Complexation of Phosphine Ligands with Peracetylated β

Feb 17, 2007 - Alessandro Galia,*,† Edward C. Navarre,† Onofrio Scialdone,† Michel Ferreira,‡. Giuseppe Filardo,† Sebastien Tilloy,‡ and E...
0 downloads 0 Views 133KB Size
J. Phys. Chem. B 2007, 111, 2573-2578

2573

Complexation of Phosphine Ligands with Peracetylated β-Cyclodextrin in Supercritical Carbon Dioxide: Spectroscopic Determination of Equilibrium Constants Alessandro Galia,*,† Edward C. Navarre,† Onofrio Scialdone,† Michel Ferreira,‡ Giuseppe Filardo,† Sebastien Tilloy,‡ and Eric Monflier‡ Dipartimento Ingegneria Chimica Processi e Materiali, UniVersita` di Palermo, Viale delle Scienze Ed. 6, 90128 Palermo, Italy, and Laboratoire de Physico-Chimie des Interfaces et Applications, UniVersite´ d’Artois, FRE CNRS 2485, Faculte´ des Sciences Jean Perrin, Rue Jean SouVraz, SP 18-62307 Lens Ce´ dex, France ReceiVed: NoVember 6, 2006; In Final Form: January 17, 2007

The interaction between peracetylated β-cyclodextrin and several triphenyl phosphine derivatives was studied in supercritical carbon dioxide (scCO2) by UV-visible spectroscopy. The equilibrium constant for a 1:1 complexation reaction was obtained from titration spectra and calculated using two established mathematical models. The values of the equilibrium constants are 1-3 orders of magnitude smaller than those obtained in aqueous solution with analogous phosphines. This is likely due to the absence in scCO2 of the hydrophobic effect, which is replaced by a corresponding, but weaker, CO2-phobic effect. The largest value of Kf was found for complexes of diphenyl(4-adamantylphenyl)phosphine, which is rationalized on the basis of the excellent fit of the phosphine in the cyclodextrin cavity, leading to enhanced host-guest van der Waals interactions. This study can be considered the first step toward the comprehension of the complexation thermodynamics of modified cyclodextrins soluble in scCO2.

Introduction Supercritical carbon dioxide (scCO2) is an appealing candidate for the replacement of organic solvents in chemical processing,1 as it offers favorable chemicophysical and technical aspects, such as nonflammability, low toxicity, large availability, low cost, absence of a gas/liquid-phase boundary, tunable solvent strength with changing density, and simplicity in workup procedures. However, the use of scCO2 as a reaction medium is often restricted by the poor solubility of substrates or catalysts in this unconventional solvent.2 In particular, the use of arylphosphine catalysts, such as triphenylphosphine and its derivatives, is not practicable at an industrial scale due to their limited solubility in scCO2.3 Several approaches have been proposed to solve this major problem, such as the modification of ligands of the catalytic systems with CO2-philic groups, such as long perfluoroalkyl chains4-6 or carbonyl groups,7,8 or the adoption of suitable surfactants that can stabilize insoluble compounds either directly or dissolved in an aqueous phase in scCO2 microemulsions.9,10 All these strategies are conditional on the fact that, to date, effective CO2-philic compounds are predominantly fluorine-based and, for this reason, quite expensive and difficult to apply to the industrial scale. Naturally occurring β-cyclodextrins (β-CD) are water-soluble, truncated, cone-shaped molecules with a tapered cavity 7.9 Å deep with top and bottom diameters of 6.0 and 6.5 Å. Together with their R- and γ- homologues, they are well-known to possess a hydrophobic cavity capable of forming inclusion complexes with many organic molecules that are otherwise poorly soluble in water. This property has been studied in aqueous systems, and it is commonly accepted that the principal factors involved * Corresponding author. Phone: 39-091-6567258. Fax: 39-091-6567280. E-mail: [email protected]. † Universita ` di Palermo. ‡ Universite ´ d’Artois.

in binding are van der Waals and hydrophobic interactions between the host β-CD cavity and the guest molecule, even if hydrogen bonding and steric effects may be present.11 The capability of forming host-guest supramolecular complexes with CDs has been used in hybrid hydrocarbon/water systems to dramatically enhance the rate of mass transfer-controlled chemical reactions.12,13 It has been recently reported that peracetylated cyclodextrins exhibit a high degree of miscibility with dense CO2 over a broad range of concentration.14,15 In previous work, we have demonstrated that β-CD derivatives can form inclusion complexes with sulfonated arylphosphines in water.16-18 By analogy, we have assumed that a similar phenomenon should occur with arylphosphines in scCO2 using highly CO2-soluble modified β-CD. This hypothesis was tested by studying the solubility of diphenyl(4-phenylphenyl)phosphine (DBP) in scCO2 both in the presence of and without peracetyl-β-cyclodextrin (per-Ac-β-CD).19 Quite interestingly, by gravimetric measurements, it was found that the solubility of DBP was increased ∼4.5 fold in scCO2 medium (40 °C, 35.2 MPa, nominal CO2 density ) 0.9 g/mL) by using per-Ac-β-CD as a solubilizer. This solubility increase was attributed to host-guest interactions in scCO2 because the phosphine was recovered in the form of an inclusion complex as demonstrated by DSC and diffuse reflectance UV spectroscopy experiments. These preliminary results suggest further study of β-CD supramolecular chemistry in scCO2 and especially the host-guest interactions in scCO2 that until now have remained a largely unexplored field. The fundamental importance of this research is that the solvent power of scCO2 can be tuned, thereby allowing the investigator to better study the role of host-solvent and guest-solvent interactions, thus giving a useful contribution to the better comprehension of the solvent effect in molecular recognition phenomena. Moreover, this research can be highly valuable for applications, given that, in principle, using scCO2-soluble CDs, it should be possible to

10.1021/jp0673112 CCC: $37.00 © 2007 American Chemical Society Published on Web 02/17/2007

2574 J. Phys. Chem. B, Vol. 111, No. 10, 2007

Galia et al.

solubilize in this unconventional solvent many species that in native form are poorly soluble in the fluid phase, thus opening a new field of application in both separations and reactions. Experimental Materials. Peracetyl β-cyclodextrin [(C12H16O8)7; Mw ) 2017.8 g/mol] purchased from Aldrich Chemicals and carbon dioxide (99.998% purity) purchased from Air Liquide were used without further purification. Diphenyl(4-phenylphenyl)phosphine (DBP) (Mw ) 338.4 g/mol) and diphenyl[4-(4-tert-butylphenyl)phenyl]phosphine (DtBBP) (Mw ) 394.5 g/mol) were prepared as reported in the literature.19,20 Diphenyl(4-adamantylphenyl)phosphine (AdTPP) (Mw ) 396.5 g/mol) was prepared from 1-(4-bromophenyl)adamantane and chlorodiphenylphosphine. The 1-(4-bromophenyl)adamantane was obtained on a multigram scale by a modified published procedure.21 The 1-bromoadamantane (11.5 g, 53.4 mmol) was dissolved in 700 mL of bromobenzene. After cooling to 0 °C with an ice-water bath, aluminum trichloride (0.79 g, 11% molar compared to 1-bromoadamantane) was introduced under nitrogen. The reaction mixture was stirred for 15 min at 0 °C under a nitrogen atmosphere and then was washed with a saturated solution of potassium carbonate up to neutral pH. The bromobenzene was then evaporated under vacuum. The residue was purified on silica gel with hexane to give 1-(4-bromophenyl)adamantane (yield: 45%) as a white solid.

1H NMR: (CDCl ) δ 1.78 (6H, m, H ), 1.90 (6H, d, H ), 3 1 3 2.12 (3H, m, H2), 7.45 (4H, dd, 3JH-H ) 9 Hz, H6,7). 13C{1H} NMR: (CDCl3) δ 29.26 (s, C2), 36.64 (s, C4), 37.07 (s, C3), 43.45 (s, C1), 119.65 (s, C5), 127.23 (s, C6), 131.49 (s, C7), 150.75 (s, C8). In order to synthesize the AdTPP, a solution of 1-(4bromophenyl)adamantane (0.99 g, 3.4 mmol, 0.33 equiv) in 10 mL of anhydrous THF was added, under nitrogen, to a suspension of magnesium (0.46 g, 20 mmol, 1.2 equiv). After a few minutes, the reaction began, and 1-(4-bromophenyl)adamantane (3.94 g, 13.6 mmol, 0.66 equiv) in 40 mL of THF was added dropwise. The reaction mixture was then heated under reflux for 1 h. After cooling, chlorodiphenylphosphine (3.97 g, 18 mmol, ∼1 equiv) in THF (5 mL) was added dropwise and then heated under reflux for 1 h. Once the reaction was complete, the mixture was poured into a mixture of ice and water. The aqueous phase was extracted with toluene (3 × 10 mL), and the combined organic layers were dried over anhydrous MgSO4, filtered, and concentrated by rotary evaporation. The resulting oil was recrystallized from methanol to give white crystals (yield: 47%).

1H NMR: (CDCl ) δ 1.78 (6H, m, H ), 1.90 (6H, d, H ), 3 1 3 2.12 (3H, m, H2), 7.30 (14H, m, H6,7,10,11,12). 13C-{1H} NMR:

Figure 1. Schematic of experimental apparatus. LS: light source, F.O.: fiber optic, M: stirring motor, H: forced-air heater, RV: reaction vessel, S: dual channel spectrometer, PC: data collection computer.

(CDCl3) δ 29.26 (s, C2), 36.64 (s, C4), 37.07 (s, C3), 43.45 (s, C1), 125.55 (d, 3JP-C ) 7.5 Hz, C6), 128.83 (d, 3JP-C ) 6.8 Hz, C11), 128.97 (s, C12), 133.20 (d, 1JP-C ) 12 Hz, C8), 134.06 (d, 2JP-C ) 19.6 Hz, C7), 134.11 (d, C10, 2JP-C ) 18.8 Hz), 138.04 (d, C9, 1JP-C ) 11.3 Hz), 152.50 (s, C5). 31P-{1H} NMR: (CDCl3) δ -5.13 (s). The 1H, 31P-{1H}, and 13C-{1H} NMR spectra were recorded at 300.13, 121.49, and 75.46 MHz, respectively, on a Bruker Avance 300 DPX instrument. Apparatus. The experiments were performed in a windowed stainless steel reaction vessel (Optek-Danulat) capable of operating at pressures up to 50 MPa and temperatures up to 240 °C. The reactor was fitted with a pressure transducer (Barksdale, model UPA3) powered by an in-house-built 15-V power supply and monitored by a digital voltmeter (model 360, Simpson). Temperature was measured with a K-type thermocouple (RS Components) with a resolution of 0.1 °C. A small magnetic stirring bar was placed into the reactor to promote mixing. The free volume of the vessel was determined to be 24.2 mL. The reaction vessel was placed in an insulated enclosure and heated by a forced-air heating element (model HGL 046, Stego Elektrotechnik) regulated by a temperature controller (model 2216e, Eurotherm). A simplified diagram of the apparatus is shown in Figure 1. Samples of the phosphines and per-Ac-β-CD were weighed and deposited into the reaction vessel as solids. The vessel was then purged with CO2 to remove air and charged to the required pressure with liquid CO2 using a high-pressure hand pump (model 750.1030.1, SITEC-Sieber Engineering AG). Operating conditions for the experiments were 30 ( 0.7 MPa and 40 ( 0.2 °C. An important aspect to be considered when spectroscopic techniques are performed in supercritical fluids is that the refractive index and solvation properties of this compressible medium change with the density; therefore, operative conditions must be selected to minimize undesirable effects, notably, fluctuations in solute molar absorptivity and bathochromic or hypsochromic shifts. Indeed, it must be emphasized that the compressibility of scCO2 becomes smaller the greater the distance from the critical point (T ) 31.1 °C and P ) 7.4 MPa). Under the conditions adopted in this study, the maximum interval for the values of density of CO2, estimated by the

Complexation of phosphine ligands with β-CD in scCO2

J. Phys. Chem. B, Vol. 111, No. 10, 2007 2575

Bender equation of state,22 was 0.911 ( 0.005 g/mL such that density-induced perturbations in the spectroscopic measurements are negligible. Absorbance Spectrometry. Absorbance measurements were made using an Avantes fiber-optic SD2000 dual spectrometer equipped with a DH2000 light source. All connections between components and optical fibers were made with SMA 905 connectors. Light from the deuterium lamp was split with a bifurcating 200-µm-diameter fiber optic into a probe channel and reference channel. The reference channel intensity was reduced with a variable attenuator (model FVA-UV, Ocean Optics) then connected to one of the spectrometers to monitor the lamp output. The probe channel was connected to the reaction vessel by an interface optic (model AF-SMA, OptekDanulat), which converts an SMA-905 connector to a 1-in. threaded connection on the reaction vessel. The interface optic collimates the input beam to an ∼10-mm-diameter spot inside the reactor and condenses the output beam to couple with the detector-side fiber optic. Optical windows on the reaction vessel are 20 mm in diameter and ∼20 mm thick, such that the path length for absorbance measurements is 5 mm. The output beam was then sent through a variable attenuator and to the second spectrometer. The signal from each spectrometer was recorded as transmitted intensity by the spectrometer manufacturer’s software (AVA-Soft version 5, Avantes) running on a PC. Absorbance spectra were calculated in GRAMS (version 7, ThermoGalactic) using a previously recorded blank spectrum of carbon dioxide at the typical operating conditions, 30 MPa and 40 °C. Mathematical Model. The mathematical model adopted to calculate the association constant (Kf) from UV-visible spectra is a rigorous treatment based on the general method for determining equilibrium constants of 1:1 complexes as described by Rose and Drago.23 This method was selected for its general validity and because it has been demonstrated in previous work19,24 that per-Ac-β-CD forms a 1:1 inclusion complex with the types of phosphines used in this study. Furthermore, the absorbances of the complexed and free phosphine are strongly overlapping, which precludes the use of simpler mathematical approaches.25 In dilute solution, the equilibrium constant of a 1:1 complexation reaction such as

per-Ac-β-CD + phosphine h complex

(1)

may be expressed in terms of the molar concentrations of the different species, since the activity coefficients can be assumed to be unity. Under these conditions, if C0P and C0CD are, respectively, the initial phosphine and initial per-Acβ-CD concentrations and Ccomplex is the equilibrium concentration of the complex, the equilibrium constant expression is

Kf )

Ccomplex (C0CD

- Ccomplex)(C0P - Ccomplex)

(2)

The concentration of the complex therefore can be found as one of the quadratic solutions of eq 2.

Ccomplex ) 1/2(Kf-1 + C0CD + C0P) 1/2x(Kf-1 + C0CD + C0P)2 - 4C0CDC0P (3) The method of calculating Kf from spectroscopic data can be briefly summarized as follows. Under dilute conditions, the Beer-Lambert law can be adopted to express the observed

absorbance, A, of the complexation equilibrium (eq 1)

A ) CD(C0CD - Ccomplex) + P(C0P - Ccomplex) + complexCcomplex (4) where CD, P, and complex are the molar absorptivities of the per-Ac-β-CD, phosphine, and inclusion complex, respectively. The path length (b ) 0.5 cm) has been omitted for clarity, but it is included as a multiplicand to all values of . It was verified experimentally that per-Ac-β-CD does not absorb light within the spectral range of the spectrometer. In this case, only the complex and the phosphine absorb in the spectral region of interest, and CD can be assumed to be zero in eq 4. Under this condition, the complex concentration can be expressed by the formula

Ccomplex )

A - PC0P complex - P

(5)

It must be noted that P can be determined as a function of wavelength from spectra recorded with the pure phosphine in scCO2 at the same temperature and pressure adopted to perform experiments with the complete system. Thus, eq 5 has only two unknowns: Ccomplex and complex. If eq 5 is substituted in the equilibrium constant expression (eq 2), the reciprocal association constant at each wavelength can be expressed as

Kf-1 )

C0PC0CD(complex - P) A - A0 - C0P - C0CD + ) complex - P A - A0 ∆ ∆A - C0P - C0CD + C0PC0CD (6) ∆ ∆A

where A0 ) PC0P is the absorbance of the phosphine alone, and A is the experimentally measured absorbance. The quantities ∆A and ∆ are defined as ∆A ) A - A0 and ∆ ) complex P which may be substituted into eq 5 to give ∆A ) ∆Ccomplex. Solutions of eq 6 for both Kf and complex at each wavelength were found by a least-squares approach searching the values of complex that minimize χ2: N

χ2 )

2 (∆Acalc - ∆Aexp ∑ i i ) i)1

(7)

is the experimenwhere N is the number of data points, ∆Aexp i tal absorbance difference for the ith spectral point, and ∆Acalc i is the analogous calculated quantity. The determination of values of Kf and complex that minimize χ2 was performed using a method that combines least-squares and an iterative minimumseeking routine as described elsewhere.26 In the same reference are also reported the equations to compute conditional and marginal standard deviation for the Kf values that give an indication of how well the model fits the data. Results from each sample and each wavelength within the spectral region demonstrating the greatest absorbance change were averaged. A second method of calculation, previously used with phosphine-β-CD inclusion complexes in aqueous solution, was employed to verify the results of the first method.27 This calculation assumes a 1:1 inclusion complex and expresses the observed complex molar absorptivity by rearrangement of eq 5, yielding

complex )

A + P(Ccomplex - C0P) Ccomplex

(8)

2576 J. Phys. Chem. B, Vol. 111, No. 10, 2007

Figure 2. Absorbance spectra of 0.10 mM DBP with varying ratios of per-Ac-β-CD in scCO2 at 40 °C and 30 MPa.

Figure 3. Absorbance spectra of 0.10 mM DtBBP with varying ratios of per-Ac-β-CD in scCO2 at 40 °C and 30 MPa.

where all variables are as before. For a chosen value of Kf, the equilibrium concentration of the complex (Ccomplex) is known from eq 3, and complex can be calculated from eq 8 for each pair of initial concentrations. The standard deviation of complex among the samples is minimized to obtain the value of Kf. In this method, integrated areas beneath the first derivative spectral curve are used in eq 8 for the value A. This approach thus calculates a single value of Kf over the selected spectral interval rather than individual values at each wavelength. Results The spectra shown in Figures 2, 3, and 4 exhibit a broad, asymmetrical absorbance with maxima at 270, 280, and 260 nm, respectively, for the phosphines DBP, DtBBP, and AdTPP. With increasing concentration of per-Ac-β-CD, a shoulder on the short-wavelength side of the spectrum appears and is attributed to the absorbance of the inclusion complex. This effect is most noticeable in the spectra from AdTPP (Figure 4), in which the shoulder exceeds the absorbance of the parent phosphine in three samples. The shape of the absorbance spectrum for DBP (Figure 2) is similar to spectra recorded in the solid phase by diffuse reflectance spectroscopy, although the absorbance maximum is blue-shifted 60 nm.19 Values of the association constant for the phosphine-perAc-β-CD complex are summarized in Table 1. The greater relative error in the calculation of Kf for DBP arises in large

Galia et al.

Figure 4. Absorbance spectra of 0.11 mM AdTPP with varying ratios of per-Ac-β-CD in scCO2 at 40 °C and 30 MPa.

part due to the close proximity of the absorbance maxima of the inclusion complex and the free phosphine, as is visible in Figure 2. As a result, it is possible to consider the Kf of DBP and DtBBP as equivalent. In both methods of calculation, only the 4-adamantyl-substituted phosphine demonstrated a significantly larger association constant. The association constants calculated with eqs 3 and 8 are slightly larger than those obtained from eq 6. This is more an artifact arising from the selection of wavelengths than a real difference in the value of the constants. In the spectra of all three compounds, the wavelength difference between complexed and free phosphine is small, thus calculations of Kf may include wavelengths with small differences in absorbance and small values of complex. The second method obtains its best fits from a portion of the derivative spectrum between two isosbestic points, such as those shown in Figure 5. Since one of these points will always fall at the absorbance maximum of the inclusion complex, the calculation is made on one-half of the absorbance peak, including data in which there is little difference in absorbance between free phosphine and the inclusion complex. In the spectra presented here, the overall result is a bias toward larger values of Kf. This is confirmed by the calculations based on eq 6, which show larger values of Kf and greater error at wavelengths away from the absorbance maximum of the inclusion complex. The two calculation methods behave differently in the presence of small differences in absorbance and spectral noise; thus, they are complementary for the determination of the Kf of phosphine-per-Ac-β-CD complexes. The second method (eq 3 and 8) includes wavelengths with a small ∆A and can be susceptible to overestimating Kf, especially in cases in which the isosbestic point is not clear, whereas it is possible to exclude these values in the first method (eq 6). Spectral noise is handled better by the second method, since it effectively removes random noise contributions by using an integrated absorbance value. The first method retains the noise contribution at each wavelength and can result in a greater relative error in some cases. For example, the relative error in the Kf of AdTPP is much less for the second than the first method. In the absence of noise, as in the case of a synthetic spectrum, both methods yield the same result, with a relative error in the range of 0.5%. The values of Kf determined in this study are approximately 2-3 orders of magnitude smaller than those of inclusion complexes of analogous sulfonated phosphines in aqueous solution.28 Similarly, the complexation equilibrium constants of

Complexation of phosphine ligands with β-CD in scCO2 TABLE 1: Association Constant (Kf, M-1) of Selected Phosphines with per-Ac-β-CD in Supercritical CO2 at 40 °C and 30 MPa compd

Kfa

χ2 spectral regiona

Kfb

spectral regionb

DBP 120 ( 30 0.1 227 - 237 nm 190 ( 30 220 - 234 nm DtBBP 100 ( 10 0.3 225 - 238 nm 190 ( 40 218 - 236 nm AdTPP 270 ( 40 0.1 232 - 241 nm 340 ( 30 222 - 236 nm a

Using method of Drago (eq 6). b Using an algorithmic method with eqs 3 and 8.

Figure 5. Derivative absorbance spectra of 0.11 mM AdTPP with varying ratios of per-Ac-β-CD in scCO2 at 40 °C and 30 MPa.

several adamantane derivatives with β-CD are 102 or even 103 M-1 greater than those shown in Table 1.29 On the other hand, 4-t-Bu-4-phenyl and 4-adamantyl groups are known to be wellaccommodated in the cavity of β-CD.30 Moreover, theoretical calculations have shown that the 4-phenylphenyl part of DBP fits tightly in the cavity of per-Ac-β-CD, leading to the formation of stable 1:1 inclusion complexes.31 In making these comparisons, it must be noted that the computational studies of DBP were performed neglecting solvent effects and that the other published data refer to aqueous systems. The other major difference between aqueous and scCO2 systems is the use of native β-CD versus per-Ac-β-CD. As estimated by computational studies,31 the height of per-Ac-βCD is notably greater than that of native β-CD (13.1 Å versus 7.9 Å). However, the part of the phosphine recognized by the CD is roughly linear. Thus, an increase in the CD height is expected to be favorable to the complexation of the guest. Consequently, the observed decrease in the association constant does not seem to be due to differences in the host molecule and may be attributed to the role of the solvent in the complexation thermodynamics. As previously reported, the most important contributions to the complexation thermodynamics of cyclodextrins in aqueous solutions are believed to originate from (a) penetration of the hydrophobic part of the guest molecule into the hydrophobic cyclodextrin cavity and (b) dehydration of the organic guest.11 The release of water molecules from the β-CD cavity into the bulk of the aqueous phase is also a significant contribution to the complexation thermodynamics, and it is one major difference with respect to the classical hydrophobic effect. One further aspect that must be considered is the conformational change or strain release of the cyclodextrin molecule upon complexation. Molecular recognition phenomena involve only noncovalent interactions, such as electrostatic (ion-ion, ion-dipole, dipoledipole, dipole-induced dipole, and higher-order terms), van der

J. Phys. Chem. B, Vol. 111, No. 10, 2007 2577 Waals, hydrophobic, hydrogen bonding, charge transfer, π-π stacking interactions, and steric effects. It is commonly accepted that complex formation is due not to a single weak interaction but rather to the simultaneous cooperation of several of them. From these concepts, it is evident that the nature of the solvent in which a host-guest interaction occurs can play an important role in determining the values of complexation equilibrium constants. In this perspective, it is important to emphasize that the solvent properties of supercritical carbon dioxide are much different from those of both water and nonpolar organic solvents. In fact, contrary to water, CO2 has no dipole moment and behaves effectively as a hydrophobic solvent.32 Even at high density, scCO2 has far weaker van der Waals forces than hydrocarbon solvents, as shown by its relatively low dielectric constant and volume polarizability. Although the solubility parameter of compressed carbon dioxide approaches that of hexane, it is well-known that strong differences have been observed in the solvent strength of these compounds.33 For example, amorphous hydrocarbon polymers, such as atactic polypropylene and polybutadiene, which are quite soluble in liquid hexane, are insoluble in scCO2. These differences in the behavior of the two compounds have been attributed to the large quadrupole moment of CO2 that gives a substantial contribution to its solubility parameter and can give rise to quadrupoledipole and quadrupole-quadrupole interactions that influence solvation and solubilities.34 It seems reasonable to hypothesize that the smaller values of Kf in scCO2 versus aqueous media arise from the absence of the hydrophobic effect, which is replaced by a weaker CO2-phobic effect. That is to say, in scCO2, the desolvation of the guest and β-CD cavity is a less energetically favorable change with respect to this process in water. As reported above, the equilibrium constant for complexation of the 4-adamantyl-substituted phosphine in scCO2 has the greatest value among the phosphines studied. In conventional media, the affinity of guest molecules toward host cyclodextrins strongly depends on the conformity in size between the guest and the cyclodextrin cavity. It has been shown in aqueous systems that the adamantyl group is one of the “best” guest moieties that fits almost perfectly into the β-CD cavity.11 This match in size is such that the adamantyl group excludes waters of solvation from the face of β-CD into which it inserts.35 Moreover, the good steric fit between the adamantyl group and the β-CD cavity allows the onset of more effective van der Waals interactions between the host and guest. Our results indicate that these considerations are valid also in the case of complexation in scCO2. This result, if further confirmed, offers a starting point for the conceptual design of functional groups, leading to high values of the equilibrium constant for complexation reactions with CO2-philic CDs in scCO2. Conclusions The complexation reaction between CO2-philic peracetylated β-cyclodextrin and several triphenyl phosphine derivatives was studied in scCO2 by UV-vis spectroscopy. The equilibrium constant, Kf, for the 1:1 complexation reaction was obtained from titration spectra using two established mathematical models published in the literature, which yield consistent results. The values of the equilibrium constants computed in scCO2 are 1-3 orders of magnitude lower than those determined in aqueous systems. This result is likely due to the absence of the hydrophobic effect, which cannot be operative in scCO2. Considering the unique solvent properties of scCO2, these results

2578 J. Phys. Chem. B, Vol. 111, No. 10, 2007 suggest that the solvent-phobic effect of CO2 during the desolvation of the guest molecule and β-CD cavity is weaker than that which occurs in water. The highest value of Kf was obtained in the case of diphenyl-(4-adamantylphenyl)phosphine, which supports the hypothesis that a good fit between the host and the guest cavity is important to favor complexation. If further confirmed, this criterion could be important in the design of functional groups capable of yielding high values of Kf for inclusion reactions in scCO2. Acknowledgment. The financial support from Universita` di Palermo is gratefully acknowledged. References and Notes (1) Beckman, E. J. J. Supercrit. Fluids 2004, 28, 121. (2) Leitner, W. Acc. Chem. Res. 2002, 35, 746. (3) Jessop, P. G.; Hsiao, Y.; Ikariya, T.; Noyori, R. J. Am. Chem. Soc. 1996, 118, 344. (4) Kainz, S.; Koch, D.; Baumann, W.; Leitner, W. Angew. Chem. Int. Ed. 1997, 36, 1628. (5) Banet Osumna, A. M.; Chen, W.; Hope, E. G.; Kemmitt, R. D. W.; Paige, D. R.; Stuart, A. M.; Xiao, J.; Xu, L. J. Chem. Soc. Dalton Trans. 2000, 22, 4052. (6) Palo, D. R.; Erkey, C. Organometallics 2000, 19, 81. (7) Chen, W.; Xu, L.; Xiao, J. Org. Lett. 2000, 2, 2675. (8) Hu, Y.; Chen, W.; Xu, L.; Xiao, J. Organometallics 2001, 20, 3206. (9) Holmes, J. D.; Steyler, D. C.; Rees, G. D.; Robinson, B. H. Langmuir 1998, 14, 6371. (10) Jacobson, G. B.; Lee, C. T., Jr.; daRocha, S. R. P.; Johnston, K. P. J. Org. Chem. 1999, 64, 1207. (11) Rekharsky, M. V.; Inoue, Y. Chem. ReV. 1998, 98, 1875. (12) Tilloy, S.; Genin, E.; Hapiot, F.; Landy, D.; Fourmentin, S.; Geneˆt, J.-P.; Michelet, V.; Monflier, E. AdV. Synth. Catal. 2006, 348, 1547. (13) Torque, C.; Sueur, B.; Cabou, J.; Bricout, H.; Hapiot, F.; Monflier, E. Tetrahedron 2005, 61, 4811. (14) Potluri, V. K.; Xu, J.; Enick, R.; Beckman, E.; Hamilton, A. D. Org. Lett. 2002, 4, 2333.

Galia et al. (15) Potluri, V. K.; Hamilton, A. D.; Karanikas, C. F.; Bane, S. E.; Xu, J.; Beckman, E. J.; Enick, R. M. Fluid Phase Equilib. 2003, 211, 211. (16) Canipelle, M.; Caron, L. Christine, C.; Tilloy, S.; Monflier, E. Carbohydr. Res. 2002, 337, 281. (17) Caron, L.; Christine, C.; Tilloy, S.; Monflier, E.; Landy, D.; Fourmentin, S.; Surpateanou, G. Supramol. Chem. 2002, 14, 11. (18) Monflier, E.; Tilloy, S.; Me´liet, C.; Mortreux, A.; Landy, D.; Fourmentin, S.; Serpateanou, G. New J. Chem. 1999, 23, 469. (19) Filardo, G.; Di Blasi, M.; Galia, A.; Ponchel, A.; Bricout, H.; Sayede, A. D.; Monflier, E. J. Supercrit. Fluids 2006, 36, 173. (20) Caron, L.; Bricout, H.; Tilloy, S.; Landy, D.; Fourmentin, S.; Monflier, E. AdV. Synth. Catal. 2004, 346, 1449. (21) Broxton, T. J.; Capper, G.; Deady, L. W.; Lenko, A.; Topsom, R. D. J. Chem. Soc. Perkin Trans. 2 1972, 9, 1237. (22) Brunner, G. Gas extraction. An introduction to Fundamentals of Supercritical Fluids and the Application to Separation Processes; Baumgartel, H., Franck, E. U., Grunbein, W., Eds.; Steinkopff Darmstadt Springer: New York, 1994; pp 15-16. (23) Rose, N. J.; Drago, R. S. J. Am. Chem. Soc. 1959, 81, 6138. (24) Canipelle, M.; Caron, L, Christine, C.; Tilloy, S.; Monflier, E. Carbohydr. Res. 2002, 337, 281. (25) Rose, N. J.; Drago, R. S. J. Am. Chem. Soc. 1959, 81, 6141. (26) Long, J. H.; Drago, R. S. J. Chem. Educ. 1982, 59, 1037. (27) Canipelle, M.; Caron, L.; Bricout, H.; Tilloy, S.; Monflier, E. New J. Chem. 2003, 27, 1603. (28) Canipelle, M.; Tilloy, S.; Ponchel, A.; Bricout, H.; Monflier, E. J. Incl. Phenom. 2005, 51, 79. (29) Carrazana, J.; Jover, A.; Meijide, F.; Soto, V. H.; Vazquez Tato, J. J. Phys. Chem. B 2005, 109, 9719. (30) Zhang, B.; Breslow, R. J. Am. Chem. Soc. 1993, 115, 9353. (31) Sayede, A. D.; Ponchel, A.; Filardo, G.; Galia, A.; Monflier, E. J. Mol. Struct. 2006, 99, 777. (32) Hyatt, J. J. Org. Chem. 1984, 49, 5097. (33) McFann, G. J.; Howdle, S. M.; Johnston, K. P. AIChE J. 1994, 40, 543. (34) Myers, A. L.; Prausnitz, J. M. Ind. Eng. Chem. Fundam. 1965, 4, 209. (35) Tellini, V. H. S.; Jover, A.; Galantini, L.; Meijide, F.; Tato, J. V. Acta Crystallogr. B 2004, 60, 204.