A. Entrainer Effect in Mixtures

was further demonstrated for several quaternary systems, each consisting of a

supercritical fluid, a cosolvent, and a nonpolar and a polar solute (139).

Mechanistically, the entrainer effect has been explained in terms of a higher

than bulk population of the cosolvent molecules in the vicinity of the solute

molecule. It may be argued that the “clustering” of cosolvent molecules about a

solute is a consequence of the local density augmentation in supercritical fluid

solutions and that the observation of the entrainer effect is a precursor to the

solute-solute clustering concept. Specific solute–cosolvent interactions such as

hydrogen bonding may also play a significant role in the observed entrainer

effect in some systems (140).

Several investigations of supercritical fluid–cosolvent systems have focused on the effects of hydrogen bonding and the role of specific intermolecular

interactions in solubility enhancements. Walsh et al. used infrared absorption

results to show that the entrainer effect in supercritical fluid–cosolvent mixtures

is due to various types of hydrogen bonding interactions (141–143). Infrared

absorption spectra have also been employed to estimate the extent of hydrogen

bonding between solutes such as benzoic acid and salicylic acid and alcohol

cosolvent molecules in supercritical CO2 (144). Bennet et al. used a supercritical fluid chromatography technique to determine the solubilities of 17 solutes

in three supercritical fluids (ethane, CO2 , and fluoroform) with eight cosolvents

(145). Their results showed that solubility enhancements are present in the supercritical fluid–cosolvent mixtures and that the enhancements become more

significant at higher densities. More quantitatively, the solubility enhancement

observed for anthracene in an ethane-ethanol mixture was predominantly due to

the change in density that occurs on going from the neat fluid to the mixture.

However, for carbazole and 2-naphthol in the same mixture, the solubility enhancements were considerably higher than those predicted on the basis of the

density change, suggesting the involvement of specific intermolecular interactions (145). Ting et al. investigated the solubility of naproxen [(S)-6-methoxy-αmethyl-2-naphthaleneacetic acid] in supercritical CO2 –cosolvent mixtures (six

different polar cosolvents at concentrations up to 5.25 mol %) at different temperatures (146). The solubility enhancements differ for the various cosolvents—in

the order of increasing enhancement, ethyl acetate, acetone, methanol, ethanol,

2-propanol, and 1-propanol. For example, the solubility of naproxen in the supercritical CO2 -1-propanol (5.25 mol %) mixture at 125 bar and 333.1 K is

about 50 times higher than that in neat CO2 under the same conditions (146).

It was estimated that the density increase from neat CO2 to the mixtures could

account for 30–70% of the observed solubility enhancements at low cosolvent

concentrations (1.75 mol %) but be less significant at higher cosolvent concentrations. It was suggested that the observed solubility enhancements in the supercritical CO2 –cosolvent mixtures were consistent with a solute-solute clustering

mechanism and were also strongly influenced by hydrogen bonding interactions



Copyright 2002 by Marcel Dekker. All Rights Reserved.



(146,147). Foster and coworkers measured the solubility of hydroxybenzoic acid

in supercritical CO2 with 3.5 mol % methanol or acetone as a cosolvent and

found enhancements that were beyond the effects of the density increases from

neat CO2 to the mixtures (148). They attributed the solubility enhancements to

a higher local concentration of cosolvent molecules around the solute and even

estimated the local mixture compositions in terms of the experimental solubility

data.



Structure 6



Molecular spectroscopy methods have also been applied to the study of

the entrainer effect in supercritical fluid–cosolvent mixtures. Again, the molecular probes employed for absorption and fluorescence measurements include the

Kamlet–Taft π∗ polarity/polarizability scale probes (13,14), pyrene (15,16), and

TICT molecules (17).

Kim and Johnston used phenol blue as a probe to investigate the local

compositions of octane, acetone, ethanol, and methanol in CO2 at 35◦ C over

the entire bulk composition range (mole fraction from 0 to 1) (149). Their

results show that the cosolvent local concentrations calculated on the basis of

absorption spectral shifts are higher than the corresponding bulk concentrations

over the entire mixture composition range. In addition, the local concentration

enhancement is more significant at low cosolvent mole fractions, although the

absolute local concentration increases with increasing bulk concentration of the

cosolvent.

Nitroanisoles have achieved popularity as probes in the study of supercritical fluid–cosolvent mixtures (150–153). For example, Yonker and Smith used

2-nitroanisole to determine local concentrations of the cosolvent 2-propanol in

supercritical CO2 at different temperatures (150). Their results are similar to

those of Kim and Johnston (149); the difference between the local and bulk

cosolvent concentrations is more significant at low pressures and decreases with

increasing pressure, approaching the bulk concentration at high pressures (Figure 18) (150). Also, results obtained in supercritical CO2 with methanol and

tetrahydrofuran (THF) as cosolvents are similar (151,152). Eckert and coworkers

investigated supercritical ethane with several cosolvents using the solvatochromatic shifts of 4-nitroanisole and 4-nitrophenol (153). When the cosolvent is

basic, the spectral shifts of 4-nitrophenol are larger than those of 4-nitroanisole



Copyright 2002 by Marcel Dekker. All Rights Reserved.



Figure 18 Local composition vs. pressure for constant temperature at 62◦ C and 122◦ C

at (ᮀ) 0.051, (×) 0.106, and (᭿) 0.132 bulk mole fraction compositions. (From Ref. 150.)



Copyright 2002 by Marcel Dekker. All Rights Reserved.



because 4-nitrophenol can participate in hydrogen bonding. In addition, for 4nitrophenol in the supercritical ethane–basic cosolvent mixtures, the spectral

shifts correlate well with the Kamlet–Taft solvent basicity parameters (153).

Many other probes have been used to study supercritical fluid–cosolvent

mixtures, including the charge transfer complexes FeII (1,10-phenanthroline)3 2+

and FeIII (2,4-pentadionate)3 (for CO2 -methanol mixtures) (154), Nile red dye

(for Freon-13, Freon-23, and CO2 with the cosolvents methanol, THF, acetonitrile, and dichloromethane) (155), benzophenone (for ethane with the cosolvents

2,2,2-trifluoroethanol, ethanol, chloroform, propionitrile, 1,2-dibromoethane, and

1,1,1-trichloroethane) (156), 4-amino-N -methylphthalimide (for CO2 –2-propanol

mixtures) (157), and other molecular probes such as 2-naphthol, 5-cyano-2naphthol, and 7-azaindole for a variety of supercritical fluid–cosolvent mixtures

(158,159).

As expected, pyrene has also been used to characterize supercritical fluid–

cosolvent mixtures. For example, Zagrobelny and Bright used the Py polarity

scale and pyrene excimer formation to study supercritical CO2 –methanol and

CO2 –acetonitrile mixtures (160). Their results suggest the clustering of cosolvent

molecules around pyrene. Similarly, Brennecke and coworkers measured Py

values in CO2 , CHF3 , and CO2 -CHF3 mixtures (43).

TICT molecules are also excellent probes for the study of supercritical

fluid–cosolvent mixtures. Sun et al. carried out a systematic investigation of supercritical CO2 -CHF3 mixtures using DMABN and DMAEB as probes (63,161).

In their experiments, shifts of the LE and TICT emission bands and TICT emission fractional contributions were determined for the probe molecules in the neat

fluids and mixtures of various CHF3 compositions (6% and 11%). The data indicate that the solute is preferentially solvated by the polar component CHF3 in

the mixtures. The preferential solvation can be observed for pyrene in the same

supercritical fluid mixtures, according to Brennecke and coworkers (43). The

results of Sun et al. also suggest that the local composition effect is more significant at lower reduced densities (161). In another experiment, DMABN was

used by Sun and Fox to determine the microscopic solvation effects in CO2 -THF

and CHF3 -hexane mixtures (162). Schulte and Kauffman have also used TICT

molecules [bis(aminophenyl)sulfone and bis(4,4 -dimethylaminophenyl)sulfone]

to characterize supercritical CO2 -ethanol mixtures (163,164). Their results, based

on the shifts of the LE and TICT emission bands, suggest that the local ethanol

concentrations are an order of magnitude higher than the bulk concentrations.

Dillow et al. investigated the tautomeric equilibrium of the Schiff base 4(methoxy)-1-(N -phenylforminidoyl)-2-naphthol in supercritical ethane with acetone, chloroform, dimethylacetamide, ethanol, 2,2,2-trifluoroethanol, and 1,1,1,

3,3,3-hexafluoro-2-propanol as cosolvents (165). Their results show that the polar cosolvents acetone, chloroform, and dimethylacetamide have little effect on

the keto-enol equilibrium but that the cosolvents capable of hydrogen bonding



Copyright 2002 by Marcel Dekker. All Rights Reserved.



Scheme 3



4-(Methoxy)-1-(N -phenylforminidoyl)-2-naphthol.



shift the equilibrium toward the keto tautomer. For ethanol and trifluoroethanol

as cosolvents, the equilibrium was found to shift back toward the enol form with

increasing density. It was also found that the position of the keto-enol equilibrium in the near-critical region of the solvent was more toward the keto form

than what would be predicted on the basis of the bulk cosolvent concentration. It

was concluded that the clustering of cosolvent molecules about the Schiff base

was responsible for these results. (See Scheme 3.)

B. Bimolecular Reactions

Studies of the entrainer effect discussed above demonstrate that the solute in supercritical fluid–cosolvent mixtures is, in many cases, surrounded preferentially

by the cosolvent molecules. Since the cosolvent may be regarded as a second

solute, the solute–cosolvent clustering may be considered as a special case of

solute-solute clustering. An important consequence of the entrainer effect is enhancement in solute–cosolvent interactions or reactions. Similarly, solute-solute

clustering in supercritical fluid solutions may enhance bimolecular reactions

between the solute molecules. Extensive investigation of the solute-solute clustering phenomenon by many research groups has been prompted by the prospect

of being able to influence bimolecular interactions and reactions under supercritical fluid conditions and, as a result, increase reaction yields and alter product

distributions. Spectroscopic and other instrumental techniques combined with

molecular probes that undergo well-characterized bimolecular processes or reactions (such as the formation of an excimer or exciplex, photodimerization, and

fluorescence quenching) have been used.



Copyright 2002 by Marcel Dekker. All Rights Reserved.



Scheme 4



1. Excimer and Exciplex

Formation of pyrene excimer (a complex between a photoexcited and a groundstate pyrene molecule; Scheme 4) is an extensively characterized and wellunderstood bimolecular process (35). Because the process is known to be diffusion controlled in normal liquid solutions, it serves as a relatively simple model

system for studying solvent effects on bimolecular reactions. In fact, it has been

widely employed in the probing of the solute-solute clustering in supercritical

fluid solutions (40–42,46,47,160,166–168). (See Scheme 4.)

Eckert’s group was the first to report pyrene-excimer formation in supercritical fluids at pyrene concentrations significantly below those required in

normal liquid solutions (Figure 19) (40,41). Taking into account the difference

in viscosity and molecular diffusion in supercritical CO2 (150 bar and 35◦ C) as

opposed to normal liquid cyclohexane, they concluded that the observed yield

for excimer formation in CO2 exceeded what might be expected from the higher



Figure 19



Excimer formation in dilute supercritical fluid solutions. (From Ref. 40.)



Copyright 2002 by Marcel Dekker. All Rights Reserved.



diffusivity. Thus, enhanced solute–solute interactions in a supercritical fluid became a possibility. According to Eckert, Brennecke, and coworkers (40,41,166),

similarly efficient pyrene-excimer formation takes place in nonpolar and polar

supercritical fluids such as ethylene and fluoroform.

Bright and coworkers investigated pyrene-excimer formation in supercritical fluids from a somewhat different angle using not only steady-state but

also time-resolved fluorescence techniques (47,167). They measured fluorescence lifetimes of the pyrene monomer and excimer at a pyrene concentration

of 100 µM in supercritical ethane, CO2 , and fluoroform at reduced densities

higher than 0.8. Since the kinetics for pyrene-excimer formation was found to

be diffusion controlled in ethane and CO2 and less than diffusion controlled

in fluoroform, they concluded that there was no evidence for enhanced pyrene–

pyrene interactions in supercritical fluids. The less efficient excimer formation in

fluoroform was discussed in terms of the influence of solute–solvent clustering

on excimer lifetime and stability. Experimentally, their fluorescence measurements were influenced by extreme inner-filter (self-absorption) effects due to

the high pyrene concentration in the supercritical fluid solutions (35).

Sun and Bunker performed a more quantitative analysis of the photophysical results of pyrene in supercritical CO2 (46). In their experiments absolute and

relative fluorescence quantum yields of the pyrene monomer and excimer were

determined in supercritical CO2 at 35◦ C and 50◦ C over the CO2 reduced-density

range of about 0.5–2 (Figures 20 and 21). Although the pyrene concentrations

were between 2 × 10−6 and 7 × 10−5 M in these supercritical CO2 solutions,

significant pyrene excimer fluorescence was observed. In an attempt to quantitatively model the experimental results in terms of the classical photophysical

mechanism established for pyrene in normal liquid solutions, they found that

the results deviate significantly from the classical mechanism. The disagreement

could be reconciled by replacing the pyrene concentration in the photophysical

model with a local pyrene concentration (the actual concentration of ground-state

pyrene molecules in the vicinity of a photoexcited pyrene molecule). In the sense

that the local concentration of pyrene is higher than the bulk concentration—

up to a factor of 9, assuming diffusion-controlled conditions—pyrene-pyrene

clustering enhances excimer formation in supercritical CO2 (Figure 22) (46).

An excimer is a special case of exciplex—a complex between an excitedstate molecule and a ground-state molecule, where the two molecules have

different identities. Exciplex formation has been used as a model bimolecular

process in the study of solute-solute clustering in supercritical fluid solutions.

Brennecke et al. reported the investigation of naphthalene-triethylamine exciplex

formation in supercritical CO2 at 35◦ C and 50◦ C (166). Their results show that

the exciplex emission can be observed, even at low triethylamine concentrations

(5 × 10−3 –5 × 10−2 M). Similarly, Inomata et al. investigated the formation

of pyrene-dimethylaniline excimer in supercritical CO2 at 45◦ C (169). They



Copyright 2002 by Marcel Dekker. All Rights Reserved.



Figure 20 Fluorescence quantum yields of pyrene in supercritical CO2 (35◦ C) at

concentrations of 2 × 10−6 M (᭺) and 6 × 10−5 M (total, ᮀ: monomer, ᭝: and excimer,

᭞) as a function of CO2 reduced densities. (From Ref. 46.)



Figure 21 Ratios of pyrene excimer and monomer fluorescence quantum yields as a

function of CO2 reduced densities at 35◦ C (6.2 × 10−5 M, ᭺) and 50◦ C (5.9 × 10−5

and 6.8 × 10−5 M, ᮀ). (From Ref. 46.)



Copyright 2002 by Marcel Dekker. All Rights Reserved.



Figure 22 Ratios of the local and bulk pyrene concentrations as a function of CO2

reduced densities at 35◦ C (2.8 × 10−5 M, ᭝; 6.2 × 10−5 M, ᭺) and 50◦ C (5.9 × 10−5

and 6.8 × 10−5 M, ᮀ). (From Ref. 46.)



found unusually efficient exciplex formation and attributed the enhancement to

preferential clustering of dimethylaniline molecules about pyrene.

Molecules capable of forming an intramolecular exciplex have also been

used in the probing of solute-solute clustering in supercritical fluid solutions

(170–172). These systems are fundamentally different from their intermolecular

counterparts because intramolecular exciplex formation is independent of both

bulk and local concentration as a result of the two participating pieces of the

complex being linked by a tether. Okada et al. investigated the intramolecular exciplex formation of p-(N ,N -dimethylaminophenyl)-(CH2 )2 -9-anthryl (DMAPA)

in supercritical ethylene and fluoroform at 30◦ C (170). No exciplex formation

was observed in the nonpolar fluid ethylene; however, in supercritical fluoroform

two emission bands (normal and exciplex) were detected. Similarly, Rice et al.

investigated the intramolecular excimer formation of 1,3-bis(1-pyrenyl)propane

in supercritical ethane and fluoroform (171). They found that the ratio of excimer emission to monomer emission increases with increasing fluid density

and that the excimer formation is at least partially dynamic in nature. Quantitative interpretation of their results was complicated by the existence of multiple

ground-state species of the probe at all fluid densities.



Copyright 2002 by Marcel Dekker. All Rights Reserved.



Structure 7



Rollins et al. investigated the intramolecular excimer formation of 1,3di(2-naphthyl)propane in supercritical CO2 (172) and compared the results with

intermolecular pyrene-excimer formation recorded under similar conditions (46).

Their results show that the ratio of excimer emission to monomer emission

decreases gradually with increasing CO2 density (Figure 23), in a pattern that

agrees well with that predicted from viscosity changes in terms of the classical

photophysical model for excimer formation (35). In a comparison of 1,3-di(2naphthyl)propane and pyrene in the same fluid, the ratio of excimer emission to



Figure 23 The FD / FM ratios (normalized at the reduced density of 1.9) for the intramolecular excimer formation in 1,3-di(2-naphthyl)propane (᭺) and the intermolecular

excimer formation in pyrene [ᮀ] in supercritical CO2 at 40◦ C. (From Ref. 172.)



Copyright 2002 by Marcel Dekker. All Rights Reserved.



monomer emission is considerably less sensitive to changes in fluid density for

the tethered system, which seems to support the conclusion that the formation

of intermolecular pyrene excimer is affected by solute-solute clustering.

2. Photodimerization

Photodimerization reactions in supercritical fluid solutions have been used to

probe the effects of possible solute-solute clustering. Kimura et al. investigated

the dimerization of 2-methyl-2-nitrosopropane in CO2 , chlorotrifluoromethane,

fluoroform, argon, and xenon (173–176). Their results show that the density

dependence of the dimerization equilibrium constant is rather complex, probably

due to the existence of various dimerization mechanisms in different density

regions.

Hrnjez et al. evaluated the product distribution of the photodimerization

of isophorone in supercritical fluoroform and CO2 (177). The reaction typically

produces a mixture of various regioisomers and stereoisomers. Relative yields

of the regioisomers are fluid density dependent in the polar fluid fluoroform

but exhibit little or no change with fluid density in CO2 . On the other hand,

relative yields of the stereoisomers are affected by changes in the fluid density

in both fluoroform and CO2 . The results were discussed in terms of solvation

and various degrees of solvent reorganization required for the various products.

(See Scheme 5.)

Tsugane et al. used Fourier transform infrared absorption spectroscopy to

investigate the dimerization reaction of benzoic acid in saturated supercritical



Scheme 5



Copyright 2002 by Marcel Dekker. All Rights Reserved.