New 19F NMR methodology reveals structures of molecules in complex mixtures of fluorinated compounds

Although the number of natural fluorinated compounds is very small, fluorinated pharmaceuticals and agrochemicals are numerous. 19F NMR spectroscopy has a great potential for the structure elucidation of fluorinated organic molecules, starting with their production by chemical or chemoenzymatic reactions, through monitoring their structural integrity, to their biotic and abiotic transformation and ultimate degradation in the environment. Additionally, choosing to incorporate 19F into any organic molecule opens a convenient route to study reaction mechanisms and kinetics. Addressing limitations of the existing 19F NMR techniques, we have developed methodology that uses 19F as a powerful spectroscopic spy to study mixtures of fluorinated molecules. The proposed 19F-centred NMR analysis utilises the substantial resolution and sensitivity of 19F to obtain a large number of NMR parameters, which enable structure determination of fluorinated compounds without the need for their separation or the use of standards. Here we illustrate the 19F-centred structure determination process and demonstrate its power by successfully elucidating the structures of chloramination disinfectant by-products of a single mono-fluorinated phenolic compound, which would have been impossible otherwise. This novel NMR approach for the structure elucidation of molecules in complex mixtures represents a major contribution towards the analysis of chemical and biological processes involving fluorinated compounds.


Introduction
While uorine-containing compounds are the least abundant natural organohalides, 1 modern society has become dependent on numerous man-made uorinated organic molecules such as pharmaceuticals and agrochemicals.
Presently, about 20% of the commercial pharmaceuticals contain uorine and the proportion of newly approved uoropharmaceuticals is rising steadily. [2][3][4] Similarly, uoroagrochemicals have become indispensable for crop production and protecting public health from parasitically transmitted infectious diseases; 5 53% of all active agrochemicals registered during 1998-2020 are classed as uoro-agrochemicals. 6 New fragrance and semiochemical molecules can also benet from uorination. 7 In addition, 18 F is the most frequently used radioisotope in positron emission tomography radiopharmaceuticals for both clinical and preclinical research, and the search for simple and efficient 18 F-labeling procedures is an active research area. 8 Reecting such interest in uorinated molecules, design of efficient and environmentally safe uorination methods 9-11 and scaled up manufacture of uorinated molecules 12 are among the most active elds of organic chemistry. Enzymatic 13 and chemoenzymatic [14][15][16] platforms for the preparation of uorinated compounds are also emerging. To support these developments, there is a need to characterise uorinated molecules using efficient analytical methods, amongst which 19 F NMR spectroscopy plays a prominent role. What makes 19 F the ideal NMR nucleus is its high sensitivity, 100% natural abundance, large chemical shi dispersion and strong and far-reaching spin-spin interactions.
An important advantage of 19 F over other nuclei is the absence of the background signal, reecting the lack of uorinated endogenous compounds. 19 F NMR has the ability to study uorinated molecules in the presence of other CHN-containing molecules and mixtures of uorinated compounds produced by chemical or chemoenzymatic reactions could in principle be analysed with minimal clean-up steps or compound separation.
In its simplest form, 1D 19 F NMR has been widely used in studies of biodegradation and biotransformation of uorinated compounds [17][18][19] and has helped to characterise their catabolic pathways [20][21][22][23][24] and identify cryptic liabilities and features with potentially problematic structural arrangements, 25 which can lead to recalcitrance and/or toxicity. 26 Nevertheless, studying biodegradation pathways still typically requires isolation of metabolites and their identication using known standards; 17 both of these steps could be problematic. Another frequent application of 19 F NMR comes from using a uorinated molecule as one of the reactants in studies of mechanisms and kinetics of chemical reactions. 27,28 The methodology presented here aims to make the process of structure elucidation of uorine-containing molecules contained in (complex) mixtures more efficient. It follows the "NMR spies" approach, where 13 C labelled tags provide information about the nuclei in their vicinity, 29,30 leading to structural characterisation of molecules. In a recent example, introduction of -O 13 CH 3 groups to a subset of molecules as NMR tags led to structural characterisation of 32 phenolic molecules, or their fragments, in a complex matrix of peat fulvic acid. 31 In the case of uorinated organic compounds, 19 F atoms provide a 100% NMR active tags already present in molecules, enabling 19 F-centred NMR structure determination. An example of this approach includes the FESTA family of NMR experiments 32-34 that provide 1 H- 19 19 F-13 C and 1 H-1 H coupling constants and 13 Cinduced 19 F isotopic shis. Put together, the obtained information allows elucidation of uorine-containing molecular moieties and in favourable cases complete structure determination of small uorinated molecules.
We have chosen to illustrate this approach on a study of disinfection by-products (DBPs) produced during water treatment. DBPs are formed when disinfectants react with naturally dissolved organic matter (DOM), anthropogenic contaminants, bromide, and iodide during the production of potable water. Approximately 600-700 DBPs have been reported in the literature so far, 35 some of which exhibit severe health effects. 36,37 Amongst halogenated DBPs, the focus so far has been on the quantication of trihalomethanes (THMs), haloacetic acids (HAAs) and total organic halides (TOXs). [38][39][40][41] As the known compounds constitute less than 50% of TOXs produced by chlorination and less than 20% by chloramination, 38 new generations of DBPs are being continually identied and clas-sied for high priority toxicity studies. 35,42 The commonly used alternative disinfectants to chlorine (ozone, chloramines, and chlorine dioxide) produce lower levels of the four regulated THMs and most HAAs as well as TOXs, however, they increase the concentration of some other priority DBPs. 35,38,43 Chloramination also incorporates nitrogen into DOM molecules 44 generating N-containing DBPs, 39,45 which can be even more toxic than those currently regulated. 37,46 Chloramination was therefore chosen for this study and 15 N labelled NH 4 Cl was used in all experiments to prepare 15 N-containing compounds amenable to NMR studies.
Analytical techniques for the structure determination of DBPs play an important role in this process. Traditional methods, such as liquid/liquid extraction, GC, GC/MS, and solid-phase extraction/MS, 47 oen produce only tentative structures that need validation through the use of authentic chemical standards. 35 Specialised MS 48,49 and MS/MS 50,51 techniques are also being used in this eld. Ultrahigh resolution Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) is making contributions to the characterisation of DBPs at the level of molecular formulae, compound class and functional group classication, including identication of compound classes with the highest DBP formation potential. [52][53][54][55][56][57][58] When ion fragmentation is used more denite structural information can be obtained by MS. 49,51,59 On the other hand, the use of NMR spectroscopy in the structure determination of DBPs is rare and usually requires some form of compound separation. [60][61][62][63] Here we illustrate the power of 19 F-centred NMR structure elucidation of uorinated molecules using a complex mixture of DBPs produced by chloramination of a single uorine-containing molecule.

Experimental methodology
Chloramination A 500 ml sample was prepared with LC-MS grade water and 50 mg L À1 of 3-uoro-4-hydroxybenzoic acid (1). The solution was buffered to pH 7.2 with phosphate buffer. A 15 Nmonochloramine solution was prepared by slow addition of sodium hypochlorite solution to 15 NH 4 Cl in a chlorine-toammonia ratio of 0.8 mol mol À1 and added to the sample in a 3 : 5 mass ratio of carbon: disinfectant, as described previously. 64 All samples were kept in the dark at 20 C for 5 days before the addition of excess Na 2 S 2 O 3 to stop the reaction. The reaction mixture was adjusted to pH 2.0 using HCl before being pumped through PPL SPE cartridges (1 g, 6 ml, Agilent) at a ow rate of $5 ml min À1 . Each cartridge was conditioned using methanol followed by acidied Milli-Q water (pH 2). Aer adsorption of the sample, the column was washed with acidied water in order to minimise the retention of inorganic species. The cartridge was then allowed to dry before being eluted with methanol. The eluent was rotary evaporated to dryness.

NMR experiments and instrumentation
Six new NMR experiments were designed and used in this work: 19 F, 13 C ( 15 N) HMBC optimised for n J FC and 1 J FC coupling constants, (Fig. S10 † and S11 †, respectively), and ⑤ 1 H-detected 2D H 1 C n F (Fig. S12 †). Apart from ② all other pulse sequences make use of a double inversion adiabatic sweep; 65,66 the pulse sequence ① uses a z-lter to deliver pure phase multiplets; 67 the pulse sequence ②, inspired by a 3D TOCSY-HSQC experiment, 68 incorporates the 1 H chemical shi labelling followed by a spin-lock period before the magnetisation is transferred to 19 F for detection; pulse sequence ③ is a simple modication of a 3D 19 F-1 H heteronuclear TOCSY edited 1 H-1 H TOCSY 69 that removes the 1 H chemical shi labelling aer the 19 F / 1 H transfer; the two HMQC based pulse sequences ④ and ④ 0 use the echo-antiecho quadrature detection as proposed by Bazzo et al. 70 but eliminate the 19 F chemical shi evolution and yield pure antiphase 13 C, 19 F doublets; experiment ⑤ is a purposely designed reduced dimensionality [71][72][73] (3,2)D 19 F-detected HCF correlation experiment with a simpli-ed polarisation transfer pathway relative to the existing 1 Hdetected triple-resonance HCF experiment. 74 The full analysis of these experiments will be published elsewhere, however, their most relevant aspect for this work, sensitivity, is analysed in the ESI †.
The reaction product mixture (30 mg) was dissolved in CD 3 OH (180 mL) and placed into a 3 mm NMR tube. Spectra involving 19 F were acquired on a 500 MHz Bruker Avance III HD NMR spectrometer equipped with a 5 mm QCI-F CryoProbe, while the 1D 1 H and a 2D 1 H, 15 N HSQC spectra were obtained on a 800 MHz AVANCE III NMR spectrometer equipped with a 5 mm TCI cryoprobe. All experiments were performed at 300 K using parameters summarised in Table S1 †.

Results and discussion
Hardware requirements and design of 19

F-centered experiments
Historically, pulsing on 1 H and 19 F in one NMR experiment, a requirement for all experiments discussed here, was only possible on a limited number of spectrometers. 75 However, this capability is much more common today. When 13 C information is sought, three channel NMR spectrometers are required for all but peruorinated molecules. To boost the sensitivity of such experiments, highly sensitive triple-or quadruple resonance cryoprobes capable of pulsing simultaneously on 1 H, 13 C and 19 F are typically required. Such systems have become more widely available, mainly due to their use in binding studies of biomacromolecules with uorinated ligands.
The chemical shi correlation experiments involving 19 F have evolved together with general improvements of liquid-state NMR methodology; 75 most notably the use of adiabatic 19 F inversion pulses is now widespread. 66,[76][77][78] Nevertheless, even some more recent 19 F experiments yield magnitude mode spectra, 76,78 provide correlation but not the values of coupling constants, 76 or contain refocusing periods that generally decrease their sensitivity. 77,78 Some phase sensitive experiments yield complicated cross peak structures, thereby lowering their sensitivity. [79][80][81] The new NMR experiments presented here build on these advances, are phase sensitive and produce cross peaks with a simple pattern that allow identication of active coupling constants. They incorporate adiabatic inversion pulses covering a 100 KHz frequency range, ensuring their optimal performance across a range of 19 F chemical shis. The use of a single polarisation transfer interval optimised for n J HF or n J FC coupling constants and the elimination of the effects of passive coupling whenever possible, means that they provide chemical shi correlations mediated by a broad range of coupling constants (4-12 Hz n J HF and 3-26 Hz for n J FC , see Tables S2 † and S3). When applicable, they also use 1 H or 19 F decoupling in the directly detected periods to simplify cross peaks and to boost the sensitivity.

Hundreds of DBPs formed by chloramination of a single molecule
DBPs are typically formed from compounds with activated aromatic rings that react with oxidants to produce modied phenolics and unsaturated aliphatic compounds leading to the generation of trihalomethanes. 82 A simple molecule, 3-uoro-4hydroxybenzoic acid (1, Fig. 1) was therefore selected as a suitable model compound for chloramination using 15 NH 4 Cl.
A 500 MHz 1 H-decoupled 1D 19 F spectrum of the reaction mixture produced by chloramination of 1 is very complex; it contains hundreds of peaks of varying intensity spread across a 90 ppm 19 F chemical shi range, with the majority and the most intense signals appearing within a 34 ppm range. A partial spectrum is shown in Fig. 2 with thirteen of the most intense resonances numbered. Fig. S1 † and S2 present vertical expansions of the full 19 F spectrum and the aromatic  part of a 1 H NMR spectrum of the reaction mixture, respectively.
Providing uorine is not removed during the reaction, chloramination products of a uorinated compound will contain at least one 19 F atom. If the reaction causes oligomerisation, molecules with several 19 F atoms will also be present. Nevertheless, these will likely be too distant to exhibit 19 F-19 F couplings and 19 F atoms will therefore only couple to protons in 12 C molecules and protons and carbons in 13 (Fig. 3) that can be exploited to yield chemical shi correlations of many nuclei.
A 2D 1 H, 19 F correlation spectrum (Fig. S3 †) illustrates the complexity of the investigated mixture. Zoomed in regions of 19 F-centred spectra acquired in this work showing the assignment of signals of compound 9 are presented in Fig. 4.
The 19 F-detected z-ltered 2D 1 H, 19 F HETCOR spectrum (①, Fig. 4) shows HF cross peaks with protons H2 and H5 whose appearance is mediated by large J HF coupling constants.

Sensitivity and resolution limits of 19 F-centered NMR
Based on the analysis of signal intensities of the thirteen most intense resonances seen in the 1 H-decoupled 1D 19 F NMR spectrum of a 30 mg mixture ( Fig. 1 and S1 †), it can be estimated that compound 11the lowest concentration compound that yielded signals in experiments involving 13 Cis present at 1 mM (or 30 mg in 180 mL of CD 3 OH in a 3 mm NMR tube assuming an average molecular weight of 170 g mol À1 for compounds in this mixture). This sensitivity limit applies to an overnight experiment on a 500 MHz NMR spectrometer equipped with a 5 mm QCI-F CryoProbe and a 3 mm sample tube.
Exploring a hypothetical scenario, 30 mg of a mixture could contain a 1000 similar size compounds at around 30 mg each. These would be amenable to the structure determination as outlined here, thanks to the remarkable sensitivity of today's NMR spectrometers and the efficiency of the 19 F-centered approach. The sensitivity of 1 H, 19 F correlation experiments is naturally higher with an estimated concentration limit of $30 mM (or 1 mg for compounds with M w ¼ 170 g mol À1 in 180 ml). This statement is supported by the appearance of hundreds of cross peaks in the 2D 1 H, 19 F HETCOR spectrum (Fig. S3 †) associated with 19 F signals that are 30 Â weaker than the signal of 11. Around 200 spin systems of these minor compounds could be identied in this spectrum. Their cross peaks were resolved due to the exquisite sensitivity of 19 F to its chemical environment. The presented analysis thus provides a glimpse into the complexity of mixtures that are amenable to structure elucidation by 19 F-centered NMR.

Structure determination process in 19 F-centred NMR
In reference (and using symbols to ⑦) to the schematic representation of 19 F-centred NMR experiments (Fig. 3) and the example spectra of the chloramination product mixture (Fig. 4), the steps involved in 19 F-centred NMR structure determination are discussed below and summarised in a owchart (Fig. 5).
The process starts with the acquisition of standard 1D 1 Hcoupled and 1 H-decoupled 19 F spectra, which provide 19 F chemical shis and values of n J HF coupling constants.
① Chemical shis of 19 F-coupled protons are determined in a 2D 19 F, 1 H HETCOR experiment; n J HF coupling constants are assigned.
② The 19 F-associated proton network is extended by protons not directly coupled to 19 F in a 2D 19 F, 1 H TOCSY-HETCOR experiment.
③ J HH coupling constants are obtained in a 2D 19 F, 1 H CP-DIPSI3-DIPSI2 experiment; extension of the proton network, established by ①and ②, is possible.
The correlated 19 F and 1 H chemical shis and homo-and heteronuclear coupling constants can now be interpreted to propose structural fragments by considering the effect of substituents, 85 values of J HF coupling constants 86,87 (Table S2 †) and J HH coupling constants.
⑤ The 2D(3,2) H 1 C n F correlation spectra provide a distinction between protonated and non-protonated 19 F-coupled carbons and chemical shi correlations of HC pairs. Experiments involving 19 F-13 C correlations are very informative and should be performed if sufficient amount of material is available. Considering the effects of substituents, 88 the sizes of J FC coupling constants 86,87 (Table S3 † ⑥ Relative sizes of molecules in a mixture are estimated by a 2D 19 F DOSY experiment. Taking advantage of the large chemical shi dispersion of 19 F, interpretation of 19 F-detected DOSY spectra 89 (Fig. S4 †) is straightforward due to minimal signal overlap. A one-shot DOSY experiment 90 with rectangular 19 F pulses was used here; for spectra covering a wider range of 19 F chemical sis, the use of adiabatic pulses is recommended. 91,92 For the studied mixture, the measured diffusion coefficients generally decreased with increasing molecular weight of compounds and their substituents in the order COOH, NO 2 and Cl. The contribution from the carboxyl groups was particularly large, presumably because of the formation of hydrogen bonds with the solvent. Assessment of the molecular weight also helps to decide if data beyond the reach of 19 F-centred experiments are required.
⑦ 2D 1 H, 13 C HSQC/HMBC spectra provide one-bond and long-range 1 H-13 C correlations beyond the reach of the 19 Fcentered experiment. 2D 19 F, 1 H HOESY experiments can also help to identify more remote protons. Fig. 4 Regions of the 500 MHz NMR spectra acquired with the pulse sequences presented in Fig. S8-S12 † showing chemical shift correlations for compound 9. In addition to 2D cross peaks, the figures display the structure of 9 with selected NMR parameters, and where appropriate, F 2 traces showing the fine structure of cross peaks. ① Overlay of the 2D 1 H, 19 F HETCOR (blue/turquoise) and ② 2D 1 H, 19 F TOCSY-HETCOR (red/ magenta) cross peaks. The TOCSY spectrum was left-shifted to facilitate identification of signals. F 2 traces through H2 and H5 cross peaks from the HETCOR spectrum are shown. 1 H chemical shifts and J HF values (bold) are displayed on the structure; ③ A 2D 19 F, 1 H CP-DIPSI3-DIPSI2 spectrum; F 2 trace at the 19 F chemical shift of 9 is shown; 1 H chemical shifts and J HH values (blue) are displayed on the structure; ④ A 2D 19 F, 13 C HMBC spectrum optimised for n J FC coupling constants. Internal F 1 and F 2 projections and F 2 traces at the 13 C chemical shifts of 9 are displayed; the J FC values are shown in red; ⑤overlay of two edited 2D(3,2) H 1 C n F correlation spectra containing individual cross peaks of the F 1 doublets that code for 13 C chemical shifts. Blue/turquoise and red/magenta colours indicate antiphase J FC F 2 doublets in each spectrum. The internal F 1 projection of one of the spectra is displayed. Vertical lines connect the corresponding signals with their midpoint marking the 1 H chemical shifts. The 1 H/ 13 C chemical shifts and J FC coupling constants (red) are indicated. These active coupling constants appear in antiphase, which can cause partial signal cancellation. Thus, to obtain more accurate values it is best to determine them from a 1 H coupled 19 F spectrum. The H6,F cross peak only appears in the 2D 1 H, 19 F TOCSY-HETCOR spectrum (②, Fig. 4) because the J H6,F coupling constant is too small to generate a response in the former experiment. A 2D 19 F, 1 H CP-DIPSI3-DIPSI2 (③, Fig. 4) serves to extend the proton networks beyond the protons coupled to 19 F, similarly to 2D 1 H, 19 F TOCSY-HETCOR experiment. However, as a 1 H-detected experiment, it provides values of J HH coupling constants that are beneficial to the structure determination process. A 2D 19 F, 13 C HMBC spectrum optimised for n J FC coupling constants (④, Fig. 4) provides the chemical shifts and n J FC coupling constants of all 19 F-coupled carbons. For one-bond 19 F-13 C correlations, the sensitivity of the experiment can be enhanced by optimising the polarisation transfer periods for 1 J FC coupling constants (pulse sequence of Fig. S11 †). If the values of 1 J FC coupling are known, the HMBC experiment can be set up to yield the one-bond correlations as well. Finally, the outcome of a simultaneous H 1 C n F correlation is illustrated in ⑤ (Fig. 4). This intrinsically 3D experiment has been modified using the principles of reduced dimensionality 83,84 to produce a (3, 2)D experiment. Here, the 13 C chemical shift is coded in the 1 H dimension by the width of the F 1 -doublet. In this experiment two interleaved spectra are acquired, which contain in-phase or antiphase F 1 doublets. Editing of these spectra increases the S/N ratio and removes half of the cross peaks in each spectrum, thus reducing spectral overlap.
Using standard 2D 1 H, 13 C one-bond and long-range correlated experiments alone to analyse complex mixtures is problematic due to the complexity of their spectra. Nevertheless, for larger molecules, which contain spin systems isolated from those containing 19 F, protons and carbons identied by 19 Fcentred experiments can act as starting points for extending the assignments through the analyses of 2D 1 H, 13  Analysis of the chloramination reaction pathways 19 F-centred NMR methodology provided a rich set of NMR parameters for the chloramination reaction product mixture (Table S5 †), which allowed the structure elucidation of eleven molecules, present in concentrations above the current sensitivity threshold, and partial structures for two additional molecules (Fig. 6). The analysed mixture was prepared in a 5 day experiment, which led to extensive modication of the starting material producing phenolic and likely also non-phenolic compounds, initially via transfer of Cl released from hypochlorous acid, HOCl. 50,56 Electrophilic substitution reactions, as the main chlorination mechanism for aromatic substitution, 94 resulted in chlorination of 1 producing 2 as the major product. Several DBPs generated by other reactions were also modied in this way-a Cl substitution at the activated ortho position next to an OH group (9/3, 8/13, 4/6). The unexpected appearance of a brominated compound formed from the starting material (1/10) can be explained by the use of NaOCl manufactured by the electrolysis of sodium chloride. Water used in this process contains small amounts of sodium bromide, 95 which led to the production of sodium bromatethe source of Br. The presence and the position of Br in compound 10 was established through a comparison of chemical shis of 2 and 10, which differ only in the nature of the halogen substituent. The experimental differences in the 1 H and 13 C NMR chemical shis at corresponding positions agreed perfectly with the values predicted by  considering the effects of Cl and Br on the chemical shi of benzene resonance. 76,79  The second reaction type observed was decarboxylative chlorination 96 (1/9 or 7/11). The halogenated sites also continued to react with monochloramine through nucleophilic substitution by H 2 N in a dechlorinative amination. 97 The generated aromatic amines were further oxidised by NH 2 Cl to form nitroso-and eventually nitro compounds, 50 (2/12, 10/ 12, 9/8, 3/13). An unexpected outcome was the appearance of compounds 4 and 5. These compounds were not part of the starting material, as conrmed by the absence of their signals in the 1 H-decoupled 19 F spectrum of 1. Their structures were veried by a comparison of NMR parameters with literature data. 98,99 Performing such checks is generally recommended, especially in instances where the appearance of the identied compounds is difficult to rationalise. Such comparisons are considered to be reliable due to sensitivity of NMR parameters to molecular structures.
Two additional compounds, containing a tri-substituted benzene ring with a carboxylic group (7) or a chlorine (11) at position C-1, were identied. The differences between the 13 C and 1 H chemical shis of the corresponding atoms of these compounds matched the differences observed for an analogous pair of molecules, 1 and 9. A possible mechanism for the formation of compounds 7 and 11 from 1 and 9, respectively, is via resonance stabilised phenoxyl radicals produced by dissociation or abstraction of the phenolic hydrogen. 100 This hypothesis is supported by the observed changes of colour of the reaction mixture over the course of 5 days, which could indicate the existence of quinone/semiquinone equilibria. Based on the 19 F DOSY spectrum (Fig. S4 †), molecules 7 and 11 are the largest, likely dimeric molecules. Attempts to extend their structures using 1 H, 13 C correlation experiments, as suggested in step ⑦ of Fig. 5, did not yield further information. A 1 H, 19 F HOESY experiment (not performed here) represents another opportunity for structural characterisation.
The origin of most but not all compounds identied in this study can thus be explained by known reaction mechanisms. It is possible that during the course of chloramination, uorine radicals were created, further modifying the pool of the produced compounds. This could help to explain the variety of 19 F containing compounds ( Fig. 1 and ES1 †) that are present in concentrations too low to currently allow their structure elucidation. The other source of heterogeneity of the nal mixture are the N-containing molecules, as indicated by the richness of its 2D 1 H, 15 N HSQC spectrum (Fig. S6 †). None of compounds 2-13 contain a protonated NH x (x ¼ 1, 2) group, indicating that the nitrogenated products of 1 are present at low concentrations.
The number of compounds obtained in our experiments, which admittedly aimed to maximise the production of DBPs, is astounding. Their structural studies will continue to attract attention due to the potential inuence of DBPs on human health and the environment.

Conclusions
By analysing a complex mixture of DBPs produced by chloramination of a single uorine-tagged molecule, we have demonstrated the feasibility of 19 F-centred NMR structure determination of small molecules without the need for compound separation. The 19 F-centred experiments correlated 19 F chemical shis with those of 1 H, 13 C and 15 N, provided values of J HF , J FC and J NF coupling constants, including 1 H-1 H chemical shi correlations and J HH coupling constants for a subset of protons. The proposed experiments, which can also be used in their own right, thus collectively represent an efficient NMR approach to the structure determination of mono-uorinated moieties and small compounds in complex mixtures.

Data availability
Data is available on request.

Author contributions
NGAB proposed the methodology, designed experiments and performed the structure elucidation. AJRS, DU and RY contributed to the implementation of the experiments and acquisition of spectra. AJRS performed the chloramination reaction. All authors contributed to the analysis of the spectra and writing of the manuscript.

Conflicts of interest
There are no conicts to declare.