Tracing the Acetalization of Cyclohexanone in CO2-Expanded Alcohols by Attenuated Total Reflection Infrared Spectroscopy

The CO2-catalyzed acetalization is regarded as a promising alternative to the conventional acid-catalyzed method from a viewpoint of green chemistry (C. A. Eckert et al., Ind. Eng. Chem. Res. 43, 2605 (2004)). We have applied in situ attenuated total reflection infrared (ATR-IR) spectroscopy for elucidating and monitoring the acetalization of cyclohexanone in CO2-expanded ethylene glycol and methanol at 50 °C and 3 MPa. The ATR-IR spectra of the reaction mixtures periodically recorded with a ZnSe crystal demonstrate that ATR-IR spectroscopy is a practical tool for tracing the kinetics of acetalizations in situ. In addition, the rate of CO2 dissolution as well as CO2 solubility into the cyclohexanone–alcohol mixtures could be evaluated from the CO2-v3-antisymmetric stretching band. The ZnSe ATR crystal, however, was corroded during longer use under the acidic conditions realized by the dissolution of CO2 in the alcohols. In contrast, the corrosion did not occur when a Ge crystal was used instead of a ZnSe crystal, and therefore the application of a Ge ATR crystal is recommended for continuous long-term experiments with these media.

INTRODUCTION CO 2 -expanded liquids (CXLs) are promising media for performing separations, extractions, and catalytic and noncatalytic reactions, as well as for other applications. 1 One of the most promising CXLs in terms of ''green'' chemical synthesis is the CO 2 -expanded alcohol, which exhibits acidity and thus can promote acid-catalyzed reactions. It is well known that water, when it dissolves CO 2 , shows acidity owing to the formation of carbonic acid (HOC(O)OH). By analogy, alkylcarbonic acid, ROC(O)OH, is considered to form in CO 2 -expanded alcohols (ROH), creating acidic environments. Eckert's group used diazodiphenylmethane as a probe molecule to trap the acidic species generated in CO 2 -expanded alcohols. 2,3 They showed that the alcohol-CO 2 complexes react similarly to carboxylic acids with the probe molecule, indicating the formation of alkylcarbonic acids in CO 2 -expanded alcohols. More recently, they developed a technique to measure the alkylcarbonic acid formations and the dissociation constants in CO 2 -expanded methanol, ethanol, and benzyl alcohol at 25 8C. 4 The utility of CO 2 -expanded alcohols not only as solvents but also as acid catalysts was first demonstrated by Eckert and co-workers. They reported that the acetalization of cyclohexanone with methanol is accelerated remarkably in the presence of CO 2 (Scheme 1); under optimal conditions, the rate in CO 2expanded methanol was 130 times faster than that in CO 2 -free methanol. [5][6][7] They also found that the hydrolysis of b-pinene to terpineol and other alcohols proceeds in CO 2 -expanded 1:1 MeOH-H 2 O mixture, whereas the same reaction did not take place in the absence of CO 2 . [6][7][8] The diazotization of aniline and subsequent coupling to give methyl yellow or substitution to give benzyl iodide was also shown to be accelerated in CO 2expanded methanol. 9 Another successful example of the use of CO 2 -expanded alcohols in organic synthesis is the Ugi reaction, which involves the addition of alkylcarbonic acid to the intermediate nitrilium ion. 10 These examples clearly emphasize not only the high utility of CO 2 -expanded alcohols for acid-catalyzed reactions, but also the potential of the CXLs as alternatives to conventional, problematic acid catalysts used in industrial processes. Disposal of solid acid catalysts or neutralization of homogeneous acid catalysts giving waste salts are still environmental issues in conventional processes, whereas the active acidic species, alkylcarbonic acids, in CO 2 -expanded alcohols can easily be decomposed to alcohols and CO 2 just by depressurization of CO 2 .
One technical problem of the reactions in CO 2 -expanded alcohols arises from the high-pressure CO 2 conditions, which require a tightly closed stainless-steel reactor. The composition of reaction mixtures is therefore known only when the reactor is opened through troublesome procedures including depressurization and decomposition of the cell. Hence, more facilitated methods that furnish information on the compositions are highly desired from a practical point of view.
In situ attenuated total reflection infrared (ATR-IR) spectroscopy, if the ATR crystal is placed at the bottom of the cell, is a powerful tool to investigate gas-expanded liquids. 11 For example, Kazarian's group [12][13][14][15] and our group 16 have used this spectroscopy to elucidate the molecular interactions in CO 2expanded ionic liquids. Here, we have applied ATR-IR spectroscopy to elucidate CO 2 -expanded liquids composed of cyclohexanone and alcohols and have evaluated the potential and limitations of this spectroscopy for tracing the acetalizations.
High-Pressure Attenutated Total Reflection Infrared Cell and Procedures. SAFETY NOTE: The present experiments involve the use of relatively high pressures and require equipment with the appropriate pressure rating. The specially constructed high-pressure cell (ETH workshop, Fig. 1) had an internal volume of 10 mL and possessed an ATR crystal (ZnSe or Ge; size 50 mm 3 20 mm 3 2 mm; angle of incidence 608; Crystran Ltd.). 16,17 The IR beam reflects six times at the interface of the CO 2 -expanded reaction mixtures with the ATR crystal surface. Spectra were recorded with a resolution of 2 cm À1 using an EQUINOX-55 Fourier transform infrared (FT-IR) spectrometer equipped with a liquid-N 2 -cooled MCT detector (Bruker Optics). High-pressure liquid CO 2 was fed into the cell using a PM-101 compressor (NWA), of which pressure was set below the limitation pressure of the whole system.
In situ ATR-IR spectra during the acetalizations were recorded as follows: 0.5 mL of cyclohexanone and 5 mL of anhydrous alcohol were loaded simultaneously in the cell using two syringes under N 2 atmosphere. The cell was sealed and heated to 50 8C. Then, the compressed liquid CO 2 was introduced until the pressure reached 3 MPa. The acetalization mixtures were not stirred unless otherwise stated.
The product liquid mixtures were analyzed also by 1  Scanning Electron Microscopy. Scanning electron microscopy (SEM) was performed on a Gemini 1530 field emission microscope (Zeiss). Secondary electron images of the asobtained samples were recorded at 1 kV.

RESULTS AND DISCUSSION
Acetalization of Cyclohexanone in CO 2 -Expanded Ethylene Glycol over ZnSe Attenuated Total Reflection Crystal. The in situ ATR-IR spectra recorded during the acetalization of cyclohexanone with ethylene glycol at 50 8C under 3 MPa CO 2 are shown in Figs. 2-5. Although ZnSe is known to react with acids to give toxic H 2 Se, 18 reasonable spectra could be obtained under the acidic conditions, indicating that the acidity of this medium is not strong enough to promote the rapid decomposition/dissolution of ZnSe. Note, however, that the surface of ZnSe crystal gradually underwent corrosion, and this was accelerated by stirring of the acetalization mixture (see the Kinetic Graphs subsection below).
The formation of the corresponding five-membered cyclic acetal involves the disappearance of two OH groups in an ethylene glycol molecule, while it is accompanied by the formation of one H 2 O molecule (Scheme 1). As shown in the behavior of band A in Fig   stretching mode band decreased drastically as the reaction proceeded, indicating that the intensity decrease of the ethylene glycol band prevailed over the intensity increase of the H 2 O band. The top parts of the C(sp 3 )-H stretching mode band B slightly changed according to the increased and decreased amounts of the acetal product and the reactants (ethylene glycol and cyclohexanone), respectively. Figure 3 shows the change of the CO 2 -m 3 -antisymmetric stretching mode band 16,19,20 with time. Band C is assigned to natural 13 CO 2 , while band B is probably due to the (m 3 þ m 2 ) À m 2 hot band. 21,22 Compared to the m 3 band of pure CO 2 at 50 8C and 3 MPa, which displays the P-and R-branches, 20 the main band A was not split and was far less intense owing to the smaller amount of dissolved CO 2 . In addition, band A was much narrower, likely due to the impeded rotational motion by the interaction with ethylene glycol. It should be noted that band A increased very slowly with time, because of the sluggish rate of CO 2 dissolution in ethylene glycol. This was also confirmed by the gradual decrease of CO 2 pressure with time, which was compensated by adding further amounts of CO 2 to the cell to keep the pressure constant. Microscopically, the slow-rate dissolution of CO 2 probably originates from the strong intermolecular interactions among ethylene glycol molecules, which form a relatively strong hydrogen-bond network and cause the relatively high viscosity of ethylene glycol. The former report by Eckert and co-workers mainly focused on acetalization with methanol, and the detailed kinetics of the reaction with ethylene glycol were not given. 5 Since the intensity of the CO 2 m 3 band increased slowly even when the acetalization mixture was stirred, the mass transfer of CO 2 also must be taken into account when the detailed kinetics of the reaction is considered.
The change of the C¼O stretching mode band of cyclohexanone is shown in Fig. 4. This band change is probably most reliable for checking the extent of the progress of acetalization. Our group, for example, has shown that the intensity change of the C¼O band actually can be used for making a calibration curve for quantitative analysis in the catalytic chemical fixation of CO 2 with epoxides. 17 In the present experiment, the relative intensity of the C¼O band just after the addition of CO 2 was lower than that at 1 h. Since the spectra were recorded under static conditions, this was probably due to a heterogeneous distribution of ethylene glycol, cyclohexanone, and CO 2 . However, after 1 h, the decrease of the band with time could clearly be seen, indicating the conversion of cyclohexanone to the corresponding acetal. Strictly, the C¼O band was composed of three bands. At the initial stage of the reaction, an intense band B was observed at 1700 cm À1 with a shoulder band C at 1685 cm À1 . Intriguingly, however, as the intensity of bands B and C decreased with the time, at the end of the reaction appeared another band A at a higher wavenumber of 1716 cm À1 . Similar C¼O band discreteness was also observed by other groups for CO 2 -free carbonyl compound-alcohol and carbonyl compound-water systems. [23][24][25] They attributed the highest wavenumber band to a non-hydrogen-bonded free C¼O bond, whereas the bands at lower wavenumbers were suggested to be due to the C¼O bond hydrogen-bonding with alcohol or water. According to their assignment, band A can be attributed to the non-hydrogenbonded C¼O bond, while band B is assignable to the monohydrogen-bonded C¼O bond. The shoulder band C is difficult to interpret. One possible assignment for this band is the C¼O bond di-hydrogen-bonding with ethylene glycol. However, this still remains inconclusive, because the earlier work by Symons and Eaton revealed that not only methanol but also ethylene glycol cannot di-hydrogen-bond with acetone at 25 8C. 24 In addition, Saied and co-workers observed a much greater shift for the C¼O bond in 3,3,5,5-tetramethylcyclohexanone doubly hydrogen-bonding with a specially designed diphenol, 2,2 0 -(1,2-ethynediyl)bis[6-(1,1-dimethylethyl)-4-methylphenol] (free C¼O, 1717 cm À1 ; di-hydrogen-bonded C¼O, 1670 cm À1 ). 26 We confirmed that band C similarly appeared even for the CO 2 -free ethylene glycol-cyclohexanone mixture, and thus this band is not related to the alkylcarbonic acids formed by the reaction between ethylene glycol and CO 2 (Scheme 1). Figure 5 shows the band changes in the finger-print region, which shows the clear increase and decrease of the bands due to the product and reactants, respectively. The intensity decrease of the broad band Z allowed the CO 2 -m 2 -bending mode band a to appear at 662 cm À1 . This wavenumber is very close to that observed for the CO 2 -expanded benzyl alcohol at 80 8C and 16 MPa, 27 and shifted to a lower wavenumber compared to that of pure CO 2 at 50 8C and 3 MPa (667 cm À1 ), indicating that the surrounding ethylene glycol molecules interacting with CO 2 suppressed the bending vibration of CO 2 .
Acetalization of Cyclohexanone in CO 2 -Expanded Methanol over the ZnSe Attenuated Total Reflection Crystal. In situ ATR-IR spectroscopy was also successfully applied for the acetalization of cyclohexanone in CO 2 -expanded methanol under otherwise similar conditions, though the change of the bands was much less drastic due to the sluggish reaction. Similar increase and decrease of the bands due to the corresponding acetal and the reactants (methanol and cyclohexanone), respectively, could be observed, as, for example, shown in Fig. 6. One remarkable difference compared to the former system with ethylene glycol was that CO 2 dissolved much faster in the liquid phase composed of methanol and cyclohexanone. As shown in Fig. 7, the intensity of the CO 2 m 3 band recorded at 10 min was already very close to that at 4.83 h, although the mixture was not stirred. It also should be noted that the amount of dissolved CO 2 was much greater for the methanol-cyclohexanone-CO 2 system than for the former ethylene glycol-cyclohexanone-CO 2 system, as evidenced from the much stronger intensity of the CO 2 m 3 band. Knowing the refractive index of methanol and ethylene glycol, it is possible to estimate the mole fraction of CO 2 in the methanolcyclohexanone-CO 2 mixture, which is approximately double the CO 2 mole fraction in the ethylene glycol-cyclohexanone-CO 2 mixture. There was also a large difference in broadness of the band between the two systems; the (m 3 þ m 2 ) À m 2 hot band was hidden by the broad CO 2 m 3 band for the present system (Fig. 7), indicating that the rotational motion of CO 2 molecules in the CO 2 -expanded methanol-cyclohexanone was not so strongly impeded as in the CO 2 -expanded ethylene glycolcyclohexanone. The much weaker strength of attractive intermolecular interactions among methanol molecules, which can also be seen in the higher volatility and lower boiling point of methanol, resulted in the faster and greater dissolution of CO 2 ; the vapor pressures (20 8C)/boiling points of methanol and ethylene glycol are 97.68 mmHg/64.7 8C and 0.08 mmHg/ 196-198 8C, respectively, according to the data published by Sigma-Aldrich. Kinetic Graphs. We have also measured the ATR-IR spectra of the methanol-cyclohexanone-CO 2 system and the ethylene glycol-cyclohexanone-CO 2 system under ''stirred'' conditions using the ZnSe ATR crystal. Stirring the methanolcyclohexanone-CO 2 system accelerated the corrosion of the crystal surface; the baseline of the spectrum suddenly moved up after 4 h, and a reasonable spectrum was no longer available. The surface and probably also the subsurface of the ZnSe crystal underwent decomposition through the reaction with the acidic CO 2 -expanded methanol, 18 and the contact of a possible acidic species, i.e., methylcarbonic acid, with the crystal surface became more intense with stirring of the mixture. SEM images of the surface of the ZnSe crystal before and after having been immersed in the CO 2 -expanded methanol at 50 8C and 3 MPa for 6 h are shown in Fig. 8. As emerges from these images, the acidity of this medium could change the clearly flat surface of the parent ZnSe crystal to an awfully rugged one. The corrosion was less prominent for the ethylene glycol-cyclohexanone-CO 2 system, which seems to be related to the smaller dissolved amount of CO 2 and to the higher viscosity of the mixture inhibiting the contact of the acidic species with the crystal surface. Until the ZnSe crystal was deteriorated, however, reasonable band-intensity changes of the reactants and products were observed, and we could successfully plot kinetic graphs for the two acetalization systems, respectively, which are shown in Fig. 9.
Further Comments. In situ ATR-IR spectroscopy thus has been demonstrated to be able to furnish information on the composition of CO 2 -expanded acetalization mixtures, which changes with time. Since the acetalizations are performed under relatively high pressure CO 2 conditions and thus require reactors that are not easily opened, in situ spectroscopy will be highly demanded if the acetalizations or their related reactions are commercialized. In addition, we could also obtain some new insight into the acetalizations that was not discussed previously by Eckert and co-workers 5-7 such as the difference in the CO 2 dissolution rate between ethylene glycol-cyclohexanone mixtures and methanol-cyclohexanone mixtures, and the discrete carbonyl bands probably indicating the presence of free cyclohexanone and mono-hydrogen-bonded cyclohexanone in the CO 2 -expanded ethylene glycol-cyclohexanone.
The method, however, did not furnish the decisive mechanistic information such as the presence of alkylcarbonic acids or possible intermediates such as hemiacetals. This was probably because these ''key'' species existed in extremely low concentrations or their lifetimes were too short to be detected by IR spectroscopy. Even for the acetalization under conventional conditions, we are unaware of any published reports showing the spectra of possible intermediates; the wellestablished mechanism has been proposed based only on the results of kinetic analyses. 28,29 The concurrent analysis of the resultant liquid phases by 1 H/ 13 C NMR, GC, and GC-MS revealed that the reactions took place in almost 100% selectivity, although, in some cases, a negligible amount of 1-methoxycyclohexene was detected for the acetalization of cyclohexanone in CO 2 -expanded methanol.
Although in situ ATR-IR spectroscopy is thus considered to be a promising tool for tracing the acetalizations, we should note a limitation of the ZnSe crystal, namely, its sensitivity to acidic conditions. As we observed in a sudden jump up of the baseline of the spectrum after the 4-h acetalization of cyclohexanone in CO 2 -expanded methanol, the surface of the ZnSe crystal was corroded under the acidic conditions. The deterioration of the crystal occurred only at the surface, because it could be reused after polishing the surface with a diamond paste. The use of a more acid-tolerant ATR crystal (e.g., Ge, diamond) is necessary to surely circumvent the corrosion problem; we, at least, observed similar band changes as observed here when a Ge crystal was used for the acetalization systems (Fig. 10). Nevertheless, the use of Ge is restricted as this material has a very high refractive index (4.0), which decreases the signal drastically compared to ZnSe or diamond and furthermore possesses a low rupture modulus, which limits the use to moderate pressure. 30

CONCLUSION
In situ attenuated total reflection infrared spectroscopy has been shown to be a useful tool for tracing the acetalization of cyclohexanone in CO 2 -expanded ethylene glycol and methanol at 50 8C and 3 MPa. The spectra of the acetalization mixtures recorded with a ZnSe crystal showed the clear increase and decrease of the bands due to the products and reactants, respectively. In addition, they revealed that the rate of CO 2 dissolution strongly depends on the properties of the alcohols. The method, however, did not afford any mechanistic information, probably due to the extremely low concentrations or short lifetimes of the possible intermediates. Finally, ZnSe was relatively unstable under the acidic conditions realized by the dissolution of CO 2 in the alcohols, and acid-tolerant ATR crystals such as Ge crystal and diamond crystal are much more appropriate for tracing the acetalizations. It was revealed that the spectrum changes with Ge crystal were similar to those recorded with ZnSe crystal.