¹⁹F-centred NMR analysis of mono-fluorinated compounds

Abstract

Addressing limitations of the existing NMR techniques for the structure determination of mono-fluorinated compounds, we have developed methodology that uses ¹⁹F as the focal point of this process. The proposed ¹⁹F-centred NMR analysis consists of a complementary set of broadband, phase-sensitive NMR experiments that utilise the substantial sensitivity of ¹⁹F and its far reaching couplings with ¹H and ¹³C to obtain a large number of NMR parameters. The assembled ¹H, ¹³C and ¹⁹F chemical shifts, values of J_HF, J_HH, and J_FC coupling constants and the size of ¹³C induced ¹⁹F isotopic shifts constitute a rich source of information that enables structure elucidation of fluorinated moieties and even complete structures of molecules. Here we introduce the methodology, provide a detailed description of each NMR experiment and illustrate their interpretation using 3-fluoro-3-deoxy-D-glucose. This novel approach performs particularly well in the structure elucidation of fluorinated compounds embedded in complex mixtures, eliminating the need for compound separation or use of standards to confirm the structures. It represents a major contribution towards the analysis of fluorinated agrochemicals and (radio)pharmaceuticals at any point during their lifetime, including preparation, use, biotransformation and biodegradation in the environment. The developed methodology can also assist with the investigations of the stability of fluoroorganics and their pharmacokinetics. Studies of reaction mechanisms using fluorinated molecules as convenient reporters of these processes, will also benefit.

---

Introduction

Fluorine's unique properties, such as high electronegativity, strength of a single fluorine–carbon bond and small atomic radius, impart significant benefits to fluorinated organic molecules.¹ Fluorination has been shown to enhance potency and/or specificity of molecular interactions, increase membrane permeability, modulate metabolism, moderate the pK_a of proximal functionalities, influence conformation, stabilise inherently reactive functionalities and produce viable bioisosteres.^2,3 Currently, about 20% of the commercial pharmaceuticals contain fluorine and the proportion of newly approved fluoro-pharmaceuticals is rising steadily.^4,5 The proportion of fluoro-agrochemicals is even larger; 53% of all active agrochemicals registered during 1998–2020 belong to this category.⁶ Similarly, ¹⁸F is the most frequently used radioisotope in positron emission tomography radiopharmaceuticals.⁷ Fluorination also has the potential to become a useful tool for improving properties of fragrance and semiochemical molecules.⁸

To capitalise on the ability of fluorine to improve molecular properties, there is a drive to design efficient and environmentally-safe chemical,^9,10 enzymatic¹¹ and chemo-enzymatic^12–14 fluorination methods. To assist these efforts, efficient analytical methods for the characterisation of fluorinated molecules are required. ¹⁹F NMR spectroscopy plays a prominent role in this area due to the favourable properties of ¹⁹F, such as its high sensitivity, 100% natural abundance, large chemical shift dispersion, large and far-reaching spin–spin interactions and ¹³C induced ¹⁹F isotopic shifts.

The lack of background ¹⁹F signals, due to the scarcity of fluorinated endogenous compounds, makes ¹⁹F NMR perfect for the analysis of mixtures produced by chemical or chemoenzymatic reactions with minimum clean-up steps or compound separation required. 1D ¹⁹F NMR spectroscopy has been widely used in studies of biodegradation and biotransformation of fluorinated compounds^15–21 mostly relying on the use of known standards¹⁵ or tabulated ¹⁹F chemical shifts. In a similar manner, ¹⁹F NMR has also been used for probing the mechanism and kinetics of chemical reactions, were fluorine is a convenient reporter of the processes taking place.^22,23

In support of such wide ranging activities, we have developed a ¹⁹F-centred NMR approach for the analysis of mono-fluorinated compounds, taking ¹⁹F NMR beyond recording simple 1D NMR spectra. Put together, the information obtained allows the structure elucidation of fluorine-containing molecular moieties and complete structure determination of small fluorine-containing molecules. It is well suited for the studies of complex mixtures. The ¹⁹F-centred NMR shares similarities to the “NMR spy” approach developed for the analysis of complex mixtures of soil organic matter, where –O¹³CH₃ tags are introduced to a subset of molecules.^24–26 Nevertheless, there are significant differences between the two approaches. Firstly, fluorinated molecules already contain ¹⁹F and therefore do not require additional chemical modifications. Secondly, the fluorine atom is typically closer to the protons and carbons of an organic molecule than are the nuclei of the –O¹³CH₃ group which, when combined with far reaching ¹⁹F couplings, allows to inspect parts of the molecule that are more remote from the ¹⁹F “tag.” The FESTA family of NMR experiments^27–29 that relies on selective manipulation of individual ¹H and ¹⁹F resonances illustrated this approach and provided ¹H–¹⁹F chemical shift correlations and ¹H–¹⁹F coupling constants when such spin manipulations were possible.

Our methodology utilises the far reaching ¹H–¹⁹F and ¹⁹F–¹³C couplings to obtain ¹H and ¹³C chemical shifts of nuclei multiple bonds away from the ¹⁹F atom, provides accurate values of numerous J_HF, J_FC, and J_HH coupling constants and ¹³C induced ¹⁹F isotopic shifts from several purposely designed nonselective 2D NMR experiments. Their advantages over similar existing NMR experiments are highlighted. The ¹⁹F-centered approach is illustrated using 3-fluoro-3-deoxy-D-glucose, 1, which can be characterized as a simple mixture of two ¹⁹F-containing molecules. Application of this methodology to a very complex mixture of compounds produced by chloramination of a single fluorinated molecule is presented elsewhere.³⁰

Experimental

The sample of 3-fluoro-3-deoxy-D-glucose (30 mg), 1, was dissolved in 600 μL of D₂O (Merck, 99.9 atom% D) and placed into a 5 mm NMR tube. Spectra involving ¹⁹F were acquired at 300 K on a 400 MHz Bruker Avance III NMR spectrometers equipped with a TBO BB-H/F-D probe. A 1D ¹H spectrum was acquired on an 800 MHz Bruker Avance III NMR spectrometer equipped with a TCI 5 mm probe. Parameters of the performed NMR experiments are presented in Table S1† and Bruker pulse sequences compatible with TopSpin 3 can be found in the ESI† (pp. 1–6).

The following symbols are used to depict the pulse sequences in Fig. 1–6: the thin and thick filled rectangles represent high power 90° (¹H, p1 or ¹⁹F, p3) and 180° (¹H, p2) pulses, respectively. 1 ms adiabatic CHIRP pulses with a peak power of 10.3 kHz (p44, shaded trapezoid with an inclined arrow) were applied to ¹⁹F. A 20 ms 60 kHz CHIRP ¹H pulse with a peak power of 2294 Hz (p32, trapezoid with inclined arrow) was used as part of the z-filter. A 500 μs CHIRP pulse (p14) and 2 ms composite CHIRP pulse (p24) were applied to ¹³C with a peak power of 9800 Hz. Unless stated otherwise, the r.f. pulses were applied from the x-axis. The 100% pulsed field gradient strength corresponds to 53.5 G cm⁻¹.

Fig. 1 (a) 400 MHz 1D ¹H spectrum of 1 (structure in the inset) with HOD suppression and resonance assignments; (b) ¹H-coupled 1D ¹⁹F spectrum of 1. Expansions of ¹⁹F multiplets of both anomeric forms from resolution-enhanced spectra produced using Lorentzian to Gaussian line shape conversion (LB = −2.0 Hz, GB = 0.5) are given in the inset.

Fig. 2 (a) Pulse sequence of a ¹⁹F-detected z-filtered 2D ¹H, ¹⁹F HETCOR. In a non z-filtered experiment, the part within the dashed rectangle is not included. For explanation of symbols used for pulses see Experimental. The NMR parameters used are given in Table S1.† The delays used were as follows: Δ₁ = p44; Δ₂ = one half of the J_HF evolution; t₁(0), the initial t₁ evolution delay time = 0.5 × in0, where in0 is the t₁ increment. The gradient strengths were as follows: G₀ = 3%; G₁ = 17%; G₂ = 31%; G₃ = 24%; G₄ = 10.0%. The following phase cycling was used: φ₁ = x, −x; φ₂ = 4x, 4(−x); φ₃ = 2y, 2(−y); Ψ = x, 2(−x), x. States-TPPI protocol was used for sign discrimination in F₁ with the phase φ₁ incremented by 90°. Purging of ¹⁹F magnetisation at the beginning of the pulse sequence by a composite 90° ¹⁹F pulse and pulsed field gradients (PFGs) minimises the cancellation artefacts. (b) An overlay of the ¹⁹F-detected 2D ¹H, ¹⁹F HETCOR spectra with (blue/turquoise) and without the z-filter (red/magenta). For clarity, the spectrum acquired without a z-filter was offset horizontally to the right. Insets show 1D F₁ traces taken at positions indicated by arrows. 1D ¹H and ¹⁹F spectra are shown along the left and top, respectively.

Fig. 3 (a) Pulse sequence of a 2D ¹H, ¹⁹F TOCSY–HETCOR. For explanation of symbols used for pulses see Experimental. The NMR parameters used are given in Table S1.†The delays were as follows: Δ₁ = p44; Δ₂ = one half of the J_HF evolution; t₁(0) is the initial t₁ evolution delay time = 0.5 × in0, where in0 is the t₁ increment. The gradient strengths were are follows: G₀ = 5%; G₁ = 17%; G₂ = 31%; G₃ = 24%. The following phase cycling was used: φ₁ = x, −x; φ₂ = 4x, 4(−x); φ₃ = 2y, 2(−y); Ψ = x, 2(−x), x, −x, 2x, −x. States-TPPI protocol was used for sign discrimination in F₁ with the phase φ₁ incremented by 90°. Purging of ¹⁹F magnetisation after the z-filter by a composite 90° ¹⁹F pulse followed by the G₂ PFG minimises the cancellation artefacts. (b) An overlay of the ¹⁹F-detected 2D ¹H,¹⁹F TOCSY–HETCOR spectrum (blue/turquoise) and a z-filtered VT ¹⁹F-detected 2D ¹H, ¹⁹F HETCOR spectrum (red/magenta, horizontally offset to the right) of 1 acquired with the pulse sequence shown in (a) and Fig. 2a, respectively. Vertical traces of the two spectra as indicated by arrows are shown in the inset. Exclusive/stronger TOCSY cross peaks are labelled in blue. 1D ¹H and ¹⁹F spectra are shown along the left and the top, respectively.

Fig. 4 (a) Pulse sequence of a 2D ¹⁹F, ¹H CP-DIPSI3–DIPSI2. For explanation of symbols used for pulses see Experimental. The NMR parameters used are given in Table S1.†The dashed line indicates signal acquisition before an optional ¹H–¹H spin-lock. For description of pulses see Experimental. The delays were as follows: δ₁ = 20 μs; δ₂ = δ₁ + (2/π) × p3; δ₃ = p2; t₁(0) is the initial t₁ evolution delay time = 0.5 × in0, where in0 is the t₁ increment. The gradient strengths were as follows: G₀ = 5%; G₁ = 17%; G₂ = 31%; G₃ = 66%. The following phase cycling was used: φ₁ = y, −y; φ₂ = 4x, 4(−x); φ₃ = 2y, 2(−y); Ψ = x, 2(−x), x. The states-TPPI protocol was used for sign discrimination in F₁ with the phase φ₁ incremented by 90°. Purging of ¹⁹F magnetisation at the beginning of the pulse sequence by a composite 90° ¹⁹F pulse and PFGs minimises the cancellation artefacts. (b) An overlay of two 2D ¹⁹F, ¹H CP-DIPSI3–DIPSI2 spectra acquired with 20 ms ¹⁹F → ¹H cross-polarisation (CP) only (red) and an additional 50 ms ¹H → ¹H spin-lock (blue) using the pulse sequence shown in (a). The red spectrum was offset vertically to facilitate visualisation of the cross peaks. The two insets show overlaid 1D F2 traces through ¹⁹F resonances of α- and β-D-glucose from both spectra. Twice as many scans were acquired for the blue spectrum as for the red spectrum. 1D ¹⁹F and ¹H projections of the blue spectrum are shown along the left and top, respectively.

Fig. 5 (a) Pulse sequence of the 2D ¹⁹F, ¹³C HMBC optimised for ⁿJ_FC correlations. For explanation of symbols used for pulses see Experimental. The NMR parameters used are given in Table S1.† The delays were as follows: d6 = 0.25/ⁿJ_FC; Δ = p44; Δ₃ = 2 × p16 + 2 × d16 + p24 + Δ + 8 μs; Δ₁ = d6 − Δ₃/2; Δ₂ = d6 + Δ₃/2 − p14 + (2/π) × p1; t₁(0) is the initial t₁ evolution delay time = 0.5 × in0, where in0 is the t₁ increment. The gradient strengths were are follows: G₁ = 80%; G₂ = cnst30 × G₁, where cnst30 = (1 − sfo2/sfo1)/(1 + sfo2/sfo1) and sfo1 and sfo2 are ¹⁹F and ¹³C frequencies, respectively. The following phase cycling was used: φ₁ = 2x, 2(−x); φ₂ = x, −x; φ₃ = 4x, 4(−x); Ψ = 2(x, −x), 2(−x, x). The echo-antiecho protocol was used with PFGs changing sign between real and imaginary increments. Phases φ₂ and Ψ were incremented by 180° together with the PFG sign change, (b) A 2D ¹⁹F, ¹³C HMBC spectrum of 1 optimised for ⁿJ_FC of 20 Hz acquired using the pulse sequence of shown in (a). The two insets show 1D F₂ traces for individual ¹³C resonances of the α- and β-forms of 1. 1D ¹H-decoupled ¹⁹F NMR spectrum and the ¹³C projection are shown on the top and along the left of the spectrum, respectively.

Fig. 6 (a) Pulse sequence of the (3, 2)D H¹CⁿF correlation experiment. For explanation of symbols used for pulses see Experimental. The NMR parameters used are given in Table S1.†The delays were as follows: d2 = 0.25/¹J_HC; d3 = 0.5/¹J_HC; d4 = 0.25/ⁿJ_FC; d6 = cnst1/¹J_HC, where cnst1 = 0.5 for CH and 0.25 for CH₂ groups; Δ₁ = d3 − p14/2; Δ₂ = d2 − p14/2 − p16 − d16; Δ₃ = d2 − p14/2 − 2t₁(0); Δ₄ = d6; Δ₅ = d4; Δ₆ = p16 + d16 − (2/π)p1 + 4 μs; Δ₇ = p16 + d16 + 4 μs, where p16 and d16 are the PFG length and the recovery time, respectively. The gradient strengths were as follows: G₁ = 40%; G₂ = 42.51%; G₃ = 13%. The following phase cycling was used: φ₁ = y, −y; φ₂ = 4x, 4(−x); φ₃ = 2x, 2(−x); φ₄ = 2y, 2(−y); Ψ = x, 2(−x), x, −x, 2x, −x. The echo-antiecho protocol was used with G₁ changing sign between real and imaginary increments. Phases φ₁ and Ψ were incremented by 180° together with the sign change. Two interleaved experiments were acquired applying either the φ₃ or φ₄ phase to the last 90° ¹³C pulse, (b) an F₁ antiphase (3, 2)D H¹CⁿF spectra of 1 acquired using the pulse sequence shown in (a) showing the cross peaks of the β-anomer of 1. Positive and negative cross peaks are shown in blue and turquoise, respectively. The insets contain vertical and horizontal traces through the H1, F cross peaks. The ¹H chemical shift of protons directly attached to ¹³C atoms and the associated κΩ_13C frequencies are indicated. Antiphase doublets in F₂ show ⁿJ_FC coupling constants. Horizontal and vertical internal projections are shown on the top and along the left side of all spectra, respectively. The editing process that simplifies this spectrum is explained in the text and shown in Fig. S5.†

Results and discussion

1D ¹H and ¹⁹F spectra of 3-fluoro-3-deoxy-D-glucose, 1

A 400 MHz 1D ¹H spectrum of 1 with the suppression of the HOD signal shows considerable overlap of ¹H resonances (Fig. 1a). A ¹H-coupled 1D ¹⁹F spectrum 1 (Fig. 1b) contains two ¹⁹F signals belonging to α- and β-anomeric forms of 1. The insets highlight numerous ¹H–¹⁹F coupling constants of 1. NMR parameters of 1 obtained using the developed experiments, including those involving ¹³C, are presented in Table S2.†

¹⁹F-centred NMR experiments – novelty and hardware requirements

Although a number of NMR experiments exist that correlate ¹⁹F chemical shifts with those of other nuclei,³¹ the majority of existing techniques yield magnitude mode spectra.^32,33 Acquisition of pure-phase absorption signals in a phase-sensitive manner is much preferred, as it provides higher sensitivity and allows for accurate determination of coupling constants, including identification of the active coupling constants. Some existing phase-sensitive experiments yield complicated cross peak structures that lower their sensitivity.^34,35

The optimal performance of experiments constituting the ¹⁹F-centred NMR approach across a range of ¹⁹F frequencies, is ensured by the use of adiabatic inversion pulses.^36,37 The experiments provide pure phase multiplets with simple structure afforded by ¹H or ¹⁹F decoupling and were designed to minimise the effect of passive spins; they do not use refocusing intervals, which maximises their sensitivity. NMR hardware capable of pulsing simultaneously on ¹H and ¹⁹F frequencies is required; fortunately, such systems are more widespread now. To access the rich information provided by ¹³C–¹⁹F interactions, a three-channel NMR spectrometer is necessary. Maximum benefits are realised on systems equipped with highly sensitive low temperature probes. These have also become more widely available, mainly due to their use in binding studies of biomacromolecules with fluorinated ligands.

Fluorine–proton and proton–proton correlation

Following the acquisition of ¹H-decoupled and ¹H-coupled 1D ¹⁹F spectra, mapping of the ¹H–¹⁹F correlations is the natural next step in investigating the structure of fluorinated compounds. For this task a choice of three types of experiment exist: hetero-COSY, HETCOR or HMBC.^31,32 Most of these can be implemented using ¹⁹F or ¹H as the directly detected nucleus. Using ¹⁹F as the directly detected nucleus, the 2D ¹H, ¹⁹F HMBC has the highest sensitivity, but yields mixed-phased multiplets. 2D ¹H, ¹⁹F hetero-COSY can be implemented with either nucleus being sampled in the directly detected (F₂) dimension. Nevertheless, sampling ¹⁹F in the F₂ dimension has a distinct advantage of acquiring spectra with the high digital resolution required for the identification of active and passive J_HF coupling constants and potentially also for their measurements. A disadvantage of COSY type spectra is the antiphase nature of their cross peaks (particularly in F₁) and their large footprint.

Choosing to obtain the ¹H–¹⁹F correlations using a phase-sensitive ¹⁹F-detected 2D ¹H, ¹⁹F HETCOR experiment (Fig. 2a) retains the advantages of ¹⁹F detection. Its uniform performance across a large ¹⁹F chemical shift range is guaranteed by the use of broadband inversion CHIRP pulses³⁸ arranged in a double inversion adiabatic sweep (Fig. S1†), a feature applied in several experiments presented here to eliminate phase evolution of the transverse magnetisation during pulses.^39–41 This allows the use of such pulses not only for spin inversion but also refocusing.

The structure of cross peaks in HETCOR spectra is simplified by the application of a 180° ¹⁹F pulse in the middle of the t₁ interval, reducing the probability of signal overlap in spectra of complex mixtures. A drawback of this experiment is the evolution of ¹H–¹H couplings during the defocusing interval 2Δ₂, which competes with the evolution of ¹H–¹⁹F couplings, decreasing its sensitivity. This decrease can often be tolerated because of the 100% natural abundance of both nuclei.

Due to diverse sizes of J_HF coupling constants, no attempt was made to refocus ¹⁹F magnetisation prior to detection and ¹H decoupling was not applied during t₂. Preserving the antiphase character of cross peaks is important, as it allows the identification of active couplings. Nevertheless, if a ¹H-coupled ¹⁹F 1D spectrum is overlap free, it is advised to read the coupling constants from this spectrum, where accurate values are readily obtained (see Fig. 1b).

In a basic HETCOR experiment,³² the evolution of ¹H–¹H couplings during the ¹H–¹⁹F defocusing interval, 2Δ₂, leads to the appearance of mixed phase proton multiplets in F₁ – a feature that is masked by the magnitude mode presentation of spectra. This issue was resolved in the proposed phase-sensitive experiment by inserting a z-filter⁴² after the t₁ period, which separates the evolution of ¹H–¹H couplings during the t₁ and the 2Δ₂ defocusing interval. Providing the t_1max is kept short (<30 ms), the cross peaks appear as singlets in F₁. The described features of the experiment are illustrated on a 2D ¹H, ¹⁹F HETCOR spectrum of 1 (Fig. 2b), where correlations with many ¹⁹F coupled protons are observed.

Protons not coupled by a sizable (>1.0 Hz) coupling constant to a ¹⁹F, but which are part of a spin system containing at least one ¹H coupled to a ¹⁹F, are detected in a 2D ¹H, ¹⁹F TOCSY–HETCOR experiment (Fig. 3a). Here, the ¹H chemical shifts are labelled before their magnetisation is spread through the network of J_HH coupled spins by a DIPSI-2 spin-lock.⁴³ Part of the magnetisation that has reached the ¹⁹F-coupled protons is then transferred to ¹⁹F for detection in a subsequent HETCOR step. An overlay of the 2D ¹H, ¹⁹F HETCOR and 2D ¹H, ¹⁹F TOCSY–HETCOR spectra (Fig. 3b) revealed several protons with a J_HF close to zero, which were not detected by the HETCOR experiment. Other protons of both anomeric forms of 1 coupled with small coupling constants to ¹⁹F showed increased intensities.

In addition to J_HF coupling constants, J_HH coupling constants provide important structural information that for complex mixtures is inaccessible by standard 2D experiments, but can be retrieved when some form of ¹⁹F editing is used. In principle, ¹H–¹H couplings modulate cross peaks in the F₁ dimension of the 2D (TOCSY–)HETCOR experiments discussed above but in practice, typical t₁ acquisition times used to record such spectra are too short to resolve them. The ¹H–¹H couplings are more likely to be resolved in the F₂ dimension of ¹H-detected experiments considering a non-refocused 2D ¹H-detected ¹⁹F, ¹H HETCOR, this experiment shows F₂ multiplets with J_HF and J_HH coupling constants as anti-phase and inphase splitting, respectively, complicating access to J_HH coupling constants (data not shown).

The J_HH coupling constants can be measured more effectively from inphase proton multiplets acquired in the presence of ¹⁹F decoupling. Developed for simple mixtures of fluorinated compounds, this reasoning has led to the design of FESTA experiments.^27–29 These 1D selective experiments require that both ¹⁹F and ¹H multiplets are amenable to selective inversion, which is rarely the case for complex mixtures; experiments that do not rely on selective manipulations of spins are more robust.

A suitable alternative involving the use of ¹⁹F → ¹H cross-polarisation (CP) that produces inphase ¹H multiplets was already proposed in the form of a 3D CP ¹⁹F, ¹H heteronuclear TOCSY experiment.⁴⁴ We did not find it necessary to label the ¹H chemical shifts after the initial ¹⁹F → ¹H magnetisation transfer and present here a 2D version of this experiment in the form of a 2D ¹⁹F, ¹H CP-DIPSI3–DIPSI2 (Fig. 4a). Here, the signal acquisition can start immediately after the z-filter⁴² that follows the CP step. Note that signals of protons not coupled to ¹⁹F can appear in the spectrum even at this point due to the ¹H–¹H TOCSY transfer that takes place simultaneously with the heteronuclear CP step.

This pulse sequence can be extended by a dedicated ¹H–¹H DIPSI-2 spin-lock propagating the magnetisation transfer to more remote parts of the spin system. Application of two z-filters and ¹⁹F decoupling ensures that pure inphase ¹H multiplets are eventually acquired. DIPSI-3,⁴⁵ using 40 μs ¹⁹F/¹H pulses, was applied for the CP step covering a ±4 kHz frequency range with >75% efficiency. A slight improvement was achieved with the FLOPSY-16 mixing scheme⁴⁶ covering ±4.7 kHz, i.e. 25 ppm of ¹⁹F resonances on a 400 MHz NMR spectrometer with >65% efficiency relative to the on-resonance signal (Fig. S2†). Further improvements, not explored here, can be achieved by using broadband pulses during the CP step.⁴⁷

An overlay of two 400 MHz 2D ¹⁹F, ¹H CP-DIPSI3–DIPSI2 spectra acquired with a 20 ms ¹⁹F → ¹H cross-polarisation (red) and an additional 50 ms ¹H → ¹H spin-lock (blue) using the pulse sequence of Fig. 4a is presented in Fig. 4b. Both spectra are suitable for the determination of the J_HH coupling constants. The former spectrum contains pure in phase multiplets of protons H2, 3 and 4 of 1, while the latter spectrum also shows all their other protons. Note the dominance of the H3 signals in the red spectra caused by an effective CP via large J_H3F3 (∼50 Hz).

Fluorine–carbon correlation

Structure determination of sparsely protonated fluorinated molecules, such as heavily substituted aromatic rings, based only on ¹H and ¹⁹F chemical shifts and coupling constants could be problematic. Thanks to the far-reaching ¹⁹F–¹³C couplings (ⁿJ_FC, n = 1–5), many ¹⁹F-coupled ¹³C atoms can be identified by 2D ¹⁹F, ¹³C correlated experiments such as HMBC or HSQC, making structure determination of such molecules possible. A 2D ¹⁹F, ¹³C HSQC experiment^33,41 was not considered in this study mainly because of a larger complexity of the double INEPT transfer. For small molecules, the slower relaxation of single-quantum (HSQC) relative to multiple-quantum (HMBC) coherences does not make a substantial difference to their sensitivity and for mono-fluorinated compounds F₁ singlets are produced by both experiments.

As the ¹J_FC coupling constants are large (∼150–250 Hz), while the ^n>1J_FC typically range from 0 to 50 Hz,⁴⁸ the one-bond (Fig. S3†) and long-range correlation (Fig. 5a) experiments are best performed separately. A single long-range optimised experiment can also yield one-bond correlations if multiple rotations of the ¹⁹F magnetisation vectors during the evolution interval fall outside of even multiples of 0.5/¹J_FC. This approach can only be used when values of ¹J_FC coupling constants are known, and if dealing with mixtures, their spread is narrow. Values of ¹J_FC coupling constants required for such optimisation can be obtained from 1D ¹H-decoupled ¹⁹F spectra acquired with a sufficient S/N ratio. Alternatively, accordion optimisation⁴⁹ can be used to obtain simultaneously both types of correlations. Both experiments perform best when ¹H decoupling is applied during most of the pulse sequence. Such decoupling removes splitting of cross peaks by ¹H–¹³C couplings in F₁ and by J_HF in F₂. Resulting F₁ singlets and F₂ anti-phase doublets split by ¹⁹F–¹³C interactions (Fig. 5b) allow accurate measurement of J_FC coupling constants that provide valuable structural information.

A comparison of ¹⁹F chemical shifts of ¹³C isotopomers obtained from 2D ¹⁹F, ¹³C HMBC spectra with the ¹⁹F signal in a 1D ¹H-decoupled ¹⁹F spectrum yields ¹³C induced ¹⁹F isotopic shifts (see a large isotopic shift of C3 resonances in Fig. 5b). In aliphatic systems these decrease with the number of bonds separating the two atoms and are generally measurable to up to four bonds separating ¹³C and ¹⁹F. A careful alignment of the one-bond correlation trace from the pure phase HMBC spectrum and the satellites from the 1D ¹⁹F spectrum is required to obtain accurate values of these isotopic shifts.

Proton–carbon–fluorine correlation

¹H–¹H and ¹H–¹³C interactions are the cornerstone of NMR structure determination of small molecules. For fluorinated compounds, the existence of ¹H–¹⁹F and ¹⁹F–¹³C couplings makes this process even more robust. However, for complex mixtures, mapping of these interactions separately, can compromise identification of the nuclei belonging to individual molecules.

This ambiguity can be avoided by correlating all three spin types in a dedicated HCF experiment. There are numerous possibilities for how such an experiment can be designed. Inspired by the 3D HNCA, a pulse sequence for assigning protein backbone resonances,⁵⁰ a ¹H-detected 3D triple-resonance ¹H, ¹³C, ¹⁹F experiment has been proposed previously.^51,52 This out-and-back 3D experiment contains J_FC defocusing and refocusing intervals, samples ¹⁹F and ¹³C chemical shifts indirectly and applies simultaneous ¹³C and ¹⁹F decoupling during the direct detection of ¹H. We prefer to use a unidirectional polarisation transfer pathway and direct detection of ¹⁹F; both of these features are well suited for molecules with a large spread of coupling constants, as is typical for ¹⁹F–¹³C interactions. The pulse sequence of such an experiment starts with a one-bond ¹H–¹³C correlation step followed by a ¹³C, ¹⁹F long-range transfer step. It incorporates a reduced dimensionality approach^53–55 and samples ¹³C chemical shifts simultaneously with the indirect labelling of ¹H resonances. The resulting 2D experiment is referred to as (3, 2)D H¹CⁿF, where the superscripts indicate the type of ¹³C and ¹⁹F interactions (1-one-bond, n-long-range) mediating the polarisation transfer (Fig. 6a).

In the (3, 2)D H¹CⁿF experiment, the ¹H chemical shifts are recorded first, while suppressing the evolution of ¹H–¹H and ¹H–¹⁹F couplings by a BIRD^r,X pulse^56,57 and a 180° ¹⁹F pulse applied in the middle of the t₁ period, respectively.

The magnetisation is then transferred in an INEPT step to ¹³C via one-bond ¹H–¹³C couplings, where it is refocused before starting ¹H decoupling. During the subsequent evolution interval, the ¹⁹F–¹³C anti-phase magnetisation is developed while the central 180° ¹³C and ¹⁹F pulses move simultaneously with the t₁ incrementation. This causes modulation of ¹H chemical shifts by ¹³C offsets, Ω_13C (=δ(¹³C) − ¹³C r.f. carrier frequency) of their directly bonded ¹³C, splitting the signals into doublets centred at the ¹H chemical shift. The size of ¹³C doublets can be scaled down relative to the t₁ evolution (κ factor), keeping the F₁ spectral width small and without any limitations for setting the length of the constant-time ¹⁹F–¹³C coupling evolution interval, 2Δ₅.

The signal is finally transferred to ¹⁹F, where it is detected during t₂ under ¹H decoupling as a pure phase doublet in anti-phase with regard to J_FC (Fig. 6b).

Interleaved acquisition of two spectra, differing by 90° in the phase of the last 90° ¹³C pulse of the pulse sequence, generates inphase and anti-phase F₁ doublets, respectively, allowing spectra to be simplified by spectral editing^58,59 as illustrated in Fig. S5.† A pulsed field gradient assisted echo-antiecho protocol is used to obtain pure phase signals in F₁.

Overall, the reduced dimensionality experiment retains the full information content of 3D spectra with substantially increased digital resolution. Due to the use of a single ⁿJ_FC evolution interval, sensitivity is also improved relative to the original 3D HCF experiment.⁵¹ Detecting ¹⁹F under ¹H decoupling during t₂ further increases sensitivity of this experiment, while providing values of J_FC coupling constants. The (3, 2)D H¹CⁿF experiment thus complements the 2D ¹⁹F, ¹³C HMBC technique discussed above and for protonated carbons correlates unambiguously three atom types, HCF, instead of aiming to achieve the same through a combined interpretation of 2D ¹H, ¹³C HSQC and 2D ¹⁹F, ¹³C HMBC spectra, which for complex mixtures, is problematic.

Structure determination process in ¹⁹F-centred NMR

This process is briefly summarised with the help of a graphical representation in Fig. 7, using the β-anomeric form of 1 as an example. The ¹⁹F–¹H correlations experiments, 2D ¹H, ¹⁹F HETCOR and 2D ¹H, ¹⁹F TOCSY–HETCOR spectra, together with 1D ¹H-coupled/decoupled ¹⁹F spectra provided the parameters summarised in Fig. 7a, while 2D ¹⁹F, ¹H CP-DIPSI3–DIPSI2 experiments extended the identified spin system to protons not directly coupled to fluorine (Fig. 7b).

Fig. 7 NMR parameters obtained by ¹⁹F-centred NMR for the β-anomeric form of 1. Chemical shifts, coupling constants and ¹³C isotopic shifts are given in ppm, Hz and ppb, respectively. (a) δ_1H/ⁿJ_HF; red and blue colours indicate correlations obtained from 2D ¹H, ¹⁹F HETCOR and 2D ¹H, ¹⁹F TOCSY–HETCOR spectra, respectively; (b) J_Hi,H(i+1), red and blue colour indicates correlation from 2D ¹⁹F, ¹H CP-DIPSI3–DIPSI2 without and with the DIPSI-2 extension; (c) ⁿJ_FC/Δ¹⁹F(¹³C); (d) δ_1H/δ_13C; n.d. – not detected.

These experiments thus provide ¹⁹F and ¹H chemical shift correlations together with ⁿJ_HF (n = 2–4)⁶⁰ and ⁿJ_HH (n = 2–3) coupling constants, enabling the start of a structure determination process.

Experiments involving ¹⁹F–¹³C correlations are very informative. Central to these is the 2D ¹⁹F, ¹³C HMBC experiment, which provides long-range ¹⁹F–¹³C correlations and ⁿJ_FC coupling constants and in conjunction with a 1D ¹H decoupled ¹⁹F spectrum also the ¹³C induced ¹⁹F isotopic chemical shifts (Fig. 7c). The subsequent (3, 2)D H¹CⁿF experiment provides correlations of HC pairs, in which the carbon is coupled to ¹⁹F, and if present, a distinction between non-protonated and protonated carbons (Fig. 7d).

Occasionally, a 2D ¹H, ¹⁹F HOESY experiment^31,61 can be used to identify protons not accessible by exploring J coupled networks of spins. In general, at this point, the chemical shift assignment and of numerous ¹H, ¹³C and ¹⁹F resonances, values of J_HF, J_HH and J_FC coupling constants and ¹³C induced ¹⁹F isotopic shifts are known and the structure determination of fluorine containing moieties can be completed.

For larger molecules, which contain spin systems isolated from those containing ¹⁹F, the ¹⁹F-centered approach provides a starting point by identifying protons and carbons that appear in both the ¹⁹F-centered and the standard ¹H–¹H and ¹H–¹³C 2D chemical shift correlated spectra. These resonances can then be used to extend the structures and connect the fluorinated and non-fluorinated parts of molecules, e.g. via ¹H–¹H NOESY experiments or ¹H–¹³C HMBC experiments, which can bridge such spin-systems. This approach is particularly beneficial for analyses of mixtures, where the identity of cross peaks belonging to the non-fluorinated parts of the molecule could be difficult to establish.

Although the discussed NMR experiments were developed for mono-fluorinated compounds, they can also be applied to compounds bearing more than one fluorine atom. Nevertheless, the presence of multiple ¹⁹F atoms should be taken into account when setting up some of the experiments, as the existence of passive ¹H–¹⁹F (or ¹⁹F–¹³C) couplings need to be reflected in the parameters used as outlined in Table S3.†

It should be emphasised, that the ¹⁹F-centered approach takes full advantage of the high sensitivity of ¹⁹F to its environment and minute differences in the ¹⁹F chemical shift of the order of few Hz are sufficient to obtain the kind of information illustrated here on a very simple mixture provided by 1. Application of the ¹⁹F-centered approach to a very complex mixture of chloramination by-products is presented elsewhere.³⁰

Conclusions

The described methodology is based on a concerted use of several NMR experiments, nevertheless, these can also be used in their own right. Collectively, these experiments represent the most effective NMR approach for the structure determination of mono-fluorinated compounds, particularly those contained in mixtures.

The ¹⁹F-centred approach developed here is applicable at any point during the lifetime of fluorinated compounds, e.g. in analysing reaction mixtures during their production, performing mechanistic studies to understand reaction mechanisms and to optimise chemical reactions, investigating their stability, pharmacokinetics, biodegradation and biotransformation and ultimately to follow their fate in the environment.⁶²

Data availability

The spectra obtained in this study are available here: https://doi.org/10.7488/ds/3422.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

NGAB would like to acknowledge NERC soil security programme NE/N020227/1 for funding. AJRS was supported by Scottish Water and EPSRC grant EP/N509644/1 and RY by the NERC Centre for Doctoral Training, E4 (NE/S007407/1). Instrument support was in part provided by the EPSRC grant EP/R030065/1. The authors would like to thank Juraj Bella and Dr Lorna Murray for maintenance of the NMR spectrometers.

References

1. K. Muller, C. Faeh and F. Diederich, Science, 2007, 317, 1881–1886.
2. E. P. Gillis, K. J. Eastman, M. D. Hill, D. J. Donnelly and N. A. Meanwell, J. Med. Chem., 2015, 58, 8315–8359.
3. B. M. Johnson, Y. Z. Shu, X. L. Zhuo and N. A. Meanwell, J. Med. Chem., 2020, 63, 6315–6386.
4. M. Inoue, Y. Sumii and N. Shibata, ACS Omega, 2020, 5, 10633–10640.
5. J. L. Han, A. M. Remete, L. S. Dobson, L. Kiss, K. Izawa, H. Moriwaki, V. A. Soloshonok and D. O’Hagan, J. Fluorine Chem., 2020, 239, 109639.
6. Y. Ogawa, E. Tokunaga, O. Kobayashi, K. Hirai and N. Shibata, iScience, 2020, 23, 101467.
7. O. Jacobson, D. O. Kiesewetter and X. Y. Chen, Bioconjugate Chem., 2015, 26, 1–18.
8. P. T. Lowe and D. O’Hagan, J. Fluorine Chem., 2020, 230, 109420.
9. T. Liang, C. N. Neumann and T. Ritter, Angew. Chem., Int. Ed., 2013, 52, 8214–8264.
10. T. Furuya, A. S. Kamlet and T. Ritter, Nature, 2011, 473, 470–477.
11. D. O'Hagan and H. Deng, Chem. Rev., 2015, 115, 634–649.
12. J. Fang, D. Hait, M. Head-Gordon and M. C. Y. Chang, Angew. Chem., Int. Ed., 2019, 58, 11841–11845.
13. T. Hayashi, G. Kehr, K. Bergander and R. Gilmour, Angew. Chem., Int. Ed., 2019, 58, 3814–3818.
14. A. Rentmeister, F. H. Arnold and R. Fasan, Nat. Chem. Biol., 2009, 5, 26–28.
15. C. D. Murphy, Biotechnol. Lett., 2010, 32, 351–359.
16. B. D. Key, R. D. Howell and C. S. Criddle, Environ. Sci. Technol., 1997, 31, 2445–2454.
17. X. J. Zhang, T. B. Lai and R. Y. C. Kong, in Fluorous Chemistry, ed. I. T. Horvath, 2012, vol. 308, pp. 365–404.
18. M. G. Boersma, T. Y. Dinarieva, W. J. Middelhoven, W. J. H. van Berkel, J. Doran, J. Vervoort and I. Rietjens, Appl. Environ. Microbiol., 1998, 64, 1256–1263.
19. V. S. Bondar, M. G. Boersma, E. L. Golovlev, J. Vervoort, W. J. H. Van Berkel, Z. I. Finkelstein, I. P. Solyanikova, L. A. Golovleva and I. Rietjens, Biodegradation, 1998, 9, 475–486.
20. M. Kiel and K. H. Engesser, Appl. Microbiol. Biotechnol., 2015, 99, 7433–7464.
21. R. Natarajan, R. Azerad, B. Badet and E. Copin, J. Fluorine Chem., 2005, 126, 425–436.
22. E. Nieto-Sepulveda, A. D. Bage, L. A. Evans, T. A. Hunt, A. G. Leach, S. P. Thomas and G. C. Lloyd-Jones, J. Am. Chem. Soc., 2019, 141, 18600–18611.
23. R. Wei, A. M. R. Hall, R. Behrens, M. S. Pritchard, E. J. King and G. C. Lloyd-Jones, Eur. J. Org. Chem., 2021, 2021, 2331–2342.
24. N. G. A. Bell, L. Murray, M. C. Graham and D. Uhrin, Chem. Commun., 2014, 50, 1694–1697.
25. N. G. A. Bell, M. C. Graham and D. Uhrin, Analyst, 2016, 141, 4614–4624.
26. G. A. Bell, A. A. L. Michalchuk, J. W. T. Blackburn, M. C. Graham and D. Uhrin, Angew. Chem., Int. Ed., 2015, 54, 8382–8385.
27. L. Castanar, P. Moutzouri, T. M. Barbosa, C. F. Tormena, R. Rittner, A. R. Phillips, S. R. Coombes, M. Nilsson and G. A. Morris, Anal. Chem., 2018, 90, 5445–5450.
28. T. M. Barbosa, L. Castanar, P. Moutzouri, M. Nilsson, G. A. Morris, R. Rittner and C. F. Tormena, Anal. Chem., 2020, 92, 2224–2228.
29. G. Dal Poggetto, J. V. Soares and C. F. Tormena, Anal. Chem., 2020, 92, 14047–14053.
30. A. J. R. Smith, R. York, D. Uhrin and N. G. A. Bell, Chem. Sci., 2021 10.1039/D1SC06057K.
31. J. Battiste and R. A. Newmark, Prog. Nucl. Magn. Reson. Spectrosc., 2006, 48, 1–23.
32. A. A. Marchione, R. J. Dooley and B. Conklin, Magn. Reson. Chem., 2014, 52, 183–189.
33. R. A. Newmark and R. J. Webb, J. Fluorine Chem., 2005, 126, 355–360.
34. K. A. M. Ampt, R. Aspers, P. Dvortsak, R. M. van der Werf, S. S. Wijmenga and M. Jaeger, J. Magn. Reson., 2012, 215, 27–33.
35. R. Aspers, K. A. M. Ampt, P. Dvortsak, M. Jaeger and S. S. Wijmenga, J. Magn. Reson., 2013, 231, 79–89.
36. A. A. Marchione and B. Conklin, Appl. Magn. Reson., 2017, 48, 485–499.
37. J. E. Power, M. Foroozandeh, P. Moutzouri, R. W. Adams, M. Nilsson, S. R. Coombes, A. R. Phillips and G. A. Morris, Chem. Commun., 2016, 52, 6892–6894.
38. J. M. Bohlen, I. Burghardt, M. Rey and G. Bodenhausen, J. Magn. Reson., 1990, 90, 183–191.
39. T. L. Hwang and A. J. Shaka, J. Magn. Reson., Ser. A, 1995, 112, 275–279.
40. M. H. Levitt and R. Freeman, J. Magn. Reson., 1981, 43, 65–80.
41. B. Adams, Magn. Reson. Chem., 2008, 46, 377–380.
42. M. J. Thrippleton and J. Keeler, Angew. Chem., Int. Ed., 2003, 42, 3938–3941.
43. S. P. Rucker and A. J. Shaka, Mol. Phys., 1989, 68, 509–517.
44. H. Hu, P. Kulanthaivel and K. Krishnamurthy, J. Org. Chem., 2007, 72, 6259–6262.
45. A. J. Shaka, C. J. Lee and A. Pines, J. Magn. Reson., 1988, 77, 274–293.
46. M. Kadkhodaie, O. Rivas, M. Tan, A. Mohebbi and A. J. Shaka, J. Magn. Reson., 1991, 91, 437–443.
47. A. A. Marchione and E. L. Diaz, J. Magn. Reson., 2018, 286, 143–147.
48. H. Duddeck and M. R. Islam, Tetrahedron, 1981, 37, 1193–1197.
49. G. Bodenhausen and R. R. Ernst, J. Am. Chem. Soc., 1982, 104, 1304–1309.
50. L. E. Kay, M. Ikura, R. Tschudin and A. Bax, J. Magn. Reson., 1990, 89, 496–514.
51. L. Li and P. L. Rinaldi, Macromolecules, 1996, 29, 4808–4810.
52. L. L. Li, B. Zhang, F. Wyzgoski, X. H. Li, E. F. McCord and P. L. Rinaldi, ACS Macro Lett., 2013, 2, 141–145.
53. G. Bodenhausen and R. R. Ernst, J. Magn. Reson., 1981, 45, 367–373.
54. Y. Shen, H. S. Atreya, G. H. Liu and T. Szyperski, J. Am. Chem. Soc., 2005, 127, 9085–9099.
55. W. Kozminski and I. Zhukov, J. Biomol. NMR, 2003, 26, 157–166.
56. J. R. Garbow, D. P. Weitekamp and A. Pines, Chem. Phys. Lett., 1982, 93, 504–509.
57. D. Uhrin, T. Liptaj and K. E. Kover, J. Magn. Reson., Ser. A, 1993, 101, 41–46.
58. N. Brodaczewska, Z. Kostalova and D. Uhrin, J. Biomol. NMR, 2018, 70, 115–122.
59. J. Sakas and N. G. A. Bell, Faraday Discuss., 2019, 218, 191–201.
60. Hans Reich's Collection, ¹⁹F NMR Spectroscopy,https://organicchemistrydata.org/hansreich/resources/nmr/?index=nmr_index%2F19F_coupling#f-data08.
61. P. L. Rinaldi, J. Am. Chem. Soc., 1983, 105, 5167–5168.
62. M. T. Anaraki, D. H. Lysak, K. Downey, F. V. C. Kock, X. You, R. D. Majumdar, A. Barison, L. M. Liao, A. G. Ferreira, V. Decker, B. Goerling, M. Spraul, M. Godejohann, P. A. Helm, S. Kleywegt, K. Jobst, R. Soong, M. J. Simpson and A. J. Simpson, Prog. Nucl. Magn. Reson. Spectrosc., 2021, 126–127, 121–180.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ra08046f

---