Dual role of azide as a quencher and stimulator of singlet oxygen generation on some manganese oxides

Abstract

Sodium azide is a known singlet oxygen (¹O₂) quencher in both homogeneous and heterogeneous catalytic transformations. We report a paradoxical trend of a strong enhancement in the ¹O₂ production during H₂O₂ disproportionation observed on some MnO_x catalysts, specifically layered δ-MnO₂. Results show that this activation is due to N₃⁻ complexation with lattice Mn^II of δ-MnO₂ that upon rapid air oxidation produces a metastable surface Mn^III–azide complex that strongly catalyzes H₂O₂ disproportionation. Due to this dual role, the use of NaN₃ as a diagnostic tool for ¹O₂ intermediacy should be carefully reconsidered.

---

Singlet oxygen (¹O₂) is a nonradical reactive oxygen species (ROS) known for its high specificity for electrophilic compounds, including biomolecules such as nucleic acids, lipids and proteins, and is exploited in several industrial transformations. Major diagnostic tools for detecting the presence of ¹O₂ in aqueous media are chemical traps and quenchers. For example, reactive dienes are commonly used traps that selectively react with ¹O₂ to form stable endoperoxide adducts that can be detected by mass spectroscopy.¹ In this process, the traps are consumed depending on the rate of ¹O₂ formation. On the other hand, physical quenchers that deactivate ¹O₂ to triplet oxygen (³O₂) through a radiationless process are the most preferred probes for ¹O₂ since the quencher is not consumed and no new products are formed that can contaminate the product stream. Sodium azide (NaN₃) is the most used physical quencher to monitor the presence of ¹O₂. The deactivation process is as follows:^2,3The rate constant for reaction (1) is high with a measured value of 5.0 × 10⁸ M⁻¹ s⁻¹ in water.⁴ The accepted quenching mechanism is that it proceeds through the formation of a singlet charge transfer complex (reaction (1)) in which the electronic charge of N₃⁻ is partially transferred to oxygen leading to the formation of radical intermediates of superoxide (O₂⁻*) and azidyl (N₃*).² This complex undergoes intersystem crossing and then dissociates to triplet oxygen (³O₂) with the regeneration of the azide anion. Several factors are known to affect this interactive bimolecular quenching rate, such as viscosity,⁵ dielectric constant (polarity),⁵ temperature,⁶ ionic strength of the protic solvents,⁷ and the degree of protonation of azide.⁴ Herein we show that NaN₃ is not a quencher for ¹O₂ produced by some MnO_x, specifically δ-MnO₂. Rather, the presence of N₃⁻ ions enhances ¹O₂ production through the formation of the Mn^III–azide complex that is highly catalytic towards H₂O₂ disproportionation.

We performed tests with several catalysts that were efficient ¹O₂ generators in the absence of azide, namely stoichiometric V₂O₅, δ-MnO₂, acid treated hexagonal δ-MnO₂ (H-δ-MnO₂), (La_0.8Sr_0.2)_0.95MnO_3−δ (henceforth LaSrMnO₃), LiMn₂O₄, and La₂NiO₄.⁸ Two ¹O₂-sensitive fluorogenic probes were used: 1,3 diphenylisobenzofuran (DPBF) and singlet oxygen sensor green (SOSG). DPBF emits a blue fluorescence (∼480 nm) that is proportionally quenched when it reacts with ¹O₂ to form an endoperoxide or 1,2-dibenzoylbenzene that is not fluorescent. The rate constant for this reaction is 2.3 × 10⁹ M⁻¹ s⁻¹.⁹ In a typical run, 1 mg of catalyst was first incubated with 30 µM of probe along with 15 mM of NaN₃ (if present) to achieve stable coverage and then reacted with 10–40 µM of H₂O₂. The emission intensity of the supernatant alone after centrifugation was measured. While a decrease in the DPBF intensity is indicative of the extent of ¹O₂ generation, a decrease can also be caused by sorption of DPBF molecules on the catalyst. For incontrovertible evidence for ¹O₂ production, we used a second probe, SOSG, which shows a positive increase in emission, unlike DPBF, which shows an emission decrease, upon reaction with ¹O₂. In this case, any SOSG adsorption on the catalyst surface will only lead to a decrease and not an increase in the emission intensity. SOSG is made up of a fluorescein-type fluorophore and an anthracene-derived trapping moiety. In the absence of ¹O₂, the emission of the fluorophore is quenched by internal transfer from the anthracene moiety. However, after the reaction of the anthracene moiety with ¹O₂, which forms an endoperoxide that is no longer an efficient intramolecular electron donor, the fluorescence of the fluorophore is restored in proportion to the concentration of ¹O₂. Furthermore, SOSG is also insensitive to ROS such as OH* or O₂⁻* and has a high fidelity unless exposed to UV radiation.^10,11 All SOSG tests were performed under a non-UV excitation wavelength of 405 nm in a 0.1 M Tris buffer prepared with D₂O or H₂O. DPBF tests were performed in a 1 : 1 (v/v) ethanol/water mixture due to its limited solubility in water.

Initial tests to study the effect of N₃⁻ on the ¹O₂ production were performed with vapor-phase grown^12,13 V₂O₅ and δ-MnO₂ as catalysts. V₂O₅ was chosen because it is known to disproportionate H₂O₂ to ¹O₂.¹⁴ Fig. 1a shows the change in the emission intensity of the ¹O₂-sensitive SOSG solution in D₂O. The baseline emission of the supernatant without any H₂O₂ was taken as zero. One observes that for both oxides in the absence of N₃⁻, the SOSG emission increases with increasing H₂O₂ concentrations (test #3 and #5), with δ-MnO₂ displaying higher activity for ¹O₂ generation than V₂O₅. In the presence of N₃⁻, contrasting trends are observed. In V₂O₅, the presence of N₃⁻, which was added prior to the addition of H₂O₂, caused significant quenching of the ¹O₂ signal (test #4), which decreased by 75% with the addition of 15 mM of NaN₃ compared to the case without N₃⁻. This is as expected and indicates that N₃⁻ acts as a ¹O₂ quencher. In contrast, the addition of H₂O₂ to an azide-containing buffer with δ-MnO₂ led to a dramatic increase of ¹O₂ signal by 140–160% (test #6), compared to the same test without azide. This suggests that N₃⁻ induces ¹O₂ production in δ-MnO₂. Note that neither N₃⁻ nor H₂O₂ alone in the absence of an oxide catalyst causes any change in the emission intensity of the probe (test #1 and #2). This shows that the N₃⁻-stimulated ¹O₂ generation observed with δ-MnO₂ is not due to the reaction with the probe. The ¹O₂ generation was confirmed using a D₂O/H₂O isotope effect. The lifetime of ¹O₂ in H₂O is only 2.5 µs but increases nearly 10 times (∼20 µs) in D₂O.^15,16 The solvent-induced quenching of ¹O₂ is most efficient when the infrared vibrational frequencies of the solvent (e.g., H₂O) coincide with the energy of ¹Δ → ³Σ transition of O₂, thus causing deactivation by transferring the correct quanta of energy.¹⁶ In D₂O, vibrational frequencies shift, and ¹O₂ energy transfer becomes less efficient.¹⁶ Consistent with this, the intensity of the ¹O₂ signal observed with δ-MnO₂ and H₂O₂ was almost 3.0–4.5 times higher in D₂O than that observed in H₂O with all other parameters being the same (test #5 and #7). This confirms that SOSG emissions observed with H₂O₂ decomposition are in fact related to ¹O₂. Furthermore, similar measurements probing N₃⁻-induced ¹O₂ generation resulted in the dramatic decrease in the enhancement of SOSG intensity when H₂O instead of D₂O was used as the solvent (test #7). The SOSG signal increased only marginally with H₂O₂ addition with H₂O as the solvent compared to large increase observed in D₂O. This is additional evidence for the ¹O₂ formation.

Fig. 1 Contrasting trends in ¹O₂-related SOSG (a) and DPBF (b) emissions as well as ¹O₂ production (c) caused by the presence of NaN₃ in the electrolyte exposed to V₂O₅, δ-MnO₂, and other oxide powders in the presence of 10–40 µM H₂O₂ concentrations. Also shown is the effect of H₂O₂ addition to the bulk Mn^III–azide complex. Oxide weight = 1 mg ml⁻¹. [NaN₃] = 15 mM.

Although SOSG tests confirmed the effect of azide-induced activation, the amount of ¹O₂ produced could not be calculated due to the lack of information on the probe sensitivity factor. Hence, runs were performed using DPBF, a more well-studied probe.⁹ The testing was also extended to LiMn₂O₄, La₂NiO₄, LaSrMnO₃, and H-δ-MnO₂ that are also good ¹O₂ generators⁸ to test the quenching/stimulating effects of N₃⁻. Fig. 1b and c compare DPBF emission changes of the supernatant and ¹O₂ produced for various oxides with and without the presence of azide. The results indicate that V₂O₅ and then H-δ-MnO₂ are the most active catalysts for the disproportionation of H₂O₂. The addition of azide (Fig. 1b and Fig. S1), stimulates ¹O₂ generation, as observed from the dramatic decrease in emission intensity with increasing concentrations of H₂O₂ in the presence of azide. However, only a marginal decrease in emission intensity is observed with other oxides, such as La₂NiO₄ (Fig. 1b and Fig. S1) and LiMn₂O₄ (Fig. S1). Thus, NaN₃ acts as a quencher for ¹O₂ on these catalysts like that observed on V₂O₅. Fig. 1c plots the ¹O₂ concentration calculated from the calibration curve (Fig. S2) and sensitivity factor⁹ as a function of H₂O₂ concentration. The presence of azide leads to an increase in ¹O₂ production only in δ-MnO₂. The acid treatment of δ-MnO₂ (H-δ-MnO₂) at pH = 3 is known to increase the Mn^II and Mn vacancy concentrations in the lattice.¹⁷ Fig. 1b and c show that, even in the absence of N₃⁻, H-δ-MnO₂ exhibits a higher catalytic activity for H₂O₂ to ¹O₂ conversion than δ-MnO₂. In the presence of N₃⁻, the ¹O₂ is further enhanced, much more than that observed with δ-MnO₂. Therefore, the presence of Mn^II/vacancy leads to an enhancement of azide stimulation.

It is known that Mn^II–azide complexes in the bulk undergo slow oxidation in air-saturated solutions to form the Mn^III–azide complex that exhibits a visible absorption peak at 430 nm.^18,19 We confirmed this by adding a 3–30 mM NaN₃ solution directly to air-saturated 0.1 M Mn^IISO₄ under ambient conditions. As shown in Fig. 2a, the initially light pink solution of Mn^II turned into increasing shades of dark orange upon oxidation, which is characteristic of the Mn^III–azide complex. The absorbance spectrum (Fig. 2b) shows a marked increase in absorbance between 430 and 530 nm that increases with an increase in NaN₃ concentration (inset of Fig. 2b). This peak is characteristic of Mn^III complexes.²⁰ Therefore, it is possible that the N₃⁻ can bind to the lattice Mn^II of the δ-MnO₂ catalyst that after rapid oxidation in air forms a metastable Mn^III–azide complex that catalyzes the disproportionation of H₂O₂ to ¹O₂. To confirm this, we monitored in situ the absorbance change of a δ-MnO₂ film with the addition of 15 mM of NaN₃ solution (Fig. 2c). The spectrum taken immediately after NaN₃ addition shows a marked increase in absorbance at a wavelength less than 530 nm, as also evident in the ΔA = A_δ-MnO2,NaN3 − A_δ-MnO2 spectrum (inset), which provides clear evidence for the fast lattice Mn^III formation. To understand why N₃⁻-induced ¹O₂ enhancement is specific to δ-MnO₂, we analysed the lattice Mn cation composition of different MnO_x through X-ray photoelectron spectroscopy and photoluminescence (PL) spectroscopy, as detailed elsewhere,²¹ and the data are provided in Fig. S3 and Fig. 2d. LaSrMnO₃ has the highest concentration of Mn^III (37%) and the lowest concentration of Mn^II (1%) in the lattice. Whereas δ-MnO₂ has the highest Mn^II content in its lattice (5%), which after acid treatment increases further to 14%. Given that H-δ-MnO₂ shows a higher enhancement in the N₃⁻-induced ¹O₂ production than that observed in δ-MnO₂, the effect of N₃⁻ is likely correlated with the increased presence of Mn^II in the lattice. After azide treatment, the lattice Mn^III concentration of H-δ-MnO₂ increased from 24% to 29%, while the Mn^II concentration decreased from 14% to 6%. This further confirmed that Mn^II after azide treatment is converted to Mn^III. The lower azide-induced ¹O₂ production observed with other MnO_x may be related to the presence of a fewer Mn^II lattice sites for azide complexation. Note that when H₂O₂ was added to the bulk Mn^III–azide complex in the presence of SOSG (test #8, Fig. 1), the intensity of the ¹O₂-related SOSG signal decreased rather than increased with H₂O₂ addition. This also caused the quenching of absorbance at 530 nm (Fig. 2b). Therefore, the bulk Mn^III–azide does not catalyze the disproportionation of H₂O₂ to ¹O₂.

Fig. 2 (a) and (b) Photographs and absorbance spectrum of bulk Mn^II solution with the addition of NaN₃ and H₂O₂; (c) in situ changes in the absorbance and differnce absorbance (inset) spectrum of δ-MnO₂ due to NaN₃ addition; (d) comparison of the lattice Mn cation composition of pristine and azide-treated Mn oxide powders obtained from PL spectroscopy; (e) Mott–Schottky plots of H-δ-MnO₂ and δ-MnO₂ (inset) in the presence/absence of azide; and (f) changes in the E_h (OCP) of the δ-MnO₂ film upon sequential addition of 3 mM NaN₃ and then 10 µM of H₂O₂ pulses.

These different quenching/stimulating effects of azide on ¹O₂ may be related to the differences in the thermodynamic redox potential. The redox potential of the ¹O₂/H₂O₂ reaction, , is 0.77 V versus the standard hydrogen electrode (SHE) at pH = 7 for H₂O₂ and ¹O₂ concentrations of 10 µM. The redox potential of the Mn^III−N₃⁻/Mn^II−N₃⁻ complex in the bulk is only 0.66 V²² and, therefore, cannot induce H₂O₂ disproportionation to ¹O₂, as observed here. On the other hand, V₂O₅, δ-MnO₂ and H-δ-MnO₂ are known to be semiconductors with high work function (Φ) and electron affinity (χ) values (5.6–6 eV).^17,20,23 The flat band potential (U_FB) values determined from the Mott–Schottky (M–S) plots (Fig. 2e) of δ-MnO₂ and H-δ-MnO₂ are 1.2 V and 1.5 V versus SHE, respectively, with the former and the latter being n-type and p-type semiconductors, respectively. These U_FB values are more positive than , which explains why both materials have a high catalytic activity for ¹O₂ production from H₂O₂. In the presence of azide, the bulk U_FB does not shift significantly, indicating that azide does not modulate the bulk electronic properties. However, its addition leads to a positive shift in the oxidation–reduction potential (E_h, Fig. 2f) or open circuit potential (OCP), which is more reflective of the surface Fermi energy (E_F) changes.²⁴ This E_h value corresponds to the relative difference between the surface potential of δ-MnO₂, equivalent to E_F under band bending conditions, and that of the reference electrode.²⁴ The concomitant increase in the lattice Mn^III content suggests that the Mn^III formation causes a decrease in the surface E_F energy relative to vacuum energy (increase in Φ), thus producing an greater driving force for e⁻ injection from the ¹O₂/H₂O₂ reaction. Mn^III is also the active site for H₂O oxidation.²⁵ Consistent with this, the addition of H₂O₂ results in a negative shift in E_h, implying an increase in the surface E_F energy (decrease in Φ). This is as expected since e⁻ injection would lead to an increase in E_F energy and a conversion of Mn^III back to Mn^II (Fig. 2b).

In summary, NaN₃ can be both a quencher and a stimulator for ¹O₂ production.

Conflicts of interest

All authors declare that they have no conflicts of interest.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: emission spectrum, calibration curves, and experimental details. See DOI: https://doi.org/10.1039/d6cc01794k.

Acknowledgements

The authors are grateful for the financial support provided by the Howard P. Isermann fellowship and the Rensselaer Polytechnic Institute.

References

1. J. Tang, J. Chen, Z. Zhang, Q. Ma, X. Hu, P. Li, Z. Liu, P. Cui, C. Wan, Q. Ke, L. Fu, J. Kim, T. Hamada, Y. Kang and Y. Yamauchi, Chem. Sci., 2023, 14, 13402–13409.
2. J. R. Harbour and S. L. Issler, J. Am. Chem. Soc., 1982, 104, 903–905.
3. C. Schweitzer and R. Schmidt, Chem. Rev., 2003, 103, 1685–1758.
4. W. R. Haag and T. Mill, Photochem. Photobiol., 1987, 45, 317–321.
5. N. Miyoshi, M. Ueda, K. Fuke, Y. Tanimoto, M. Itoh and G. Tomita, Z. Naturforsch. B, 1982, 37, 649–652.
6. M. Li, C. S. Cline, E. B. Koker, H. H. Carmichael, C. F. Chignell and P. Bilski, Photochem. Photobiol., 2001, 74, 760–764.
7. M. A. Rubio, D. O. Mártire, S. E. Braslavsky and E. A. Lissi, J. Photochem. Photobiol., A, 1992, 66, 153–157.
8. S. Mangu and V. Chakrapani, J. Phys. Chem. C, 2026, 130, 6180–6191.
9. T. Entradas, S. Waldron and M. Volk, J. Photochem. Photobiol., B, 2020, 204, 111787.
10. X. Ragàs, A. Jiménez-Banzo, D. Sánchez-García, X. Batllori and S. Nonell, Chem. Commun., 2009, 2920–2922,  10.1039/b822776d.
11. H. Liu, P. J. Carter, A. C. Laan, R. Eelkema and A. G. Denkova, Sci. Rep., 2019, 9, 8393.
12. S. Lee, Q. Wang and V. Chakrapani, Phys. Rev. Mater., 2024, 8, 125801.
13. V. Chakrapani, M. Brier, A. Puntambekar and T. DiGiovanni, J. Mater. Res., 2016, 31, 17.
14. F. Natalio, R. André, A. F. Hartog, B. Stoll, K. P. Jochum, R. Wever and W. Tremel, Nat. Nanotechnol., 2012, 7, 530–535.
15. N. Miyoshi, M. Ueda, K. Fuke, Y. Tanimoto, M. Itoh and G. Tomita, Z. Naturforsch. B, 1982, 37, 649–652.
16. P. B. Merkel and D. R. Kearns, J. Am. Chem. Soc., 1972, 94, 7244–7253.
17. C. Wang, N. Smieszek and V. Chakrapani, Chem. Mater., 2021, 33, 7805–7817.
18. H. D. Moya, E. A. Neves, M. E. Vázquez and N. Coichev, Talanta, 1996, 43, 67–72.
19. H. D. Moya, E. A. Neves and N. Coichev, Spectrosc. Lett., 2001, 34, 537–547.
20. C. Wang and V. Chakrapani, ACS Earth Space Chem., 2023, 7, 774–787.
21. V. Chakrapani, C. Wang, Q. Wang and N. Smieszek, Surf. Interface Anal., 2022, 54, 1192–1202.
22. H. D. Moya, E. A. Neves and N. Coichev, Talanta, 1997, 44, 797–803.
23. S. Lee, Q. Wang and V. Chakrapani, J. Appl. Phys., 2024, 136, 205101.
24. V. Chakrapani, in Metal Ions and the Route to Life, ed. W. Nitschke and S. Duval, CRC Press, Boca Raton, 1st edn, 2025, p. 24 DOI:10.1201/9781003459910-7.
25. I. Roy, C. Wang, N. Smieszek, X. Li, L. Tsapatsaris and V. Chakrapani, ChemSusChem, 2022, 15, e202200062.

---