Solvent-dependent reactivity of azo-BF₂ switches

Abstract

Azo-BF₂ complexes are visible and near-infrared (NIR) light-activated photoswitches showcasing versatility in applications ranging from energy storage to three-dimensional displays. While highly appealing as molecular switches, factors that affect their photophysical properties and solution-state reactivity still need to be teased out. In this paper, we present the synthesis and characterization of two azo-BF₂ analogues (1 and 2), each modified with structurally distinct dimethylamine-substituted naphthalene moieties. In addition to investigating the effect of the π-system expansion on the photophysical and photoswitching properties of the system, we also elaborate on the stability (i.e., whether they undergo solvolysis or 1,2-BF₂ shift) of the switches in different solvents. Specifically, switch 1 absorbs in the NIR region, enabling its activation with 700 nm light, while switch 2 exhibits enhanced separation between the trans and cis isomer absorption bands, resulting in an improved photostationary state. As for the solvent effect, we discovered that polar aprotic solvents induce an intramolecular 1,2-BF₂ shift in both switches, transforming the azo-BF₂ photoswitches into boron difluoride hydrazone fluorophores, whereas polar protic solvents facilitate the solvolysis of the azo-BF₂ into the starting hydrazone derivative. In the former, the donor number of the solvent is a major factor in determining the obtained outcome, while in the latter, it is the solvent's hydrogen-bond donation capability. These insights into the design strategy and solvent-mediated reactivity of azo-BF₂s will contribute to their further development into efficient NIR-responsive photoswitches, paving the way for innovative applications in smart materials and molecular devices.

---

Introduction

Visible and near-infrared (NIR) light-activated photoswitches have gained popularity in the fields of photopharmacology and adaptive materials in recent years,^1–8 because they enable deep substrate penetration while maintaining low phototoxicity. This interest resulted in the development of an array of red-light and NIR-activated photoswitches, such as azobenzenes,^9–17 diarylethenes,^18–21 indigoids,^22–25 dihydropyrenes,^26,27 donor–acceptor Stenhouse adducts,^28–32 and heteroaryl azoswitches.^33,34 Our contribution to this research area was the introduction of azo-BF₂ photoswitches,³⁵ which were the first azobenzene derivatives to be activated with NIR light.³⁶ These switches also exhibit other sought-after characteristics, such as high photostationary states (PSS), enhanced isomerization quantum yields (Φ), and negative photochromism. We recently took advantage of these unique features to develop a rewritable, solid-state, hand-held volumetric 3D photochromic display capable of showcasing both static 3D images and dynamic 2D animations.³⁷ We also demonstrated how such photoswitches can be used in molecular solar thermal energy storage applications.³⁸ Beyond these applications, structural modifications of azo-BF₂ switches have revealed several key insights: (1) extending the π-system from quinolinyl to phenanthridinyl results in self-aggregates that enhance the half-life (τ_1/2) of its thermal relaxation;³⁹ (2) an oxygen atom adjacent to the BF₂ bridge can induce a thermal 1,2-BF₂ shift, converting the azo-BF₂ photoswitch into a borondifluorohydrazone (BODIHY) in both solution and the solid state;⁴⁰ and (3) increasing the electron density through para-substitution causes a bathochromic shift in the activation wavelength, enabling control over the trans/cis isomerization process using NIR light.³⁶ Nonetheless, and despite the many promising features of azo-BF₂ switches, practical applications in the solution state are hindered by their reactivity, especially in polar media, particularly water. Hence, understanding the solvent interactions with these switches is an essential step towards extending their utility, especially in vivo settings.

Here, we report on two azo-BF₂ analogues (1 and 2) that incorporate an N,N-dimethylnaphthalen-1-amine or N,N-dimethylnaphthalen-2-amine moiety attached at the 5 or 6-position, respectively (Fig. 1a). Our initial goal was to investigate how modifications to the π-system will influence the activation wavelength and overall photoswitching efficiency compared to the parent azo-BF₂. Structural analysis revealed that these two switches exhibit distinct photophysical properties and photoswitching performances. Specifically, the absorption spectrum of 1 extends into the NIR region, thus allowing NIR activation. In contrast, switch 2 has a more pronounced separation between the trans and cis absorption bands, resulting in an improved PSS ratio.

Fig. 1 (a) The structures of the two azo-BF₂ derivatives 1 and 2. (b) Photographs of the solution of trans-1 and (c) trans-2 before and after the 1,2-BF₂ shift in various solvents, along with their corresponding molecular structures. (d) Kinetic profiles of the 1,2-BF₂ shift of trans-1 and (e) trans-2 in DMSO, DMF, and DMAc (4 × 10⁻⁵ M).

To gain a deeper understanding of how polar solvent environments affect the molecular properties of the switches, we systematically examined the behavior of both derivatives under various solvent environments. Our focus was on understanding the correlation between their stability and key solvent parameters, including donor number (DN), dipolarity/polarizability (π*), basicity, and hydrogen bond donation ability (α).⁴¹ We discovered that polar aprotic solvents such as dimethyl sulfoxide (DMSO), dimethylformamide (DMF), and dimethylacetamide (DMAc) induce the intramolecular 1,2-BF₂ shift in both derivatives, leading to the formation of the BODIHY product (Fig. 1b and c). In solvents such as acetone, tetrahydrofuran (THF), ethyl acetate (EA) and polar protic alcohols both switches undergo solvolysis yielding their starting hydrazones, though at different time-scales. A comprehensive analysis of solvent parameters revealed a strong correlation between the DN of the solvents and the rate of BODIHY formation, suggesting that solvents with a greater lone pair donation ability result in a more efficient 1,2-BF₂ shift. In solvents with lower DN (e.g., acetone, THF, and EA), both switches undergo solvolysis, regenerating their original hydrazones on a timescale comparable to that of the BF₂ shift. We hypothesize that the trace amounts of water (<50 ppm) present in these solvents facilitate solvolysis; in contrast, in non-nucleophilic solvents such as toluene and dichloromethane (DCM), no reaction is observed. On the other hand, the relatively fast solvolysis of the azo-BF₂ switches to their hydrazone precursors in polar protic alcohols is primarily driven by the solvents' hydrogen bond donation ability (α).

Results and discussion

Azo-BF₂s 1 and 2 were synthesized by reacting their respective hydrazone precursors with boron trifluoride diethyl etherate, in the presence of N,N-diisopropylethylamine (Schemes S1 and S2). The azo-BF₂ switches and their precursors were fully characterized using NMR spectroscopy and mass spectrometry (Fig. S1–S13).

The photoisomerization performance of the switches in DCM was evaluated using UV-Vis and NMR spectroscopies (Fig. S14–S20), and the data are summarized in Table 1. Switch 1 predominantly exists in its thermodynamically stable trans form in the dark (95%), with a maximum UV-Vis absorbance (λ_max) at 675 nm (ε = 34 300 M⁻¹ cm⁻¹, Fig. 2c) and an absorption tail extending to approximately 800 nm. While the λ_max is slightly blue-shifted relative to the parent system (λ_max = 680 nm), the absorption tail extends further by 50 nm into the near-infrared region, making it the most red-shifted azo-BF₂ system reported to date. These results indicate that attaching a naphthyl group to the azo-BF₂ core through the β-position effectively elongates the π-conjugation. Upon irradiation with 700 nm light, the cis isomer (λ_max = 620 nm; ε = 17 500 M⁻¹ cm⁻¹, Fig. 2c) becomes predominant, accompanied by a slight color change of the solution from greenish cyan to cyan blue (Fig. 2a). The thermal half-life (τ_1/2) of the metastable cis-1 isomer in aerated DCM was measured to be 2.2 ± 0.01 min at 294 K (Fig. S25). Because of the rapid back isomerization, accurate PSS₇₀₀ could not be determined. The lowest estimated amount of the cis isomer at PSS₇₀₀ is 47% (Fig. S16), with a quantum yield (Φ_trans→cis) of 13.6 ± 1.7% (Fig. S21). This value is underestimated because of the fast back isomerization process. A deduced absorption curve for cis-1, based on previously reported methods,^42,43 gives a λ_max of 595 nm with an ε of 15 800 M⁻¹ cm⁻¹ (Fig. 2c), yielding a Δλ_max value of 80 nm. Further irradiation at 480 nm results in a PSS₄₈₀ of 63% trans (Fig. S16) and a Φ_cis→trans of 35.1 ± 2.7% (Fig. S22). This value is overestimated because of the fast back isomerization process. The relatively low PSS values for both the forward and backward photoisomerization processes are attributed to the significant overlap of the absorption bands of the two isomers and the short τ_1/2. The photoswitching can be cycled multiple times with no signs of photodegradation (Fig. S14c).

Table 1 Summary of photophysical data for azo-BF₂s 1 and 2 in DCM

Azo-BF2	λ abs (nm)/ε^b (M^−1cm^−1)		PSS (%) @ λirr^c (nm)		Φ isom (%)^d		τ 1/2 (min)
	trans	cis	trans : cis	trans : cis	trans → cis	cis → trans
a Absorption maximum. b Extinction coefficient. c Photostationary state measured at the optimal irradiation wavelength. d Photoisomerization quantum yield. e Thermal half-life of the cis → trans transformation at 298 K. f Deduced using previously reported methods.^41,42
1	675/34 300	595/15 800^f	25 : 75 @ 700^f	63 : 37 @ 480	13.6 ± 1.7	35.1 ± 2.7	2.2 ± 0.1
2	578/21 300	473/19 800^f	21 : 79 @ 650^f	67 : 33 @ 442	6.3 ± 0.1	24.7 ± 2.3	384 ± 4

---

Fig. 2 Light-induced trans/cis isomerization of (a) 1 and (b) 2, along with the corresponding UV/vis absorption changes (c) and (d), respectively, in DCM (4 × 10⁻⁵ M).

Switch 2 exists predominantly in the trans form (62–64%, Fig. S53 and S54) when equilibrated under dark (λ_max = 578 nm, ε = 21 300 M⁻¹ cm⁻¹, Fig. 2d). The λ_max of trans-2 is blue shifted by around 100 nm from that of the parent compound (λ_max = 680 nm) and trans-1 (λ_max = 675 nm), as the NMe₂ group is not in direct conjugation with the azo core. Upon irradiation with 650 nm light, trans-2 switches to its cis form (λ_max = 475 nm, ε = 17 500 M⁻¹ cm⁻¹; deduced λ_max = 473 nm, ε = 19 800 M⁻¹ cm⁻¹; Fig. 2d), accompanied by a drastic color change of the solution from purple to brown (Fig. 2b). The PSS₆₅₀ for trans-2 was measured to be 74% cis (Fig. S17), with a Φ_trans→cis of 6.3 ± 0.1% (Fig. S23). Switch 2 exhibits a better band separation between the trans and cis isomers than switch 1 (Δλ_max: 105 vs. 80 nm, respectively). The reverse process (cis → trans) can be triggered using a 442 nm light source, resulting in a PSS₄₄₂ of 67% trans (Fig. S17) and a Φ_cis→trans of 24.7 ± 2.3% (Fig. S24). The τ_1/2 for the cis → trans process of switch 2 was calculated to be 384 ± 4 min (Fig. S26), approximately 175 times longer than that of switch 1. Again, and as previously discussed, this difference is attributed to the fact that the NMe₂ group is not in direct conjugation with the azo core. The longer τ_1/2 allows for a better assessment of the system's photoswitching efficacy. The photoswitching can be cycled numerous times with no signs of photodegradation (Fig. S15c).

The major challenge with using azo-BF₂ switches in the solution state lies in the stability of the N → B coordination bond, which can undergo 1,2-BF₂ shift and/or hydrolysis. It should be noted here that this reactivity is mainly observed for the trans form, while the cis isomer is far more stable.^36,40 To probe the effect of solvent polarity and proticity on the stability of this bond, we conducted a comprehensive investigation across a range of solvents (Fig. 1d and e). Both switches exhibit good stability, i.e., undergo no reaction, in non-polar solvents such as DCM (Fig. S41 and S42) and toluene (Fig. S53 and S54). On the other hand, polar aprotic solvents such as DMSO, DMF, and DMAc initiate an intramolecular 1,2-BF₂ shift in both switches, leading to the formation of BODIHY products (Fig. S35–S37 and S44–S46). We hypothesize that these polar aprotic solvents coordinate with the boron core of the azo-BF₂ switches, leading to the cleavage of the intramolecular N → B coordination bond (Fig. 3a). The resulting bond dissociation increases the rotational freedom of the C–N single bond, allowing it to adopt a conformation that favors the formation of a new and more stable intramolecular N → B bond. This transformation is reminiscent of our earlier finding, where an oxygen atom located near the BF₂ group catalyzed the 1,2-BF₂ shift, resulting in the formation of a BODIHY product.⁴⁰ The rate of the shift in trans-1 based on UV/Vis spectroscopy studies follows the order DMSO > DMF > DMAc, with a τ_1/2 of 0.68 ± 0.03, 1.8 ± 0.2, and 3.3 ± 0.4 h, respectively (Table 2). The formation of BODIHY was confirmed using ¹H and ¹⁹F NMR spectroscopies in DMSO-d₆ (Fig. S27). Specifically, the ¹⁹F spectra showed the disappearance of the initial azo-BF₂ resonance signal at −145.0 ppm and the simultaneous appearance of a new signal at −123.5 ppm, which corresponds to the corresponding BODIHY dye.⁴⁰ The formation rate of BODIHY was determined to be 0.53 h, which aligns well with the result obtained from the UV/Vis measurements. In aprotic solvents with lower donor numbers, such as THF, acetone, and EA (Fig. S38–S40 and S48–S50), no 1,2-BF₂ is observed. Nonetheless, even though ultra-dry (H₂O < 50 ppm) solvents are used, the switches undergo slow solvent-assisted solvolysis (Fig. 3b). The reaction follows the order: THF > acetone > EA, with τ_1/2 of, 44 ± 3, 53 ± 7, and 82 ± 15 h, respectively for trans-1. Similar trends are observed for trans-2 (Table 2).

Fig. 3 Proposed reaction mechanisms for azo-BF₂ compounds. (a) 1,2-BF₂ shift in polar aprotic solvents having high DN (e.g., DMSO); (b) solvent-assisted slow solvolysis in polar aprotic solvents with low DN (e.g., acetone); and (c) rapid solvolysis in polar protic solvents (e.g., MeOH).

Table 2 Summary of decomposition data for both azo-BF₂s in aprotic solvents

Azo-BF2s	Donor number	1		2
		τ 1/2 (h)	k (s^−1)	τ 1/2 (h)	k (s^−1)
a Trans-1 decomposition half-life. b The reaction rate of decomposition. c 1,2-BF2 Shift. d Solvolysis.
DMSO^c	29.8	0.68 ± 0.03	2.8 ± 0.1 × 10^−4	6.2 ± 0.6	3.1 ± 0.3 × 10^−5
DMF^c	26.6	1.8 ± 0.2	1.1 ± 0.1 × 10^−4	16 ± 1	1.2 ± 0.1 × 10^−5
DMAc^c	27.8	3.3 ± 0.4	5.8 ± 0.6 × 10^−5	17 ± 2	1.1 ± 0.1 × 10^−5
THF^d	20.0	44 ± 3	4.4 ± 0.3 × 10^−6	36 ± 5	5.5 ± 0.8 × 10^−6
Acetone^d	17.0	53 ± 7	3.7 ± 0.4 × 10^−6	109 ± 9	1.8 ± 0.2 × 10^−6
EA^d	17.1	82 ± 15	2.4 ± 0.4 × 10^−6	184 ± 26	1.1 ± 0.1 × 10^−6

---

Finally, in alcohols methanol (MeOH), ethanol (EtOH), 1-propanol (n-PrOH), 2-propanol (i-PrOH), 2-methyl-1-propanol (i-BuOH), 2-butanol (sec-BuOH), and benzyl alcohol (BnOH, Fig. S28–S34 and S47) the switches quickly revert to their hydrazone precursors (Table 3 and Fig. 3c).⁴⁰

Table 3 Summary of solvolysis data for trans-1 in polar protic alcohols

Solvent	MeOH	EtOH	n-PrOH	i-PrOH	i-BuOH	sec-BuOH	BnOH
a Trans-1 decomposition half-life. b The reaction rate of decomposition.
τ 1/2 (h)	0.063 ± 0.005	0.13 ± 0.01	0.16 ± 0.01	0.38 ± 0.04	0.20 ± 0.01	0.42 ± 0.04	0.82 ± 0.09
k (s^−1)	3.1 ± 0.3 × 10^−3	1.5 ± 0.1 × 10^−3	1.2 ± 0.1 × 10^−3	5.2 ± 0.6 × 10^−4	9.4 ± 0.2 × 10^−4	4.6 ± 0.4 × 10^−4	2.4 ± 0.3 × 10^−4

---

To understand the different solvent-dependent outcomes, we correlated the reaction rates to various solvent parameters, including α, π*, DN, acceptor number (AN), and the acidity, basicity, and polarity index (Z). In the case of trans-1, the best correlations for the 1,2-BF₂-shift rates were with the DN (R² = 0.95, Fig. 4a) and π* (R² = 0.92, Fig. S51a).⁴⁴ Similarly, trans-2 exhibited the best correlation with DN (R² = 0.91, Fig. 4b) followed by π* (R² = 0.74, Fig. S51b). Other solvent parameters show much weaker correlations. These findings strongly suggest that the solvent donor number has the most substantial influence on the molecular reaction pathway in 1 and 2. That is, solvents that donate lone pair electrons to the boron in the azo-BF₂ switches more readily result in faster 1,2-BF₂ shifts. As for the hydrolysis in alcohols, surprisingly, the process is not governed by the pK_a of the solvents, but rather by their hydrogen bond donating ability (α, Fig. 4c, R² = 0.97). Hydrogen bond donation ability is determined not solely by the intrinsic acidity (pK_a) of the solvent, but by a combination of factors, including the electronegativity of the hydrogen-bearing atom, the polarity of the hydrogen-donating bond, steric hindrance, and the overall solvent structure (including resonance effects). Under our experimental conditions, it appears that steric effects in protic solvents play a decisive role in modulating the reaction rate by influencing the accessibility and strength of hydrogen bonding at the coordination site. These findings highlight that the multifaceted nature of solvent interactions must be carefully considered when evaluating the stability of the N → B coordination bond in azo-BF₂ switches.

Fig. 4 Correlation of reaction rate with the most relevant parameter: (a) trans-1 in aprotic solvents versus DN, (b) trans-2 in aprotic solvents versus DN, and (c) trans-1 in protic solvents versus α.

To elucidate the mechanism of the BF₂-shift and solvolysis reactions in trans-1 and trans-2, we conducted DFT-based quantum chemical calculations in three representative solvents: dichloromethane (DCM, a weakly polar solvent), dimethyl sulfoxide (DMSO, a strong polar aprotic Lewis base), and methanol (MeOH, a protic solvent with moderate Lewis base strength). These calculations were performed at M06-2X/def2-TZVP/SMD level for single-point energies, with initial geometries optimized at the B3LYP-D3/def2-SVP/SMD level (for more computational details, please see the SI).

Starting with trans-1 in DCM, the direct 1,2-BF₂ shift was found to have a high free energy activation barrier of 34.5 kcal mol⁻¹ through transition state TS1a (Fig. 5a). This substantial barrier suggests that the 1,2-BF₂ shift will unlikely proceed in DCM, aligning with experimental observations. It is important to note that these calculations were performed using an implicit solvent model, as the 1,2-BF₂ shift in DCM is proposed to be a single-step process with no solvent mediation (Fig. S55).

Fig. 5 Computed reaction mechanisms illustrating the 1,2-BF₂ shift in (a) DCM and (b) DMSO, as well as (c) the solvolysis of trans-1 with MeOH. All calculations were performed at the M06-2X/def2-TZVP/SMD//B3LYP-D3/def2-SVP/SMD level of theory. The subpanel in (c) shows a comparison of the energy barriers at TS1c in methanol and ethanol. The free energy activation barriers (ΔG^‡) and free energy differences (ΔG⁰) are relative to the individual starting species trans-1, and explicit solvent molecules are used when applicable. Values are free energies in kcal mol⁻¹. Panel (c) is not drawn to scale to improve clarity.

The energy barrier for the 1,2-BF₂ shift reaction is significantly lower in DMSO than in DCM. Our modelling suggests that DMSO, a strong Lewis base with its high donor number, participates directly in the reaction, playing a critical role in the 1,2-BF₂ shift, i.e., DMSO directly attacks the boron atom, displacing the azo group bound to it. Computations show that this substitution proceeds via transition state TS1b, requiring a free energy barrier of only 22.5 kcal mol⁻¹ (Fig. 5b and S56), leading to intermediate Int1b, which is endergonic by 4.8 kcal mol⁻¹. Analysis of the structure of Int1b reveals a key bond distance of 2.91 Å for the broken B–N bond and 1.53 Å for the newly formed B–O bond, creating sufficient space for the rotation of the azo group, which occurs via the transition state TS2b (ΔG^‡ = 22.3 kcal mol⁻¹) resulting in intermediate Int2b (11.5 kcal mol⁻¹). Finally, dissociation of DMSO viaTS3b (ΔG^‡ = 23.5 kcal mol⁻¹) and rebinding of the rotated azo group to the boron leads to the formation of the exergonic product 2b (−13.3 kcal mol⁻¹). These calculated moderate energy barriers in DMSO explain why the 1,2-BF₂ shift occurs readily in this solvent, as observed experimentally.

The DFT calculations also indicate that in methanol, solvolysis is more facile than the 1,2-BF₂ shift (Fig. 5c and S57). Methanol, being a protic solvent, initiates the reaction by attacking the boron atom, leading to the displacement of the azo group via the transition state TS1c (ΔG^‡ = 20.5 kcal mol⁻¹ relative to the free starting materials). This process is followed by a rearrangement through proton transfer (facilitated by surrounding solvent molecules) to Int2c. Subsequently, the dissociation of BF₂OMe occurs viaTS2c (ΔG^‡ = 1.1 kcal mol⁻¹), resulting in the intermediate Int3c that tautomerizes to Int4c.⁴⁵

The H-bonds between the two fluorine atoms in the BF₂ bridge and surrounding methanol molecules result in a lower azo group displacement energy barrier (20.5 kcal mol⁻¹) compared to DMSO (22.5 kcal mol⁻¹, Fig. 5b and c). Electrostatic potential energy surface analysis of trans-1 derivatives reveals significant negative charge density at the fluorine atoms (Fig. S58 and Table S1). These H-bond interactions, partially accounted for in the solvent model, stabilize the transition states and lower the energy barriers. More importantly, the protic nature of methanol promotes solvolysis by donating a proton to trans-1 and supports the formation of Int2c by enabling efficient proton diffusion. Notably, alcohol molecules act as H-bond donors, and their donating ability is a key factor governing the solvolysis rate of the azo-BF₂ compounds. When we modeled a similar solvolysis reaction in ethanol (α = 0.86), which has a weaker H-bond donating ability than methanol (α = 0.98), the energy barrier of TS1 increased from 20.5 to 22.3 kcal mol⁻¹ (Fig. 5c and Table S2), respectively. These findings demonstrate that the H-bond donating ability of the protic solvent significantly influences reaction kinetics, which is consistent with experimental observations. It is also worth mentioning that methanol is more effective than ethanol at facilitating proton transfer, owing to its greater polarity and smaller steric hindrance. These properties enhance the formation of Int2c and promote more efficient solvolysis. The calculated results for trans-2 showed similar trends (Fig. S59–S64).

Conclusions

In summary, the synthesis and characterization of two structurally distinct azo-BF₂ photoswitches presented in this work underscore the significant potential of structural modulation in enhancing their photophysical performance. We demonstrate that expanding the π-conjugation profoundly influences the activation wavelength and photoswitching performance, with switch 1 achieving an extended NIR absorption profile, and switch 2 benefiting from improved separation of the absorption bands. The solvent-dependent behavior, particularly the 1,2-BF₂ shift, highlights the critical role of solvent donor number in dictating reaction pathways and product stability. On the other hand, the observed solvent-mediated solvolysis showcases the importance of the solvent's hydrogen bond donation ability in driving the process. Importantly, the interplay between solvent proticity and steric effects can be leveraged to mitigate the reaction. Moreover, the data indicate that lowering accessibility to the N → B bond through steric hindrance can lower its reactivity. Overall, the gained insights can be harnessed to further the design of these efficient visible and NIR-responsive photoswitches, and expand opportunities in adaptive materials, molecular devices, and energy storage applications.

Author contributions

Q. Qi conceived the initial project, synthesized the compounds, and conducted their photophysical and structural characterization. Q. Qi also wrote the original draft of the manuscript. H. Fu investigated and analyzed the solvent-dependent behavior of the compounds and co-wrote the original draft. L. Peng performed and summarized the DFT calculations under the guidance of X. Liu. S. P. assisted with UV-Vis absorption measurements. I. A. supervised the overall project and provided critical revisions to the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data are available in the main text and the SI. See DOI: https://doi.org/10.1039/d5sc05042a.

Acknowledgements

X. L. thank the Ministry of Education, Singapore (MOE-T2EP10222-0001), for financial support. The authors are also grateful for the computational resources provided by the Singapore University of Technology and Design (SUTD) and the National Supercomputing Centre, Singapore. Q. Q. acknowledges Jiangsu Specially Appointed Professorship.

References

1. W. A. Velema, W. Szymanski and B. L. Feringa, J. Am. Chem. Soc., 2014, 136, 2178.
2. D. Bléger and S. Hecht, Angew. Chem., Int. Ed., 2015, 54, 11338.
3. M. M. Lerch, M. J. Hansen, G. M. van Dam, W. Szymanski and B. L. Feringa, Angew. Chem., Int. Ed., 2016, 55, 10978.
4. M. Wegener, M. J. Hansen, A. J. M. Driessen, W. Szymanski and B. L. Feringa, J. Am. Chem. Soc., 2017, 139, 17979.
5. K. Hüll, J. Morstein and D. Trauner, Chem. Rev., 2018, 118, 10710.
6. I. M. Welleman, M. W. H. Hoorens, B. L. Feringa, H. H. Boersma and W. Szymański, Chem. Sci., 2020, 11, 11672.
7. H. Wang, H. K. Bisoyi, X. Zhang, F. Hassan and Q. Li, Chem.–Eur. J., 2022, 28, e202103906.
8. Y. Xu, Y. Tang and Q. Li, Adv. Funct. Mater., 2024, 2416359.
9. R. Siewertsen, H. Neumann, B. Buchheim-Stehn, R. Herges, C. Näther, F. Renth and F. Temps, J. Am. Chem. Soc., 2009, 131, 15594.
10. A. A. Beharry, O. Sadovski and G. A. Woolley, J. Am. Chem. Soc., 2011, 133, 19684.
11. H. M. D. Bandara and S. C. Burdette, Chem. Soc. Rev., 2012, 41, 1809.
12. D. Bléger, J. Schwarz, A. M. Brouwer and S. Hecht, J. Am. Chem. Soc., 2012, 134, 20597–20600.
13. S. Samanta, A. A. Beharry, O. Sadovski, T. M. McCormick, A. Babalhavaeji, V. Tropepe and G. A. Woolley, J. Am. Chem. Soc., 2013, 135, 9777.
14. M. Dong, A. Babalhavaeji, S. Samanta, A. A. Beharry and G. A. Woolley, Acc. Chem. Res., 2015, 48, 2662.
15. M. Hammerich, C. Schütt, C. Stähler, P. Lentes, F. Röhricht, R. Höppner and R. Herges, J. Am. Chem. Soc., 2016, 138, 13111.
16. P. Lentes, E. Stadler, F. Röhricht, A. Brahms, J. Gröbner, F. D. Sönnichsen, G. Gescheidt and R. Herges, J. Am. Chem. Soc., 2019, 141, 13592.
17. L. N. Lameijer, S. Budzak, N. A. Simeth, M. J. Hansen, B. L. Feringa, D. Jacquemin and W. Szymanski, Angew. Chem., Int. Ed., 2020, 59, 21663.
18. T. Fukaminato, T. Hirose, T. Doi, M. Hazama, K. Matsuda and M. Irie, J. Am. Chem. Soc., 2014, 136, 17145.
19. N. M.-W. Wu, M. Ng, W. H. Lam, H.-L. Wong and V. W.-W. Yam, J. Am. Chem. Soc., 2017, 139, 15142.
20. H. Xi, Z. Zhang, W. Zhang, M. Li, C. Lian, Q. Luo, H. Tian and W.-H. Zhu, J. Am. Chem. Soc., 2019, 141, 18467.
21. Z. Zhang, W. Wang, P. Jin, J. Xue, L. Sun, J. Huang, J. Zhang and H. Tian, Nat. Commun., 2019, 10, 4232.
22. S. Wiedbrauk and H. Dube, Tetrahedron Lett., 2015, 56, 4266.
23. C.-Y. Huang, A. Bonasera, L. Hristov, Y. Garmshausen, B. M. Schmidt, D. Jacquemin and S. Hecht, J. Am. Chem. Soc., 2017, 139, 15205.
24. L. Köttner, E. Ciekalski and H. Dube, Angew. Chem., Int. Ed., 2023, 62, e202312955.
25. M. Zitzmann, M. Fröhling and H. Dube, Angew. Chem., Int. Ed., 2024, 63, e202318767.
26. M. Jacquet, L. M. Uriarte, F. Lafolet, M. Boggio-Pasqua, M. Sliwa, F. Loiseau, E. Saint-Aman, S. Cobo and G. Royal, J. Phys. Chem. Lett., 2020, 11, 2682.
27. K. Klaue, W. Han, P. Liesfeld, F. Berger, Y. Garmshausen and S. Hecht, J. Am. Chem. Soc., 2020, 142, 11857.
28. S. Helmy, F. A. Leibfarth, S. Oh, J. E. Poelma, C. J. Hawker and J. Read de Alaniz, J. Am. Chem. Soc., 2014, 136, 8169.
29. M. M. Lerch, W. Szymański and B. L. Feringa, Chem. Soc. Rev., 2018, 47, 1910.
30. R. Castagna, G. Maleeva, D. Pirovano, C. Matera and P. Gorostiza, J. Am. Chem. Soc., 2022, 144, 15595.
31. M. Clerc, S. Sandlass, O. Rifaie-Graham, J. A. Peterson, N. Bruns, J. R. de Alaniz and L. F. Boesel, Chem. Soc. Rev., 2023, 52, 8245.
32. C. A. Reyes, A. Karr, C. A. Ramsperger, A. T. G. K, H. J. Lee and E. Picazo, J. Am. Chem. Soc., 2025, 147, 10.
33. S. Crespi, N. A. Simeth and B. König, Nat. Rev. Chem., 2019, 3, 133.
34. T. Dang, Z.-Y. Zhang and T. Li, J. Am. Chem. Soc., 2024, 146, 19609.
35. Y. Yang, R. P. Hughes and I. Aprahamian, J. Am. Chem. Soc., 2012, 134, 15221.
36. Y. Yang, R. P. Hughes and I. Aprahamian, J. Am. Chem. Soc., 2014, 136, 13190.
37. Q. Qi, J. T. Plank, A. R. Lippert and I. Aprahamian, Chem, 2024, 10, 3575.
38. Q. Qiu, Q. Qi, J. Usuba, K. Lee, I. Aprahamian and G. G. D. Han, Chem. Sci., 2023, 14, 11359.
39. H. Qian, Y.-Y. Wang, D.-S. Guo and I. Aprahamian, J. Am. Chem. Soc., 2017, 139, 1037.
40. Q. Qi, S. Huang, X. Liu and I. Aprahamian, J. Am. Chem. Soc., 2024, 146, 6471.
41. Y. Marcus, Chem. Soc. Rev., 1993, 22, 409.
42. C. Y. Huang, A. Bonasera, L. Hristov, Y. Garmshausen, B. M. Schmidt, D. Jacquemin and S. Hecht, J. Am. Chem. Soc., 2017, 139, 15205.
43. B. Shao, H. Qian, Q. Li and I. Aprahamian, J. Am. Chem. Soc., 2019, 141, 8364.
44. Acetone, EA, and THF were included in the DN correlation analysis even though they do not induce an observable BF₂ shift. We hypothesize that the rate determining step in both reactions is the solvent attack on the boron center, and so we are using the hydrolysis rate in these solvents as place holders for the fastest rate at which such a shift can take place.
45. S. M. Landge, E. Tkatchouk, D. Benítez, D. A. Lanfranchi, M. Elhabiri, W. A. Goddard and I. Aprahamian, J. Am. Chem. Soc., 2011, 133, 9812.

---

Footnote

† These three authors contributed equally.

---