Computational evaluation of RM resistance to oxidation by 1O2.
To confirm that the bridged BAC molecule we envisioned would be resistant to parasitic oxidation processes, we first performed DFT calculations on BAC, BP55 and BP66 to assess the activation barriers for their oxidative degradation through Cα hydrogen abstraction by 1O2 (Fig. 3). Cα hydrogen abstraction by 1O2 of alkylamines can proceed on either the closed-shell singlet (CSS) or the open-shell singlet (OSS) surface (Fig. 3a). After the formation of an initial encounter complex I, the CSS process (via TS-1) produces an ion pair consisting of an iminium and the OOH anion, which then combine in a barrierless or near-barrierless fashion to generate the zwitterionic structure III. The OSS process (via TS-2) produces a carbon-centered radical and the OOH radical, which can either recombine, undergo a radical-polar crossover to the CSS surface (vide infra), or dissociate to react with other chemical species present in the system.
Fig. 3b shows the free energy evolution along the two possible Cα hydrogen abstraction pathways for BAC, BP55 and BP66. Our calculations showed that for all three RMs surveyed, the OSS Cα hydrogen abstraction transition state TS-2 is more favorable than the CSS transition state TS-1. BAC was also found to have much higher Cα hydrogen abstraction barriers than either BP55 or BP66. (It is worth noting that for all three RMs, the CSS and OSS surfaces intersect after TS-2; it is therefore plausible that a radical-polar crossover event will follow TS-2, leading to intermediate III.)
The lower energy of TS-2a compared to TS-1a can be rationalized by noting that TS-1a momentarily produces an iminium ion with the C=N double bond at a bridgehead position, which is highly strained according to Bredt’s rule. In contrast, TS-2a produces a carbon-centered radical at the bridgehead position instead of forming a full double bond, lessening the strain. Despite being the more favorable pathway of the two, TS-2a still has a prohibitively high activation barrier of 37.9 kcal/mol, which lends support to our expectation that the bridged structure of BAC could confer exceptional stability under oxidative conditions. In contrast, the OSS Cα hydrogen abstraction barriers for BP55 and BP66 were only 20.4 and 22.6 kcal/mol, respectively (Fig. 3c), indicating that these two RM species are susceptible to oxidative degradation by 1O2 at room temperature.
We next considered possible degradation of the RM molecules by 1O2 through concerted H2 abstraction processes on the CSS surface (Fig. 4). This process leads to the generation of H2O2 and an enamine (Pathway A) or alkene (Pathway B) depending on the location of H2 abstraction (Fig. 4a). Fig. 4b shows the free energy evolution along the possible concerted H2 abstraction pathways for BAC, BP55 and BP66. For both BP55 and BP66, the enamine-generating Pathway A was found to be more favorable, consistent with prior literature findings that neighboring amine substituents accelerate oxidative degradation through concerted H2 abstraction27. For BAC, on the other hand, Pathway A would lead to a highly strained "bridgehead" double bond, resulting in an extremely high 52.2 kcal/mol barrier. TS-4a, which avoids placing the double bond in the bridgehead position, has a much lower barrier of 35.4 kcal/mol (Fig. 4c), albeit still prohibitively high for room-temperature conditions. These calculations further established that the bridged bicyclic structure of BAC would render it resistant to oxidative degradation by 1O2 through multiple possible mechanisms.
A direct comparison of OSS (radical) oxidation of the three RMs by 1O2 is shown in Fig. 5a. The calculated energies indicate that BAC is dramatically more resistant to OSS oxidation by 1O2 than BP55 or BP66. The unusually high resistance to OSS oxidation for BAC can be attributed to the geometry of its radical form IVa after hydrogen atom transfer. In the BP55- and BP66-derived radicals IVb and IVc, the carbon radical centers are able to attain mostly planar geometries, as would be preferred in unstrained model systems (see Supplementary Fig. 1 for comparison). In the BAC-derived radical IVa, however, planarization is energetically extremely costly for the bridgehead carbon, leading to the carbon radical center being much more nonplanar (Fig. 5b). This leads to poor delocalization of the unpaired electron and higher energies as a result. Similarly, the high energy cost of planarizing a bridgehead carbon also renders CSS oxidation of BAC (through TS-1a or TS-3a) extremely unfavorable. In addition, O2•− degradation of BAC was also found to be energetically all the way uphill. Overall, our computational analyses confirmed the soundness of our Bredt’s-rule-based design principle and predicted that BAC would be exceptionally stable to oxidative degradation by 1O2 at room temperature.
Investigation of reactivity of RMs with 1O2.
To substantiate the computational findings regarding the stability of BP55, BP66, and BAC against 1O2, each RM was synthesized16,28, and a method involving 1O2-enriched environment is employed for degradation monitoring; upon exposure to 525 nm LED, tetraphenylporphyrin (TPP) is initially excited to its singlet state and undergoes intersystem crossing (ISC) to the triplet state, which, in the presence of proximal 3O2, facilitates energy transfer, leading to the formation of the highly reactive 1O2 species29,30 (Fig. 6a). In the 1O2 experiment, NMR analysis was concurrently performed to monitor the degradation of RMs, with Fig. 6b distinctly emphasizing the C–H bonds designated for tracing across RMs. The 1H NMR spectra, displayed in black and bluish purple in the first and second rows (Figs. 6c–e and Figs. 6f–h), depict the conditions of each RM prior to exposure to 1O2, without and with the electrolytes and photocatalyst, respectively; the elevated, dense peaks in chemical shift ranging from 3.3 to 3.7 represent the proton patterns of tetraethyleneglycol dimethylether (TEGDME) as one of the electrolyte molecules. The analytes containing the RMs, the electrolytes, and the photocatalyst were exposed to 1O2 environment, and the corresponding spectra are obtained and shown in Figs. 6i–k. Characteristic proton peaks—eight of each Hα and Hβ for BP55 and eight of each Hα and Hβ with four of Hγ for BP66—completely disappeared while new peaks, belonging to an unidentified byproduct(s), were detected in the spectra (Figs. 6i,k). These results were consistent with our computational prediction that oxidative degradation through the OSS pathway would be feasible at room temperature for both BP55 (20.4 kcal/mol barrier, corresponding to the difference between II and TS-2) and BP66 (22.6 kcal/mol). However, BAC demonstrated chemical stability without unintended deterioration, following exposure to 1O2 as shown in Fig. 6k.
Electrochemical behavior of RMs.
The durability of RMs towards ROS was evaluated via CV by comparing the electrochemical activity of RMs before and after exposure to ROS. All examined RMs demonstrated a redox potential under 3.54 V, conforming to the criteria for diminishing 1O2 evolution during charging process (Figs. 7a–c). Nevertheless, both BP55 and BP66 were found to be devoid of redox-active properties after exposure to 1O2, indicating a considerable loss of electrochemical activity due to the oxidative effects of 1O2 (Figs. 7d,e). This observation reveals the irreversible reaction triggered by 1O2 resulting in the formation of byproducts lacking in redox activity, which is characteristic of trap-type-RM behavior10. One notable is that no discernible changes in CV profiles of the RMs upon exposure to O2•− were shown implying the resilience of BP55 and BP66 to O2•−, compared to their sensitivity to 1O2 (Supplementary Figs. 12a,b). Remarkably, BAC maintains reversible CV profiles even after exposure to either 1O2 or O2•−, sustaining redox activity without any potential shift (Figs. 7c,f and Supplementary Fig. 12c). This persistence is attributed to the chemical robustness of BAC, as proven by NMR analysis, which also underlines high durability of BAC in maintaining electrochemical activity.
The stability of RMs during galvanostatic cycling test was verified in LOB configuration (Supplementary Fig. 13), with the initial charging potential of BP66 approaching 3.50 V and a longer plateau than BP55. In general, the relationship between the redox potentials of RMs and kinetic of Li2O2 decomposition adheres to an inverted parabola shape in accordance with Marcus theory16. Given that the kinetic of Li2O2 decomposition approaches saturation around 3.70 V10, 31, it is reasonable to deduce that BP66, with its higher redox potential compared to BP55, exhibits enhanced kinetics for Li2O2 decomposition, resulting in a longer charging voltage plateau. However, the terminal charging voltage for both BP55 and BP66 had progressively increased by the third cycle, reaching the typical charging voltage of LOBs without a redox catalyst. For both BP55 and BP66, a decline in the retention of the charging voltage was observed, indicative of the RMs undergoing degradation concurrently with the cycling process. Moreover, amount of evolved 1O2 accumulated over cycles, which affects the capacity of both BP55 and BP66 to sustain the low charging voltages. Differing from BP55 and BP66, BAC retained a stable redox potential before and after exposure to 1O2. The chemical stability of BAC, anchored in its resistance to 1O2, further delineates the proficiency of BAC in oxidizing Li2O2.
To confirm the resilience of BAC against 1O2, relying solely on electrochemical measurements proves insufficient. Hence, to validate the continuous performance of BAC as an RM post 1O2 exposure, the species of gases evolved were analyzed by DEMS during the initial step of the charging process in the presence of three RMs: BP55, BP66, and BAC. Fig. 8a shows the voltage and gas evolution profiles comparing BP55 before and after 1O2 exposure. Before 1O2 exposure, BP55 exhibited charging plateau at 3.25 V (vs Li+/Li) producing certain amount of O2 with suppressed H2 and O2 gas. However, 1O2 exposed BP55 only gave concentrated O2 evolution during the first hour of charging and nearly no O2 evolved. Large amount of CO2 rather evolved after indicating extensive side reactions occurred at high charging potential32–3334 and BP55 lost function as RM by 1O2. BP66 demonstrated results analogous to those of BP55 (Fig. 8b). BP66 before 1O2 exposure exhibited higher charging potential than BP55, which is in accord with Fig.6. A little increased amount of O2 than that with BP55 was detected and it was due to the higher charging kinetics of BP66 with higher charging potential. After 1O2 exposure, BP66 also reacted with 1O2 and lost catalytic function, therefore the O2 evolution profile changed to unstable and large amount of CO2 was detected. That is, BP55 and BP66 has trap-type RMs characteristics and 1O2 critically affects to their function as RM eventually deactivating them. Fig. 7c shows the behavior of BAC which contrasts to BP55 or BP66. In both before and after 1O2 exposure, BAC gave catalytic O2 evolution profiles implying highly 1O2 resistive characteristic of BAC (Fig. 8c). Particularly, CO2 evolution was suppressed even after 1O2 exposure which is indicative of well-preserved functionality of BAC as RM. Based on the gas evolution profiles in Fig. 8, accumulated gas evolution rate was recalculated (Supplementary Fig. 15 and Supplementary Table 1). Proportion of O2 before 1O2 exposure was comparable for BP55 and BP66, 47 and 43% respectively, but still lower than BAC suggesting comparatively low Li2O2 decomposition kinetic of BP55 and BP66. The solution of BP55 and BP66, following exposure to 1O2, gave a depressed ratio of O2, with CO2 levels rising from 42 to 74% and from 40 to 67%, respectively. Increment of CO2 demonstrates that BP55 and BP66 lose function as RM by the aggressive attack of 1O2 and side reactions occurred intensively at high charging potential. In contrast, BAC showed highly catalytic behavior with the ratio of evolved O2 even after 1O2 exposure comparable to before 1O2 exposure (79 and 82%, respectively). The solution of BAC, after exposure to 1O2, exhibited an increased H2 ratio compared to its pre-exposure state. However, the total ratios of H2 and CO2 in the solution of BAC before and after exposure were nearly identical, suggesting that minimal side reactions occurred at similar rates in both environments. Electrochemical measurements consistently point out that BP55 and BP66 have no durability against 1O2, thereby both are significantly fragile to 1O2-involved oxidation. More importantly, such 1O2 reaction makes BP55 and BP66 lose their electrochemical activity. BAC, on the other hand, has highly 1O2 durable molecular structure and therefore promise preservation of electrochemical activity even under the continuously 1O2 evolving environment, the LOBs.