ABSTRACT
Trialkyl oxonium salts are among the most powerful electrophiles known. They are immediately quenched by nucleophiles like alcohols and halide ions, and are therefore used in unreactive solvents and have non‐nucleophilic counterions. Any proposal to incorporate a nucleophilic hydroxy substituent into a trialkyl oxonium species would thus be ill‐considered, and pairing an R3O+ cation with an alkoxide counterion would be an abstract notion, at best. Here, we describe how these combinations are made possible within the context of the robust environment of the heterotriquinane ring system. A stable, C3 symmetric oxatriquinanetriol can be simply accessed by reaction of the tris‐epoxide of 1,4,7‐cyclononatriene with aqueous acid. The three OH groups of this hemispherical cation point into the endo space of the bowl and participate in strong intramolecular hydrogen bonding. A close balance in the dynamics of proton exchange between these hydroxy functions and amine bases is predicted by modeling. In practice, treatment of oxatriquinanetriol with the non‐nucleophilic base 2‐tert‐butyl‐1,1,3,3‐tetramethylguanidine (Barton's base) leads to a zwitterionic oxonium alkoxide.
Keywords: hemispherical polycycles, hypervalent compounds, molecular modeling, oxonium ions, zwitterions
A new take on frustration: How can a trialkyloxonium super‐electrophile coexist with an OH nucleophile in the same molecule? To take a step further into chemical no‐man's land, what if the hydroxy group is deprotonated to give an oxonium alkoxide zwitterion? Impossible? Triquinane ring systems have high fragmentation barriers and provide an effective platform for stabilizing combinations of otherwise incompatible functional groups, leading to a remarkable feat of rule‐bending.

1. Introduction
From a chemical perspective, the term “hydroxy‐substituted trialkyloxonium ion” is incongruous. Trialkyloxonium salts are non‐selective alkyl transfer agents that react even with weak nucleophiles at rates approaching diffusion‐controlled limits [1]. Steps beyond a hydroxy‐substituted trialkyloxonium ion, in terms of plausibility, would be a trialkyloxonium alkoxide ion pair, followed by a trialkyloxonium oxide zwitterion (Figure 1). Even if the intermolecular reactions in the latter species could somehow be suppressed, there would still be intramolecular fragmentation pathways available. Incorporation of otherwise incompatible moieties into ring systems with high steric or scission barriers is a strategy that can however be taken to address the chemical challenge represented by Figure 1d.
FIGURE 1.

Top: (a) Triakyloxonium ion; (b) hydroxy‐substituted trialkyloxonium ion; (c) trialkyloxonium alkoxide ion pair; (d) trialkyloxonium alkoxide zwitterion. A‒ represents a generic anion, n is ≥ 1. Bottom: Structures of the heterotriquinanes.
We have hitherto described the synthesis of azatriquinane 1 (Figure 1) and multiple derivatives thereof [2]. The trefoil fusion of three five‐membered rings in 1 results in a puckered structure with an acutely pyramidalized, non‐invertible, apical heteroatom, which has variously enabled the design of chelates for metals and protons, fullerene clasps, and curved aromatics [2]. Incorporation of oxygen into the ring system gives an oxatriquinane (2), a trialkyloxonium species that is markedly stabilized relative to common R3O+ salts, such that it can be chromatographed on silica or even refluxed in water uneventfully [3]. Indeed, even tertiary (3) and bis‐allylic (4) oxonium cations, which would otherwise spontaneously dissociate, are stable and isolable within the triquinane framework [4].
The installation of an α‐hydroxy group into oxatriquinane 2 has previously given 5, which resulted from the protonation of ketone 6 [5]. This unusual juxtaposition of functionalities led to a record C–O+ bond length of 1.66 Å, where a typical ether C─O bond generally measures 1.43 Å. However, what could not be claimed in 5 was the synthesis of a true hydroxy‐substituted trialkyloxonium species, since the OH lone pairs are strongly contracted by delocalization into the C─O+ σ* orbital, as indicated in Figure 2. In fact, this is how such a species became stable enough to permit isolation and even analysis by x‐ray crystallography [5]. Movement of the OH group one carbon along, from the α‐ to the β‐position, results in 7, a much more unlikely species for the reasons discussed above. Nevertheless, we have exploited the oxatriquinane platform for probing a wide range of “rule‐bending” chemistries [2], and thus 7, and ultimately its deprotonated zwitterion, were looked upon as approachable targets for synthesis.
FIGURE 2.

Top: α‐ and β‐Hydroxy oxonium functions in the triquinane ring system, with synthetic derivations. Bottom: Synthesis of oxatriquinanetriol 11. Reagents and conditions: (a) mCPBA, CH2Cl2, 12 h, 67%; (b) aq. H2SO4, rt, 12 h; (c) aq. NaOH, 96%; (d) TfOH or Tf2NH, MeCN, >90%.
The most reasonable path to 7 is as shown in Figure 2 (top), where a precursor such as 8 would ultimately derive from cyclononatriene 9, which is analogous to our published route to 2 [3]. However, for synthetic expediency, we became more attracted to the possibility that the symmetric tris‐epoxide of cyclononatriene 10 could undergo a cascade toward an oxatriquinanetriol 11 as seen in Figure 2 (bottom).
2. Results and Discussion
Oxidation of 1,4,7‐cyclononatriene 9 to the tris‐epoxide is known chemistry, and gives the C 3 v symmetric product 10 [6]. Remarkably, simple reaction of 10 with aqueous sulfuric acid followed by basic workup gave bicyclic tetrol 12 which, we proposed, was itself derived from the hydrolysis of target molecule 11 during workup (Figure 2). Indeed, treatment of 12 with either triflic acid or trifluoromethanesulfonimide gave 11 in high yield as the TfO− or Tf2N– salt, respectively. Oxonium salt 11 Tf2N− is organic‐soluble and stable in MeCN at room temperature, however heating the solution to 70°C leads to a clean transformation over the course of an hour. To our surprise, this was not the result of a dimerization or intramolecular fragmentation, but rather a solvolysis giving rise to product 13 (Figure 3). This appears to proceed via nucleophilic ring opening by MeCN to produce a Ritter‐like intermediate, on which the neighboring hydroxy group closes to give the oxazoline ring. Analogous chemistry is observed with acetone, producing 14 through a similar mechanism. Reaction of 11 Tf2N− with lithium bromide led to facile ring opening to give 15, which reverts to 11 on treatment with silver trifluoromethanesulfonimide.
FIGURE 3.

Structures of derivatives of 11.
We were unable to induce 11 Tf2N– to crystallize by itself, but could leverage the all‐endo stereochemistry of the OH groups to obtain a lithium complex, as the oxygens are spatially situated in a nearly identical pattern to 9‐crown‐3, which selectively forms a 2:1 host‐guest complex with Li+ [7]. Thus, diffusion of diethyl ether into a solution of 11 Tf2N– and LiTf2N in acetonitrile at low temperature produced crystals suitable for x‐ray diffraction. Here, two 11 cations complex Li+ in an octahedral arrangement, with Li─O distances of 2.06 Å and lithium at the center of threefold rotational symmetry axis in the rhombohedral crystal system (Figure 4). While complexes of Li+ with polyethers are common, only a single example of an [R(OH)3]2Li+ structure is seen in the Cambridge Structural Database, in this case involving two inositols sandwiching the cation [8].
FIGURE 4.

ORTEP diagram of the x‐ray crystal structure of [(11)2Li]3+ with Tf2N− counterions omitted for clarity. Left: Side‐view showing two units of 11 coordinated to a lithium ion in an octahedral arrangement, with average Li─O bond lengths of 2.06 Å. Right: Top view depicting a slightly staggered arrangement with respect to the two molecules of 11 composing C 3 symmetry.
We expected computational modeling of the deprotonation of cation 11 to its zwitterion 11z to lead down low‐barrier fragmentation pathways, either toward rearranged products or the transition states (or higher‐order saddle points) en route to them. However, geometry optimization of 11z, both in the gas phase and in solvent models, produced energy‐minimized structures with all substituents axial and evidence of strong hydrogen bonding between the O– and the two OH groups in the endo space of the bowl (Figure 5), with O─O distances as short as 2.52 Å. An argument could thus be made that the resulting distribution of negative charge between the three oxygens would limit the reactivity of 11z.
FIGURE 5.

Graphical representations of the calculated energy‐minimized structures of 11 and 11z at the MP2/6‐311++G(d,p) level of theory. Left: Structure of 11 where the hydroxy groups assume axial conformations, resulting in an H‐bonding triad with average calculated H···O distances of 2.02 Å. Right: Structure of 11z. One oxygen is deprotonated and the remaining two hydroxy groups H‐bond to the anionic center with average calculated H···O distances of 1.59 Å.
Initial efforts to deprotonate 11 with the strong, non‐nucleophilic base LHMDS led to decomposition, even at −78°C. One potential concern under these circumstances was the availability of hydrogens α to the oxonium center, the deprotonation of which would lead to a highly reactive oxonium ylide. In an attempt to differentiate between these sites and the OH groups, we sought to identify a more selective base. Protonation energies based on high‐level ab initio calculations indicated that the hydroxy groups were more acidic than typical alcohols, which could be expected based on the structure and cationic nature of 11. Interestingly, the protonation of bases as weak as trialkyl amines was shown to be a close energetic match to the protonation of 11z. Using B3LYP/6‐311++G(d,p) and MP2/6‐311++G(d,p) methods with continuum solvent models [9], triethylamine (TEA) actually showed a slightly more negative (exothermic) protonation energy than 11z by 1.32 and 0.83 kcal mol−1, respectively (see Table S1); values which are close to the generally accepted experimental error of the model chemistries (±1 kcal/mol).
TEA was thus titrated into a solution of 11 Tf2N– in D8‐THF at −78°C. Consistent with modeling, the 1H signals of TEA were shifted toward those of the triethylammonium cation, although less than halfway between that of the free base and the fully protonated form (Figure S1), suggesting a K eq in favor of the reactants. Nevertheless, line broadening observed in both 11 and TEA showed that both compounds were jointly engaged in slow and reversible proton exchange.
Next, we turned to the non‐nucleophilic base 2‐tert‐butyl‐1,1,3,3‐tetramethylguanidine or “Barton's base” (BB) [10], which is stronger than TEA (pKa of ammonium salts 23.6 vs. 17.2 in MeCN), but much less basic than LHMDS [11, 12]. Using the B3LYP/6‐311++G(d,p) method both in the gas phase and with a continuum solvent model of THF (CPCM), BB showed more negative protonation energies than 11z by 9.5 and 13.3 kcal mol−1, respectively (Table S1). With energy differences on this order, quantitative deprotonation of 11 by BB is expected. A highly useful feature that differentiates BB from conventional amine bases is its ability to function as an acid‐base indicator, since the free base and protonated forms of BB are clearly distinguishable by 1H‐NMR at low temperature. In order to rule out the possibility of adventitious acid confounding the titration, we synthesized 11 Tf2N– by reaction of pure bromotriol 15 with freshly‐prepared AgTf2N (see Supporting Information). Thus, an NMR tube containing a solution of 11 Tf2N– in D8‐THF was cooled in a methanol‐liquid N2 bath (−98°C) and a solution of 0.5 eq BB in D8‐THF was added via syringe. The mixture was immediately placed inside an NMR probe pre‐cooled to −100°C and allowed to come gradually to −60°C while the sample tube was spun to promote mixing. In contrast to the broad resonance signals of TEA that result from equilibrium with 11, the signals of BBH+ were sharp and well defined, indicating that quantitative protonation occurs with an absence of slow exchange between BBH+ and zwitterion 11z (Figure 6). Changes in the chemical shift of the oxatriquinane protons were small, but this observation was not unexpected. Whereas acid‐base titrations monitored by NMR typically lead to distinctive chemical shift changes due to electronic and conformational perturbations, these effects will be more subtle in the oxatriquinane ring system, where hydrogen bonding delocalizes the negative charge and modeling suggests little in the way of structural reorganization. Despite the minor differences in chemical shifts between 11 and the 11/11z mixture, measurable changes were observed in the coupling constants, which are more sensitive to torsional perturbations. Furthermore, line broadening was observed in the oxatriquinane protons, consistent with a slow proton exchange process between 11 and 11z. Indeed, efforts to fully deprotonate 11 with 1.0 equivalent of BB under the same conditions resulted in decomposition, suggesting that interactions between 11 and 11z produced a necessary stabilizing effect. This reliance on weak interactions for stability is a property shared for example by tetraalkylammonium hydroxides, which owe their existence to solvation of the hydroxide in protic media; once the solvent is removed they decompose via Hofmann elimination [13]. Finally, allowing the temperature of the mixture in Figure 6d to warm above −60°C also results in gradual decomposition which, given the exotic nature of 11z, was also not unexpected.
FIGURE 6.

Titration of 11 with Barton's base (BB). 500 MHz 1H‐NMR spectra in D8‐THF at −60°C of (a) BB with colored annotations of unique protons; (b) BBH+ reference generated by treatment of BB with AcOH. Note the spectroscopic desymmetrization that occurs as a consequence of protonation. (c) Oxonium 11 Tf2N– ; (d) Treatment of 11 Tf2N– with 0.5 eq BB, resulting in quantitative protonation of BB and the production of an equilibrium mixture of 0.5 eq each of 11 Tf2N– and 11z, as evidenced by line broadening, chemical shift changes, and a change in the coupling constant of the bridgehead methine proton.
As noted above, we have largely attributed the persistence of zwitterion 11z in solution to the distribution of the negative charge through dynamic hydrogen bond exchange, where the charge can migrate between oxygens in such close proximity. The contribution of this effect could be examined by subjecting 18, a dimethylated version of 11, to the same conditions used to produce 11z. Access to 18 is straightforward and starts with ketal 14, which had been previously obtained by solvolysis of 11 in acetone. Methylation of the hydroxy groups with methyl iodide produced 16, after which deprotection with 3N HCl gave diol 17. Finally, treatment of 17 with Tf2NH in MeCN produced oxonium salt 18 Tf2N– in nearly quantitative yield (Figure 7).
FIGURE 7.

Synthesis of 18, the bis(O‐methylated) derivative of 11. Reagents and conditions: (a) MeI, NaH, THF, rt, 12 h, 90%; (b) 3N aq. HCl, rt, 12 h, 58%; (c) Tf2NH, MeCN, rt, 20 min, 98%.
When 18 was subjected to the same reaction conditions used to produce zwitterion 11z from 11 and the progress monitored by 1H‐NMR, strikingly, 18 underwent rapid decomposition to give a mixture of products (Figures S2 and S3). Results from this experiment emphasize the importance of hydrogen bonding toward the stabilization of zwitterion 11z. Once 18 is deprotonated, the H‐bond‐enforced proximity of the oxygens in the endospace is lost and a more splayed out arrangement is adopted, according to modeling (Supporting Information Section 5). This frees the alkoxide function to potentially operate as a nucleophile toward other molecules of 18. Another likely path to decomposition involves internal fragmentations. Support for this outcome could be derived from the NMR of the product mixture (Supporting Information Section 4) as well as the energy minimized structure of the zwitterion of 18 (Figure S5), which showed a strong contraction of the C─O‒ bond distance to 1.31 Å (compared to the C─OMe bonds at 1.41 Å), with the adjacent C─C bonds alternatively elongated (1.56–1.59 Å) and shortened (1.50 Å). The two C─O+ bonds toward which the electrons ultimately flow were also longer (1.54 Å) than the other C─O+ bond (1.50 Å). No such bond alterations were observed in the modeling of 11z.
3. Conclusion
In summary, oxatriquinanetriol 11 can be synthesized in three simple steps from cyclononatriene 9. To our knowledge, it represents the first alkyloxonium ion bearing a nucleophilic OH function. It is stabilized both by the robust nature of the triquinane ring system and strong intramolecular hydrogen bonding, to the extent that it reacts with relatively non‐nucleophilic solvents (MeCN, Me2CO) at elevated temperatures before reacting with itself, either intermolecularly or via fragmentation. Most remarkably, triol 11 can be deprotonated with a base to give a charge‐compensated trialkyloxonium alkoxide 11z, a species containing both cationic and anionic oxygen in the same molecule, which is persistent at low temperatures and can be characterized by NMR. A bis‐O‐methylated analogue of 11, that is, 18, decomposes immediately upon deprotonation, which is attributed to its inability to form hydrogen bonds, showcasing the stabilizing influence of effective charge dispersion in this hemispherical tricycle.
Author Contributions
Andrew L. Otsuki: investigation, methodology. Gizem Atakan Alkan: investigation, methodology. Shuhei Manabe: investigation, methodology. Wondimagegn T. Shewa: investigation, methodology. Gorkem Gunbas: investigation, methodology. James C. Fettinger: data curation, formal analysis. Nema Hafezi: investigation, methodology, writing – review and editing. Mark Mascal: conceptualization, writing – original draft, writing – review and editing, supervision, funding acquisition, project administration.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Supporting File: anie72258‐sup‐0001‐SuppMat.pdf.
Acknowledgments
This work was financially supported by National Science Foundation Grants CHE‐1362164 and CHE‐2153258. We thank Drs. Ping Yu and Bennett Addison for help with low‐temperature NMR experiments. NMR data were obtained on an NMR instrument made available by National Science Foundation grant DBI‐9724412. Molecular graphics for the computational models were generated using the UCSF Chimera program, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from NIH P41‐GM103311.
Otsuki A. L., Alkan G. A., and Manabe S., et al., Angewandte Chemie International Edition (65, no. 29 (2026): e2727232, 10.1002/anie.2727232
Andrew L. Otsuki, Gizem Atakan Alkan, and Shuhei Manabe are joint first authors.
Contributor Information
Mark Mascal, Email: mjmascal@ucdavis.edu.
Nema Hafezi, Email: nhafezi@ucdavis.edu.
Data Availability Statement
The data that support the findings of this study are available in the Supporting Information of this article.
References
- 1. Lewis E. S. and Vanderpool S. H., “Reactivity in Methyl Transfer Reactions. 2. Leaving Group Effect on Rates With Substituted Thiophenoxides,” Journal of the American Chemical Society 100 (1978): 6421–6424, 10.1021/ja00488a025. [DOI] [Google Scholar]
- 2. Hafezi N. and Mascal M., “Chemistry of the Heterotriquinanes and Heterotriquinacenes,” Synlett 33 (2022): 231–246, 10.1055/s-0040-1719847. [DOI] [Google Scholar]
- 3. Mascal M., Hafezi N., Meher N. K., and Fettinger J. C., “Oxatriquinane and Oxatriquinacene: Extraordinary Oxonium Ions,” Journal of the American Chemical Society 130 (2008): 13532–13533, 10.1021/ja805686u. [DOI] [PubMed] [Google Scholar]
- 4. Gunbas G., Hafezi N., Sheppard W. L., et al., “Extreme Oxatriquinanes and a Record C–O Bond Length,” Nature Chemistry 4 (2012): 1018–1023, 10.1038/nchem.1502. [DOI] [PubMed] [Google Scholar]
- 5. Gunbas G., Sheppard W. L., Fettinger J. C., Olmstead M. M., and Mascal M., “Extreme Oxatriquinanes: Structural Characterization of α‐Oxyoxonium Species With Extraordinarily Long Carbon‐Oxygen Bonds,” Journal of the American Chemical Society 135 (2013): 8173–8176, 10.1021/ja4032715. [DOI] [PubMed] [Google Scholar]
- 6. Prinzbach H., Wessely V., and Fritz H., “Valenzisomere Trioxa‐hexa‐σ/π‐homobenzole,” Tetrahedron Letters 17 (1976): 2765–2768, 10.1016/S0040-4039(00)93833-4. [DOI] [Google Scholar]
- 7. Oral I. and Abetz V. A., “Highly Selective Polymer Material Using Benzo‐9‐crown‐3 for the Extraction of Lithium in Presence of Other Interfering Alkali Metal Ions,” Macromolecular Rapid Communications 42 (2021): 2000746, 10.1002/marc.202000746. [DOI] [PubMed] [Google Scholar]
- 8. Neis C., Kradolfer T., and Hegetschweiler K., “Formation of Lithium, Sodium and Potassium Complexes With 1,3,5‐Triamino‐1,3,5‐trideoxy‐cis‐inositol (Taci),” Acta Crystallographica Section C: Crystal Structure Communication 70 (2014): 632–637, 10.1107/S2053229614011036. [DOI] [PubMed] [Google Scholar]
- 9. Marenich A. V., Cramer C. J., and Truhlar D. G., “Universal Solvation Model Based on Solute Electron Density and on a Continuum Model of the Solvent Defined by the Bulk Dielectric Constant and Atomic Surface Tensions,” Journal of Physical Chemistry B 113 (2009): 6378–6396, 10.1021/jp810292n. [DOI] [PubMed] [Google Scholar]
- 10. Barton D. H. R., Elliott J. D., and Géro S. D., “Synthesis and Properties of a Series of Sterically Hindered Guanidine Bases,” Journal of the Chemical Society, Perkin Transactions 1 (1982): 2085–2090, 10.1039/P19820002085. [DOI] [Google Scholar]
- 11. Grimm D. T. and Bartmess J. E., “Gas‐Phase Acidities of Amines: Proton Affinities and Thermochemistry of Deprotonation Reactions,” Journal of the American Chemical Society 114 (1992): 1227–1231, 10.1021/ja00030a016. [DOI] [Google Scholar]
- 12. Ahlbrecht H. and Schneider G., “Bestimmung der Ionenpaar‐Basizität von Lithium‐ und Kaliumamiden,” Tetrahedron 42 (1986): 4729–4741, 10.1016/S0040-4020(01)82054-8. [DOI] [Google Scholar]
- 13. Edson J. B., Macomber C. S., Pivovar B. S., and Boncella J. M., “Decomposition Pathways of an Alkaline Fuel Cell Membrane Material Component via Evolved Gas Analysis,” Journal of Membrane Science 399–400 (2012): 49–59, 10.1016/j.memsci.2012.01.025. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting File: anie72258‐sup‐0001‐SuppMat.pdf.
Data Availability Statement
The data that support the findings of this study are available in the Supporting Information of this article.
