Compound 1 was obtained out of a mixture of volatile, low-molecular weight (< 1000 g/mole) cyclosiloxanes, resulting from vacuum stripping of a high molecular weight, vinyl-terminated, linear methyl(3,3,3-trifluoropropyl)siloxane polymer. The distribution of cyclic methyl(3,3,3-trifluoropropyl)siloxanes in equilibrium with linear homopolymer, contains a majority of cyclic siloxane species, and of these, dominantly the cyclic tetramers, pentamers and hexamers. A typical distribution (wt%) comprises 41-47% D4F, 31-36% D5F, and 7-8 % D6F isomers, with only 13-17% linear polymer [23]. Further, in the most volatile portion, obtained by distillation under reduced pressure, the cyclic tetramer siloxanes make up an even greater proportion of the mixture. Yuzhelevskii reported distillable fractions from a caustic hydrolytic polymerization mixture as containing »16% trimers, 75 % tetramers, and 9 % pentamers [6].
However, given the presence of multiple cyclosiloxane oligomers even in the distillate (with molecular weights varying from 625-937 g/mole), plus the possibility of numerous stereoisomeric forms for each given formula weight which could broaden or depress the melting points of the other individual components, the likelihood of resolving any one specific compound appeared slim. It is therefore surprising that a solid, crystalline precipitate was observed to form from such a complex mixture, where all the congeneric and isomeric siloxanes share a common repeat unit, have similar polarity and solubility parameters, and also likely have a high mutual solubility in one another. Based on the dominant cyclosiloxane oligomer in both the polymerization reaction and the resultant distillate mixture, on a mass basis alone it was likely that the solid material which crystallized out was a cyclotetrasiloxane.
Workers at Dow-Corning first revealed significant differences in Si-O stretching frequencies between various cyclic diorganosiloxane oligomers, that could be used to identify families of specific isomers based on Si-O ring sizes [24]. It was subsequently more generally found that both symmetric [R2SiO]x and asymmetric [RRʹSiO]x diorganosubstituted cyclosiloxanes, possessed strong asymmetric Si-O stretching frequencies of the siloxane rings which fell into specific ranges for cyclic trimers (x = 3) and for cyclic tetramers (x =4) [25]. Thus, on the basis of that work, the vibrational spectrum of crystalline solid 1, was examined. A strong absorption at 1030 cm-1 was seen in the IR spectrum, which was taken as indicative of a cyclotetrasiloxane ring in comparison to the mixture of [CF3C2H4(CH3)SiO]3 isomers, as well as D3 and D4 reference samples [26]. Knowing now that the isolated solid material was very likely a tetrasiloxane, it then remained to identify which geometric cyclotetrasiloxane Isomer A-D corresponded to Compound 1.
In the family of D4F cyclosiloxanes, there are 4 possible stereoisomers, denoted A-D in Figure 1 below [27]. Isomer A is the all-cis molecule, where all of the trifluoropropyl groups lie on the same side of the Si4O4 ring (all –C2H4CF3 up or down). Isomer B has the arrangement with three trifluoropropyl groups on the same side of the ring, and the fourth one oriented in the opposite direction (three trifluoropropyls up and one down; cis-cis-trans). Isomers C and D both have two CF3C2H4- groups on one side of the ring and the other two trifluoropropyl groups on the opposite side. Isomer C is the cyclic siloxane variant where the same-side-oriented trifluoropropyl groups are on adjacent Si atoms (two fluoropropyl groups up and two down; cis-trans-cis orientation); Isomer D is similar, except the same-side oriented trifluoroalkyl groups are on Si atoms located across from each other on the Si-O ring and the orientation of the substituents are alternately above and then below the Si-O ring (two trifluoropropyl groups up and two down; trans, trans, trans – or “all-trans” isomer).
In the process of chain growth by condensation of an unsymmetric diorganosilane diol (alternatively, formation of cyclic oligomers by cycloreversion in an equilibration polymerization), the arrangement of each siloxane repeat unit with respect to the successive one in a sequence gives rise to the possibility of stereoisomers. Assuming random probability of a given incoming methyl(trifluoropropyl) siloxane unit being in any specific orientation with respect to the first one, and little if any energetic difference in the orientation of a successive unit being added, one can calculate reasonable cyclic isomer probability distributions, based solely on the cumulative probabilities [28,29]. From these considerations, the relative abundance of the four isomers A to D as indicated above, should be in the ratio 1:4:2:1, respectively.
Cai and co-workers observed four such isomers in reference mixtures of D4F (and also in environmental samples), which were separated by gas chromatography and in all fractions, their molecular weights each corresponding to [CF3C2H4(CH3)SiO]4 were verified by a mass-selective detector [30]. They were not able to directly determine the stereochemistry using GC-MS data alone. However, the authors assigned stereochemistry for each cyclotetrasiloxane component in the mixture, using the relative signal intensity in the total ion chromatogram, which qualitatively matched the expected 1:4:2:1 molar ratio derived from purely random statistical considerations. Accordingly in order of increasing retention time, tr, the separated isomeric, D4F cyclic siloxanes were assigned as tr = 12.027 min = (cis, cis, cis) Isomer A; tr = 12.496 min = (cis, cis, trans) Isomer B; tr = 12.643 min = (cis, trans, cis) Isomer C; and tr = 12.827 min = (trans, trans, trans) Isomer D. This work established proof that all four possible D4F isomers can be formed in industrial polymerization processes, and that their amounts followed a predicted distribution. However, insufficiently small quantities were separated for additional study, further details about other spectroscopic and physical properties of each separated isomer were not investigated, and more direct confirmation of the assigned stereochemistry was still lacking.
Before turning to more detailed spectroscopic investigations, we note that the observed melting point of Compound 1 isolated in this work was 73-75 °C, which fairly closely matches the 76.5-77 °C value for a room-temperature solid fraction, denoted as “Substance 1” by Yuzhelevski [6]. Substance 1 was the highest melting point compound isolated from a distillate cut containing tetrasiloxane isomeric materials [CF3C2H4(CH3)SiO]4 (characterized by its solution molecular weight). We therefore believe Compound 1 (this work) likely corresponds to Substance 1 from earlier work. One could rationalize that in the solid state, Substance 1 and Compound 1 here, based on their high melting points, might represent the most densely packable arrangement of the four isomeric cyclotetrasiloxanes A-D. However, which of these four stereoisomers would pack the most efficiently is not intuitively obvious, as it is not always the case that the most symmetric isomer packs more densely in the solid state and/or retains its solid structure to a higher temperature. As an example, the higher melting form of the related cyclic trimer, trans-D3F has nominally Cs point group symmetry, while the lower melting point, cis-D3F isomer belongs to the much higher symmetry, C3v point group class. In contrast, in the methyl(phenyl)cyclotrisiloxane system, the higher symmetry cis-[CH3(C6H5)SiO]3 has a melting point of » 100 °C, while the lower symmetry trans-[CH3(C6H5)SiO]3 melts at 40 °C [27].
We next turned to NMR spectroscopy to further characterize Compound 1 and differentiate it amongst the possible Isomers A-D, based on spectra obtained for the 1H, 13C, 29Si, and 19F nuclei present. In contrast to the commercially important D3F cyclotrisiloxanes, for which both pure isomeric forms have been isolated and 1H, 13C, 29Si, and 19F NMR data tabulated [31,32,33,34,35], much less information on the congeneric D4F cyclotetrasiloxanes is known. The 1H NMR of 1 shows a singlet Si-CH3 resonance (d = 0.270 ppm), as well as only two sets of multiplet resonances associated with the –C2H4- portion of the 3,3,3-trifluoropropyl group, bearing coupling to 19F of the terminal –CF3 group (see Supplementary Information (SI), Fig. S1). Kalig and coworkers [34] found only three resolvable Si-CH3 peaks in the range 0.23 to 0.25 ppm in a commercial reference mixture of D4F isomers, and tentatively assigned other species at 0.26 and 0.28 ppm as arising from the higher molecular weight homologs D5F and D6F of undefined stereochemistry [36].
The peak assignments in the 13C NMR for the -CH3, and C1, C2, and C3 of the 3,3,3-trifluoropropyl groups in 1 were first made on the assumption of: 1) the Si-CH3 chemical shift being the closest to the TMS reference, and 2) then going downfield for the trifluoropropyl carbons C1-C3 in that order, the latter being deshielded by their increasing proximity to the highly electron-withdrawing, terminal -CF3 group (see SI, Fig. S2). The 13C-NMR chemical shift peak assignments were confirmed through the use of an 1H-13C HSQC experiment (see SI, Fig. S3). Finally, the assignments were in accord with the expectation that the 13C-19F coupling constants would increase in the order 1J13C-19F > 2J13C-19F > 3J13C-19F, with the observed chemical shifts/coupling constants being C1 (9.392 ppm, no apparent 13C-19F coupling), C2 (28.289 ppm, 30.21 Hz), and C3 (128.95 ppm, 276.1 Hz). We note in passing that the 13C NMR spectrum observed here for Compound 1, and those reported for both cis- and trans-D3F [34], showed 13C-19F coupling to all three carbons of the trifluoropropyl groups, which was of comparable magnitude in both sets of siloxanes.
Yuzhelevskii et al. attempted empirical, group contribution additivity calculations, based on perturbation of the 29Si NMR chemical shift relative to the parent all-methyl siloxanes, in a series of mixed DxDFy cyclics (x + y = 3, 4, 5), but the agreement between calculated and observed chemical shifts for the D4F isomers was not very good [31]. They further expected that with six potential stereochemically different Si environments in the four possible D4F stereoisomers, there might be only five distinct chemical shifts seen, because of the similarity in calculated resonance values for Isomers C and D. In reality, these workers observed just two resolved peaks in the experimental 29Si spectrum of a D4F mixture, at -19.84 and -20.06 ppm [37]. Other researchers observed two separate resonance envelopes, each one consisting of at least two partially overlapping single peaks, at approximately -20.1 and -20.3 ppm [35]. The 29Si-NMR spectrum of Compound 1, which showed only a single resonance at -20.38 ppm, corresponds well with the more upfield resonance of this latter work (see SI, Fig. S5). As with other researchers who had previously examined the cis- or trans-D3F compounds (or a mixture thereof) by 29Si-NMR [31,33,34,35], it is noteworthy to elaborate that we were also unable to either resolve separate signals for the two types of inequivalent Si atoms in trans-D3F, or to observe different chemical shifts attributable to the two stereoisomers, in a cis-trans-D3F mixture. Even when measured at higher magnetic field than in previous work (14.1 T vs 9.4, 7.0, or 2.35 T), we still observed only a single, unresolved resonance at –9.61 ppm for all Si-atoms in the cis-trans-D3F isomer mixture.
Finally, it has been evident in work on the stereoregularity of linear siloxane polymers prepared from cis-D3F [7a], that 19F NMR has the highest sensitivity and ability to discriminate subtle polymer microstructure and tacticity resulting from the combination of methyl and 3,3,3-trifluoropropyl groups on each Si atom. Examination of a reference mixture of D4F isomers by 19F NMR [34], showed good resolution of five distinct single peaks spanning a narrow range of < 0.1 ppm around -69.77 ppm. We observed a single 19F NMR resonance for the -CF3 group in 1, split by geminal 19F-1H coupling to the adjacent methylene, at -69.53 ppm, which is slightly downfield of all the peaks in the reported D4F isomer mixture (see SI, Fig. S4). From the totality of our NMR data, all of the spectra for 1 are relatively simple, and show an apparent single environment for the -[CF3C2H4(CH3)SiO]- unit and its methyl/fluoroalkyl group sub-components. Accordingly, the structures of Isomer A, C, or D would be consistent with the observed spectra, and only Isomer B might be ruled out by its magnetically inequivalent alkyl groups [38]. Assignment of the stereochemistry of 1, therefore necessitated the use of a single crystal X-ray diffraction study for a definitive result.