In Section 2, we describe a number of designs of high-pressure
cells for I R and Raman spectroscopy. Then, in Section 3, we
describe the use of scXe as a solvent for both spectroscopy
and chromatography and compare its properties with those
of scC0,. In Section 4, we outline the role of Raman spectroscopy and show how it can be applied to the study of
supercritical fluid/gas mixtures. Sections 5-7 give an illustrative range of applications to hydrogen bonding (9, to
polymers (6), and to organometallic chemistry (7). Finally,
in Section 8, we show how IR spectroscopy is playing an important role in the development of laboratory-scale flowreactors.
Martyn Poliakoff was born in 1947 in London
[ U K ) , studied chemistry at Cambridge University
( B A , 1969) and received his P1i.D.for work on the
matrix isolation of large molecules under the supervision of J. J. Turner, with whom he has subsequently
had a long and productive scientific collaboration.
In 1972 he moved to the University of Newcastle
upon Tyne and in 1979 to the University of Nottingham, where he is currently Professor of Chemistry
and holder of an EPSRC Fellowship in Clean TechM. Poliakoft
S . M. Howdle
S. Kazarian
nology. For the past eight years his research has
,focussed increasingly on the chemical and analytical
applications of supercriticalJ7uid.s. He has been awarded the Mrldola Medal (1976) and Tilden Medal (1990) o f t h e Royal
Society of Chemistry and has helda Nuffield Foundation University Fellowship (198819) anda Royal Society Leverhulme Trust
Senior Fellowship (199314) . He has longstanding scientific connections with Germany; Stipendiat of the Alexander von
Humboldt StiJtung (Lasecforschung in Garching) , 1978 and 1979, and,fruitful collaborations with the Mas-Planck-Institut fur
Strahlenchemie, Miilheim, and the Universitji of Wiirzburg.
Steve Howdie was born in 1964 in Rotherham, Yorkshire. He studied chemistry at the Universiiy of Munchester (1986) and then
moved to Nottingham to study for his Ph.D. (1989) under the supervision of M . Poliakoff: For the next two years, he extended
his Ph.D. work on chemical reactions in supercriticalfluids, as BP Venture Research Fellow at Nottingham. In 1991, he M'US
awarded a Royal Society Universitj>Research Fellowship, which he has held at Nottingham. Over the past three years, he has
investigated the supercritical impregnation of polymers porous solids and his interests have broadened to encompass the
preparation and characterization ofbiocompatible coatings. Recently, he and M . Poliakoff have built up a strong and extremely
fruitful collaboration with research groups in Russia.
Sergei Kazarian born in 1958 in East Kazakhstan ( U S S R ) , studied physics at the Novosibirsk State and Yerevan State
Universities. After completing his diploma in phj.sics (1980) he moved into phjjsical chemistry and received his doctorate in 1987
under the supervision of B. Z
! Lokshin at the Nesnieyanov Institute of Organoelenient Compounds [Moscow) f o r the work on
the interactions of transition metal complexes with proton donors in liquefied noble gases. I n 1988, he moved to the Institute of
Spectroscopy in Troitzk, where he continued this work. He spent six months at the Ma.x-Planck-Institutfur Strahlenchemie,
Miilheim a d f r o m 1991 - 1994 he worked as Visiting Researcher in M . Poliakqfls group ut the University of Nottingham ( U K ) .
Currently, he is a Research Associate with C . A . Eckert at Georgia Institute of' Technology, Atlanta ( U S A ) . His research
interests include the spectroscopic study of intermolecular interactions in liquid xenon, supercritical,fluids, and polymers.
1276
A n g m . Clwm. Int. Ed. EngI 1995, 34, 1275-1295
Supercritical Fluids
2. The Design of High-pressure Cells
2. I . General Considerations
The design of the high-pressure cell needed for a particular IR
experiment clearly depends on the nature of that experiment.
Important points include the maximum operating pressure and
temperature. the optical pathlength required, the particular
range of the IR spectrum, which will restrict the choice of window material and, finally, whether the contents of the cell need
to be stirred. There is also a number of more subjective questions: Are the measurements to be strictly quantitative? Is it
likely the cell will have to be dismantled frequently (for example,
to remove reaction products)?
Although the pressures involved in experiments in supercritical fluids (usually less than 300 atm) are sometimes considered
to be modest by those who specialize in “high-pressure” chemistry, the gaslike nature of the fluids impose quite severe constraints on the design of the cells. As far as we are aware, there
are few, if any, cells which are commercially available. However,
a considerable variety of designs for high-pressure cells for use
with supercritical fluids have already been published[61including some elegant designs in Buback’s review.14]All of these cells
share a number of general features. Clearly, the cell has to be
safe for high-pressure operation, there has to be at least one IR
transparent window, although most designs accommodate two,
and at least one port for filling/venting the fluid.
In general. the IR transparent windows are mechanically the
weakest part of any cell and sealing these windows to the cell
body is usually the most difficult engineering task. Sherman and
Stadtmuller have provided a very detailed discussion of the design of high-pressure cells, although much of their bookr7]is
devoted to pressures greatly in excess of those required for supercritical fluids, and Whyman has reviewed the design of highpressure IR cells for conventional solvents.[81Both the book and
the review discuss, in some detail, the safety margins needed for
the IR windows. Three of the key parameters are the ratio of the
total area of window to the unsupported area (i.e. the area
through which the IR light passes), the thickness of the window,
and the rupture modulus of the window material. By applying
the appropriate formulae, one finds that surprisingly thin windows, only 2 or 3 mm thick, are required for miniature cells with
very small unsupported areas, such as the chromatography cell
shown in Figure Ib. All modern Fourier transform (FT) IR
instruments have a circular IR beam profile so it is no longer
necessary to use rectangular windows,r91which improved the
performance of high-pressure cells in older dispersive IR spectrometers. Even so, it should be borne in mind that most commercial suppliers have a range of standard diameters for IR
windows and apparently minor changes in design can alter the
cost of a cell disproportionately, depending on whether the windows are standard or non-standard sizes.
Sherman and Stadtmuller have described a number of approaches to the design of the pressure seal between optical windows and the cell body,[’] two of which are most appropriate for
supercritical fluids. Both approaches work well but each has
problems. The more straightforward approach, the “opposed
force” method, involves sealing the cell body to the inner face of
the window (i.e the face which is in contact with the fluid) with
Aii~cii~
C 7. i c i i i . I i i r
Ed. En,$ 1995, 34, 1275-1295
REVIEWS
a gasket, O-ring, or similar. The main drawback is that windows
often break during assembly; most window materials are brittle,
and mechanical pressure must be applied to the window to make
the seal. Supercritical fluids can cause swelling in some polymers
and, in smaller cells, the swelling of polymer O-rings can be
sufficient to shatter the windows. The alternative approach, the
“Poulter” method, has been pioneered for liquefied gases by
Bulanin in Russia.[1o1This approach involves sealing a window
on its planar outer surface; the internal pressure of the fluid then
forces the window into the seal, and leaks are less frequent.
However, it is often difficult with this design to avoid crevices
and dead-volumes which can be more easily penetrated by supercritical fluids than by conventional liquids. Such crevices
preclude quantitative measurements and, in evacuable cells, additional precautions are needed to prevent atmospheric pressure
pushing the windows inwards away from their seals. We now
describe a selection of cells that illustrate these general design
principles.
2.2. Selected Cell Designs for IR Spectroscopy
The cell shown in Figure l a was developed by Johnston and
Kim for investigating solvatochromism and hydrogen bonding“’] (see Section 5 ) . The pathlength is long (70 mm) because
the cell is intended for use with dilute solutes in fluids (e.g. scSF,
or scC0,) which d o not themselves absorb in the region of
interest. Since the studies were to be quantitative, the overriding
requirement was to eliminate dead volumes where undissolved
solute might accumulate. Thus, the windows are sealed on their
inside edges, and a mechanical stirrer is provided to ensure that
the contents of the cell are homogeneous. An absence of dead
volume is also a crucial requirement for the cell shown in Figure Ib. This cell, developed by Jenkins et al. is intended for
FTIR identification of analytes separated by capillary supercritical fluid chromatography (cSFC,[”] see Section 3). The volumes of fluid involved in such a separation are very small, because the capillary is only 50-100 pm in diameter, so the
internal volume of the cell has to be extremely small to preserve
the separation of the analytes. At the same time, the optical
70 rnm
10 rnrn
Fig I. Two high-pressure IR cells for use with supercritical fluids. hoth with windows sealed by the “opposed force” method. a) Schematic cross-section of a stainless steel cell designed by Johnston and Kim for studying solvatochromism and
hydrogen bonding [ I l l . P = fill port; PTFE = polymer seals; CaF, = calcium fluoride windows. The cell has two electric cartridge heaters (not shown) embedded in
the walls for raising the temperature of the cell. b) A brass cell for cSFC-FTIR
measurements [12]. P = ports connected to a silica capillary tube (50 pm i.d.);
F = graphite ferrule for sealing the capillary to the cell body; O-ring = Viton seals;
ZnSe = zinc selenide windows. Note that the two parts of the Figure are drawn to
different scales. The cell a) has an internal volume of about 5 mL, while cell b) has
a volume of only 500 nL.
1217
M. Poliakoff et al.
REVIEWS
pathlength must be as long as possible to give increased sensitivity and good limits of detectability. The design is therefore a
carefully engineered compromise between these conflicting requirements: Although the internal volume of the cSFC cell in
Figure l b is 500 nL,
times that of Johnston’s cell,[”] the
pathlength of the cell is 4.4 mm, that is only 1/16th of that of
Johnston’s. With appropriate beam condensing lenses on the
cSFC cell,[’21 the optical throughputs of the two cells are remarkably similar.
Figure 2 illustrates a completely different approach adopted
by Smith and co-workers for the study of reverse micelles in
supercritical and near-critical solution.[’ 31 In this case, the optical pathlength needed to be short because the mixtures under
ZnS
.....
22
V
Fig. 2. Variable pathlength IR reflectance cell devised by Smith and co-workers. P = fill ports;
IR = beam of FTIR interferometer; M = external
mirrors; MI = internal mirror, which can he moved
up and down to vary the pathlength by rotating the
handle of the modified high-pressure valve V ;
ZnS = zinc sulfide window; W = observation window; from ref. [13].
study were highly IR absorbing but, at the same time, the vessel
itself had to be relatively large so that the components could be
properly mixed. The design satisfies these requirements in that
the spectra are measured in reflectance rather than in transmission, with a single IR-transparent window and a highly polished
steel mirror, which can be moved to adjust the optical pathlength. Although this design worked well for their particular
experiments, quantitative measurements are quite difficult to
make in such a cell because one has to correct for the effect of
the light which is reflected from the outside face of the window
without passing through the solution.
The three cells described so far were designed, at least initially, for very specific experiments. A much simpler and more
general design is shown in Figure 3. This cell, developed at
Nottingham, is specifically designed to be easy to dismantle and
reassemble.[’41 The windows are therefore sealed on their outer
edges (one into the body of the cell and the other into the
removable holder), and the main seal is made between metal
surfaces and a PTFE spacer. Unlike the Smith cell (see Fig. 2),
the pathlength cannot be varied during an experiment, but a
selection of different pathlengths is available merely by using a
number of different holders, each with the window held a t a
different depth. The simplicity of the cell means that it can easily
be adapted for experiments as varied as impregnation of polymers (Section 6) o r photochemistry of organometallic compounds (Section 7). The cell is however too small to allow effective stirring, although it can be used as a component of flow
systems (see Section 8).
Any of these cells can be heated and varying the temperature
of the fluid is frequently a key feature in supercritical experiments. It should, however, be remembered that even quite modest rises in temperature can cause substantial rises in pressure. In
general, it is more difficult to cool IR cells than to heat them.
The Russian cell designs, described in some detail by Bulanin[lol
have usually included cooling, largely because the cells were
primarily intended for spectroscopy in liquid solution a t cryogenic temperatures. Figure 4 shows a relatively simple miniature
Fig. 4. External view (a) and cross-section (b) of a miniature high-pressure IR cell.
made of copper and capable of being cooled to cryogenic temperatures.
C. F. = mounting stud for cold finger ofcryogeniccooler; In = indium metal which
seals the CaF, window to the Cu cell body by the “Poulter Principle”; Pb = main
lead seal between the two halves of the cell; and 1 = gas fill ports.
high-pressure cell,[’51 small enough to mount directly on a
closed cycle-cryogenic cooler, and suitable for use in the temperature range 30-300 K. This design will allow fluids with a critical temperature T, well below room temperature to be studied
close to their critical points as well as at ambient temperatures.
2.3. Cells for Raman Spectroscopy
Fig. 3. Cross-section (a) and external view (b) of a stainless steel general purpose
high-pressure 1R cell. P = fill ports, threaded for standard 1/16th inch 0.d. pipe
fittings; E = epoxy resin, which is used to seal the CaF, windows (15 mni diameter.
10 mm thickness) by the “Poulter principle”: PTFE = principal seal between the
two parts of the cell, which is made metal +metal. For further details, see ref. [14]
1278
A number of 1R cells have been based on high-pressure T- o r
X-pieces,[16] adapted so that the ports can take I R windows
rather than pipes. This design approach has also been taken in
the construction of cells for Raman spectroscopy, where the
optical and geometric requirements are somewhat different
from those of IR. Briefly, Raman experiments in supercritical
fluids usually involve excitation of the sample by a focussed and
relatively intense laser beam; the scattered light is then detected
in a direction perpendicular to the laser beam. Raman scattering
is a relatively weak effect and so this scattered light has to be
Angeu Clien7 In1 Ed. Engl. 1995, 34, 1275-1295
Supercritical Fluids
collected effjcicntly which is difficult to achieve in practice.
Thus, the number of published Raman studies on supercritical
systems is much smaller than that for the corresponding IR
studies. Ben-Amotz and co-workers have designed an elegant
cell with a single window;[”] the scattered light is collected at
180‘ to the incident laser beam. Although similar in concept to
Smith’s design for I R cells (see Fig. 2 ) , the Raman cell does not
require a mirror since Raman scattering occurs in all directions.
A number of groups have used optic fibers to introduce the laser
light and this considerably simplifies the Raman experiment.
Figures Sa and b show two such designs; a) is a cell intended for
Fig 5 . Three high-pressure cells for Raman spectroscopy of supercritical fluids.
a) Cell for studying reactions in scH,O. The laser beam enters the cell through an
optic fiber, and A parallel fiber is used t o collect the scattered light; from ref. [lS]
b) A cell, based on high-pressure capillary tubing, C, for Ranian measurements with
an Ar ion (visible wavelength) laser. The cell is mounted directly between the entrance slits S of the spectrometer. F = optic fibers for the input of laser light;
T = thermocouple. scC0, = ports for the supercritical fluid (for more complete
diagrams, see ref [19]). c) Cell for FT Raman, based on a capillary, C. similar to
that in Fig. 5b. The opaque polyimide coating of the capillary, (thick black bar) IS
removed at the point where the near-IR laser irradiates the cell (similarly, the
coating IS removed i i i cell b). The capillary is mounted directly in the holder supplied
with the Perkin-Elmer System 2000 FT Raman instrument, which has a silvered
reflector designed to optimize collection of the scattered light. (Note the different
hcales of the drawings of the three cells.)
experiments in scH,O, fibers are used both for laser excitation
and collection of the scattered light[’*] and b) is a sub-miniature
capillary
devised by Howdle et al. for studying gases
dissolved in scC0,. Raman spectroscopy has distinct advantages over I R spectroscopy for experiments in scH,O, largely
because the fluid is highly absorbing over much of the range of
the I R spectrum and so corrosive that most IR window materials are rapidly destroyed. The ultimate in simplicity is achieved
by the cell for FT-Ramanr201shown in Figure 5c. This cell is
made merely by removing the polyimide coating from standard
silica capillary tubing of the type normally used for chromatography. The small diameter of these capillary cells ensures that
the scattered light can be collected extremely efficiently and,
furthermore, one does not need to worry about cleaning the cell
since it is essentially disposable!
Safety Note: All of the illustrations in Figures 1 to 5 are
purely schematic and should not be regarded as engineering
drawings for the construction of safe high-pressure cells. Those
wishing to build such equipment for themselves must carry out
their own safety evaluation.
3. Supercritical Xenon
Ever since the classic work by Andrews on the critical
point,’”] a huge proportion of the published research on super
A n g w . Chon. Inr Ed Eiigl. 1995, 34, 1275-1295
REVIEWS
critical fluids has involved SCCO,, most recently because of its
potential as an environmentally acceptable substitute for organic solvents in extraction and in synthetic chemistry.[’~21
From
the spectroscopic standpoint, scC0, has the distinct disadvantage that a number of key regions in the IR spectrum are completely obscured by absorptions of scC0, itself. This problem
has led to considerable interest in liquefied and supercritical
noble gases[1oh1which constitute a unique set of solvents for
spectroscopy, totally transparent throughout the spectrum from
vacuum-UV to far-IR.
The first systematic IR studies in liquid noble gas solution
were initiated in Leningrad by Bulanin and co-workers over 20
years ago.[loalTheir primary interests have centered on cryogenic noble gas solutions, an area which they have recently
reviewed in some detail[’0b.221but, as mentioned in Section 2,
some of their cells had a sufficiently high-pressure rating
(i.e, > 60 atm) to be used with supercritical xenon scXe. Maier
et al. also developed the use of liquid noble gases in the USA but
his equipment could not be used for supercritical fluids because
it was not suitable for use at pressures above 20 atm.[gl
scXe is a fascinating material. Its critical temperature, 16.7 “C
(see Table l ) , is very close to ambient temperature, making scXe
a convenient medium for investigations of the spectroscopic
effects associated with the transition from the liquid to the
supercritical state. The critical density p c , 1.11 gmL-‘, is
greater than the density not only of water but also of many
common organic compounds, both liquid and solid. scXe is a
moderately good solvent for nonpolar compounds and, although there have been few quantitative comparisons, the solubility in scXe is somewhat higher than in scCO,, in at least some
cases (e.g. polycyclic aromatic compounds) .cZ3, 241 In general,
the solvent properties of scXe are comparable to but not quite
as good as those of hexane or other alkanes. scXe can therefore
be used as a solvent for chemical reactions and for chromatographic separations. F o r a given compound, solubility is nearly
always greater in scXe near room temperature than in liquid Xe
(LXe) at cryogenic temperatures; this increase in solubility is
probably due more to the higher temperature of the scXe rather
than to some inherent solvent property of the fluid itself.
Whatever its cause, the effect is substantial. For example,
[(q6-C,H,)Cr(CO),] is virtually insoluble in LXe even at
- 30 “C, but usable concentrations can be obtained[251in solution in scXe at 25 “C,even in the presence of “antisolvents” such
as H, or N,. As with other supercritical fluids, solubility of a
compound in scXe at a particular temperature increases with
increasing pressure and hence density. This property is a key
but it can
factor in the use of scXe for
complicate some studies such as the investigation of hydrogen
bonding (see Section 5).[16]
Liquid Xe (LXe) has been extremely successful as a solvent
for studying weak intermolecular interactions,[”” for example
for proton donors,[221 or for stabilizing highly labile
organometallic compounds.[261 This success largely derives
from the cryogenic temperatures of LXe, a property which
clearly cannot be matched by scXe, although scXe still retains
the chemical inertness of LXe. Nevertheless, IR spectra of stable
solutes reveal few differences between the properties of LXe and
scXe. Some I R bands, for example, v ( C - 0 ) vibrations of transition metal carbonyl compounds are significantly broader in
1279
REVIEWS
M. Poliakoff et al.
scXe than in cryogenic LXe, possibly a consequence of the lower
viscosity of scXe. On the other hand, inherently broader
bands,[*’] for example the v(H -H) band of [W(CO),(v2-H,)],
can have identical FWHM (full width half maximum) values in
scXe at 25 “C and LXe at -90 “C. The transition through the
critical point from LXe + scXe has almost no effect on some
vibrational properties. For example, Moustakas and Weitz
that the rate of vibrational relaxation of HCI
(v = 1 --+ v = 0) in Xe is unchanged on heating at constant density from 289 K (liquid phase) to 291 K (supercritical), possibly
because bimolecular encounters HCl + HCI are not a major
relaxation pathway.
In general, as shown in Figure 6, the IR bands of a particular
solute in scXe are slightly narrower than the bands of the same
fluid, and hence its density, is then gradually increased under
computer control, and the various components of the mixture
elute from the end of the column sequentially, with the most
soluble compound eluting first. Thus, unlike GC, the whole
separation is carried out isothermally and cSFC is particularly
suited to separating thermally labile compounds of high molecular weight (e.g. polygly~erides).[~’l
One of the attractions of
cSFC over LC is that one can use most of the detectors normally
used for G C (i.e. FTIR, flame ionization, e t ~ ) . [ 321
~ ~Such
+ detectors are not easily used with LC because signals from the
liquid mobile phase swamp those due to the analytes. cSFCFTIR, therefore, offers excellent opportunities not only to separate but also to fingerprint the components of a complex mixture and Novotny and co-workers were the first to suggest that
the IR transparency of scXe would make it preferable to scC0,
as a mobile phase for c S F C - F T I R . ~ ~ ~ ]
The general requirements of a high-pressure IR cell for cSFCFTIR were discussed in Section 2. The differences between the
cell designs from different groups have lain in the volume and
beam optics rather than in the overall concept but it is these
differences which have determined the sensitivity and detection
limits of the various cells.[12~ 341 The more sensitive cells can
be used to record spectra “on the fly” so that spectra of all of the
components in a mixture can be obtained in a single run.
Figure 7 shows a chromatogram for the separation of a mixture of polycyclic aromatic compounds (PAH) and the FTIR
2 3 3
i
f”
-
21 70
I
2110
5/ cm-‘
Fig. 6. Comparison of the IR bands of [Cp’Ir(CO)(H),] (Cp* = $-C,Me,) dissolved in different
supercritical fluids, all of which are doped with H,:
a) Xe. b) C 0 2 ,c) C,H,. and d) CHF, In each case,
[Cp*Ir(CO)(H),] was generated in situ by irradiation
of [Cp*Ir(CO),] with UV light in the presence ofH,.
The IR band illustrated is in the v(1r-H) region and
corresponds to the unresolved a‘ and a” vibrations of
the H-Ir-H moiety; from ref. [29].
solute dissolved in other supercritical fluids at the same temperatureJ2’] As might be expected the width of the bands is dependent on the polarity of the fluid; the broadest band in Figure 6
is observed in the most polar fluid CHF,, although it is possible
that some form of hydrogen bonding could be occurring in this
case. The overriding advantages of scXe remain its inertness and
its transparency (lack of absorption bands) which can transform
IR spectroscopy into a surprisingly sensitive and versatile tool
for structural chemistry. This transparency of scXe has also
been a key factor in its use as a mobile phase for supercritical
fluid chromatography (SFC).
SFC is a relatively specialized chromatographic technique
that shares features of both gas and liquid chromatography (GC
and LC). The techniques of SFC have been described in detail
elsewhere[301but in principle they are quite simple. In its capillary form (cSFC), separation occurs on a microbore silica
column suitably coated on the inside surface with a stationary
phase similar to that used for capillary GC. The supercritical
fluid mobile phase flows down the column and then through an
extremely narrow bore restrictor which throttles the flow of the
fluid sufficiently to maintain a high pressure in the column. To
carry out a separation, the fluid is made to flow through the
column at a relatively low pressure and the mixture of analytes
is injected onto the column so that the components partition
between the mobile and stationary phases. The pressure of the
1280
--
0
10 tlrnin-
40
90
90
platrn-
310
I
4000
2000
c--.i. / cm-’
700
Fig. 7. cSfC-FTIR separation of a test mixture of eight polycyclic aromatic compounds (PAH) (IOmgmL-’ in CH,CI,) using scXe as a mobile phase. a) Chromatogram. reconstructed from the total IR absorption spectrum; the first peak is
due to the solvent CH,CI,, and the other peaks are due to the different PAH
compounds which elute from the column in order of increasing molecular weight.
b) IR spectra corresponding respectively to peak 1 and to the leading (2a) and
trailing edges (2b) of peak 2 of the chromatogram; the spectra are readily identifiable as those of chrysene (1). 9,10-diphenylanthracene (2a), and perylene (2b).
Spectra were recorded in 15 s using the cell illustrated in Fig. 1 b; from ref. [12].
spectra corresponding to some of the peaks.[121In this case, the
most characteristic bands of the different PAH compounds lie
below 800 cm-’, a region which is obscured in scC0, by absorptions of the CO, itself. On the other hand, the absorptions
of CO, mask relatively limited regions of the spectrum (38003500,2250-2100, and < 800 cm-’, see Fig. 9a) and, for many
compounds, spectra adequate for identification can be obtained
with scC0, as eluent. A quite detailed comparison has been
made between scXe and scC0, as mobile phases.’351The concluAngm.. Clietn. Int. Ed. Engl. 1995, 34. 127551295
Supercritical Fluids
REVIEWS
sion was that there are minor differences. Analytes do not necessarily elute in precisely the same order; the more polar analytes taking longer to elute in scXe than in scC0,. There are
modest shifts in the wavenumbers of IR bands of solutes as the
density of the fluid is increased but the effect is similar in magnitude for both fluids. These shifts present problems when analytes are to be identified by matching the observed spectra
against a library of spectra, which are unlikely to have been
recorded under supercritical conditions.[36.3 7 1 It seems, however, that particular classes of bands (e.g. v ( C - 0 ) bands of
ketones) all display comparable shifts and so it should be relatively simple to adapt the standard algorithms for library
searching to allow for these supercritical shifts.
The sensitivity of the cSFC-FTIR method is surprisingly high
and, outside the regions where scC0, absorbs, there is little
difference between scXe and scC0,. Using the cell illustrated in
Figure 1 b, the minimum detectable injected quantity of caffeine
in suC0, is only 95 pg. and the cell gives a linear response over
a wide range of concentrations[361 (Fig. 8). Over the past
decade, the price of Xe has been highly volatile but, in recent
years. the price has stabilized at a level which permits scXe to be
used in the laboratory whenever cSFC-FTIR data are required
over the entire region. cSFC columns have a very small volume
and Cook and co-workers quote a usage of about 3 mL scXe per
day[231(although they did not define how many separations
constituted a "day"). With larger scale usage, there is always the
possibility of recycling the Xe. for example, as is done by Bulanin and co-workers.f'ol Fortunately, cSFC-FTIR equipment can
easily be switched between scC0, and scXe so that the chromatographer can choose the more appropriate mobile phase for
a particular separation.
tection, there are some experiments where it is a distinct advantage, particularly those where the properties of pure fluids are to
be probed. In such cases, the IR absorption bands associated
with the fundamental vibrations of the molecules (CO,, C,H,,
etc) are so intense that extraordinarily short optical pathlengths
would be needed for usable measurements. This difference can
be seen clearly in Figure 9, where the Raman and FTIR spectra
of scC0, are compared.
1.6-
z
L
0
3600
2800
2000
1600
1200
800
G/cm-'
Fig. 9. Comparison o f the FTIR spectrum (bottom) and the FT Raman (top) spectrum of scC0,. The IR spectrum was recorded in the cSFC cell (Fig. lb) at 250 bar
pressure and ca. 32 C with a 4.4 mm optical pathlength, while the Raman spectrum
was recorded at ca. 100 bar pressure in the cell constructed from high-pressure
capillary tubing (Fig. 5c). The dotted lines indicate the two Fermi resonance bands
in the IR spectrum which coincide with the principal bands in the FT Raman
spectrum. The two weak features, labeled H in the Raman spectrum. are due to hot
bands. while the band labeled N, is due to atmospheric nitrogen i n the light path of
the F T Raman spectrometer.
4. Raman Spectroscopy in Supercritical Fluids
There have been a number of Raman investigations where
changes in either the contour or the position of Raman bands
have been measured as a function of fluid density and these
Raman scattering is a relatively weak effect compared to IR
changes have then been related to the intermolecular interacabsorption. Although this weakness can create problems in detions within fluids such as S C C O , , ' ~ ~ ]
SCC,H,,['~~
and recently, near-critical
b)
H2S.1391Figure 10 shows data obtained by Ben-Amotz et al. for C2H,
just below its critical point, a regime
+
where the fluid compressibility is al0.200ready anomalously high." 'I The data
are striking because they show that the
lo4
behavior of two vibrational modes of
0.0202'
the same molecule can be substantially
different under identical conditions.
A
Nevertheless, their behavior can be
modeled
reasonably well by using a
0.002'i
hard sphere model and the differences
between the modes is ascribed to the
fact that the v(C-H) vibration is more
04'
,E
,f +
polarizable than the v(C-C). Ben1900
1500
0
1
10 100 1000
;/cm-'
m l n g ---+
Amotz makes the interesting point
that these differences might be reflectFig. X. a ) Spectra illustrating the sensitivity of the cSFC-FTIR method with scC0, as the mobile phase and using
the 500 n L IR cell illustrated in Fig. 1 b. r(C-0) region of the spectrum recorded for injections of solutions
ed in a chemical reaction with two or
containing reqectively 580, 290, and 95 pg of caffeine. b) Beer's Law plot (IR peak absorbance vs. injected
more
possible products. If branching
concentration) for caffeine in scC0,. Note that. somewhat unusually, both axes have been split so as to accommoin such a reaction were to involve sigdate the very large range of concentration (95 pg-170 ng) on a single Figure; from ref. [36].
1
T
-
A n g i w C h i n . lnf. Ed. EngI. 1995. 34, 1275-1295
1281
REVIRNS
M. Poliakoff et al.
-2
t
-6
A; /cm-'
-1 0
0
5
-
10
15
p I mol L-'
Fig. 10. The shifts in frequency A; of two vibrational modes with increasing density
p of sub-critical C,H, a t 26 'C. The lines show the shifts computed with a hard fluid
model; from ref, [17]
spectrum which are obscured by 1R absorptions of scC0,
(Fig. 9 bottom) are essentially free of bands in the Raman spectrum (Fig. 9 top). The only exception is in the region, 14001200 cm-', where two bands, formally forbidden in IR spectroscopy, become allowed at high pressures and are observed as
relatively weak, pressure-dependent absorptions. (The presence
of these two bands can cause some difficulty in cSFC-FTIR with
scC0, as the mobile phase but the bands can be removed from
the spectra of the analytes by the appropriate computer subtraction of "blank" spectra of scC0, itself.[361)A detailed FTIR
study1431of these bands
has recently been published. The advantage of
mutual exclusion can be
seen in the Raman spectrum of naphthalene in
scC0, (Fig. 12), which
shows two bands in a region which would be ob500
800
scured in the IR specG/cm-'trum.['']
Fig. 12. Part of the Raman spectrum, vu and
the low sensitivity of
v g vibrations, o f a Saturated solution of ndphmost R~~~~ experithalene in scC0, (165 atm and 39°C) record-
,
nificantly different vibrational coordinates, changes in fluid
density might change the relative yields of the products. Density
effects have been observed in a number of reactions, most dramatically for the photodimerization of i s o p h o r ~ n e , [ but
~ ~ Ithe
changes in product distribution have been explained satisfactorily on the basis of solvation effects in bimolecular intermediates. Thus, Ben-Amotz's intriguing prediction remains to be
verified experimentally.
An alternative approach to studying interactions is to use
Raman spectroscopy to monitor bands of solute molecules dissolved within the fluid. The problem is one of sensitivity; relatively concentrated solutions must be used. Elegant Raman
spectroscopic work has been carried out by Akimoto and Kajimot0 in acetone and acetonitrile, both of which have quite high
solubility in a variety of fluids (scCO,, scCF,Cl, e t ~ ) . [ ~They
']
monitored the position of the v(C=O) and v(C=N) bands as a
function of fluid density and interpreted the shifts of these bands
in terms of the solute-fluid interaction potential. In general,
however, such studies are often easier to carry out by IR spectroscopy, as can be seen from the spectra in Figure 11, comparing the effects of scC,H, and scC,F, on the v, (bending vibration) band of CO,. These
spectra were recorded by
Yee et al. as part of an investigati~nr~
aimed
~ l at establishing whether the high solubility of fluorocarbons in
scC0, (see Section 5 ) might
be due to a strong
C-F . . CO, interaction.
However, the similarity of
r 0 . 5 bar
the spectra of CO, in
scC,H, and in scC,F,
b'
shows that, in this case at
least, there is no exceptional
C - F . . . CO, interaction.
Even a cursory glance at
Figure 9 reveals the conse699
669
639
quences of the centrosym-G/cm-'
metric structure of CO, and
Fig. 11. I R spectra showing the effect of
fluid pressure on the v,(O=C=O) bendthe mutual exclusion prining vibration of CO, (0.02 M) in a) C,H,
ciple of IR- and Raman-acand b) C,F, from 0.5 bar (pure CO, gas)
tive bands; regions of the
to 400 bar; from ref. [42].
t
1
1282
I
merits means that spec-
tra of solutes can only be
recorded in relatively
U
ed with 488 nm excitation (30 mW) and using
the cell shown in Fig. Sb. Note that this region of the IR spectrum would be totally obscured by intense absorptions of scC0, itself;
from ref. [19].
concentrated solutions.
A more detailed Raman
of naphthalene/scCO, was carried out by Zerda et al.
who deduced from the shifts in the wavenumbers of three different bands of naphthalene that there was a modest anisotropic
interaction between naphthalene and CO, . Their calculations
suggested that this interaction might be due to the quadrupolequadrupole CO,/C,,H,
interaction (UQQ= 19.1 kJmol-')
shown in 1. These conclusions have been supo=c=o
ported by recent MIND0/3 calculations
from Dean and c o - ~ o r k e r s . [ ~ ~ ]
A more practical application of Raman
spectroscopy is for monitoring reaction mix1
tures. Some Raman spectroscopic studies['8s461
have been reported for scH,O, where technical problems preclude the use of IR spectroscopy; apart from the
absence of IR windows which are not corroded rapidly by
scH,O, high temperatures can cause problems in FTIR, and
scH,O itself has extremely intense absorptions over much of the
mid-IR region. Masten et al. have followed the decomposition
of N,H, in near-critical and S C H , O , [ ~while,
~]
using the cell
shown in Figure 5a, Kolis and co-workers were able to monitor
the catalytic dehydrogenation of cyclohexene given in Equation
(a) (ca. 12% in scH,O; Fig. 13).['81
a
The complete miscibility of H, and N, with supercritical fluids can be exploited for a variety of new or more efficient reactions (see Section 7). Raman spectroscopy is proving to be the
most convenient method of detecting these gases;[1g,"1 the viAngen. Cliem. I n l .
Ed. Engl. 1995. 34, 1275-1295
REVIEWS
Supercritical Fluids
Raman studies generally involve observation of the spectrum of
a pure fluid o r very concentrated solution. Thus, the spectra will
be dominated by features from the bulk of the fluid. and the
effects of clustering may be difficult or impossible to observe.
Thus, it is not surprising that Raman studies such as that by
Ben-Amotz et al. d o not show evidence for clustering.“ ’I
Solute/solvent clustering is discussed further in Section 5 in
‘00
800
900
1000
20
60
100
;/cm-’t I min
the context of hydrogen bonding, but first we highlight a
Fig. 13 a ) Raman spectrum recorded during the catalytic reaction of cyclohexene
promising application of Raman spectroscopy to the study of
in acH,O. see Equation (a). The unkdbeled bands are due to unconverted cyclohexdissociation processes in supercritical fluids. One of the most
ene. those marked are due to cyclohexane and benzene. The intensity of the bands
fascinating features of supercritical fluids is that phenomena can
for cyclohcxane and benzene are recorded as a function of time in Fig. 13 b ; from
ref [18].
be studied as a function of solvent density, from gaslike to
liquidlike values. Homolytic dissociation of molecules and the
subsequent recombination of radicals (or atoms) is a process
bration of an isolated homonuclear diatomic does not give rise
which is likely to be heavily influenced by solvent density; at
to an IR absorption and, even though the vibrations may behigh densities, geminate recombination within the solvent cage
come IR active in solution or at high pressures, only very weak
should be significant but, at low densities, the process is highly
IR absorptions can be observed.[251The vibrational Raman
improbable. Ravi et al. have modeled the effects of solvent denbands“’] of both N, and H, show modest shifts and broadening
sity on the dissociation of Br, in noble gas solvents.[521
on going from the pure gas at high pressure to a mixture of the
Clearly, geminate recombination is an extremely rapid progas with scCOz at a similar pressure (Fig. 14 a). The broadening,
cess. Recently, Knopf and co-workers have begun experihowever, is considerably more dramatic and highly pressure-dem e n t ~ [aimed
~ ~ ~at] detecting vibrationally excited I, formed by
pendent for the rotational bands of H, in near- and supercritical
geminate recombination in scC0, on a picosecond timescale
CO, (see Fig. 14b); this broadening has been tentatively at[Eq. (b1.1
tributed to clustering of the solvent around the H, “ s o l ~ t e ” . ~ ~ ’ ~
-
I,(c
4145
4155
540
580
’
620
G1cm-lFig. 14. Raman spectra showing how the Q‘,” vibrational band of H, is broadened
and shifted to lower frequency in scC0, (a) (23 “Cand 175 atm; mole ratio H, :CO,,
1 :4.8)compared to the same band (b) of H, gas at the same total pressure (175 atm).
By contrast. there is almost no frequency shift in the S\” rotational band (c) but
there is a strong pressure dependence in width; 1 : pure H, gas (87 atm); H, gas
dissolved in scC0, at 60 atm (2) and 160 atm (3). The band marked with an * is a
laser plasma line: from ref. [19).
Solutejsolvent clustering[481(sometimes also referred to as
“Charisma”[481)in supercritical systems is still a subject of much
Near the critical point, local fluctuations of density
undoubtedly occur even in pure fluids and give rise to the wellknown effect of “critical opalescence”. This can cause such
strong scattering of light that the fluid is rendered opaque even
in the IR region. The scientific debate, however, centers on
whether clustering can occur specifically around solute molecules. Evidence for clustering comes largely from UV/Vis fluorescence measurements where fluorescence lifetimes are extremely sensitive to the concentration of the fluorescer;
unexpectedly short lifetimes close to the critical point are taken
to indicate a local increase in the concentration of solute. In
fluorescence studies, the behavior of solute molecules is monitored in dilute solution and, for precisely this reason, effects due
to clustering are relatively easy to detect.‘50,5 1 1 By contrast,
= 0)
- I1 I.
scco,
21‘
I,(G
>> I)
-+
12(G
= 0)
(bf
In principle, the requisite sensitivity can be achieved because
the dissociation is completely reversible and spectra can be accumulated over many shots at 200 Hz. Preliminary experiments in
cyclohexane have been successful, revealing the generation of
very highly excited I, (v < 52); work on scC0, solutions is in
progress. The dynamics of the geminate recombination itself
have been studied in scAr on much shorter ( i t . sub-picosecond)
timescales by Zewail and c o - w ~ r k e r susing
~ ~ ~ UV/Vis
~]
detection which has much higher inherent sensitivity than Raman
spectroscopy but which cannot provide as detailed information
as Raman spectroscopy on the vibrational distribution in condensed phases.
5. Hydrogen Bonding
5.1. The Need for IR Monitoring
The original impetus for studying hydrogen bonding in supercritical fluids was twofold; first, the possibility of establishing
whether equilibria involving hydrogen bonding were sensitive to
solvent density and, second, the fact that hydrogen bonding is
frequently a key factor in the action of so-called “modifiers”,
compounds added to supercritical fluids to enhance the solubility of particular
It is occasionally possible to use
UV/Vis spectroscopy for monitoring equilibria of compounds
with hydrogen bonds. This has obvious advantages because UV
spectrometers have high sensitivity and few, if any, of the common supercritical fluids absorb in the UV range. Thus, the absorption spectra of the two tautomers in equibrium, 2 and 3, are
significantly different.I5 In scC2H, (and in liquid higher alkanes), the azo tautomer, 2, predominates at all fluid densities
1283
M. Poliakoff et al.
REVIEWS
while, in scC0, , increased fluid density increases the concentration of hydrazone 3 relative to that of 2, to a point where the
concentrations of the two tautomers are almost equal. This effect is attributed to a quadrupolar interaction of 3 with CO,
(analogous to the interaction between naphthalene and CO, in
1). By contrast, 3 is favored at all densities in the more polar
fluid scCHF,, presumably because a hydrogen bond between
the solvent and solute (F,C-H.. . O=C), can stabilize 3 relative
to 2.
OH
H
2
3
On the other hand, significant hydrogen bonding has been
excluded by Tomasko et al. as a possible explanation of their
fluorescence experiment^[^ 'I with 4 and related compounds in
pure and modified scC0,. In general, however, hydrogen bonding cannot be detected by UV spectroscopy and vibrational
spectroscopy is needed.
5.2. Hydrogen Bonding and IR Spectroscopy
Unlike UV/vis spectroscopy, IR spectroscopy is widely applicable for studing hydrogen bonding. Thus, in compounds with
0 - H groups, hydrogen bonding causes the v(0-H) band, typically at 3600cm-', to broaden, increase in intensity, and to
shift down in wavenumber;[561 both qualitatively and semiquantitatively, the stronger the interaction the greater the increase in intensity and the greater the wavenumber shift.["]
Most studies of hydrogen bonding in supercritical fluids have
been carried out with alcohols and have involved measuring the
change in equilibrium constant as a function of temperature,
pressure, or density for a particular equilibrium [Eq. (c)].
RO-H
+B
e RO-H-B
[RO - H B]
-
Although the v(0-H) region of the spectrum contained no
solvent absorptions in scXe, the bands of the free and hydrogenbonded rotamers overlapped so that computer subtraction was
still required to disentangle the spectra. Furthermore, the solubility of the compound increased dramatically with increasing
density of scXe so that spectra required additional normalization. Nevertheless, their results reveal a striking change in partial molar volume associated with the rotamerization close to
the critical point, changes which are over two orders of magnitude larger than those observed in conventional solvents. This
anomaly was attributed to solvent/solute clustering.
Surprisingly, no such clustering effects were observed by
Saito and c o - ~ o r k e r s , [who
~ ~ ] investigated the tautomerism of
2,4-pentanedione in scC0, , which also involves intramolecular
hydrogen bonding [Eq. (e)]. The ratio of tautomers was mea-
0
0
O/H-. 0
sured based on the v(C=O) region of the spectrum, where scC0,
does not absorb significantly. Increasing density was found to
favor the keto (non-H-bonded) tautomer, although the enol
form remained predominant at all densities. This decrease in
hydrogen bonding (intra- or intermolecular) as the fluid density
increases appears to occur in nearly all of the systems studied
so far.
5.3. Hydrogen Bonding of Methanol
Kc = [RO-H][B]
However, the variation in intensity of the v(0-H) band of
hydrogen-bonded 0 - H groups makes it difficult to measure the
concentrations of both the alcohol [ROH] and that of the hydrogen-bonded complex [RO- H-B]. To circumvent this problem,
quantitative studies have often been made with the total concentration of alcohol constant (i.e. [ROH] + [RO-H-B] =
constant) whereupon it is only necessary to monitor the amount
of free alcohol [ROH] by way of the v(0-H) band. Unfortunately, scC0, has a strong IR combination band in precisely the
region where this v(0-H) band occurs (see Fig. 9 bottom), and
v(0-H) vibrations are poor Raman scatterers so that Raman
spectroscopy is not a viable alternative in this case. A number of
strategies have been tried to overcome this difficulty. One approach.[lbltried by Ikawa and Fujita, is to use scXe which has
1284
no IR absorptions there (see Section 3 ) . They studied 4-hydroxy-4-methyl-pentan-2-one, one rotamer of which has intramolecular hydrogen bonding [Eq. (d)] .
The most extensive work on hydrogen bonding in supercritical fluids has been carried out by the groups of Johnston and his
collaborators and of Smith, using the cells illustrated in Figures
l a and 2, respectively. Much of their attention has been focussed on CH,OH which is the most commonly used modifier
for scC0, ; indeed, chromatographers often use commercial
cylinders of CO, premixed with CH,0H.[591 Smith and coworkers[60,6 1 1 overcame the difficulty of observing v(0 -H) vibrations in scC0, by using deuterated CH,OD which has its
v(0-D) band in the region, about 2600 cm-'. unobscured by
absorptions of scC0, (Fig. 15). They were helped by the design
of their cell which allowed high concentrations of CH,OD to be
used, because the use of low concentrations of deuterated alcohols in metal apparatus is frequently plagued by D / H exchange
with residual traces of H,O. Some of their results are shown
Angrbi. Clwm. 1171. Ed. Eflgl. 1995, 34. 1275-1295
REVIRNS
Supercritical Fluids
I
3000
investigating the effect of solvent density on hydrogen bonding
in scC0,. The results were encouraging; increased solvent
density appeared to shift the equilibrium in favor of the unbonded species. This study was made with scSF, which, unlike scCO,, has no IR bands in the v(0-H) region so that undeuterated CH,OH could be used. The v(0-H) bands of both the free
and the hydrogen-bonded alcohol could also be monitored, but
the extinction coefficient of the band of hydrogen-bonded complex was difficult to calibrate. More importantly, the vapor
pressures of CH,OH and NEt, were sufficiently low that care
was needed to keep the CH,OH/NEt,/CO, mixture as a single
phase.
I
C-ti comb.
-
2800
2600
G/cm-’
Fig. 15. The FTlR spectrum of CH,OD in scCO,, illustrating the strategy by Smith
and co-workers of using deuterated alcohols to avoid the spectral regions obscured
by absorptions of scC0,. The spectrum was recorded in the cell shown in Fig. 2 at
40°C and 400 bar. The band marked “C-H comb.” is a composite absorption
resultng from scveral v(C - H ) combinations; from ref. [61].
schematically in Figure 16. The phase diagram shown there
indicates that it is possible for higher concentrations of CH,OD
to be dissolved in the single supercritical phase than in the gas
phase at the same temperature.L611Thus, supercritical mixtures
of scC0, and CH,OH can contain predominantly hydrogenbonded alcohol. Surprisingly, their experiments showed that
there was relatively little difference in aggregation between any
of the CH,(CH,),OH alcohols 0 < n < 9. Figure 16 illustrates
XJ J
O.I5
0.05
1
O.*
XM~OD-
Fig. 16. Phase diagram (pressure p vs. mole fraction x) for
CH,OD in CO, at 4 0 T , with
schematic representations of
the aggregation of CH,OD in
the vapor, liquid, and single
phase regions of the diagram.
a = critical point; b = vapor
phase; c = liquid phase; from
ref. [61].
two further points: First it is extremely important in such studies, indeed in any supercritical study, to know the precise phase
behavior of the system in question. Second the self-association
of CH,OH and other alcohols consists of a series of coupled
equilibria [Eq. (f)], which are complicated, if not impossible, to
separate for quantitative study.
Gupta et al. largely avoided this problem of self association of
CH,OH by using a considerable excess of the base Et,N.[621
They then tried to use the equilibrium in Equation (g) for
CH,OH t Et,N
An,q,ii C’heni In1
CH30-H-NEt3
Ed Engl 1995, 34, 1275-1295
(g)
5.4. Hydrogen Bonding of Perfluorinated Alcohols
Smith and co-workers were the first to use perfluoroalcohols
in the study of hydrogen bonding in supercritical systems.[421
These alcohols had been used in related studies in the non-supercritical range over a number of years prior to this,[22h1
particularly because they are much more acidic than their nonfluorinated analogues[631and therefore form stronger hydrogen bonds
and can be used in greater dilution in the supercritical solution.
In scCO,, fluoroalcohols have the added attraction that they
are considerably more soluble than the corresponding nonfluorinated compounds. Studies of the type illustrated in Figure 11,
suggest that there is no enhanced interaction between CO, and
C - F bonds. Smith and his co-workers therefore proposed that
the reason for the high solubility lay in the formation of weaker
hydrogen bonds by the fluoroalcohols than by the nonfluorinated alcohols.[421However, this suggestion does not explain why
other fluorocarbons that are not alcohols also show enhanced
solubility in scC0,. Whatever the reason, the phenomenon is
quite widespread and has considerable promise, as shown by
DeSimone et al., for the synthesis of f l u o r ~ p o l y m e rand
s ~ ~by
~~
Wai et al., for the extraction of metals with fluorinated chelating
agents in supercritical
Smith’s experiments on hydrogen bonding were performed
with unbranched p e r f l u o r o a l c ~ h o l s , [ which
~ ~ ~ retain the
propensity to aggregate and therefore the equilibria are just as
complicated as those of the unfluorinated compounds. By contrast, branched perfluoroalcohols such as perfluoro-rut-butyl
alcohol (PFTB), (CF,),COH, have a very small degree of selfeven at low temperature in cryogenic solvents.
This means that PFTB has quite a low boiling point and vapor
pressure (ca. 40 Torr at room temperature) relatively high in
comparison to the unbranched alcohols. In collaboration with
Gupta and Johnston, we have exploited the volatility of PFTB
to study its hydrogen bonding to (CH,),O, an even more
volatile base (vapor pressure at room temperature ca. 4 atm) .[661
With PFTB and (CH,),O in the cell shown in Figure la, it was
possible to obtain the required concentrations of acid and base
in the gas phase without any added fluid. We were therefore able
to study the effect of fluid density on this equilibrium between
the components without and with hydrogen bonds from zero up
to liquidlike densities. Again SCSF,was used as the fluid because
it has no significant IR absorptions above 2500 cm-’ even with
an optical pathlength of 7 cm. As pointed out by Inomata et
a1.,[671the key problem in such studies is that IR extinction
1285
M. Poliakoff et al.
REVIEWS
clustering.[661As in other studies, it was important to establish
whether the presence of PFTB and (CH,),O affected the critical
point of SF,. It was shown that T, was unaffected within experimental error by PFTB (ca. l m ~ and
) (CH,),O (ca. 3 0 m ~ )
using the relatively uncommon method of measuring the
velocity of sound in the fluid.[661The acoustic properties of a
fluid change very substantially in the region of the critical point
and, hence, even small changes in T, and p , can be detected.LZoa,681 The method promises to have quite widespread applications.
Recently, Ochel and S~hneider[~'I
have monitored the hydrogen bonding and solubility of polar organic acids and esters in
scCClF, . In an elegant application of near-IR spectroscopy,
they used the 2 x v(0-H) bands to quantify the amount of
hydrogen bonding and the 2 x v(C-H) bands, which are essentially unaffected by hydrogen bonding to measure the total
amount of solute in the fluid. Their conclusion was that the
more important factors for determining solubility were molecular size, polarity, and self-association.
coefficients of solutes can vary with fluid density and, furthermore, the effect is not necessarily linear[661nor is it the same for
different solutes. Thus, for the PFTB/(CH,),O/SF, system, a
considerable amount of calibration was required to allow for the
nonlinear variation of IR extinction coefficients with fluid density and to avoid spectroscopic artefacts. Spectra, such as those
in Figure 17, fully supported the results of earlier workers,
I
I
3700
3100
3400
c- V
/ cm"
5.5. Hydrogen Bonding and Protonation
of Organometallic Compounds
Fig. 17. I R absorption spectra in the v(0-H) region showing the effect of adding
SF, to a mixture of (CF,),COH and (CH,),O at 5 0 ' C Trace a was recorded after
addition of 560 psi ofSF, and trace h with 645 psi total pressure of SF,. The increase
in pressure causes a reduction in intensity of the broad band of the complex with
hydrogen bonds with a concomitant increase in the sharp band of the free
(CF,),COH. The spectra were recorded in the cell shown in Fig. la; from ref. [66].
Kazarian et al.r631made a qualitative study in scC,H, of
the interaction between (CF,),CHOH and [Cp*Ir(CO),]
(Cp* = C,Me,) [ E q . (h)]. The effect of fluid density was similar
to that for other systems; increased density disfavors hydrogen
bonding. Equilibrium (h) i s interesting, however, because it is
one of the rare cases where such an equilibrium has been monitored by way of the base rather than the acid, using the extremely intense v(C-0) bands of the metal compound.
namely that increased solvent density shifts the equilibrium in
favor of the free acid and base. It can be seen from these spectra
that, close to the critical pressure, even quite modest changes in
pressure can cause a significant shift in equilibrium. Gupta et al.
have outlined the thermodynamics behind this shift.[621Very
tritely, their explanation can be summarised as stating that solvation energies become increasingly important as the density of
the fluid increases, and that the solvation energy of the complex
with hydrogen bonds is less than the sum of the solvation energies of the free alcohol and free base. This behavior has been
successfully modeled and In K, decreases almost linearly with
increasing fluid density (Fig. 18 left). Perhaps the most interesting finding was the observation of nonlinear behavior of In K, of
the PFTB/(CH,),O/SF, system at temperatures close to critical
(Fig. 18 right). This deviation was attributed to enhanced solute/solute interaction, probably mediated by solute/solvent
I
t
"
A similar study["] in scXe of the interaction of [Cp*Ir(CO),]
with the stronger proton donor HCI produced an unexpected
result; the IR spectra showed only the known bands of the
protonated ion 5.['01 No IR absorptions which could be assigned to a counter anion were observed but C1- does not absorb IR radiation, and the bands of [HCI,]- would probably
have been much weaker than the v(C-0) bands
of 5, which themselves were very weak in this
experiment. Although the observation of ionic
species in a nonpolar fluid is unusual, Lokshin et
al. reported some years ago that protonation of
neutral organometallic compounds can lead to
5
ionic species in LXe solution at cryogenic temperatures[22b*
"1 and, presumably, such charged species are
present as ion pairs or small aggregates in both LXe and scXe.
Therefore, it is not entirely surprising that solutions of 5 in scXe
proved to be unstable, with the v(C -0) IR bands disappearing
within a few minutes of the HC1 being added, possibly because
of precipitation o r further reaction with HCI.
o&C&+
0
2
4
p / rnol L-'
6
1
8 1 0
0
2
(J
4 6
I rnol L-'
8
1
Fig. 1 8 . Plots for the (CF,),COH/(CH,),O system, showing the Variation of InK,
with density of SF, at eonstunt tpmperuturr. See Equation (c) for definition of K , .
Left: 60 C, [PFTB] 1 . 2 m ~[ .( C H , ) , 0 ] 3 0 m ~ ;right: 50'C.pointsmarked o [PFTB]
Z.OmM. [(CH,),O] 1 6 m and
~ thosemarkede.[PFTB] 1 . 5 m ~ . [ ( C H , ) , O ] 2 5 m ~In.
both parts of the Figure, the solid lines indicate the predictions of the modified
lattice fluid hydrogen bonding model; from ref. 1661.
1286
0
Angew Chrtn. I n t . Ed. Engl. 1995, 34, 1275-1295
Supercritical Fluids
RRIIEWS
5.6. Reverse Micelles in Supercritical Fluids
Very recent, and as yet unpublished, work by Eckert and
c o - w o r k e r ~ [has
~ ~ ~shown that phase transfer catalysts can
transport ions into scC0,. However, most of the research effort
to solubilize ionic compounds in supercritical fluids has involved surfactants and reverse micelles. Indeed, much of the IR
work of Smith’s group has been aimed directly or indirectly at
the study of reverse micelles in supercritical
Apart
from scCHF,, most fluids with a critical temperature T, close to
room temperature are nonpolar. This means that a vast range of
hydrophilic and ionic compounds cannot be dissolved o r processed in these fluids. Smith and co-workers have shown that
some of these compounds (e.g. cytochrome c) can be solubilized
by adding surfactants which allow reverse micelles to form in the
supercritical
Such micelles have been observed in
scC,H,. in near-critical and scC,H,, and in scXe[74.751but
crucially no suitable surfactant has yet been found which can
form reverse micelles in scC0,. The search for such a surfactant
has been intense because, once identified, it will open up whole
new areas of chemistry in scC0,. Unfortunately, as Smith and
co-workers wryly
“the list of conventional (ionic)
surfactants that are insoluble in CO, is indeed impressive but
not surprising because these (surfactants) have been developed
for ‘oil’ phases of much different character”. Therefore, FTIR
studies of hydrogen bonding have been an important tool in the
search for non-ionic surfactants for scC0,. The role of IR spectroscopy has been twofold, to estimate the degree of aggregation
of the surfactants and to establish whether water has indeed
been incorporated into the aggregates. Smith and co-worke r ~have
[ used
~ ~the~ approach
~
already outlined above, namely
their variable path IR cell and deuterated alcohols, to compare
hydrogen bonding of polyethylene glycol dodecyl ethers in
scC,H, (where reverse micelles have already been observed) and
in scC0, (where they have not). Their results indicate that, in
the absence of added water, the solubility of these surfactants is
higher in scCOz than in scC,H, but that there is less aggregation
0.5
I
A
0i
5400
5200
5000
4800
1800
1600
1400
1200
.i.t cm-‘
Fig. 19. FTIR spectra o f a mixture containing 2.5 % surfactant, 6.0% pentanol and
H,O in scC02 at 60 C and 5000 psi pressure. The absorption hands of H,O are the
weak combination band (arrowed) and the stronger band, ca. 1600 cm-I. shown
separately with an expanded wavenumber scale. This band is broader in the presence of surfactant than in pure scC0,. as indicated by the shading. Other bands are
surfactant: from ref. [76].
labeled as follows: $: xC0,; A : pentanol;
+,
of the molecules in scC0, than in scC,H,. Ciearly, a tendency
to aggregate is an important factor in the formation of reverse
micelles and so the difference between scC0, and scC,H, is
consistent with the lack of micelles in scC0,.
In a similar study, McFann et al. investigated the non-ionic
s~rfactant,[~
pentaethylene
~l
glycol n-octyl ether, dissolved in
scCO,/n-pentanol/H,O. FTIR spectra recorded in the presence
of the surfactant showed an unresolved broadening to high
wavenumber of the band at approximately 1600cm-‘, compared to spectra recorded in the absence of surfactant (Fig. 19).
This broadening is attributed to water within surfactant aggregates but the level of water present in the surfactant/scCO,
phase did not appear to be much higher than that found in the
absence of the surfactant.
6. Impregnation and Extraction of Polymers
Supercritical fluid extraction (SFE) is one of the major uses of
scC0, both on the industrial scale[’] and in the analytical labor a t ~ r y , [ ~ ’where
” ] it is rapidly replacing more traditional extraction (e.g. Soxhlet) in many “standard” methods. Analytical SFE
not only eliminates the use of potentially toxic organic solvents
but also it is usually faster than conventional extraction techniques. Equally important, SFE delivers the extracted analytes
free of solvent for subsequent analysis, allowing measurement
to be made from smaller samples than could be handled by
conventional means.
One of the major problems with any analytical extraction
method lies in knowing when an extraction is complete. Of
course, one can deliberately “spike” a matrix with some material and then establish how well the spike is recovered. However,
one cannot know, by definition, how much analyte is contained
in an “unknown” sample; all that can be done is to compare the
amounts of analyte extracted from the same sample by different
methods. SFE appears to be efficient; there have even been
occasions when SFE has extracted more analyte from a certified
sample than that sample was supposed to contain. Nevertheless,
Clifford and co-workers are attempting to develop modelsr771
that will allow analysts to use the accumulated results of several
sequential extractions to estimate the probable total content of
the sample.
There is clearly a need to test such models. Spectroscopic
monitoring of extraction requires high sensitivity because the
extracted analyte is usually present in the fluid at relatively low
concentrations. The aim is to monitor the levels of both the
extracted analyte in solution and of the residual analyte in the
matrix. FTIR spectroscopy is probably a better method than
UV/Vis spectroscopy for carrying out this monitoring because
IR absorption bands are usually more sensitive to environment
than are UV/Vis bands but the detection limits of IR spectroscopy are much worse. Although attempts are being made to
couple FTIR detection to SFE,[321the technique is less developed than the cSFC-FTIR method (see Section 3). Nevertheless, progress is being made. Hawthorne and co-workers
have reported the use of IR optic fibers for monitoring SFE[32b1
and recent work at N ~ t t i n g h a m [ ~ ~has
~ . ’shown
~ ] that FTIR
spectroscopy can be used for monitoring the extraction of transition metal carbonyl compounds, for example [CpMn(CO),]
1287
M. Poliakoff et al.
REVIEWS
(Cp = q'-C,H,), with supercritical phases from and impregnation into polymers. The important feature of this work is that
such compounds have unusually narrow and intense v(C-0) IR
bands, the precise wavenumbers of which are particularly sensitive to the environment. Thus, one can easily distinguish between the IR absorptions of [CpMn(CO),] in scC0, solution
and those of [CpMn(CO),] isolated within a polymer matrix.
Using a cell similar to that in Figure 3, Howdle and co-worke r ~ [ ' ~have
] shown that FTIR spectroscopy can be used to follow the impregnation of [CpMn(CO),] into polyethylene (PE)
and its subsequent extraction (Fig. 20). So far, the technique has
only been used with static extraction (i.e. in a closed cell without
any flow of scC0,) but even in these circumstances there is
potential for accumulating useful data.
*
12
I
A
I
2040
2034
I
I
I
2028
2022
2016
i. I cm-'
Fig. 20. "Real-time" FTIR monitoring of extraction of [CpMn(CO),] from
polyethylene (PE) with supercritical scC0,. The extraction can be monitored because the v(C-0) bands of [CpMn(CO),] have slightly higher wavenumber when
the compound is dissolved in scC0, solution than when it is in a PE matrix. The
spectra were obtained by placing four pieces of PE film (total thickness 1 mm) in a
cell and then pressurising with scC0, (200atm 32°C). The absorption due to
[CpMn(CO),] in scC0, grows as a shoulder, marked with a A, on the side of the
band due to [CpMn(CO),] within the PE. labeled with an *. The spectra in the inset
were obtained by computer subtraction and show the growth of the band of "free"
[CpMn(CO),] dissolved in scC0,; from ref. [78].
The impregnation of organometallic compounds into a polymer matrix with supercritical phases leaves no solvent residues
within the polymer because the CO, for example can easily
diffuse out. This diffusion process can also be monitored by
FTIR spectroscopy but Briscoe and co-workers have devised
ultrasonic methods that are considerably more sensitive.[791
Once impregnated into the polymer, the organometallic compounds can be induced to react with the polymer itself without
any side reactions with solvent residues. One such reaction,[801
the C-H activation of PE by [Cp*Ir(CO),] is illustrated schematically in Figure 21. This photochemical reaction can be followed easily by FTIR spectroscopy because reactant and
product have substantially different spectra in the v(C-0) region, where PE itself does not absorb significantly. Furthermore, the fact that the product is bound to the polymer was
established by showing that excess reactant can be extracted by
scCO,, while the product cannot. Unfortunately, this reaction
1288
\
/
Fig. 21. Schematic representation of the three stages of the C-H activation of
polyethylene (PE), the chains of which are represented by zigzag lines, by
(Cp*lr(CO),]. a) Impregnation of [Cp*Ir(CO),] into PE using scC0,: b) C-H
activation by UV irradiation of the impregnated film. and c) extraction of unconverted [Cp*Ir(CO),] by scC0,. The whole experiment was carried out in a cell
similar to that shown in Fig. 3; from ref. [SO].
suffers from the same limitation as the C-H activation of lighter
hydrocarbons by [Cp*Ir(CO),J; once [Cp*Ir(CO)H(R)J has
been formed, it is extremely difficult to develop the chemistry by
exploiting the activated C-H bond to make new organic compounds.["] Interestingly, hydrogen bonding of the Ir center by
(CF,),COH was found to reduce significantly the efficiency of
the C-H activation.[*'] It is not clear, however, whether this
reduction in efficiency is due to the change in the electron density on the metal center or merely due to a reduction in the UV
extinction of the complex with hydrogen bonds, leading to less
UV light absorbed in a given irradiation time. Clearly, the impregnation of other organometallic compounds into polymers
offers considerable possibilities for the modification or tailoring
of both bulk and surface properties of the polymers.
Over the past decade, Rest and co-workers have pioneered the
use of impregnated polymer films as a substitute for frozen
noble gases in matrix isolation experiments,[831to create what
they rather self-depreciatively describe as "infra-dig" matrices.
Impregnation using supercritical fluids extends the possibilities
of Rest's technique because some of his original experiments
were bedevilled by problems of solvent residues. Recently, the
high-pressure/low-temperature cell, illustrated in Figure 4, has
been used[' 5al to synthesize the highly unstable molecule
[Fe(CO),N,] from [Fe(CO),] impregnated into PE and then to
follow its subsequent thermal reaction with H,. Both the photochemical and thermal steps were followed by FTIR spectroscopy [Eq. (i)].
-
70 atm
a"''
*
The same high-pressure/low-temperature cell has also been
used by Kazarian et al. to investigate the reaction of HCI with
[Cp*Ir(CO),J in PE.t15blThis reaction led to the formation of
[Cp*Ir(CO),H]+, again with no detectable counterion, but unlike scXe solution at room temperature (see Section 5 ) ,
Angew. Chem. Int. Ed. Engl. 1995, 34, 1275-1295
Supercritical Fluids
REVIEWS
[Cp*Ir(CO),H]+ was quite stable in PE at 250 K. More surprisingly, venting the HCI and warming the impregnated PE film
back to room temperature caused up to 70 % regeneration of the
neutral [Cp*Ir(CO),] [Eq. (j)]. Under these conditions,
[Cp*Ir(CO),] appears to be isolated within PE and its u(C-0)
bands are only slightly broader than those observed in alkane
solution. Thus, the fact that the protonation of [Cp*Ir(CO),] in
PE is largely reversible strongly suggests that [Cp*Ir(CO),H]+
is present as ion pairs rather than as large aggregates.
A rather different non-spectroscopic study involving ions,
polymers. and scC0, has just been reported by Sullenberger et
al.. who fabricated two platinum microelectrodes, embedded
next to each other in the end of a glass rod.[841The electrodes
were then covered with a 50 pm film of polyethylene oxide
(PEO) doped with an appropriate electrolyte (i.e. LiCF,SO,).
The coated electrode assembly was then mounted in a high-pressure cell, filled with scC0, containing dissoked anthracene or
benzoquinone. The aromatic compound was thus impregnated
into the polymer where it be could be oxidized or reduced electrochemically. When the PEO film was additionally doped with
[Ru(bpy),12 +.there was evidence for the electrochemical activation of CO, and its reaction with benzoquinone.
7. Reactions of Organometallic Compounds
For over three decades. IR spectroscopy has played a central
role in the mechanistic study of photochemical reactions of
organometallic compounds, particularly of transition metal carb o n y l ~ . [Techniques
~~]
such as frozen hydrocarbon glasses, matrix isolation. liquefied noble gases, and time-resolved IR spectroscopy, have been developed and optimized through such
studies, sometimes exclusively so. The emphasis on metal carbonyls and the closely related dinitrogen and nitrosyl compounds derives partly from inherent chemical interest but equally importantly from the enormous amount of structural
information that can be derived merely from recording IR spectra in the v(C-0). v(N-N), and v(N-0) regions. Molecular
geometries, bond angles and, on occasions, even estimates of
bond lengths can be derived from these spectra. Improved instrumentation, particularly FTIR instruments, has allowed at
least some of these techniques to be extended in recent years to
non-carbonyl organometallic compounds.[861
The original impetus for studying photochemical reactions of
organometallic compounds in supercritical fluids arose from the
very low solubility of some substituted metal carbonyl compounds in liquid xenon.[*'] Thus, even though [Ru(CO),(PPh,)]
is virtually insoluble in LXe, it has modest solubility in scXe at
25 "C. Figure 22 illustrates spectra recorded by Upmacis et al.
during the reaction of [Ru(CO),(PPh,)] with H, in scXe, one of
the first reactions of organometallic compounds to be carried
out in supercritical fluid solution.[871The reaction product,
[H,Ru(CO),(PPh,)] had previously only been observed by
Whyman[*'"] in a thermal reaction in heptane under very much
harsher conditions, 100 "C and 300 atm H,. Once generatedta7]
2100
2000
1800
1900
i./cm-'
Fig. 22. FTIR spectra showing the photochemical synthesis of [H,Ru(CO),(PPh,)]
by irradiation of [Ru(CO),(PPh,)] and H, with UV light in scXe at 25°C; a)
spectrum before irradiation; b) spectrum after 9 min phorolysis. Bands due to
[H,Ru(CO),(PPh,)] are marked with arrows. The P(Ru-H) region of the spectrum
is shown with a 17-fold expansion of the absorbance scale: from ref. [87].
in scXe, [H,Ru(CO),(PPh,)] decayed over a period of about six
hours even under a high pressure of H,, possibly b y reaction
with the photoejected CO. This reactivity is considerably greater
than that of [H,Ru(CO),(PPh,),] which had been isolated many
years
This reaction of [Ru(CO),(PPh,)] led to the realization that
the irradiation of metal carbonyl compounds with UV light in
supercritical solution in the presence of H, might be a relatively
general route to either dihydrogen or dihydrido complexes, depending on the particular metal centerCz5]
[Eq. (k)].
Using this route, Howdle et al. generated a series of compounds, including the half-sandwich dihydrogen complexes, 6 8, formed from the corresponding [ (q"-C,H,)M(CO),] compounds in scXe solution.[251Such compounds should, in
I
OCycrj,,H
OC
H
6
I
I
Mn
ocvMnvH
OC
H
7
0
~ y S7
iR3
OC
H
7'
ocyFey
OC
H
8
principle, have a weak IR band within the region 34002400 cm-', associated with the v(H-H) vibration of the q2-Hz
Although the v(H -H) band of [W(CO),(q2-H,)] has
been observed[27]in scXe using a cell similar to that in Figure 3,
no band was found for any of the compounds, 6-8, presumably
because the bands were extremely weak. Nevertheless, there was
sufficient circumstantial evidence to confirm that each of these
compounds did indeed contain an q2-H, group. The compounds
6-8 are of interest because 7 was the first nonclassical dihydrogen complex of manganese to be identified, and 8 was the first
1289
M. Poliakoff et al.
REVIEWS
dihydrogen compound of iron also to contain a coordinated
olefin or dime ligand. Compound 7 appeared to be relatively
stable even after the H, had been vented (see Section 8). The
arrested oxidation of the Mn center is not unprecedented in view
of the extensive series of known q2-H -SIR, complexes of Mn
(e.g 7’).[’’]
H, is significantly less soluble than other gases (e.g. N,, CO,
etc) in conventional organic solvents. However, H, is completely miscible with a supercritical solvent, such as scXe, since the
fluid is merely a dense gas. Thus, if properly mixed with the fluid
(see Section 8), a given pressure of H, will give an effective
concentration of “dissolved” H, nearly an order of magnitude
higher than in a conventional solvent under a similar pressure of
H,. The predominant reaction pathway for labile dihydrogen
complexes appears to involve dissociative loss of the q2-H2
group and so this very high concentration of H, can increase the
lifetime of such complexes in scXe solution.[251Nevertheless,
even under a such pressure of H,, 8 reacted further within a few
minutes, possibly through transfer of H, to the C4H4ring, while
6 decayed only slowly under the same conditions.
The miscibility of H, and scXe has an additional benefit
namely that, in a closed vessel, the concentration of dissolved
H, is independent of temperature. By contrast, the solubility of
H, in a conventional solvent will show a temperature dependence as dictated by Henry’s Law. Howdle et al.[251exploited
this constant concentration of H, in scXe to investigate the
kinetics of the reaction of CO with [(q6-C6H,Me)Cr(CO),(q2-H2)] [Eq. (I)]. The reaction was monitored via the a”
Equation (I) is, perhaps, slightly misleading because there
is increasing evidence that reaction intermediates such as
[ (q6-C6H,Me)Cr(CO),] are not genuinely coordinatively unsaturated but are weakly solvated by a molecule of solvent acting
as a “token” ligand in the otherwise vacant coordination site.[851
Even solvents, apparently as inert as Xe, can act as token ligands
and the strength of the M-Xe interaction has been estimated for
the group 6 metals in gaseous and liquid Xe.[’21 In recent experiments we have begun using nanosecond time-resolved IR spectroscopy to monitor these interactions in scXe and scC0, .[931
The high concentrations of H, in supercritical solution have
since been exploited by other groups. Rathke and co-worke r ~ [ ’ ~reported
]
a series of elegant experiments where thermal
hydroformylation reactions catalyzed by [Co,(CO),] in scC0,
were monitored by NMR spectroscopy. The product distribution was rather different from that found in conventional solvents and a patent has been granted on the process.~’4c1The very
low viscosity of supercritical solvents is of considerable benefit
for NMR spectroscopy and nuclei with spin 2 1 have shorter
correlation times and correspondingly sharper resonances under supercritical conditions.
Very recently, Noyori and C O - W O ~ ~ ~ ~ Shave
[ ’ ~ ] reported the
catalytic hydrogenation of scC0, itself [Eq. (m)]. This experiment has aroused considerable excitement, not because the
product (formic acid) is any way different from that in T H F
solution but because the reaction is over an order of magnitude
faster in scC0,. Presumably, this acceleration is the result of the
greater concentration of H, in the supercritical fluid.
COz + H , (85 atm) --
[ R u W M e i h H ~ l+ NEl)
S C C O ~ . ~=
,,~
2 0 5 a h . 50’C
v(C-0) band of [(~6-C,H,Me)Cr(CO),(q2-H,)]; it showed
pseudo-first order kinetics with an activation energy of
70+5 kJmol-’ over the temperature range I1 to 80°C
(Fig. 23). If loss of H, is assumed to be the rate-determining
step, this activation energy can be taken as an estimate of the
Cr-(q2-H,) bond dissociation energy and, perhaps fortuitously,
the value is almost identical to that estimated by photoacoustic
and calorimetric techniques for related Cr-(q2-H,) bonds.[”]
t
lamp off
(4
N, is also miscible with supercritical fluids, and Howdle et al.
have again exploited this for the generation of new compounds.
Their most striking reactionr96a1was the total substitution of
CO by N, in the molecule [CpRe(CO),] to generate [CpRe(N,),]
[Eq. (n)]. As with many of the reactions described previously,
+
..
HCOOH
cXN*
+ Ng
CO
>
this substitution reaction was followed by FTIR spectroscopy
but this time in the v(N-N) region (Fig. 24). The assignment of
A
-6
the bands was confirmed by comparison with earlier experiments in solid N, matrices doped with ”N, at 20 K. Neither
-7
[CpRe(CO)(N,),] nor [CpRe(N,),] had been observed previous0
-8
ly at room temperature. However, once generated [CpRe(N,),]
3.1
3.3
3.5
0
15
30
was remarkably stable, persisting for more than 24 h in scXe
t l min
llTx103
solution
under a pressure of N,, and then taking nearly 1 h to
Fig. 23. a) An IR kinetic trace showing the photochemical generation and thermal
react with added CO at room temperature. [CpRe(CO)(N,),]
decay of [(‘16-C,H,Me)Cr(CO),(‘1z-Hz)l in scXe:H, at 70°C. see Equation (I). The
decay follows pseudo-first order kinetics. b) Arrhenius plot for the rate of decay of
was even more robust and, although only spectroscopic
[(‘lh-C,H,Me)Cr(CO),(qZ-H,)l measured in the presence of two concentrations of
amounts
were generated, the experiments suggested that
added CO. Note that, within experimental error, the two curves have the same
gradient. corresponding to an activation energy of 70 kJmo1-l; from ref. [25].
[CpRe(CO)(N,),] might be air
-5
1290
Angcw. Cliem. Int. Ed, En$
1995, 34. 1275-1295
Supercritical Fluids
REVIEWS
n
v(lr-H)
v(C-0)
P
2160
2130
2100
f--
2070
cm-'
Fig. 24. FTIR spectra in the L,(N-N) region recorded during the UV photolysis of
[CpRe(CO),] in scXeiN, (total pressure 2600 psi). The spectra show the sequential
formation of [CpRe(CO),(N,)] (white bands). [CpRe(CO)(N,),] (shaded bands)
and. finally, [CpRe(N,),] (black bands). The total irradiation time was 2 h but
subsequent experiments with more powerful lamps show that the reaction can be
carried out lar more rapidly; from ref. [96a].
2140
2070
2000
1930
91 cm-'
A number of other dinitrogen compounds have been generated by a similar route but there are limitations; UV photolysis
of [(C,H,Me)Mn(CO),(PR,)] compounds with N, in supercritical solution leads to considerable loss of the PR, ligand
and, only with PMe,, were significant amounts of the
[(C,H,Me)Mn(CO)(N,)(PR,)] compound observed.[96b1
Apart from Noyori's hydrogenation e~periment"~]
with scCO, , most attempts to activate scC0, have been disappointing.l9'I In fact, scC0, is a remarkably inert solvent and Jobling
et a]. have taken advantage of this inertness[y81to use scC0, for
photochemical activation of C-H bonds by [Cp*Ir(CO),]. CH activation of alkanes is an area ofconsiderable research activity but one continuing problem has been to find suitably inert
solvents in which to carry out activation of light alkanes such as
CH, or C,H,. Graham and co-workers were successful in using
fluorocarbon solvents[811and Bergman and co-workers used
LXer9"' but neither solvent was ideal. Jobling et al.[981showed
that photochemical C-H activation of scC,H, could be carried
out satisfactorily in scC0,. It quickly became apparent, however, that the scC0, was superfluous because the critical temperature of C,H, is so close to that of CO, (see Table 1). Thus,
scC,H, could be used as both solvent and reactant but the C-H
activation did not proceed far because a photostationary state
was established rapidly (Fig. 25 a). Surprisingly, addition of
high pressures of H, to the mixture not only promoted formation of [Cp*Ir(CO)(H),] but led to greatly increased amounts of
C-H activation [Fig. 25b and Eq. (o)].
Since the critical pressure of C,H, is considerably lower than
that of CO, (see Table l), it was possible to use much higher
pressures of H, with C,H, without exceeding the pressure rating
of the cell shown in Figure 3, and even to reach approximately
Angrw. Clim Ifit t r l Engl 1995. 34, 1275-1295
Fig. 25. FTIR spectra illustrating the dramatic effect of H, on the C - H activation
of scC2H6by [Cp*Ir(CO),]. a) IR spectrum recorded after 3 minutes UV irradiation
of [Cp*Ir(CO),] in scC,H, (68 arm); b) IR spectrum showing the strikingly different
result of 10 minutes UV photolysis of [Cp*Ir(CO),] in scC,H, (50 atm) with added
H, to give a total pressure of 179 atm. The bands are labeled as follows:
E = [Cp*Ir(CO)(H)(C,H,)]; H = [Cp*Ir(CO)(H),]; P = residual [Cp*Ir(CO),];
? = an unidentified labile photoproduct. from ref. [98].
equimolar concentrations of H, and C,H, in the solution. The
effect of H, on C-H activation is chemical rather than physical
because high pressures of He are ineffective; experiments with
D, indicated that no deuterium was incorporated into
[Cp*Ir(CO)H(Et)]. Sequential experiments in scC0, (addition
of H, to generate [Cp*Ir(CO)(H),] followed by addition C,H,)
showed that C-H activation by [Cp*Ir(CO)(H),] is more efficient than by [Cp*Ir(CO),] itself. The reason for this still remains unclear but the overall effect is dramatic.
CH, has its critical temperature at 190.5 K well below that
of C,H,. Nevertheless, it is possible to dissolve small amounts
of [Cp*Ir(CO),] in scCH, at temperatures close to ambient,
provided that the pressure is sufficiently high to give a fluid
density close to p c , 0.162 gmL-'. This density can be achieved
at a pressure of about 3750 psi at 297 K. Under these conditions,['OO1
UV irradiation of [Cp*Ir(CO),] leads to activation of
CH, and formation of [Cp*Ir(CO)H(CH,)] , identifiable by
comparison with IR data from low temperature matrices.
Irradiation of [Cp*Ir(CO),] in scC,H, leads to two processes,12'] substitution of CO by C,H, and C-H activation to generate a hydrido(viny1) complex [Eq. (p)]. As in the other C-H
activation experiments, both compounds were identified by
FTIR spectroscopy; scC,H, has strong IR bands which obscure
considerable regions of the spectrum but, fortuitously, it has a
series of "windows" in each of the regions, v(Ir -H), v(C-0),
v(C=C), and 6(C-H), needed to characterize [Cp*Ir(CO)H(q'C,HJI.
+ ++
I
&
hv
l
+
I
(P)
1291
M. Poliakoff et al.
REVIEWS
These experiments are interesting not only because C,H, or
C,H, can be activated but also because the reaction products
can be recovered. Product recovery has been a substantial problem in C-H activation experiments in conventional solvents;
attempts to remove the solvent frequently lead to decomposition of the compound. By contrast, a reduction in applied pressure will result in precipitation of solid compounds from supercritical solution. Figure 26 shows how such precipitation can be
P
R
V
Product
Fig. 26. Schematic illustration of how small quantities of reaction product can be
recovered from a cell, R (such as that shown in Fig. 3 ) . A syringe pump P forces
scC0, through the system to flush the reaction mixture (shown black) through an
expansion valve V. The reduction in pressure across the expansion valve causes the
dissolved solids to precipitate from solution allowing collection.
achieved in a controlled manner by pumping scC0, to flush the
contents of the reaction cell through an expansion valve.[93a1In
this way, it was possible to recover sufficient [Cp*Ir(CO)H(R)]
(R = Et or q1-C,H3) for ' H N M R confirmation of the structures, already identified by IR spectroscopy. Alternatively, the
compounds can be redissolved in scXe to obtain IR spectra free
from interfering IR absorptions of the solvent.[291This method
of precipitating a product from supercritical solution is a key
feature of the miniature flow reactors under development for
continuous reactions in supercritical fluids.
IR spectroscopy can be used, of course, to monitor a much
wider range of processes than just reactions of organometallic
compounds. Buback describes a number of applications of IR
and near-IR spectroscopy for following polymerization react i o n ~ . [Saito
~ ] and co-workersI'ol] have recently reported the use
of TR spectroscopy for monitoring Diels-Alder reactions between isoprene and methyl acrylate, which has relatively intense
v(C=O) bands. In fact, most reactant/product combinations are
likely to have absorption bands somewhere within the mid-IR
region, suitable for spectroscopic monitoring. However, the
particular advantage of IR spectroscopy in reactions of
organometallic compounds is both the high sensitivity provided
by the v(C-0) bands and the possibility of identifying previously unknown compounds with a considerable degree of certainty
merely from their IR spectra.
8. Reactions with Supercritical Fluids
on a Preparative Scale
The ultimate objective of many chemists is to harness new
techniques to make new compounds on a preparative scale.
However, most of the reactions and processes described in the
previous sections of this review have been carried out in dilute
solutions and on a relatively small scale. These reactions, therefore, need to be scaled up. Vibrational spectroscopy is playing
1292
a key role in the scale-up of reactions of organometallic compounds in supercritical fluids.
Apart from any safety considerations, the scale-up of reactions in supercritical fluids presents a number of unusual problems. Most substrates have critical temperatures far higher than
that of CO, and so to maintain supercritical conditions reactions must either be run in very dilute solution or at high temperatures. However, dilute solutions prevent the processing of large
quantities of material unless the reaction vessel is extremely
large. High temperatures are often undesirable because reaction
products, if labile, may degrade. Furthermore, many of the
more interesting phenomena associated with supercritical fluids
occur relatively close to the critical point (see for example,
Fig. 18). In addition, the composition of a reaction mixture
changes as the reaction proceeds and so the critical point will
also change, making it difficult to maintain the desired near-critical reaction conditions.
These problems can be minimized, if not removed completely,
by use of a continuous flow reactor because at any particular
point in such a reactor the composition remains constant
throughout the reaction. Equally, the gaslike nature of supercritical fluids (low viscosity, good heat transfer, etc) readily
lends itself to flow reactors.
Pickel and Steinerl'ozl have
pioneered the use of flow reactors for industrial-scale processes in the pharmaceutical
industry, particularly catalytic hydrogenation. Their use of
such reactors in one process
has resulted in a 1/250 reduction in reactor volume compared to the batch process in
conventional solvents.
At Nottingham, we are deFig. 27. A schematic diagram of the
flow reactor used for the synthesis
veloping miniature flow reac[lo31 of [Cr(CO),(C,H,)]. scC,H, is
tors for laboratory-scale
pumped through a reservoir containpreparative reactions. Our
ing solid [Cr(CO),], which dissolves in
the scC,H,. The supercritical solution
prototype reactor was dethen passes through a UV irradiation
signed for reactions in
cell (a modified version of that in
scC,H,, where the fluid acted
Fig. 3 ) where reactant is converted into product, through an IR cell for
both as solvent and as one of
FTIR monitoring and finally through
the reactants." 03] Figure 27
an expansion valve (analogous to
Fig. 26) to precipitate the product as a
shows a schematic view of the
dry solid.
reactor; a solid organometallic compound is dissolved in
scC,H,, converted into product by UV irradiation, monitored
by IR spectroscopy, and finally precipitated as a fine powder, all
as a single continuous process. Once the fluid pump is running,
reactant i s converted into product automatically, while the operator uses IR monitoring to optimize the reaction conditions.
Using this reactor, it was possible to isolate [Cr(CO),(C,H,)]
(> 80 % purity) directly from the reaction of [Cr(CO),J and
scC,H, at a rate of approximately 40 mgh-' [Eq. (q)].
"'"
Although [Cr(CO),(C,H,)] was already known in solut i ~ n , " ~previous
~]
attempts to isolate it had been thwarted by
Angew. Chrm. Int. Ed. Engl. 1995, 34. 1275-1295
REVIEWS
Supercritical Fluids
the lability ofthe C,H,; removal of the solvent also removed the
ligand. Expansion of scC,H, is more rapid than conventional
solvent evaporation; thus the solution does not need to be heated and a high concentration of the free ligand C,H, is maintained right up to the moment at which the solid is formed. Once
precipitated. [Cr(CO),(C,H,)] appears to be moderately stable
in the solid state."""] The relative intensities of the v(C-0) IR
bands of [Cr(CO),(C,H,)] suggest that the C,H, group is
aligned parallel to a trans pair of CO groups (as in 9 not 10) as
found by Grevels and co-workers in
the crystal structure of the related
,,,.,. I
compound [Cr(CO),(C,H,),I .r104b1
I .,,d!
/\
-%
Success with [Cr(CO),(C,H,)],
encouraged us to build a more com10
9
plex reactor using H, mixed with
scC0, (Fig. 28) for the isolation on a larger scale of the dihydrogen compounds which we had previously made only on a spectroscopic scale (see Section 7)
In this case, the mixing of the
-
@
I
I
,I I , . . ,
..,Is,
I
(FT-Rrn)
Gas M
i
J
FJ-IR
x
\
Product
Fig. 28. A simplified schematic diagram of the flow reactor used for photochemical
reactions of organometallic compounds in scCO,!H, mixtures. Initially, highpressure H 2 and s K 0 , are mixed. The organometallic compound (e.g. [CpMn(CO),])
is then dissolwd in the inixture and the H, content can be verified. as necessary. by
FT Raman speclroscopy using the cell shown in Fig.5c. After UV irradiation, the
fluid i s inonitored by FTlR spectroscopy and finally the product is precipitated, as
i n Figs.2h and 77. by expansion through a high-pressure valve.
H, and scCO, is monitored in-line by FT-Raman spectroscopy
using the cell shown in Figure 5c, and the photochemistry is
followed by FTIR spectroscopy, as before. This reactor has
been used successfully to prepare 7 I 1 O 6 ] [Eq. (r)]. This com-
7
pound has never been isolated before and transpires to be remarkably robust; it is a pale yellow solid that melts to a brownish liquid at room temperature. The thermal stability contrasts
with the relative lability of many other dihydrogen complexes.18Ql
A/igcii. < l i i w i .
hi.Ed. .Gig/. 1995. 34, 1275-1295
9. Conclusions
Supercritical fluids have a range of properties that are qualitatively different from those found in conventional solvents, including "tunable" density, high solubility of gases, low viscosity, etc. High-pressure conditions make chemistry in supercritical fluids necessarily more complicated than in conventional
solvents, although the technology for microscale reactions is
rapidly becoming simpler. We remind readers. however, that
false expectations have sometimes been aroused in the past by
over-optimistic articles and patents. There are many things
which supercritical fluids cannot do. They are not magic solvents that will dissolve materials which have defied all other
attempts to dissolve them. Of course, there are unexpected surprises such as the ability of scCOz to dissolve fluoropolymers
but such cases are exceptions. Similarly, the behavior of supercritical fluids cannot contravene fundamental laws of thermodynamics, although the different solvation properties may alter
the precise energetics of a process. Nevertheless, it is clearly
important to identify those areas of chemistry where the special
properties of supercritical fluids bring real benefits. Fortunately, diagnostic tools and laboratory techniques have now developed to the point where chemists can place their investigations
on a systematic basis.
This review has shown that vibrational spectroscopy already
plays a significant role in a very varied range of chemistry in
supercritical fluids and clearly the range will become still broader in the future. In some areas the role of spectroscopy is central
to the whole investigation while, in others, the role is to provide
diagnostics for optimizing a process. In selecting our examples,
we have not tried to be encyclopaedic but rather we hope to have
described a sufficient number of applications for individual
readers to judge whether such measurements might usefully he
applied to their own particular problems. It is important, however, to repeat the fact that we have largely ignored those areas
already covered in Buback's review.[41Neither have we had the
space to discuss the applications of UV/Vis spectroscopy and
also some important areas of chemistry in supercritical fluids
where vibrational spectroscopy has not yet been exploited fully.
For example, supercritical fluids permit the use ofenzymes in an
essentially non-aqueous environment for reactions such as esteri f i ~ a t i o n . [ ' ~Severs
'~
and co-workers have shown that supercriticai fluids have considerable promise for the deposition of thin
films and coatings because their use eliminates the need for
solvents which might leave residues and contaminate the
resulting film.[1081Furthermore, the solvent power of supercritical fluids increases the range of possible precursor molecules
for MOCVD (metal organic chemical vapor deposition
because molecules can be induced into the gas phase at
relatively low temperatures.[1091Lynch and co-workers have
found that scC0, can be used for removal of solvent from
thin films, originally deposited by wet chemistry, without destroying the integrity of the film.['101 Finally, we have not
described the chemical possibilities of fluids with critical points
at rather higher temperatures. For example, recent work by
Wood et al. has shown that previously unknown metal
sulfide complexes can be synthesized relatively simply
from K,S, and [W(CO),] in supercritical ethylene diamine
= 320"C).""1
(c
1293
REVIRNS
M. Poliakoff et al.
We hope that our review has shown how many of the
boundaries in traditlonal chemistry, analytical/spectroscopy/
polymerjsynthetic become blurred in supercritical chemistry or.
like the liquid/gas boundary, may even disappear completely.
The study of supercritical fluids should not be considered as the
exclusive preserve of physical chemists and chemical engineers.
When chemists have collaborated with scientists working with
supercritical fluids, for example, Fox and Johnston, the outcome has been extremely successful.[40J There are now many
laboratories throughout the world with expertise in handling
supercritical fluids and we urge synthetic chemists to begin collaborating with them. Exciting new science will emerge.
We thank our co-workers, colleagues, and technical staff at
Nottingham and our UK and overseas collahorators j b r their contribution to our work described within this review. We grateful1.v
acknoM,lcdgefinancial support,for this work from EPSRC Chemistry and Clean Technology (Grant nos. CR!H95464 and GRi
J9.5065) ,from EC Human Capital and Mobility and COSTprograms, the Volkswagen Stiftung, Perkin-Elmer, BP International,
BP Chemicals, Zmeca, NATO Laboratory Linkage Program and
particularly to the Royal Society for providing a Leverhulme Senior Research Fellowship to M . P., a University Research Fellowship to S. M . H., and a Visiting Fellowship to S. G. K .
Received July 6, 1994 [A 72 I E]
German version: Angew. Cliern. 1995, 107, 1409-1432
[I] M. A. McHugh, V. J. Krukonis. Supercriticul FluidEsfrucrion: Principles und
Pructice. 2nd ed.. Butterworth-Heinemann. Boston, 1994.
12) a) Supercritical FIuid Technology (Eds.: F. V . Bright, M. E. P. McNally) (ACS
Symp. Ser. 1992, 488); b) Extraction of Nutural Products Using Near-Critical
Solvents (Eds.: M. B. King, T. R. Bott), Chapman & Hall, Glasgow, 1993.
(31 R. W. Shaw, T. B. Brill, A. A. Clifford, C. A. Eckert, E. U . Franck, Chem.
Eng. News 1991, 69. 26.
[4] M. Buback, Angew. Chem. 1991, 103.658; Angew. Chrm. Int. Ed. Engl. 1991.
30, 641.
[5] a) Supercritical Fluid Science und Tecknology (Eds.: K. P. Johnston, I. M. L.
Penninger) ( A C S Symp. Ser. 1989. 406); b) Supercritical Fluid Engineering
Science: Fundamentals and Applications (Eds.: E. Kiran, I. F. Brennecke)
(ihid. 1993.541):c) Sitpercritica/ Fluids - Fundumentuls for Applicarions (Eds.:
E. Kiran. J. M. H. Sengers) (NATO ASI Ser. Ser. E 1994, 273).
[6] See C. J. Gregg, F. P. Stein, C. K. Morgan, M. Radosz, J. CAem. Eng. Dutu
1994, 39, 219.
[7] W F. Sherman, A. A. Stadtmuller. E.Yperinienta1 Teeliniques in High-pressure
Research. Wiley. London, 1987.
[XI R. Whyman in Techniquesjor Vibrational Spectroscopic Measurements under
High Pressures, in Laboratory Methods in Vibrutionul Spectroscopy (Eds.:
H. L. Willis), Wiley. London, 1987, p. 281.
[9] W. H. Beattie. W. B. Maier. R. F. Holland. S. M. Freund. B. Stewart, Proc.
SPIE Inr. Sue. Opt. Eng. 1978, 158, 113.
[lo] a) M. 0. Bulanin, J. Mol. Stract. 1973. 19, 59; b) V. V. Bertsev in Molecular
Cr,~~ospecrro.rcop~,
Advunces in Speclroscopy (Eds.: R. J. H. Clark, R. E.
Hester), Wiley. London, 1994, chapter 1 .
[ll] S. Kim. K. P. Johnston, Ind. Eng. Chcm. Res. 1987, 26, 1206.
1121 M. A. Healy, T. J. Jenkins. M. Poliakoff, TrAC Trends A n d . Chem. Pers. Ed.
1991, 10, 92.
[13] G. G. Yee. J. L. Fulton, J. P. Blitz, R. D. Smith, J. Pliys. Chem. 1991,95, 1403.
[14] a) S. M. Howdle, M. A. Healy, M. Poliakoff, J. M. Whalley in Proc. lnt.
Symp. Supercrit. Fluids (Eds.: M. Perrut), Societe Francaise de Chimie, Paris,
1988, p. 967; b) S. M. Howdle. M. Poliakoff. Supercritical Fluids- Fundamentulsjor Applrcations (Eds.: E. Kiran, J. M. H. Sengers) ( N A T O AS1 Ser. Ser. E
1994. 273. 527).
[I51 a) A. I. Cooper, M. Poliakoff, unpublished; b) S. G. Kazarian, A. I. Cooper,
M. Poliakoff, Opt. Spectrmc. 1994, 76, 242.
[16] See S. Ikawa. Y. Fujita. J. Pliys. Chem. 1993. 97, 10607.
[17] D. Ben-Amotz, F. LaPlant, D. Shea. J. Gardecki, D. List, ACS Sl'mp. Ser.
1992,488, 18.
[18] M. L. Myrick, J. Kolis, E. Parsons, K. Chilke. M. Lovelace, W. Scrivens. J.
Rmnnn Specrrosc. 1994. 25, 59.
[I91 S. M. Howdle. K. Stanley. V. K. Popov. V. N. Bdgratashvili, Appl. Spertrarc.
1994, 48, 214.
1294
[2O] a) V. N. Bagrutashvili, V. K. Popov. D. G. Robertson, S. M. Howdle, M.
Poiiakoff, M. W. George, E. Walsh, ThirdInr. Symp. Superrrit. Fluids, Strasbourg, France. 1994, 1 , 3 3 7 ; b) D. G. Robertson, S . M. Howdle. unpublished.
[21J T. Andrews. Pro<.R. Soc. London 1875, 24.455.
[221 a) S. G . Kazarian. B V. Lokshin, Y M. Kimelfeld, R. B. Materikova, I;v.
Acrrd. Nuuk S S S R Ser. Kh6n. 1986, 2603; b) B. V. Lokshin, S. G. Kazarian,
A. G. Ginzburg, .I
Mol. Srruct. 1988, 174,29.
[23] M. W Raynor. G . F. Shilstone. K. D. Bartle, A. A. Clifford, M. Cleary, B. W.
Cook. J. High Retulut. Chromutogr. 1989, 12. 301.
[24] V. J. Krukonis. M. A. McHugh. A. J. Seckner, J. Phys. Chem. 1984, 88,
2687.
[25] S. M. Howdle. M. A. Healy. M. Poliakoff. J. Am. Chem. Soc. 1990, 112,4804.
[261 M. Poliakoff. J. J. Turner. in ref. [lob], Chapter 8.
(27) J. J. Turner, M. Poliakoff, S. M. Howdle. S . A. Jackson. J. G. McLaughhn,
Furaduy Discuss. Chrm. Soc. 1988,86, 271.
[281 A. Moustakas. E. Weitz. Chrm. Phys. Lett. 1992, l9f, 264.
[2Y] M. Jobling, PhD Thesis. University of Nottingham. UK, 1992.
[30] R. M. Smith, Superirirrcul Fluid Chromulopruphy. Royal Society of Chemistry, London, 1988.
[31] a ) .4nul~ticulSupercritical Fluid Cliromutogruphy and E.rrraction (Eds.: M. L.
Lee. K. E. Markides), Chromatography Conferences. Provo. UT. 1990; b)
Applicurrons f. Supercritical Fluidr in Industriul Anu!ysis (Hrsg. : J. R.
Dean), Chapman & Hall. Glasgow. 1993: c) Ana/l-:ris with Supercritical
F/utd.s: Errrucrion and Chromutogruph.r.( E d . : B. Wenclawiak). Springer, Berlin, 1992.
[32] a) Hyphenarid 7kcliniques in Supercriticul Fluid Chromutopraph),and E-rtrurtion (Ed.: K. Jinno) ( J . Chromutogr. L i b . 1992. 53); b) D. L.Heglund, D. C.
Tilotta, S. B. Hawthorne. D. J. Miller, Anal. Chem. 1994, 66. 3543.
[33] a) S. 9 . French, M. Novotny. Anat. Chem. 1986,58, 164; b) S. V. Olesik, S. B.
French. M. Novotny. Chromutoxraphiu 1984, 18. 489.
[34] S. Shah, M. Ashraf-Khorassani, L. T. Taylor. Chromutogruphiu 1988,25,631;
L. T. Taylor. E. M. Calvey, Chem. Rev. 1989, 89, 321.
[35] T. J. Jenkins, G. Davidson. M. A. Healy, M. Poliakoff. J. High Re.dur. Chromutopr. 1992, f5,819.
[36] T. J. Jenkins. M. Kaplan. G. Davidson, M. A. Healy, M. Poliakoff, f Chromutogr. 1992. 53. 626.
[37] P. Morin, C. Beccard, M. Caude, R. Rosset. HRC & CC J. High Resolur.
Chromutogr. Chrontatogr. Commun. 1988, 11. 697.
[38] Y. Garrabos. V. Chandrasekharan, M. A. Echargui, F. Marsault-Herail,
Chem. Phys. Lett. 1989, 160, 250.
[39] F. Salmoun, J. Dubessy, Y Garrabos, F. Marsault-Herail. J. Ruman Spectrosc.
1994, 25, 281.
[40] B. J. Hrnjez, A. J. Mehta, M. A. Fox, K. P. Johnston. J. Am. Chem. Soc. 1989,
111. 2662.
[41] S. Akimoto, 0. Kajimoto. Cheni. Pliys. Lett. 1993, 209, 263.
[42] G. G. Yee. J. L. Fulton, R. D. Smith, J. Phys. Chem. 1992, 96, 6172.
I
Supercrit. Fluids 1993. 6 ,
[43] Y. Yagi, H. Tsugane. H. Inomata, S . Saito. .
139.
[44] 7. W. Zerda, X. Song, J. Jonas, Appl. Spectrosc. 1986, 40. 1194.
[45] P. Battersby. J. R. Dean, S . M. Hitchen, W. R. Tomlinson, P. M. Myers, J.
Comput. Chem. 1994, 15, 580.
[46] D. A. Masten, B. R. Foy, D. M. Harradine. R. B. Dyer, J. P h p . Chem. 1993,
97. 8557.
[47] S. M. Howdle, V. N. Bagratashvili, Chem. Phys. Lett. 1993. 214, 215.
[48] C. A. Eckert, B. L. Knutson, Fluid Phase Equilib. 1993, 83, 93.
[49] a) R . 4 . Wu, L. L. Lee, H. D. Cochran, fnd. Eng. Chem. Res. 1990, 29, 977;
b) C. B. Roberts, J. E. Chateauneuf, J. F. Brennecke, J. Am. Chem. Soc. 1992.
114. 8455; c) Y-P. Sun, ibid. 1993, f15,3340; c) J. Zagrobelny, F. V. Bright,
;bid. 1992, 114.7821; d) K. P. Johnston, S. Kim. AIChE J. 1987,33,1603;e)
J. A. O'Brien, T. W. Randolph. C. Carlier. S . Ganapathy, ibid. 1993,39, 1061;
f ) C. Carlier. T. W. Randolph, ibid. 1993. 39, 876; g) T. W. Randolph. J. A.
O'Brien, S. Ganapathy, J. Phys. Chem. 1994, 98. 4173; h) A. Morita. 0.
Kajimoto, ibid. 1990.94.6420; i) P. G. Debenedetti, Chem. Eng. Sci. 1987.42,
2203; k) P. G. Debenedetti, I. B. Petsche, R. S. Mohamed, Fluid Phase Equilih. 1989, 52. 347.
[50] a) J. F. Brennecke. D. L. Tomasko, J. Peshkin, C . A. Eckert, Ind. Eng. Chem.
Res. 1990, 29, 1682: b) J. F. Brennecke, C. A. Eckert, ACS Symp. Ser. 1989,
406, 14; c) Y. P. Sun, G. Bennett, K. P. Johnston. M. A. Fox, Anal. Chem.
1992,64.1763;d) Y-P. Sun, M. A. Fox, 1 Phys. Chem. 1993,97,282;e) D. L.
Tomasko, B. L. Knutson. J. M. Coppom, W Windsor. B. West, C. A. Eckert,
ACS Symp. Ser. 1993, 541, 220; f ) J. Zagrobelny, T. A. Betts, F. V. Bright, J.
Am. Chem. Soc. 1992,114,5249;g ) J. Zhang, F. V. Bright, J Phys. Chem. 1992,
96, 5633.
[51] D. L. Tomasko. B. L. Knutson. F. Pouillot. C. L. Liotta, C. A. Eckert. J.
Ph.ys. Chem. 1993, 97, 1 1 823.
[52] R. Ravi, L. E. S. deSouza, D. Ben-Amotz. J. Ph.vs. Chem. 1993, 97,
11 835.
[53] a) F. C. Knopf, X. Xu, S.-C. Yu, R. Lingle in Second int. Symp. Supercrit.
F/uid.s (Ed: M. A. McHugh). Johns Hopkins University. Baltimore. 1991,
p. 154: b) C. Lienau, J. C. Williamson, A. H. Zewail, Chem. P h w . Lett. 1993.
213. 289.
Angen. Chem. I n t . Ed. Engl. 1995.34, 1275-1295
Supercritical Fluids
a ) J M Walbh. M. D. Donohue, F h i d Phrrrr Eyrrrlih. 1989, 52, 397; b) A. 1.
Coopei-. S. M. Howdle. C. Hughes, M . Jobling. S. G . Kazarian, M. Poliakoff.
L. A Shepherd, K. P. Johnston. AnulI,.st 1993. 118. 1111.
K. E O'Shea. K. M. Kirmse. M. A. Fox. K. P. Johnston, J Phrs. Chrm. 1991.
9.7. 7863.
G C' Pimentrl. A . L . McClellan. The H.v~lrngerrBond. Freeman. San Francisco. 1960.
21) l i i t c i nio/(.cu/ur F(irccs: an Iirtrodurrion 10 Modern Methods und Re.wlt.s
(Eds.: P L Huyskens. W. A. P. Luck, T. Zeegers-Huyskens). Springer. Berlin. 1991. b) A. V. logansen. G. A. Kurkchi. V. M. Furman. V. P. GbdZuIIOV.
S E. Odinokov. Zh. Prikl. Spemosk. 1980. 33. 460.
Y . Xigi, S. Saito. H. Inomata. J. Chmr. Eng. Jpn. 1993, 26, 116.
J. Via. L. T. Taylor. F. K. Schweighardt, A n d . Cbem. 1994, 66, 1459.
J. L. Fulton. G. G. Yee, R. D. Smith. J. Am. Chem. Soc. 1991. 113, 8327.
J. L Fulton. G . G. Yee, R. D. Smith. ACS Sjr77p. Ser. 1993, 541, 175.
R. B. Gupta. J. R. Combes. K. P. Johnston, J. Pbjs. C/reni. 1993. 97, 707.
S. <i. Kanlrian, P. A. Hamley. M. Poliakoff. J. Am. Cheni. Soc. 1993, 115,
9069
1. M. DeSimone. Z Gum, C. S. Elsbernd. Sric,nce 1992. 257. 945.
11) K. E. Laintr, C . M. Wai, C. R. Yonker. R. D. Smith, J. Superrrir. Fluids
1991. 4. 194: b) Y. Lin. R. D. Brauer. K. E. Laintz. C. M. Wdi. Anal. Chin.
1993. 65. 2549.
S. ti. Kararian, R. G. Gupta, M. J. Clarke, K. P. Johnston, M. Poliakoff. J
A m . Chciir S i r . 1993, 115. 11099.
H. Inomata. Y Yagi. M. Saito. S. Saito, J. Supercrit. Nuids 1993. 6, 237.
V. K. Popov. J. A. Banister. V. N. Bagratashvili. S. M. Howdle, M. Poliakoff,
J Sirprwrit. Fhrids 1994. 7. 69.
H. Ochel. Ci M. Sohneider, Bm. Bunsrnge.r. PhI,.s.
Cbem. 1994. YN. 610.
S. G. Kararian, M. Jobling. M. Poliakoff, Mendelrev Coininun. 1993. 148.
B. V. Lokshin. S. G. Kazarian. A. G. Ginzburg. k v . Akuci. Nuuk SSSR Scr.
Khirii. 1988. 333.
C. A. Eckert. D. Suleiman, C. L. Liotta, D. L. Bodtwright. unpublished.
a ) R. D. Smith. J. P. Blitz. J. L. Fulton. ACS Symp. Ser. 1989, 406. 1 ; b) J. P.
Blitz. J. L. Fulton. R. D. Smith, Appl. Spectrosr. 1989. 43.812; c) G. G. Yee.
J. L. Fulton. R. D. Smith. Luugmuir 1992, 8. 377.
R. D. Smith. J L. Fulton. J. P. Blitz, J. M. Tingey. J. PIrjs. Cliern. 1990, 94,
781.
J. L. Fulton. J P. Blitz, J M . Tingey, R. D. Smith. J. Plrys. Cbrrn. 1989, Y3,
4198.
G. J. McFann. K. P. Johnston. S. M. Howdle, AIChE J. 1994, 40, 543.
a ) K. D. Bartle. A. A. Clifford. S. B. Hawthorne. J. 1. Lagenfield. D. J. Miller.
R. Robinson. J Sirparerif. F/uul.s 1990.3, 143; b) K. D. Bartle. T. Boddington,
A. A. C'liffbrd. N. J. Cotton, C. J. Dowle, A n d . Cliern. 199L63.2371; c) K . D.
Bartle, T. Boddington. A. A. Clifford. S. B. Hawthorne. J Suprrcrrl. Fluids
1992. 5. 207.
S M . Howdle. J. M. Ramsay. A. I. Cooper. J. Pofym Sci. Purt B ' Po/yni.
P / i n 1994. 32. 541.
B. J. Briscoe. S. Zakaria. Pulvrirer 1990, 31, 440.
M. Jobling. S. M. Howdle. M. Poliakoff, J. Cbem. Soc. Chem. Comniun. 1990.
1762
R. G Ball. W. A. G. Graham, D. M. Heinekey. J. K. Hoyano, A. D. McMaster. B. M. Mattson. S. T. Michel, Inorg. Chein. 1990, 29. 2023.
L r t f .1993,206. 175.
A I . Cooper. S. G. Kazarian, M . Poliakoff, Chem. Phy~.
:i) A. I . Rest. J Mol. Sfrucl. 1990. 222. 87; b) A. K . Campen, A. J. Rest, K .
Yoshihara, J Phorochein. Phorohiol. A 1991. 55, 301.
E. V. Sullenberger. S. F. Dressman, A. C. Michael. J. Phys. Cbeni. 1994, 98.
5347.
J. 1. Turner in Pho/oprocersesin Transilion Merul Cornp/c,w.v,Bioswtems und
otirrr SI,\rmi.\ (Ed.: E. Kocbanski), Kluwer, Dordrecht. 1992. p. 113.
R N . Perulz. Chem. Soc. Rev. 1993. 22. 361.
R K. Upmacis. M. Poliakoff, J. J. Turner, J. Arn. Cheni. Soc. 1986, 108, 3645.
Angcw. Cbt~ni.Int. Ed. Engl. 1995, 34. 1275-1295
REVIEWS
1881 a) R. Whyman. J Orgunornet. Chern. 1973. 56, 339: b) F. L'Eplattenier, F.
CdlderdZZo, InOrg. Cbein. 1968. 7,1290.
[89] T,".srrion M m l Hydrides (Ed.: A Dedieu). VCH. Ne* York, 1992.
[YO] For a recent review o n silane compexes: U . Schubert, A h . O F ~ U I U JCbem.
~C~.
1990. 30. 151.
[91] a ) A. A. Gonrdler. K. Zhang. S. P. Nolan. R. L. Vega. S. L. Mukherjee, C. D.
Hoff. G. J. Kubas. 0r~unornedlic.r1988. 7, 2429; b) A. A. Gonzalez. C. D.
Hoff, Inorg. Chcm. 1989, 2X. 4295.
[92] a) J. R. Wells, E. Weitr, J. Am. C/iem. Sor. 1992. / l 4 . 2783: b) B. H. Weiller,
ACS Symp. Srr. 1993. 530, 164.
(931 a) M . Poliakoff, S. M . Howdle, M. Jobling. M. W. George. Secondlrir. Sjinp.
Supererr/. Fluid.s (Ed.: M. A. McHugh). Johns Hopkins University. Baltimore. 1991, p. 189; b) M. W. George. M. Poliakoff, J. J. Turner. Anulysr 1994,
f19,551; c ) S. G. Kazarian. R. B. Gupta. K. P. Johnston. M. J. Clarke, M. W.
George, M. Poliakoff, Third Int. Svinp. Supercril. Fluuk, Strasbourg, France
1994, 1. 343.
[94] a) J. W. Rathke. R. J. Klingler, T. R. Krause. Orgunomr,tu/lrcs 1991, 10, 1350;
US-A 5 198 589.
b) ihid. 1992. 1 1 , 5 8 5 ; c) J. W. Rathke. R. J. Klingler .firm..
1994; d) R. J. Klingler. J. W. Rathke, J Am. Chrni. Soc. 1994. 116. 4772.
1951 P. G. Jessop. T. Ikariya. R. Noyori. Narure /London) 1994, 368. 231
[96] a) S. M. Howdle, P. Grebenik. R. N. Perutz, M. Poliakoff. J. Cbem. So<.
Clirrn. Corirmurz. 1989. 1517; b) J. A. Banister, M . W. George. S. Grubert.
S. M. Howdle, M. Jobling, F. P. A. Johnston. S. L. Morrison, M. Poliakoff,
U . Schubert. J. R. Westwell, J Orgunoma. Chrm., 1994. 484, 129.
[97] An exception, for example. is the reaction of hex-3-yne with scC0, catalyzed
by a nickel complex: M. T. Reetz, W. Konen. T. Strdck. Chrmiu 1993, 47,493.
[98] M. Jobling, S. M. Howdle, M. A . Hedly, M. Poliakoff. J Chem. So?. Chem.
Commim. 1990. 1287.
[99] M . B. Sponsler. B. H. Weiller, P. 0. Stoutland, R. G. Bergman. J. An!. Chem.
Soc. 1989, f/1, 6841.
[loo] A. 1. Cooper, M. Poliakoff, unpublished.
[ l o l l a ) Y. Ikushima, N . Saito, M . Arai, J. Phjs. Chein. 1992. 96. 2293: b) Bull.
Chem. Soc. Jpn. 1991, 64. 282.
11021 a) ..Drucksachen; ein Rundgdng durch das Druck- und Hydrierlabor Roche
Kaiseraugst", Rorhe Muguzin 1992, 41, 2: b) K. H. Pickel, K. Steiner. Third
In/.S y n p . Supercrit. Fluids, Strasbourg, France, 1994. 3. 25.
(1031 J. A. Banister. S. M. Howdle. M. Poliakoff. J. Cbem. SOL..Chem. Commun.
1993. 1814.
[104] a) M. E Gregory, S. A. Jackson. M. Poliakoff. J. J. Turner, J. Chrm.
Sor. Chein. Conimun. 1986, 1175; b) F.-W. Grevels. J. Jacke. S. Ozkar, J. Am.
Chein. Soc. 1987. 109. 7536; c) B. H . Weiller, E. R. Grunt. ihicl. 1987, 109.
1252.
11051 S. M. Howdle, M. Poliakoff, J. Chem. Soc. C/rern. Comniun. 1989. 1099.
I1061 M. Poliakoff, J. A. Banister. P. D. Lee in XVIrh Orgunornet. Cbrrn. Con/:
1994. U . K ; Orgunomrrullics, in press.
[107] a)S. V. Kamat, B. Iwaskewycz. E. J. Beckman, A. J. Russell. Proc. Nut/.Acud.
Srr. U S A 1993.90,2940; b) 0. Aaltonen, M . Rantakyla. CHEMTECH 1991,
240; c) A. J. Russell. E. J. Beckman, A. K . Chaudhary. hid. 1994, 33.
[lo81 B. N. Hansen. 8. M. Hybertson, R. M. Barkley, R. E Sievers. Chem. Muter.
1992. 4, 749.
[lo91 V. K. Popov. V. N. Bagratashvili. S. M. Howdle, 0 A. Louchev. E. A. Antonov, G. V. Mishakov. Third In/.Symp. Supercrit.Fluids. Strasbourg. France.
1994. 3, 369.
[I101 L. T. Canham, A. G. Cullis. C. Pickering. 0. D. Dosser, T. I. Cox, T. P.
Lynch, Nature (London) 1994, 368, 133.
[ill] a) P. T. Wood, W. T. Pennington, J. W. Kolis, B. Wu, C. J. O C o n n o r , Inorg.
Cheni. 1993.32, 129; b) P. T. Wood, W. T. Pennington. J. M. Kolis, J. Chrm.
Sue. C/irm. Commun. 1993. 235.
[112] R. C. Reid, J. M. Prausnitz. B. E. Poling, The Propertie.7 of G r m s undLiquids.
4th ed.. McGraw-Hill. New York. 1986.
[113] See, for example: D. E. Raynie, Anal. Chem. 1993, 65. 3127.
1295
Документ
Категория
Без категории
Просмотров
4
Размер файла
2 604 Кб
Теги
polymer, hydrogen, bonding, spectroscopy, synthesis, supercritical, vibrations, analysis, fluid
1/--страниц
Пожаловаться на содержимое документа