Introduction

Oral health is highly important for a person’s overall physical health and well-being1,2,3,4. Toothpaste is the most commonly used dental hygiene product to maintain oral health by preventing and protecting against oral diseases, such as caries (i.e., cavities/tooth decay) and gingivitis (i.e., gum disease)5. Within commercial toothpastes, fluoride is the most effective active ingredient to prevent caries/cavities1,5,6,7,8,9,10. Stannous fluoride (SnF2), one of three fluoride sources recognized by the United State Food and Drug Administration, has been used since the 1950’s as an effective way to deliver fluoride ions to tooth enamel and dentin11,12,13,14. However, early toothpastes exhibited challenges associated with the stability of SnF2 and its compatibility with other ingredients14,15,16,17. Fortunately, advances in formulation technologies over the last couple decades have resulted in toothpastes with reportedly stable SnF214,17,18,19,20,21.

The use of SnF2 as a fluoride source is highly appealing because SnF2 has been shown to be an effective antimicrobial agent; antimicrobial agents help reduce gingivitis and plaque formation15,21,22,23,24,25,26,27,28,29,30. The antimicrobial properties associated with SnF2 in toothpaste is thought to arise from the presence of Sn cations with an oxidation state of +2 [Sn(II)]21,23,31,32,33. Therefore, it is important that the majority of Sn within commercially available toothpaste maintains the +2 oxidation state21,23,31,32,33. However, Sn(II) cations are generally unstable and will readily oxidize to the more stable Sn(IV) cations upon air exposure. Commercial toothpaste formulations have various additives designed to stabilize Sn(II) compounds within the formulation. But, it is challenging to directly determine the Sn speciation and oxidation state within commercially available toothpaste due to the low weight loading of SnF2 (0.454 wt%, 0.34 wt% Sn) and the amorphous semi-solid nature of the toothpaste. In 2019, Myers and co-workers used Sn K-edge X-ray Absorption Near Edge Spectroscopy (XANES) to determine that Colgate TotalSF contains approximately 85% of its tin in the Sn(II) oxidation state17. More recently, Desmau and co-workers probed the Sn oxidation state within commercially available toothpastes via Sn K-edge XAS34. In both reports, the relative populations of Sn(II) and Sn(IV) species were estimated by fitting the experimental XAS spectra with XAS spectra of Sn(II) and Sn(IV) standards. However, XAS provides only a partial picture of chemical structure, and energy differences of the Sn(II) and Sn(IV) Sn K-edge spectral features are relatively small (~2.5 eV) and overlapping.

Magic-angle spinning (MAS) solid-state NMR spectroscopy is a powerful technique to determine structure within crystalline and amorphous solids. Sn possess three NMR active isotopes (115Sn, 117Sn, and 119Sn), with 119Sn generally being the preferred nucleus to probe because it has the largest gyromagnetic ratio (2.7 times lower than 1H) and natural isotopic abundance (8.59%). 119Sn exhibits a large isotropic chemical shift range that is highly dependent on the local electronic structure surrounding the Sn atom35,36,37,38,39,40,41,42,43,44,45. In addition, the magnitude of the chemical shift anisotropy (CSA) is highly dependent on the symmetry at the Sn atom; asymmetric Sn coordination environments yield large CSA and broad NMR spectra. The magnitude of CSA is often quantified with the span (Ω) which is calculated from the difference of the largest and smallest principal components of the magnetic shielding tensor (Ω = σ33 – σ11) or the chemical shift tensor (Ω = δ11 – δ33)46,47. In general, Sn(IV) adopts a much more symmetric structure than its corresponding Sn(II) analogs and likely has larger differences in energy between occupied and unoccupied orbitals. For these reasons, Sn(IV) compounds tend to have smaller CSA than Sn(II) compounds40,48,49. For example, 119Sn NMR spectra of Sn(IV) systems typically reveal spans of 0–200 ppm36,40,48,49,50,51,52,53,54,55, whereas for Sn(II) compounds, the span is often between ~ 700–1000 ppm and can be upwards of ~4000 ppm39,40,41,48,49,53,56,57,58,59,60,61,62. We note that there have been a few reports of Sn(IV) or Sn(II) exhibiting relatively large or small spans, respectively, due to either distorted or highly symmetric coordination environments, respectively49,62,63,64,65. Unfortunately, conventional room temperature 119Sn MAS NMR spectroscopy of commercial toothpaste is not practically feasible due to the low weight loading of Sn within commercial toothpastes (~ 0.34 wt% Sn). In addition, the semi-solid nature of toothpaste will likely result in molecular mobility that causes the full or partial averaging of the 119Sn CSA at room temperature. Additionally, solution 119Sn NMR is also hindered by the semi-solid nature of toothpaste.

Here, we apply cryogenic MAS dynamic nuclear polarization (DNP)66,67,68,69 to enhance 119Sn solid-state NMR signals of frozen toothpastes by one to two orders of magnitude, allowing 1D 119Sn solid-state NMR spectra to be obtained in minutes from dilute commercial toothpaste formulations. Conventional room temperature solution and solid-state NMR experiments are performed on SnF2 and SnF4. In a MAS DNP experiment, microwave irradiation is used to saturate electron paramagnetic resonance transitions, resulting in the subsequent transfer of electron spin polarization from stable free radicals to the 1H spins of the solvent matrix and/or analyte66,67,68. Toothpastes typically contain high amounts of water and/or glycerol; water/glycerol mixtures have been shown to be an ideal matrix for MAS DNP experiments66. We note that the use of DNP to enable the acquisition of 119Sn solid-state NMR spectra of dilute Sn(IV) species has been previously demonstrated55,70,71,72,73,74,75. To enable DNP experiments the AMUPol biradical76 was directly dissolved in the toothpaste formulation. The DNP experiments were performed at ca. 110 K on frozen toothpaste. Freezing the toothpaste is beneficial because it eliminates any molecular motion, allowing measurement of 119Sn CSA and enables 1H-119Sn cross-polarization (CP) to transfer the DNP-enhanced 1H polarization to 119Sn nuclei. The 119Sn CSA is shown to be a sensitive probe of the Sn oxidation state and from fits of the 1D 119Sn solid-state NMR spectra the relative amounts of Sn(II) and Sn(IV) within the formulation can be estimated.

Results and discussion

Sn(II) and Sn(IV) Fluoride—19F and 119Sn Chemical Shifts and Sample Purity Analysis. We first performed room temperature 19F and 119Sn solid-state NMR spectroscopy on SnF2 and SnF4 to determine chemical shift tensor parameters of Sn in the II or IV oxidation state, respectively. SnF4 features Sn in the more stable IV oxidation state, with each Sn atom residing in a symmetric octahedral environment coordinated by 6 F atoms (Fig. 1A)77. Consequently, no spinning sidebands are observed in the 119Sn solid-state NMR spectrum of SnF4 recorded with a 25 kHz MAS frequency, indicating Ω is less than 170 ppm (Fig. 1B, upper). Note throughout this entire manuscript, the Herzfeld-Berger convention is used to report the CSA46,47. α-SnF2 contains Sn in the less stable, II oxidation state and there are two unique Sn sites in asymmetric environments that are coordinated by three or five F atoms (Fig. 1A)78. 1D 119Sn NMR spectra of SnF2 were recorded for samples purchased from two different suppliers. The 119Sn NMR spectrum of SnF2 (supplier b) reveals two isotropic 119Sn NMR signals with δiso = –948 ppm or –1023 ppm and Ω = 990 ppm or 930 ppm, respectively (skew = κ = 1.0 in both cases; Fig. 2B, lower). The 119Sn NMR spectrum of SnF2 (supplier b) is consistent with prior reports and the structure determined from single-crystal X-ray diffraction59,79. On the other hand, the sample from supplier a shows additional 19F and 119Sn NMR signals. These additional NMR signals are attributed to Sn(IV) fluoride impurities because the observed 19F chemical shifts, 119Sn CSA, and correlations observed in the 2D 19F{119Sn} J-HMQC spectrum are similar to those observed for SnF4 (Fig. 1B and 2 and Supplementary Fig. 2). We note that the 19F chemical shift of tin fluoride materials appear to be sensitive to the Sn oxidation state; SnF2 and SnF4 exhibit a Δδiso(19F) of 100 ppm (Fig. 2A).

Fig. 1: Crystal structures, 119Sn solid-state NMR spectra and 19F{119Sn} J-HMQC solid-state NMR spectra.
figure 1

A Crystal structures of (left) SnF4 and (right) SnF277,78. B 1D 119Sn spin echo NMR spectra of (upper to lower) SnF4, SnF2 from supplier a and SnF2 from supplier b recorded with a 25 kHz MAS frequency. An analytical simulation of the 119Sn solid-state NMR spectrum of SnF2 is shown below with fits of sideband intensities shown for the SnF3 and SnF5 sites (solid green and orange peaks, respectively). Estimated and fitted chemical shift tensor spans (Ω) are indicated. C 2D 19F{119Sn} J-HMQC NMR spectra of SnF2 from (upper) supplier a and (lower) supplier b. Spectra were recorded with J-evolution times (τJ) of 100 μs. Asterisks (*) indicate spinning sidebands.

Full size image
Fig. 2: Solution and solid-state 19F NMR spectra and solution 19F{119Sn} HSQC NMR spectrum.
figure 2

A 19F spin echo NMR spectra of (upper to lower) the model toothpaste, SnF4 and SnF2 (supplier b) recorded in either (upper) solution or (lower two) the solid-state. Integrated peak intensities are indicated on the solution 19F NMR spectrum (red). Asterisks indicate spinning sidebands. B Modified 19F{119Sn} J-HSQC pulse sequence that allows for evolution of the 19F-119Sn J-coupling during the 119Sn t1-evolution period. τJ denotes time periods for evolution of heteronuclear J-couplings. Note the absence of the 19F π-pulse in the center of the t1-evolution period. C Solution 2D 19F{119Sn} J-HSQC NMR spectrum of the model toothpaste acquired with the pulse sequence shown in (B). Splittings due to 19F-119Sn one-bond J-couplings (1J) are indicated.

Full size image

Periodic plane-wave density-functional theory (DFT) calculations utilizing the gauge-including projector-augmented wave (GIPAW) method predicts that the 119Sn isotropic shielding (σiso) and Ω of two Sn species in SnF2 differ by 76 and 68 ppm, respectively, and that the most shielded (i.e., most negatively shifted) Sn site exhibits the smaller Ω (Table S1). The most shielded Sn site is coordinated by 5 F atoms. The predicted difference in σiso/δiso and Ω is in excellent agreement with that observed experimentally, where the δiso and Ω differ by ca. 75 and 60 ppm, respectively. Therefore, we assign the two 119Sn NMR signals with δiso = –948 ppm (Ω = 990 ppm) or –1023 ppm (Ω = 930 ppm) to Sn coordinated by three or five F atoms, respectively. We note that while the difference in the DFT calculated Ω is in excellent agreement with that observed experimentally, the magnitude of the DFT calculated Ω is smaller (Table S1). GIPAW calculations do not account for relativistic effects, which are likely needed to accurately calculate 119Sn CS tensors80,81. GIPAW calculated 19F chemical shielding values are in reasonable agreement with experiment (Supplementary Fig. 1). The calculations predict that SnF2 should have 19F chemical shifts which are approximately 100 ppm more positive than those of SnF4.

The 119Sn NMR spectrum of SnF4 shows multiple 119Sn NMR signals that all exhibit small CSA, revealing that this SnF4 sample contains impurities (Fig. 1B, upper). A 1D 19F spin echo NMR spectrum reveals three NMR signals between ca. –120 ppm to –160 ppm (Fig. 2A); but, only two 19F NMR signals are expected for the terminal and bridging F atoms in a 1:2 ratio, respectively (Fig. 1A). We recorded a 2D 19F{119Sn} J-HMQC NMR spectrum to better probe the SnF4 species from the impurities within this material (see Supplementary Material, Supplementary Fig. 1). Only the 119Sn NMR signals at ca. –750 ppm and –785 ppm show correlations to two unique 19F NMR signals at ca. –130 ppm and –150 ppm; the difference in the 19F isotropic chemical shift [Δδiso(19F)] for the latter two sites agrees with that predicted by GIPAW DFT calculations (Δσiso(19F) = 26 ppm, Supplementary Table 1). Therefore, both 119Sn NMR signals at ca. –750 ppm and –785 ppm could plausibly be ascribed to SnF4. However, we tentatively assign the −785 ppm 119Sn NMR signal to SnF4 based on the much larger signal intensity observed in the 1D 119Sn spin echo NMR spectrum and the 19F chemical shifts observed in the 1D 19F spin echo NMR spectrum (Fig. 1B and S1). We suspect that the other unassigned signals are derived from impurities, such as hydrated phases or other polymorphic forms of SnF4.

2D 19F{119Sn} J-HMQC NMR spectra of SnF2 from supplier b reveals the expected correlations between all 19F and 119Sn NMR signals associated with SnF2 (Fig. 1C, lower and Supplementary Fig. 3). The 2D 19F{119Sn} J-HMQC NMR spectrum of SnF2 from supplier a reveals additional correlations between different 19F and 119Sn NMR signals that were not observed for SnF2 from supplier b (Fig. 1C, upper). Notably, two unique sets of 19F NMR signals of SnF2 from supplier a resonate at significantly lower δiso(19F) than expected for SnF2; the δiso(19F) is near that observed for SnF4 (Supplementary Fig. 2). The lowest frequency 19F NMR signal correlates with a 119Sn that exhibits no observable CSA with a 25 kHz MAS frequency. Based on the low frequency 19F NMR signal and small 119Sn CSA, we assign these NMR signals to Sn(IV) fluoride impurities. We note that the observed 19F and 119Sn NMR signals are different than those assigned to SnF4.

In summary, SnF2 exhibits a significantly larger 119Sn CSA than SnF4, illustrating that the 119Sn CSA is a good probe of the Sn oxidation state, consistent with prior literature. Additionally, the 19F chemical shifts for SnF2 are much more positive than those for SnF4, suggesting that 19F NMR can also provide insight into the Sn oxidation state. Interestingly, impurities in the SnF2 and SnF4 samples from supplier a were detected in 1D 19F and 119Sn NMR spectra. Notably, a 2D 19F{119Sn} J-HMQC spectrum of SnF2 from supplier a revealed an Sn(IV) fluoride impurity. The observation of an Sn(IV)-based impurity is important because SnF2-based toothpastes rely on maximum Sn(II) availability for optimal performance. Impure SnF2 starting materials will lead to a less effective toothpaste. 19F and 119Sn MAS solid-state NMR spectroscopies are good tools to determine the purity of tin fluoride materials, which may be used as precursors in toothpaste products.

Solution 19F and 119Sn NMR Spectroscopy. We performed room temperature solution 19F and 119Sn NMR spectroscopy experiments on a model toothpaste to initially probe all Sn and F atoms before studying commercially available toothpastes with DNP-enhanced 119Sn NMR spectroscopy. The model toothpaste consisted of ca. 2 wt% SnF2 in a 1:2 mixture of D2O:glycerold-8; the solution was prepared ca. 1 month before running NMR experiments. The use of SnF2 in D2O:glycerold-8 makes this a simplified version of most SnF2-based toothpastes, allowing for an easier assessment of the Sn speciation before studying more complex commercial toothpaste formulations. We note that the solution 19F and 119Sn NMR experiments suggest that the majority of F atoms are dissociated from Sn. As discussed in more detail below, we assume that only Sn atoms fully dissociated from F will be primarily observed in the DNP-enhanced 119Sn solid-state NMR spectra because we lack the capability to decouple 19F. The sizeable 19F heteronuclear couplings for F coordinated tin ions could lead to reduced 119Sn homogeneous transverse relaxation time constants (T2’) and low 119Sn CPMG NMR signal intensities.

A 1D 19F solution NMR spectrum of the model toothpaste reveals primarily two 19F NMR signals at –88.3 ppm and –157.4 ppm (Fig. 2A). The broad hump from –150 ppm to –200 ppm is a probe background 19F NMR signal (Supplementary Fig. 4). Closer examination of the –157.4 ppm 19F NMR signal reveals three set of doublets centered around the isotropic NMR signal which correspond to 1-bond J-couplings of 19F-119Sn (1571 Hz), 19F-117Sn (1505 Hz) and 19F-115Sn (1385 Hz) (1J; Fig. 2C, upper and Supplementary Fig. 4). The integral of each doublet matches with the corresponding Sn isotopic abundances and the J-couplings scale with the gyromagnetic ratios of the tin isotopes. We recorded a 2D 19F{119Sn} J-HMQC solution NMR spectrum to probe all F atoms bonded to Sn (Supplementary Fig. 5). Only the 19F NMR signals near –155 ppm were observed in the 2D J-HMQC NMR spectrum. Therefore, we assign the 19F NMR signal at –90 ppm to free fluoride ions. Integration of the 19F NMR signals reveals that at least ca. 92% of F is present as ions within the D2O/glycerold-8 mixture (Fig. 2A). We note that the population of solvated F ions is likely even higher as the 19F transmitter was on resonance with the F-Sn 19F NMR signal and due to overlap of the –157.4 ppm signal with the probe background 19F NMR signals (see Methods).

The 19F NMR signal assigned to an F-Sn species resonates at a similar shift as was observed for SnF4 (Fig. 2A). To confirm the origin of the –157.4 ppm 19F NMR signal we recorded a 2D 19F{119Sn} J-HSQC solution NMR spectrum using the modified pulse sequence shown in Fig. 2B. The modified pulse sequence does not have a central refocusing (π) pulse applied to the 19F spins during 119Sn t1 evolution, which causes evolution of 19F-119Sn J-couplings and 119Sn chemical shifts in the indirect dimension, resulting in a multiplet dependent on the number of attached F atoms to Sn. The modified 2D 19F{119Sn} J-HSQC solution NMR spectrum reveals a septet like pattern in the 119Sn dimension and a doublet in the 19F dimension (119Sn decoupling was not performed during acquisition, Fig. 2C). Note the central line of the septet pattern is absent because the central line of the multiplet arises from 119Sn spins coupled to 3 19F spin-up and 3 19F spin-down leading to an effective J-coupling of 0 Hz. The observed 119Sn septet matches exactly to that of a numerical simulation for Sn attached to six F atoms; the three doublets exhibit splittings at 2, 4, or 6 times 1JSn-F and the intensities are consistent with an AX6 spin system determined from a modified Pascal triangle82. Therefore, the 19F NMR signal at –157.4 ppm is assigned to the (SnF6)−2 anions.

In summary, integration of the 1D 19F solution NMR spectrum suggests at most ca. 8% of F is directly bonded to Sn (Fig. 2A). However, assuming that the two 19F NMR signals correspond to free fluoride anions (F–) and (SnF6)–2, then only ca. 2% of Sn within the model toothpaste contains F bonds, consistent with a 1D 119Sn solution NMR spectrum that does not show appreciable amounts of Sn-F species (Supplementary Fig. 6). This observation is important because it implies essentially all Sn atoms are observable in the DNP 119Sn solid-state NMR experiments. Sn species exhibiting F bonds would exhibit significant 119Sn NMR signal attenuation in DNP solid-state NMR due to large 19F-119Sn dipolar couplings that will not be effectively averaged in the absence of 19F heteronuclear decoupling.

Dynamic Nuclear Polarization 119Sn Solid-State NMR Spectroscopy. We performed dynamic nuclear polarization (DNP) enhanced 119Sn solid-state NMR spectroscopy on the model toothpaste, a commercially available preventative gel (pg1) and four commercially available toothpastes (t1-t4). The weight loadings of SnF2 in the model toothpaste, pg1 and t1-t4 are ca. 2 wt%, 0.40 wt% or 0.454 wt%, corresponding to absolute Sn loadings of ca. 1.5 wt%, 0.30 wt% or 0.34 wt%, respectively. pg1 contains primarily glycerol, similar to the model toothpaste (glycerol and water, Supplementary Table 2). t1 and t3 are glycerol-based, while t2 and t4 are water and glycerol and/or sorbitol-based. In addition the toothpastes contain other ingredients, such as abrasives (Table S2). t4 included SnCl2 as an SnF2 stabilizer. Samples were prepared for DNP experiments by directly dissolving the AMUPol biradical at a final concentration of 11 mM within the toothpastes (see Methods). For the model toothpaste, H2O was added to increase DNP enhancements (10:30:60 ratio of H2O:D2O:gylcerold-8). All DNP experiments were performed immediately after biradical addition. Prolonged storage of the DNP samples in a –20 °C freezer for days resulted in no DNP enhancements, likely due to reduction of the nitroxide biradical caused by conversion of Sn(II) to Sn(IV). In the absence of dissolved SnF2, the nitroxide biradical water glycerol solutions retain their DNP enhancements after months of storage in a freezer, consistent with reaction of Sn(II) with the nitroxide radicals causing the loss in DNP enhancements.

1H → 13C cross-polarization magic-angle spinning (CPMAS) DNP enhancements (ε) of the model toothpaste and pg1 were ca. 84 and 105, respectively (Fig. 3A and Supplementary Fig. 7). This means that acquisition of solid-state NMR spectra with the same signal-to-noise (SNR) ratio would take more than a thousand times longer to acquire without MW irradiation (i.e., no DNP). The DNP enhancement (ε) measured with 1H → 119Sn CP-CPMG experiments was estimated to be between 12–45 for the model toothpaste and t1-t4 (Fig. 3B and Supplementary Fig. 8). However, a 119Sn NMR spectrum could not be recorded without MW irradiation in a reasonable amount of time (ca. 1 h). Therefore, we are likely under-estimating the 119Sn DNP enhancements. The 119Sn DNP enhancements are likely similar to those measured for 13C since in both 119Sn and 13C cross-polarization NMR experiments, the magnetization is derived from the same bath of 1H spins. The high DNP enhancements enables the acquisition of 2D 1H-119Sn CP heteronuclear correlation (HETCOR) and magic-angle turning (MAT) NMR spectra of all samples.

Fig. 3: DNP-enhanced 13C and 119Sn solid-state NMR spectra.
figure 3

Comparison of (A) 1H → 13C CPMAS and (B) 1H → 119Sn CP-CPMG NMR spectra of the model toothpaste recorded (black) with or (red) without microwave (MW) irradiation. The DNP enhancements (ε) are given in the figure. # denotes the truncated glycerol signal. DNP-enhanced (C) 1H → 13C and (D) 1H → 119Sn 2D CP-HETCOR NMR spectra of the model toothpaste. Spectra were recorded with a 10 kHz MAS frequency, eDUMBO1–22 1H homonuclear decoupling during 1H indirect dimension evolution time, and CPMG detection of 119Sn. The orange band illustrates that the same 1H chemical shift is present in both 1H NMR spectra. CP contact times (τCP) and total experiment times are indicated. Asterisks denote quadrature artifacts that occur at 1H transmitter frequency.

Full size image

A 2D 1H → 13C CP-HETCOR NMR spectrum of the model toothpaste reveals correlations between the -OH 1H NMR signals of glycerol or H2O and the 13C NMR signals of glycerol (Fig. 3C). The glycerol CHx 1H NMR signals were not observed as the glycerol was fully deuterated. A 2D 1H → 13C CP-HETCOR NMR spectrum of preventative gel 1 reveals the expected correlations between all the 1H and 13C NMR signals of the CHx and -OH groups of the glycerol solvent (Supplementary Fig. 9A). Interestingly, 2D 1H → 119Sn CP-CPMG HETCOR NMR spectra of the model toothpaste and pg1 both display correlations between all 1H NMR signals observed in the 2D 1H → 13C CP-HETCOR NMR spectra with a broad 119Sn NMR signal (Fig. 3D and Supplementary Fig. 9B). The observed 1H-119Sn correlations reveal that the Sn is present as ions that are likely solvated by H2O and/or glycerol. The interaction between Sn(II) and glycerol was also observed in liquid chromatography mass spectrometry (LCMS) experiments (Supplementary Fig. 10).

1D 1H → 119Sn CP-CPMG NMR spectra of all samples were acquired with multiple 119Sn transmitter offsets (VOCS acquisition83,84) due to the large breadth of the 119Sn NMR signals and relatively low 119Sn NMR excitation bandwidth (Supplementary Fig. 11). The 1D 1H → 119Sn CP-CPMG NMR spectra of the model toothpaste and pg1 reveal primarily a broad 119Sn NMR signal ca. 1500 ppm in breadth, a larger range than was observed for SnF2 (ca. 1000 ppm). Likewise, the 1D 1H → 119Sn CP-CPMG NMR spectra of t1-t4 reveal broad 119Sn NMR signals that cover a range of ca. 1000–1500 ppm in addition to sharper 119Sn NMR features. The broad 119Sn NMR signals likely correspond to Sn(II) species whereas the sharper features likely correspond to Sn(IV). However, identification of Sn(II) and Sn(IV) species is fairly ambiguous from the 1D NMR spectra alone. The 1D 119Sn MAS solid-state NMR spectra are likely broad and featureless because of the presence of isotropic chemical shift distributions, which are typical of disordered and amorphous systems. There is probably a distribution in the number of water, hydroxide, and glycerol (or glyceroxide) molecules coordinated to each tin ion.

To resolve Sn(II) from Sn(IV) species based on their CSA, we recorded DNP-enhanced 2D 119Sn adiabatic magic angle turning (aMAT) NMR spectra of all samples (Fig. 4 and S12–S15)85,86,87,88. These 2D NMR experiments required only ca. 6 h for the model toothpaste (Sn loading ~ 1.5 wt%) and ca. 16–17 h for pg1 and t1-t4 (Sn loading = 0.3 or 0.34 wt%, respectively). In a MAT NMR experiment, an isotropic NMR spectrum free of spinning sidebands (indirect dimension) is correlated with its corresponding anisotropic MAS NMR spectrum (direct dimension). Therefore, the anisotropic NMR spectra extracted at specific isotropic chemical shifts (δiso) can be easily fit to determine the span (Ω) and skew (κ) because the δiso is known.

Fig. 4: DNP-enhanced 119Sn aMAT NMR spectra.
figure 4

2D 119Sn aMAT NMR spectra of (A) the model toothpaste and (C) t1 acquired with a 10 kHz MAS frequency, 1H → 119Sn CP at the start of the experiment, and CPMG for 119Sn detection. B, D 119Sn solid-state NMR spectra extracted from the 2D aMAT NMR spectra at the indicated 119Sn isotropic chemical shifts (δiso). Analytically simulated spectra are shown (colored) below the (black) experimental MAS spectra. The values of the isotropic chemical shift (δiso), span (Ω) and skew (κ) used in the analytical simulations are indicated next to each row.

Full size image

2D 119Sn aMAT NMR spectra of the model toothpaste and pg1 are near identical and reveal broad isotropic 119Sn NMR spectra in the region of ca. –600 to –1000 ppm (Fig. 4A and S12A). The broad isotropic 119Sn NMR spectra illustrate that there are large distributions in the 119Sn δiso, likely due to differences in the Sn ion coordination from the solvent matrix. We note that the center of the isotropic 119Sn NMR spectrum of the model toothpaste appears at the same 119Sn chemical shift observed in solution, however, the breadth of the signal is much narrower at room temperature due to dynamics of the Sn ions in solution (Supplementary Fig. 6). Anisotropic 119Sn NMR spectra extracted from the isotropic 119Sn dimension reveals primarily Sn sites with a Ω of ca. 1200 ppm and a κ of +0.7, which are assigned to Sn(II) species based on the large CSA (Fig. 4B and S12B). However, anisotropic 119Sn NMR spectra extracted at higher 119Sn δiso of ca. –750 ppm or –800 ppm to –600 ppm for the model toothpaste or pg1, respectively, show an increased intensity for the isotropic (center band) NMR signals. An increase in intensity for the isotropic NMR signal reveals there are additional sites with spans of 150 ppm or less, but with identical 119Sn isotropic chemical shift to sites that have a larger span; the small CSA sites should correspond to Sn(IV). Therefore, the 2D 119Sn aMAT NMR spectra clearly reveal both Sn(II) and Sn(IV) sites based on their CSA. We note that the experimental anisotropic NMR spectra show distorted signal intensities at lower 119Sn chemical shifts due to limited 119Sn NMR excitation bandwidth. Nevertheless, Ω and κ can still be accurately determined by fitting the most intense spinning sidebands with the isotropic shift as a fixed constraint.

DNP-enhanced 2D 119Sn aMAT NMR spectra of t1, t2, and t3 are near identical and display three relatively sharp isotropic 119Sn NMR signals at ca. –762 to –775 ppm, –732 to –744 ppm and –665 ppm, in addition to a broad isotropic 119Sn NMR signal from ca. –500 to –600 ppm (Fig. 4C, S13A and S14A). The relatively sharp 119Sn NMR signals at ca. –762 to –775 ppm and –732 to –744 ppm clearly show intense isotropic 119Sn NMR signals, which are assigned to Sn(IV) species based on the small Ω (ca. 150 ppm; Fig. 4A, S13B and S14B). There are additional weak sidebands associated with Sn that have spans on the order of ca. 1200 ppm, suggesting some Sn(II) species are present at these isotropic shifts. Anisotropic 119Sn NMR spectra extracted at more positive 119Sn δiso of ca. –665 ppm and –600 ppm reveal significantly more intense broad 119Sn NMR spectra with spans of ca. 1200 ppm (Fig. 4A, S13B and S14B). At the lower 119Sn δiso of ca. –665 ppm, the isotropic NMR signal has increased signal intensity, consistent with additional small CSA sites (Ω ≈ 150–250 ppm). However, the more positively shifted isotropic 119Sn NMR signals are clearly primarily associated with Sn(II) sites. We note that the isotropic 119Sn NMR spectra are not representative of the Sn(II) and Sn(IV) populations due to differences in MAT efficiencies for high or low CSA sites, respectively. The similarities in the 2D 119Sn aMAT spectra of t1 and t3 are not surprising because they are both primarily glycerol-based. However, the similarities between the MAT NMR spectrum of t2 with the MAT NMR spectra of t1 and t3 are interesting because t2 contains a significant amount of water in addition to glycerol.

Interestingly, the 2D 119Sn aMAT NMR spectrum of t4 is significantly different from that of t1-t3 (Supplementary Fig. 15A). As mentioned above, t4 contains primarily water and sorbitol (similar to t2), in addition to SnCl2 as a SnF2 stabilizer (Table S2). The isotropic 119Sn NMR spectrum of t4 shows primarily three isotropic 119Sn NMR signals at ca. –665 ppm, –550 ppm and –475 ppm (Supplementary Fig. 15B). Similar 119Sn isotropic NMR signals were observed for t1-t3, however, for t4 the 119Sn isotropic NMR signals at ca. –665 and –550 ppm correspond to predominantly small CSA sites (Ω ~ 150–220 ppm; Supplementary Fig. 15B). The 119Sn isotropic NMR signal at ca. –475 ppm clearly shows sites with both large and small CSA (Ω ≈ 220 ppm and 1100 ppm).

With knowledge of all 119Sn chemical shift tensors, the 1D 1H → 119Sn CP-CPMG NMR spectra of all samples could be analytically simulated (Fig. 5 and Supplementary Table 3). The 2D 119Sn aMAT NMR spectra revealed large distributions in the 119Sn δiso. Therefore, we fit the 1D 1H → 119Sn CP-CPMG NMR spectra to sites containing large amounts of Gaussian line broadening to represent the distributions in the isotropic chemical shifts.

Fig. 5: DNP-enhanced MAS 1H → 119Sn CP-CPMG solid-state NMR spectra.
figure 5

(upper to lower) 1H → 119Sn CP-CPMG NMR spectra for the model toothpaste, pg1, and t1-t4. Multiple CP-CPMG NMR spectra were recorded with different 119Sn transmitter frequencies (i.e., VOCS style acquisition, Supplementary Fig. 11). All spectra were recorded with a 10 kHz MAS frequency and a sample temperature of approximately 110 K. Total fits are indicated by red dashed lines. Solid blue, brown, orange, and green peaks correspond to the individual sites used in fitting.

Full size image

The model toothpaste and pg1 display similar 1D 1H → 119Sn CP-CPMG NMR spectra, where the populations of Sn(II) were determined to be ca. 92 or 93%, respectively (Fig. 5). t1-t3 also display similar 1D 1H → 119Sn CP-CPMG NMR spectra (Fig. 5). The populations of Sn(II) were determined to be ca. 80% for t1 and t3 and 90% for t2. We also recorded a 1D 1H → 119Sn CP-CPMG NMR spectrum of t2 after allowing the toothpaste to dry out and exposing it to air over the course of 1 day (Fig. 5). Interestingly, the population of Sn(II) decreased from ca. 90 to 83%, resulting from oxidation of Sn(II) to Sn(IV) from O2 in the atmosphere. This experiment was an important control that confirms our hypothesis that the 119Sn NMR signals of Sn(IV) primarily have small spans, while those associated with Sn(II) have larger spans. Consistent with the differences observed in the 2D 119Sn aMAT NMR spectrum, t4 displays a different 1D 1H → 119Sn CP-CPMG NMR spectrum with a significantly lower amount of Sn(II) (ca. 68%; Fig. 5). We note that the relative populations of Sn(II) for t1-t4 determined here are generally consistent with prior measurements made in Sn K-edge X-ray absorption studies17,34.

There are three main mechanisms that can lead to inaccurate Sn(II) and Sn(IV) populations determined from the 1D 1H → 119Sn CP-CPMG NMR spectra: (1) differences in DNP enhancement, (2) difference in 1H → 119Sn CP dynamics, and (3) differences in 119Sn refocused transverse relaxation time constants (T2’). DNP enhancements for Sn(II) and Sn(IV) should be identical since 2D 1H → 119Sn CP-HETCOR NMR spectra revealed that Sn is present as ions within the solvent matrix and 1H spin diffusion should distribute the DNP enhanced 1H polarization homogeneously across the frozen solution (Fig. 3D and Supplementary Fig. 9B). 1H → 119Sn CP dynamics are likely similar for Sn(II) and Sn(IV) sites since they are both likely coordinated by water, hydroxide ions and/or glycerol molecules. However, 1H → 119Sn CP is likely less efficient for sites with high CSA because the CSA is comparable to or larger than the RF field used for the 119Sn spin-lock pulse. From this perspective, the Sn(II) populations are likely a lower bound. To assess differences in 119Sn T2’, we investigated the effect that the number of CPMG echoes used during that acquisition of 1H → 119Sn CP-CPMG NMR spectra of the model toothpaste and t1 had on the determined Sn(II) and Sn(IV) populations (Supplementary Fig. 16). 1H → 119Sn CP-CPMG NMR spectra processed with 1 to 100 spin echoes in the CPMG train reveal near identical populations of Sn(II) and Sn(IV), confirming that the 119Sn T2’ must be similar for all species. The observation of similar 119Sn T2’ for all Sn species is also consistent with minimal Sn sites exhibiting F bonds, as those sites would exhibit a shorter 119Sn T2’. Therefore, analytical simulations of the 1H → 119Sn CP-CPMG NMR spectra likely reveal relatively accurate populations of Sn(II) and Sn(IV), where the population of Sn(II) should be taken as a lower bound due to differences in CP efficiencies.

In conclusion, we applied dynamic nuclear polarization (DNP) enhanced 119Sn solid-state NMR spectroscopy to determine the Sn oxidation state and speciation within commercially available SnF2-based toothpastes that contain loadings of less than 0.5 wt%. We first obtained room-temperature 19F and 119Sn solid-state NMR spectra of SnF2 and SnF4. These experiments confirmed Sn(II) exhibits a near order of magnitude larger span than that of Sn(IV), consistent with prior literature. NMR studies of SnF2 purchased from two different suppliers revealed a significant amount of Sn and F-based impurities in one of the samples. Notably, 2D 19F{119Sn} J-HMQC NMR spectra revealed that the impure SnF2 sample contains Sn(IV) fluoride-based impurities. 19F and 119Sn solid-state NMR are good probes of the purity of SnF2 precursors used in the production of toothpastes. Solution 19F and 119Sn NMR studies on a model toothpaste consisting of ca. 2 wt% SnF2 in a 50:50 mixture of D2O:glycerold-8 suggested that only ca. 2% of the dissolved Sn ions contain F bonds. DNP experiments of model and commercially available toothpastes were enabled by directly mixing the DNP polarizing agent (AMUPol biradical) within the toothpaste. Importantly, the sensitivity gains offered by DNP enabled detection of 119Sn NMR signals from toothpastes with loadings of 0.34 wt% Sn. 2D 1H → 13C and 1H → 119Sn CP-HETCOR NMR spectra of the model toothpaste and pg1 suggested that all Sn is present as ions that are solvated by water, hydroxide anions and/or glycerol. 1D 1H → 119Sn CP-CPMG NMR spectra of all toothpastes revealed broad 119Sn NMR spectra, with some additional sharper features. Acquisition of 2D 119Sn magic-angle turning (MAT) NMR spectra of all samples allowed for the unambiguous identification of Sn(II) and Sn(IV) species based on their CSA. With knowledge of the 119Sn chemical shift tensor parameters, 1D 1H → 119Sn CP-CPMG NMR spectra were fit to estimate the populations of Sn(II) and Sn(IV) within the toothpastes. Notably, three of the four commercially available toothpastes contained at least 80% Sn(II), whereas one of the toothpaste contained a significantly higher amount of Sn(IV).

We have demonstrated that DNP-enhanced 119Sn solid-state NMR spectroscopy is an ideal technique to probe the Sn speciation with commercially available toothpastes. The determination of the Sn(II) and Sn(IV) populations within commercially available toothpastes is important both to assess the quality of current formulations and to develop new and improved formulations. Increasing the amount of Sn(II) should increase the antimicrobial properties of SnF2-based toothpastes. We observed that both glycerol-based toothpastes (t1 and t3) and toothpastes containing high amounts of water and glycerol (t2) can exhibit high amounts of Sn(II) (ca. 80–90%). However, the Sn(II) is readily oxidized to Sn(IV) after prolonged air-exposure. More detailed studies on the specific coordination of Sn and their interactions with other common toothpaste ingredients are on-going in our labs. By better understanding how common toothpaste ingredients interact with Sn, and specifically Sn(II), DNP-enhanced 119Sn solid-state NMR spectroscopy will enable the rational design and development of next-generation SnF2-based toothpastes that exhibit increased Sn(II) availability and long-term oxidation stability.

Methods

Samples of SnF4 and SnF2 (supplier a) were purchased from Sigma Aldrich, Inc. SnF2 supplier b corresponds to the SnF2 that is used in commercial toothpaste products. AMUPol was purchased from CortecNet Inc. Glycerol-d8 and D2O were purchased from Sigma-Aldrich Inc. All materials were used as received without further purification. Two different model toothpastes were prepared for solution NMR experiments and DNP solid-state NMR experiments, respectively. A micropipette was used to transfer the water-glycerol solutions. However, due to the high viscosity of the glycerol solutions, an analytical balance was used to weigh the amount of solution transferred to ensure that the desired volumes of water, glycerol and water-glycerol solutions were transferred. For the solution NMR experiments, a 2 wt% SnF2 solution was prepared by dissolving 7.1 mg of SnF2 in 347.6 mg of a 1:2 volume fraction solution of D2O:glycerol-d8. The solution was stored for 1 month at room temperature before running solution NMR experiments. The model toothpaste for DNP solid-state NMR experiments was prepared by first making approximately 200 μL (260 mg) of a stock solution of 10:30:60 volume fraction H2O:D2O:glycerol-d8. 2.2 mg of SnF2 was weighed out and 107.9 mg of the stock H2O:D2O:glycerol-d8 solution was added to give a final SnF2 concentration of 2 wt%. The SnF2 was allowed to dissolve over the course of approximately 2 h. This solution as then stored in a freezer until it was needed. To prepare samples for DNP experiments, a few mg of the AMUPol biradical were then weighed out in a glass vial and SnF2 H2O:D2O:glycerol-d8 stock solution was then added to obtain a final AMUPol concentration of 10 ± 2 mM. Approximately 20 μL of this solution was then transferred to a sapphire DNP rotor which was then capped with a silicone plug and a zirconia drive cap. The packed rotor was then transferred into the pre-cooled DNP probe within 20 min to limit oxidation of the Sn2+ ions by reaction with AMUPol. AMUPol was used for DNP experiments because this radical gives high DNP enhancements and sensitivity gains at 9.4 T for water-glycerol based mixtures66.

Samples of commercial toothpastes were prepared for DNP experiments by directly mixing the AMUPol biradical within the toothpastes to obtain a final AMUPol concentration of ca. 10 mM. A typical sample preparation of the commercial toothpastes for DNP consisted of weighing out the AMUPol biradical in a vial (ca. 1.3–2.3 mg), adding the proper amount of toothpaste to reach a concentration of ca. 10 mM, and then vigorously stirring the toothpaste for ca. 10 min to ensure the radical was homogenously mixed throughout the toothpaste. The densities of the toothpastes were assumed to be ca. 1.3 g cm−3. We note that samples of toothpastes 1–4 were taken from the middle of a fresh toothpaste tube, while preventative gel 1 was taken from the top of a fresh tube. During the mixing step, the vial was periodically held under a stream of warm water for short time periods (a maximum time of ca. 10 s) to decrease the viscosity of the toothpaste and facilitate better mixing and dissolution of the radical. Once the radical was homogenously mixed with the toothpaste, the sample was immediately packed into a 3.2 mm sapphire rotor. The sapphire rotor was sealed with a silicone soft plug and capped with a zirconia drive cap. All DNP samples were prepared immediately before performing NMR experiments. The maximum time it took to prepare the sample and insert the rotor into the spectrometer was ca. 20–30 min. We note that prolonged storage (1–2 weeks) of the prepared samples at ca. 0 °C gave no DNP enhancements due to reduction of the biradical, presumably caused by oxidation of Sn(II) to Sn(IV).

Room temperature solution 19F and 119Sn solution NMR experiments were performed on the model toothpaste and recorded on a 9.4 T (ν0(1H) = 400 MHz) Bruker standard-bore magnet equipped with a AVANCE NEO console and a liquid-N2 cooled Bruker Prodigy HXY NMR probe. 19F and 119Sn chemical shifts were referenced to CCl3F and SnMe4, respectively, with D2O as the lock signal. The 19F π/2 and π pulses were 15 and 30 μs in duration, corresponding to a 16.7 kHz radio frequency (RF) field. We note that the 19F NMR spectrum was recorded with the 19F transmitter on resonance with the SnF6−2 19F NMR signal. The 19F NMR signals of the free F ions were ca. 26 kHz away from the 19F transmitter. The 19F spin echo NMR spectrum was recorded with 100 μs delays on each side of the 19F π pulse. The 19F solution NMR spectra were acquired with different recycle delays to ensure that quantitative relative peak intensities were obtained. Recycle delays used for solution NMR experiments are given in Supplementary Table 4. The 119Sn π/2 and π pulses were 12.5 μs and 25 μs in duration, corresponding to a 20 kHz RF field. The 119Sn spin echo NMR spectrum was recorded with 10 μs echo delays on each side of the 119Sn π pulse.

Room Temperature solid-state NMR spectroscopy experiments were performed with a 9.4 T (ν0(1H) = 400 MHz) Bruker wide-bore magnet equipped with a Bruker Avance III HD console and a 2.5 mm HXY magic-angle spinning (MAS) NMR probe configured in triple resonance mode. We note that the 19F and 119Sn match was relatively poor (ca. 30 and 60% for 19F and 119Sn, respectively) when tuned simultaneously to 19F and 119Sn on the 1H and X channel, respectively. The magnetic field was referenced to 1% tetramethyl silane (TMS) in CDCl3 with adamantane (δ(1H) = 1.71 ppm) as a secondary chemical shift reference. 19F and 119Sn chemical shifts were indirectly referenced to CCl3F and SnMe4, respectively, using the previously published IUPAC recommended relative NMR frequencies89.

The 19F π/2 and π pulses were 4 and 8 μs in duration, corresponding to a 62.5 kHz radio frequency (RF) field. The 119Sn π/2 and π pulses were 3.5 and 7 μs in duration, corresponding to a 71 kHz RF field. 2D 19F{119Sn} J-based heteronuclear multiple quantum correlation (J-HMQC) experiments were recorded with the arbitrary indirect dwell (AID) HMQC pulse sequence88. SPINAL-64 heteronuclear decoupling with a 50 kHz 19F RF field was performed during the acquisition of 119Sn NMR signals90. 119Sn and 19F solid-state NMR spectra were acquired at multiple MAS frequencies to confirm the assignment of isotropic and sideband NMR signals (Supplementary Fig. 17).

DNP-enhanced solid-state NMR spectroscopy experiments were performed with a 9.4 T (ν0(1H) = 400 MHz) Bruker wide-bore magnet equipped with a 263 GHz gyrotron, a Bruker AVANCE III console and a 3.2 mm HXY MAS DNP NMR probe67. MAS solid-state NMR experiments were performed with a sample temperature of ca. 110 K.

The magnetic field was referenced to 1% TMS in CDCl3 with the 1H shift of the silicone soft plug (δ(1H) = 0.24 ppm) as a secondary chemical shift reference. The 1H shift of the silicone soft plug was determined based on the 1H shift of frozen tetrachloroethane (TCE, δ(1H) = 6.2 ppm). 13C and 119Sn shifts were indirectly referenced to SiMe4 or SnMe4, respectively, using the previously published IUPAC recommended relative NMR frequencies89. All NMR spectra were initially processed and referenced with the Bruker Topspin 3.6.1 software. Carr-Purcell Meiboom-Gill (CPMG) echo trains were co-added using the NUTs NMR software (Acorn, Inc.). The 119Sn solid-state NMR spectra were analytically fit using the open-source ssNake NMR software91.

All experimental NMR parameters (MAS frequency, recycle delay (τrec. delay), number of scans, t1 dwell (Δt1), t1 TD points, t1 acqusition time (t1 AQ), CP/J-evolution durations (τCP/J-evolv.) and total experimental times are given in Supplementary Table 4. DNP NMR experiments were performed with the NMR probe configured in either HXY triple-resonance mode (tuned to 1H-119Sn-13C) or HX double-resonance mode (tuned to 1H-119Sn). In all probe configurations, the 1H π/2 and π pulse lengths were 2.5 and 5 μs in duration, corresponding to a 100 kHz RF field. The 13C π/2 and π pulse lengths were 4 and 8 μs in duration, corresponding to a 62.5 kHz RF field. In triple-resonance HXY mode, the 119Sn π/2 and π pulse lengths were 4 and 8 μs in duration, corresponding to a 62.5 kHz RF field. In double-resonance HX mode, the 119Sn π/2 and π pulse lengths were 3 and 6 μs in duration, corresponding to an 83.3 kHz RF field. 1H → 13C cross-polarization (CP) was achieved with a 10 kHz MAS frequency with simultaneous 1H and 13C spin-lock pulses with RF fields of ca. 62 kHz (ramped from 56–62 kHz) and 64 kHz, respectively. In triple-resonance HXY mode (10 kHz MAS frequency), 1H → 119Sn CP was achieved with simultaneous 1H and 119Sn spin-lock pulses with RF fields of ca. 72 kHz (ramped from 65–72 kHz) and 56 kHz, respectively. In double-resonance HX mode (10 kHz MAS frequency), 1H → 119Sn CP was achieved with simultaneous 1H and 119Sn spin-lock pulses with RF fields of ca. 76 kHz (ramped from 69 – 76 kHz) and 80 kHz, respectively. Optimization of the 1H → 119Sn CP contact time showed that 6 ms was optimal.

All 119Sn NMR spectra were acquired with CPMG detection to increase sensitivity. π/2 (1D NMR spectra of t1–t4) or π (all other spectra) pulses were implemented in the CPMG trains92,93. 1D 1H → 119Sn CP-CPMG NMR spectra were acquired with multiple 119Sn transmitter offsets due to the large breadth of the 119Sn NMR spectra (i.e., VOCS style acquisition; Supplementary Fig. 11)83,84. 2D 1H → 13C and 1H → 119Sn CP-HETCOR NMR spectra were recorded with 100 kHz 1H RF field of eDUMBO1–22 homonuclear dipolar decoupling applied during the t1-evolution period94. Each pulse in the homonuclear dipolar decoupling train was 32 μs in duration. 2D 119Sn adiabatic magic-angle turning (aMAT) NMR spectra were recorded with the published pulse sequences85,86,87,88. Frequency swept tanh/tan inversion pulses were 100 μs in duration (i.e., 1 rotor-cycle for a 10 kHz MAS frequency) with an ca. 90 kHz RF field, a 2 MHz sweep width and 400 points. All 119Sn aMAT spectra were recorded with 1H → 119Sn CP at the start of the experiment and with the arbitrary indirect dwell (AID) t1 acquisition mode to increase sensitivity88. 100 kHz 1H RF field of SPINAL-64 heteronuclear decoupling was performed during the acquisition of 13C and 119Sn90.