Promoting the Furan Ring‐Opening Reaction to Access New Donor–Acceptor Stenhouse Adducts with Hexafluoroisopropanol

Abstract Donor–acceptor Stenhouse adducts (DASAs) are visible‐light‐responsive photoswitches with a variety of emerging applications in photoresponsive materials. Their two‐step modular synthesis, centered on the nucleophilic ring opening of an activated furan, makes DASAs readily accessible. However, the use of less reactive donors or acceptors renders the process slow and low yielding, which has limited their development. We demonstrate here that 1,1,1,3,3,3‐hexafluoro‐2‐propanol (HFIP) promotes the ring‐opening reaction and stabilizes the open isomer, allowing greatly reduced reaction times and increased yields for known derivatives. In addition, it provides access to previously unattainable DASA‐based photoswitches and DASA–polymer conjugates. The role of HFIP and the photochromic properties of a set of new DASAs is probed using a combination of 1H NMR and UV/Vis spectroscopy. The use of sterically hindered, electron‐poor amines enabled the dark equilibrium to be decoupled from closed‐isomer half‐lives for the first time.


Chemicals
All commercially obtained reagents were bought from Sigma Aldrich, TCI Europe or Fisher Scientific and were used without purification, except furfural, which was distilled prior to usage. Size exclusion beads (Bio-Beads S-X1 Support, 0.6-14 kDa) were obtained from Bio-Rad Laboratories. Anhydrous solvents were either obtained from Sigma Aldrich or from a solvent purification system.

Instruments and Methods
Room temperature reactions were carried out between 22-25 ºC. Thin layer chromatography (TLC) was performed using Merck TLC plates (silica gel 60 F254 on aluminum) and visualized by UV light (254/ 366 nm) or staining with KMnO4/NaOH. Silica gel chromatography was performed using silica gel from Sigma Aldrich (technical grade, 60 Å pore size, 40-63 μm particle size). Size exclusion chromatography (SEC) was performed on Bio-Beads S-X1 Support beads using distilled THF as mobile phase. 1 H and 13 C nuclear magnetic resonance (NMR) spectra were measured at 298 K on a Bruker Avance III 400 (400 MHz) NMR spectrometer, a Varian Unity Inova 500 MHz, or a Varian Unity Inova AS600 600 MHz spectrometer. 19 F NMR spectra were recorded on a Bruker Avance III 400 (400 MHz) NMR spectrometer. Chemical shifts (δ) are reported in ppm and referenced internally from the proteo-solvent resonance. Coupling constants (J) are reported in Hz. Abbreviations for the peak multiplicities are s (singlet), d (doublet), dd (doublet of doublet), t (triplet), q (quadruplet) and m (multiplet). For diffusion-edited 1 H NMR spectra, 40% gradient strengths were applied to selectively suppress the signals of low molecular weight species. Gas chromatography/electron impact ionization-mass spectrometry (GC/EI-MS) was measured on a Thermo Scientific ISQ GC/MS equipped with an ISQ 7000 and Trace 1300 GC using a Thermo Scientific TG-SQC capillary column (15 m, 0.25 mm I.D., 0.25 m thickness). Split/splitless injector at 280 °C; flowrate at 1 mL min -1 ; gradient set to 20 °C min -1 from 30 °C to 300 °C, then isothermal for another 4 min. EI set to 70 eV; single stage quadruple mass analyzer; mass range 35-600 amu at 2 scans min -1 in full scan mode. The retention time (Rt) is reported in min, the mass of molecular ions and characteristic fragments with >15 rel.% are reported as m/z (rel.%). High resolution mass spectrometry (HR-MS) was measured on a Waters LCT Premier ESI TOF. Attenuated total reflection Fourier-transform infrared (ATR FT-IR) spectra were recorded on a Varian 640-IR FT-IR spectrometer equipped with an ATR (attenuated total reflection) accessory or a Thermo Nicolet iS10 FTIR Spectrometer with a Smart Diamond ATR; applied as neat samples and absorbance bands reported as 1/λ in cm -1 . Abbreviations for the relative band intensities are s (strong), m (medium), w (weak). Gel permeation chromatography (GPC) was measured on an Agilent 1100 Series high-performance liquid chromatography (HPLC) system on serial coupled PSS SDV 5 m 100 Å and PSS SDV 5 m 1000 Å columns maintained at 30 °C (allows separation from ca. 1-1000 kDa). Signals were recorded on a diode array detector (235 nm/ 360 nm) and a refractive index (RI) detector (at 35 °C). Measurements were performed in THF as an eluent relative to narrow molecular weight PS standards. UV-Vis absorbance spectra were recorded on an Agilent Cary 4000 UV-Visible spectrophotometer or an Agilent 8453 UV-Visible Spectrophotometer G1103A. Details for UV-Vis absorption and photoswitching measurements are presented in the respective section 8. Photoluminescence spectra were measured on a Horiba Scientific Fluoromax-Plus fluorescence spectrometer at room temperature (excitation and emission slit widths were set to 2 nm unless otherwise stated). Figure S1: Similarities between the proposed mechanism of DASA synthesis, aza-Piancatelli rearrangement [1,2] and Stenhouse reaction. [3]

4-(furan-2-ylmethylene)-3-(trifluoromethyl)isoxazol-5(4H)-one (S4).
Hydroxylamine hydrochloride (1.9 g, 27 mmol, 2.3 eq.), 4,4,4-trifluoroacetoacetetate (2.4 g, 12 mmol, 1.0 eq.) and K2CO3 (3.8 g, 27 mmol, 2.3 eq.) were heated to reflux in EtOH (15 mL) for 3.5 hours while stirring. The reaction mixture was allowed to cool to room temperature, and the solvent was removed under reduced pressure. The yellowish residue was re-dissolved in 10 mL of an aqueous solution of NaOH (4 M) and stirred at 23 °C for 10 min. This solution was then acidified with conc. HCl to a pH value of 2 and extracted with DCM (3 x 50 mL). The organic phase was dried with MgSO4, filtered and the solvent was removed in vacuo to afford a yellow oil (1.8 g). This oil was re-dissolved in DCM (20 mL), 2-furaldehyde (1.4 g, 15 mmol) was added, and it was stirred at 23 °C for 14 h until the mixture turned brown. H2O (10 mL) was added and DCM was removed in vacuo. The precipitated brown solid was collected by filtration, rinsed with H2O and dried in a desiccator. Purification was done by passing the product through a silica plug with DCM as an eluent. After removal of the solvent in vacuo, the product S4 was obtained as a yellow solid (1.7 g, 7.4 mmol, 62%). 1

Amines
Compound S5 and S7 were prepared according to literature procedure and spectral analysis matched literature data. [6,7]

N-(4-Methoxyphenyl)propane-1,3-diamine (S5).
4-Iodoanisole (3.9 g, 17 mmol, 1.0 eq.), CuCl (freshly recrystallized from MeOH and dry, 0.17 g, 1.7 mmol, 0.1 eq.) and a fine and well dried powder of KOH (1.6 g, 28 mmol, 1.6 eq.) were added to a dry 100 mL round-bottom flask equipped with a stirrer and a septum under N2. The flask was immersed into an ice-water bath and 1,3-diaminopropane (6.3 mL, 5.6 g, 76 mmol, 4.5 eq.) was added slowly at 0 °C. The reaction mixture turned dark blue overtime. After stirring at 0 °C for 5.5 h, the mixture was exposed to air and 30 mL of H2O were added. It was extracted with DCM (3 x 150 mL) and the organic phase was dried over MgSO4, filtered and solvent was removed in vacuo to yield the crude product as a brown oil (2.5 g). The product was purified via column chromatography on silica gel (DCM/methanol/ammonia 20:10:1) to afford S5 as brownish oil (1.8 g, 9.9 mmol, 58%). 1

4-(3-Aminopropoxy)-diphenylamine (S6).
3-Bormopropylamine hydrobromide was N-boc protected as described in literature. [8] 4-Hydroxydiphenylamine (2.50 g, 13.5 mmol, 1.0 eq.), Cs2CO3 (14.6 g, 44.7 mmol, 3.3 eq.) and 3-(N-Boc)aminopropyl bromide (3.50 g, 14.7 mmol, 1.1 eq.) were added to a two-neck 250 mL round bottom flask equipped with a magnetic stir bar, a septum and a reflux condenser under N2. CH3CN (anhydrous, 75 mL) was added and the obtained suspension was heated to 60 °C under N2 and stirred for 17 h. The mixture was then allowed to cool to room temperature and the solvent was removed in vacuo. The residue was taken up in AcOEt (150 mL) and extracted with H2O (2 x 50 mL) and brine (50 mL). The organic layer was dried over MgSO4, filtered and the solvent removed in vacuo to yield a brown solid, which was re-dissolved in MeOH (15 mL) acidified with conc. HCl (2 mL). This solution was stirred at 23 °C under N2 for 20 h when reaction control by TLC (AcOEt/heptane 1:2) showed complete deprotection. The solution was then cooled to 0 °C in an ice-water bath and basified with an aqueous solution of NaOH (4 M) to a pH value of 9. H2O (15 mL) was added and a beige precipitate formed, which was collected by filtration. The product was purified via column chromatography on silica gel (DCM/methanol/ammonia 20:10:1) to afford S6 as brownish solid (1.05 g, 4.30 mmol, 32%). 1

DASA-13
This compound was prepared analogously to literature procedures. Spectroscopic data matched literature. [11]

DASA-14
This compound was prepared analogously to literature procedures. Spectroscopic data matched literature. [5]

UV-Vis Spectroscopy
In situ UV-Vis kinetic experiments were performed between 23-25 ºC under pseudo-first order conditions using a 100-fold excess of amine reagent. Stock solutions of amine and furan adduct in DCM were freshly prepared and measurements were performed by adding small amounts of stock solutions to the respective DCM/HFIP mixture in a 3 mL quartz cuvette. The absorbance was monitored with an Agilent Cary 4000 UV-Vis spectrophotometer overtime.

Rates as Function of HFIP Concentration
Kinetic experiments were done using N-methylaniline and 1 for different volumetric ratios of HFIP in DCM (0-90 vol%). Measurements were performed at concentrations of 5x10 -3 M (amine) and 5x10 -5 M (furan adduct) at the absorption maximum of 1 (375 nm) and at 565 nm over the time course of 15 to 800 min ( Figure S2). Apparent rate constants were obtained by fitting the absorbance changes to one-phase exponential decay functions using Origin 2018 software. R 2 -values reached >0.99 for HFIP concentrations <50 vol%. At concentrations >50 vol% the absorbance changes are not strictly mono-exponential anymore and the rate of DASA formation decreases suggesting the presence of side reactions or degradation of the activated furan. Relative rates (krelative) correspond to the apparent rate constants at a given HFIP concentration relative to the apparent rate constant of the reaction in pure DCM (Figure S2d). Note: due to the lower concentrations used in UV-Vis spectroscopic measurements the amount of HFIP relative to 1 was considerably higher than in the NMR kinetic experiments (i.e. 1 vol% HFIP corresponds to a large excess of ~1900 equivalents).

NMR Spectroscopy
In situ NMR experiments were performed at a concentration of 25 mM and 298 K. A solution of 1 in CD2Cl2 (0.35 mL, 50 mM, 1.0 eq.) and a solution of N-methylaniline in CD2Cl2 (0.35 mL, targeted at 55 mM, 1.1 eq.) were prepared and to the latter was added deuterated HFIP (0, 1 and 5 eq. relative to 1 or 0, ~0.2 and ~1 vol%). These solutions were combined in a standard 5 mm NMR tube and a series of 300 spectra was acquired on a time course of up to 15 hours on a Varian Unity Inova AS600 600 MHz spectrometer (delay before start of the first scan was 5 min). Conversion plots were calculated from the integrals of the signals of 1 at δ = 8.89 ppm and the signals of the N-methyl groups in the open (δ = 3.59 ppm) and closed (δ = 2.85 ppm) DASA respectively as given by: The measurements were performed in triplicates. The actual amine concentrations (as determined by 1 H-NMR) varied between 1.1-1.5 equivalents relative to 1. Analogous measurements were performed in acetonitrile-d3 or by exchanging HFIP-d2 with isopropanol or the methyl-ether of HFIP (HFIPME) (Figure S4 and Figures S8-S9). Similar measurements on the uncatalyzed runs at higher concentration and longer reaction time (50 mM, 15 h) additionally confirmed that similar conversions are reached when the curve plateaus as for the runs with HFIP (80-85%). Also, it was found that higher excess of amine can push the conversion to >95%.

Kinetic Fitting
To determine rate constants second-order kinetics was assumed. For this the following differential equation applies: The corresponding linearized integrated rate equation is:  S5-S9). The initial ratio of N-methylaniline to 1 was calculated through the x-axis intersection. Table S1: Determined rate constants in dichloromethane.
[b] Mean values of three measurements ± standard deviation.

DFT Calculations
Density functional theory (DFT) calculations at B3LYP-GD3BJ/6-31G(d) level of theory were used for geometry optimizations of the open form DASAs to determine the dihedral angle between the acceptor and donor groups, as described before for second generation DASAs. [11] Calculations at B3LYP-GD3BJ/6-311++(d,p) level of theory were performed to predict the character of the frontier orbitals and the optical excitation energy. The calculations were performed with the Gaussian 16 Rev.A.03 software and solvent effects were considered using a conductor-like polarizable continuum model (CPCM, toluene). The calculated HOMO-LUMO gaps correlate well with the relative ordering of the experimentally observed absorption peaks, however, the absolute energies are systematically overpredicted by 0.4 eV (~ 100 nm blue-shifted λmax relative to experimentally determined value in toluene), which is in accordance with previous findings for second generation DASAs. [11] The calculated HOMO-LUMO gaps listed in Table S2 were adjusted by this amount. Time-dependent DFT methods and other functionals were also tested but the predictions of the ordering of the HOMO-LUMO gaps in the series were less consistent with experimental results. Dihedral angles (ΦD-A) between acceptor and donor groups were determined from the optimized geometries and frontier orbitals graphically represented by using the Avogadro software (Figure S41-S42).

Photoswitching Experiments
The photoinduced optical absorption kinetics were measured on a pump-probe setup. The pump beam was generated by a light emitting diode (LED) source (Thorlabs) coupled into a multimode optical fiber terminated with an output collimator. The LED intensity was controlled through a digital-to-analog converter (National Instruments USB-6009) using LabVIEW. The probe beam was produced by High Power MINI Deuterium Tungsten Halogen Source w/shutter 200-2000 nm (Ocean Optics DH-MINI) coupled into a multimode fiber with an output collimator for the light delivery. The probe light was modulated by a shutter (Uniblitz CS25) which could be controlled manually or through a digital output port (National Instruments USB-6009) using LabVIEW. Pump and probe beams were overlapped using steering and focusing optics at a 90° angle inside a sample holder, which allowed for a 10x10 mm rectangular spectrophotometer cells that was connected to a circulating bath for temperature control. Additionally, the solutions were stirred during the measurements by a miniature stirring plate inserted into the sample holder (Starna Cells SCS 1.11). The sample holder was placed into a metal enclosure to prevent exposure to ambient light. Both pump and probe beams were nearly collimated inside the cell with a diameter of about 2 mm. The pump beam was blocked after passing through the sample and the probe beam was directed by a system of lenses into the detector (Ocean Optics Flame-S1-XR spectrometer), which acquired spectra of the probe light. The detector was connected to a PC via USB port. The experiment was controlled by a National Instrument LabVIEW program which collected the probe light spectra, determined sample optical absorption spectra, controlled pump and probe light sources, and stored the data on the computer S3 hard drive according to the experimental protocol. Experiments were performed in at 10 µM concentration unless otherwise stated. Samples were left to equilibrate overnight prior to measurements unless otherwise stated.

Thermodynamic Equilibrium NMR Spectroscopy
Samples were stored in the dark at room temperature overnight. Closed and open isomer were identified by 1 H NMR spectroscopy (Figures S62-S73).

Probing the H-bonding in the Presence of HFIP
As confirmed in previous studies by other groups, [10] the hydroxy protons in DASAs are easily exchangeable and their 1 H-NMR signals are completely absent when the spectra are measured in, e.g., deuterated alcohols. We also observed the disappearance of these signals when measuring 1 H-NMR spectra of DASAs in solutions containing larger amounts of HFIP (deuterated and non-deuterated). In the following Figure S80, the changes of the hydroxy proton 1 H-NMR signal under different conditions are displayed. Here, it is visible that HFIP leads to a disappearance of the hydroxy proton signal, whereas additives such as HFIPMe at similar concentrations have none (or less, only slight broadening is observed) of an effect. As also shown in Figure S81, the presence of HFIP, however, seems to not effect a change of the chemical shift of the OH-signal. On the other hand, the chemical shifts of the polyene and aromatic protons do change, which we believe can also be a result of simple solvent effects and not necessarily specific interactions (it is also established that different solvents favor different resonance structures of the DASA open form). To further probe the potential of H-bonding, IR spectroscopic measurements in DCM vs. HFIP on a Meldrum's acid based DASA (DASA-6) were conducted. As indicated in Figure S82 below, the observation of two bands for the carbonyl stretching modes in the Meldrum's acid moiety can be found for measurements in DCM or in the solid state (~1615 cm -1 : H-bonded ring carbonyl stretch, ~1700 cm -1 second non H-bonded carbonyl, in agreement with previous assignments from literature). [10] On the other hand, measurements in HFIP did not show a pronounced band at 1700 cm -1 for a non H-bonded carbonyl and the band previously observed at ~1615 cm -1 redshifts pointing towards a large change in the strength of the H-bond. Although further studies are required, this does provide additional supporting evidence for the proposed modulation of the H-bonding.                In solution the DASAs are in equilibrium between open and closed state resulting in a complex spectrum. [12,16] Characteristic peaks are marked by colored dots.