Enolization rates control mono- versus di-fluorination of 1,3-dicarbonyl derivatives† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c9sc04185k

All rate constants for fluorination and enolization are determined.


Experimental
Compound 4a was purchased from Sigma Aldrich and was recrystallized (hexane) and dried under vacuum before use in kinetic measurements. The 1,3-diaryl-1,3-propanediones 4b-d were synthesised according to literature procedures 1   To confirm that the additional peak at δ = −111.9 ppm observed in the 19 F NMR spectra for reactions conducted in water/CD3CN mixtures was a hydrate of 6a, we obtained an NMR spectrum of an authentic sample of 6a in 20% water in CD3CN (v/v). The peak at δ = −111.9 ppm was present ( Figure   3). This hydrate was not isolated after work-up in synthetic reactions.

Confirmation of purities of compounds 5a-d
All kinetic studies of keto-enol relaxation and fluorination processes were performed by monitoring the appearance or disappearance of keto and enol species by UV-vis spectrophotometry. Given that compounds 5a-d were prepared from 4a-d, we were concerned that small amounts of residual 4a-d in our preparations of 5a-d could interfere with our kinetic studies. In order to confirm the spectrophotometric purities of 5a-d,    and 5d (bottom). S12

Methods
Kinetics studies were carried out using a Varian Cary-100 Bio UV-vis Spectrophotometer equipped with a Cary Temperature Controller unit, or a Varian Cary-50 Bio UV-vis Spectrophotometer connected to a Varian Cary PCB-150 Water Peltier system. Samples were contained in quartz absorption cuvettes with a path length of 1 cm. All spectra were zeroed against air. Reactions were followed by monitoring the disappearance of the enol at a fixed wavelength corresponding to the maximum absorbance (λmax) of the relevant enol ( Table 1) Kinetics studies were carried out using the "Scanning Kinetics" or "Single Wavelength Kinetics" programs.

Photoketonization and relaxation experiments
Solutions of 4a-d and 5a-d were prepared at the required concentration in quartz cuvettes, equipped with stirrer bars. The stirred solutions were irradiated with a 365 nm UV lamp for 3 h, at room temperature. The UV lamp was then removed, and if required, the additive was transferred to the cuvette. Time arrayed multi-wavelength scans were acquired every 15 min, unless stated otherwise, to avoid continuous irradiation of the cuvette at smaller time intervals, which would slow down the rate of relaxation. Time-arrayed single-wavelength scans were conducted, as required. S13

Keto:enol ratios in the presence of additives determined by NMR spectroscopy
The keto:enol ratios in CD3CN were obtained using 1 H NMR spectroscopy in the case of compounds 4a-d and 19 F NMR spectroscopy for compounds 5a-d. In order to obtain quantitative integral values, relaxation delays of 20 s and 8 s was employed for 1 H NMR and 19 F NMR experiments, respectively.
The concentrations of 4a-d and 5a-d were 25 mM for each NMR experiment. Unless otherwise stated, the solutions were allowed to equilibrate for 10 half-lives before NMR spectra were acquired.
Ratios were determined using keto and enol peak integrals. For example, with 4a, peaks corresponding to the enol form (δ = 7.08 ppm) and the keto form (δ = 4.72 ppm) were integrated across a 0.05 ppm range.

De-fluorination of 5a
The NMR spectra below ( Figure 9) correspond to the mixture of 5a (25 mM) and DABCO (25 mM) after an incubation time of ~30 min. In the 1 H NMR spectrum, the peak at δ = 6.93 ppm which corresponds to the fluoroketo tautomer has almost disappeared. In the 19 F NMR spectrum, peaks at δ = −189.8 ppm (fluoroketo) and δ = −169.5 ppm (fluoroenol) have also disappeared and a new peak at δ = +16.5 ppm is present, which may indicate the formation of an N−F species. Other smaller peaks have also appeared between −90 ppm and −150 ppm.

Photoketonization spectra
The photoketonizations of solutions of 4a-d and 5a-d were carried out using the method discussed in Section 3.1.1. We discontinuously monitored the progress of the photoketonizations by acquiring UV-vis spectra at various time intervals, shown below.
Absorbances below 300 nm were saturated due to the high concentration of 5a-keto and are therefore not shown.  Table 1 of the main text.  Table 1 of the main text.  Table 1 of the main text.            Table 1 in the main text.  Table 1 in the main text.  Table 1 in the main text.  Table 1 in the main text.  Table 1 in the main text.  The fitting delivered k1 -> 3.6591 ´ 10 −8 , k2 -> 0.0158053, where k1 represents the first order rate constant for uncatalysed enolization (s -1 ) and k2 represents the second order rate constant for autocatalysed enolization (M -1 s -1 ). Reverse rate constants for the processes described by k1 and k2

Additives
were obtained via Ke.   Table 2 of the main text.    Table 2 of the main text.

Other additives: formic acid, DABCO
were obtained via Ke.

Fluorination of 5b
At 20 °C:          In order to confirm that the product of this reaction is indeed 6b, LC-MS analysis was carried out on the reaction mixture in the cuvette at the end of the reaction (shown on page S44, 30 °C). The peak at Rt = 2.63 min (Figure 53) corresponds to 5b-keto. As expected, this remains unreacted as it comprises ~98% of the keto-enol equilibrium, and as relaxation is slow it does not occur on the timescale of our fluorination reactions. The peak at Rt = 2.99 min corresponds to 6b (m/z = 321.29).

With 20% water in MeCN: linear analysis
Since the decays in absorbance of 5a-enol ( Figure 62a)           The rate of fluorination of 5a-enol by Selectfluor™ (0.25 mM) without additives at 20 °C was estimated to be 7.5 × 10 −6 s −1 using the results in Section 3.13.1. Compared to the rate of fluorination in the presence of Bu4N + BF4 − , 19.8 × 10 −3 s −1 , this is a 2640-fold difference in reactivities.

Kinetics of fluorination of 4a-enol by Selectfluor™ in H 2 O/MeCN mixtures
We monitored the kinetics of fluorination of 4a by Selectfluor™ in water/MeCN mixtures at λmax = 341 nm. However, non-first order kinetics were observed. LC-MS analysis of the reaction mixtures showed the presence of 6a. Since Figure 68b did not intercept the origin, we conducted experiments using a different approach, described in Section 3.18.2.

With 20% water in MeCN: linear analysis
The first 10% of the reactions were monitored by UV-vis spectrophotometry and plots of ln(A−Ainf) vs. time were linear. Gradients of the plots at each Selectfluor™ concentration gave the kobs values (Table 30)

Difluorination of 4a-enol via Selectfluor™, 20% water in MeCN-d 3
The graphs below correspond to the experiment discussed in Section 2.4 of the main text, which was monitored by 19 F NMR spectroscopy. Integral intensities were converted to concentrations for use in the model.

Scheme 1:
Full reaction scheme for the conversion of 4a-enol to 6a using an excess of Selectfluor™.  at both high and low chemical shifts, i.e. acquiring spectra between −70 ppm and −230 ppm. We also increased the relaxation delays to 8 s. An error of ± 10% is associated with NMR integrals, which explains the slightly higher concentration of 5a-keto produced (65 mM) in the reaction with 20% water than would be expected given the starting concentration of 4a (59.5 mM).