A single batch of HA discs (diameter 12.5 mm, thickness 1.0–1.3 mm; ca. 0.46 g) was supplied by Himed HiMed, (Old Bethpage, NY, USA). Prior to use, discs (n = 6) were etched × 2 in 500 mL of 10 g/L citric acid (adjusted to pH 3.5) for 30 min at room temperature. Excess acid was removed by rinsing with deionised (DI) water before storage in a humidified atmosphere.

Decay-free bovine incisors were sterilised by immersion in 50 g/L NaOCl for 2 min, followed by 300 mL/L H₂O₂ for a further 2 min. Thereafter, specimens were stored in 9 g/kg NaCl for a maximum of six weeks. Tooth sections (ca. 15 mm × 10 mm × 5 mm; 1.52 g) were cut from the buccal aspect using an Isomet™ 1000 precision saw (Buehler^®, Düsseldorf, Germany). Since one larger face of each specimen consisted of exposed dentine whilst the other was unpolished enamel, some variability was to be expected both within and between the specimens [28,29]. The amorphous smear layer, formed by cutting, was removed by placing the tooth sections in 100 g/L citric acid (pH 2.2) for 2 min. After rinsing in DI water, the specimens were stored in sterile saline. Extracted human molars and pre-molars were stored in an aqueous solution of saturated thymol for up to two weeks after removal of the roots and pulp. Prior to use, teeth were thoroughly rinsed with DI water and visually examined for evidence of damage, decay and white spot lesions. Transverse sections of coronal dentine, 100 μm in thickness, were cut with an Isomet 1000 precision saw and etched (citric acid, 100 g/L; 2 min) to remove the smear layer. Specimens (bovine and human) were sputter-coated with a palladium-gold alloy before imaging using scanning electron microscopy (SEM) at an operating potential of 10 kV (Jeol JSM 6100, Welwyn Garden City, Hertfordshire, UK). Imaging of uncoated bovine specimens was performed by atomic force microscopy (AFM) in contact mode, using a MultiMode Nanoscope IV Scanning Probe Microscope (Digital Instruments, Santa Barbara, CA, USA).

Typical SEM images of transverse sections of human and bovine dentine are shown in Figures 1(a) and Figures 1(b) and an AFM image from bovine dentine is shown in Figure 1(c). Tubule diameters in the bovine specimens (1.7–2.6 μm, c.f. lit. [30-33] 1.8–3.5 μm) were similar to those observed in human specimens (2.2–3.3 μm, c.f. lit. [30-33] 1.9–4.0 μm). Tubules in bovine specimens were less closely packed (3.1 × 10⁴ mm ⁻², c.f. lit. [30-33] 2.1 × 10⁴−3.5×10⁴ mm ⁻²) than those in human specimens (7.9 × 10⁴ mm ⁻², c.f. lit. [30-33] 2.1 × 10⁴–8.3 × 10⁴ mm ⁻²). AFM images, Figure 2, also established an optimum time of 2 min for exposure of freshly cut sections to citric acid (100 g/L), just sufficient for the removal of the smear layer and of debris within tubules. Longer exposures resulted in the formation of craters in the tubule structure whereas shorter exposures did not completely remove the smear layer.

Poly(alkyl methacrylate)s were supplied by Sigma-Aldrich (Poole, Dorset, UK or Schnelldorf, Germany). The generic structure of these polymers and associated acronyms are shown in Figure 3, and their key physicochemical parameters summarised in Table 1. PBMA and PEMA were obtained as solids whilst POMA, PLMA, PHMA, PHxMA were supplied as solutions in toluene. Pure samples of solid polymers were recovered by repeated precipitation from cooled methanol.

Emulsions (28 g/kg) were prepared by adding each polymer (450 mg) in solution (dichloromethane (PEMA) or diethyl ether (PBMA) or low-boiling petroleum ether, 3.0 g) drop-wise (2 min) into aqueous sodium dodecyl sulphate (SDS, Acros Organics, 5 g/kg, 15 g) under sonication (Ultrasonic Processor, amplitude 37%, 7 min). The organic solvent was evaporated from each batch by stirring at ca. 5 K above the solvent boiling point for 4–6 h. To compensate for the concurrent evaporation of water, 500 μL aliquots of DI water were added at 30 min intervals (after cooling to ambient temperature) to give a final mass of 15.45 g. To determine the dry mass, the aqueous phase was removed following separation by centrifugation at 4,000 rpm for 5 min. Particle sizes were determined using a Coulter N4MD Sub-Micron Particle Analyser (Beckman Coulter, High Wycombe, Buckinghamshire, UK) with multiple scattering angle detection and size distribution analysis yielding the diameters and SDs indicated in Table 1. HA discs and bovine tooth specimens were coated by immersion (2 min) in the appropriate polymer-containing emulsion, followed by rinsing with DI water.

The demineralisation experiments were performed using a 12-station synthesis carousel supplied by Radleys Discovery Technologies (Saffron Waldon, Essex, UK). Untreated pre-weighed HA discs or bovine enamel sections were individually placed on a wire mesh support at a fixed distance above the stirrer bar in a reaction tube containing 15.0 mL of 10 g/L citric acid adjusted to pH 3.75 thermostatted to 37 °C. The baskets were then lowered so that the specimens were fully immersed in the erosive medium; typically 4-6 specimens were employed for each experiment. Samples (100 μL aliquots) for the determination of dissolved phosphate were taken at 5 or 10 min intervals up to 40 min. Specimens were then removed, rinsed with DI water, and immersed in the test treatment (28 g/kg polymer latex; 120 s, 20 °C), DI water negative control or one of a range of benchmark control treatments comprising 250–2,000 mg/kg fluoride as NaF. In the latter case exposure times of both 30 s and 120 s were employed. Together with a fresh portion of erosion medium (15.0 mL), the treated specimens were then returned to a cleaned tube in the reaction carousel. Discontinuous sampling and subsequent analysis of phosphate and phosphorous by spectrophotometry or ICP-MS respectively were used to create plots of dissolution as a function of time and thus dissolution rate. The efficacy of the treatment to inhibit erosive demineralisation was determined by using each specimen as its own control. Using a one-way ANOVA with a post hoc Fisher test statistical analysis of the results was performed to compare the effects of the putative protective treatments with that of the fluoride control.

The rates of erosion of individual untreated dental specimens by citric acid (10 g/L, pH 3.75; 37 °C) have been measured by analysis of dissolved phosphorus, for comparison with rates obtained under the same conditions but after exposure of the same specimens to a putative protective treatment: the “in vitro demineralisation method”. These data have been used to assess the variability of specimens and to compare the two methods of analysis of aqueous samples for dissolved phosphate, the data are summarised graphically in Figure 4.

For HA discs, the initial rates of erosion were almost constant (0 to 10 min: as observed 1.24 ± 0.08 mg/L min, relative to the mass of the specimen 0.041 ± 0.002 mg/g min, relative to the area from the dimensions 5.6(±0.3) × 10⁻⁵ mg/mm² min), decreasing by ca. 15% over each complete time profile (20 to 30 min: 1.06 ± 0.01 mg/L min, 0.035 ± 0.001 mg/g min, 4.8(±0.1) × 10⁻⁵ mg/mm² min). For bovine tooth sections the initial rates of dissolution were higher (8.0 mg/L min estimated from a tangent to curve (i) at the origin, Figure 4; average over 0 to 5 min 5.8 mg/L min, 0.058 mg/g min, 1.6 × 10⁻⁴ mg/mm² min). In all experiments the highest total amounts of dissolved phosphorus after 30 min were equivalent to 4.6(±0.9)% of the mass of the HA specimens and 2.5(±0.3)% of the mass of the of bovine tooth specimens. The corresponding proportions reacted of citric acid were estimated respectively as 6(±1)% and 11(±1)%.

In comparing the erosion results from the two methods (15 HA specimens each), Figure 4 lines (ii) and (iii)), the errors were similar for results from both analytical methods (coefficients of variation, cv, 16–26% for ICP-MS and 18–25% for the phosphovanadomolybdate method). The mean values were consistent to within 5–11%, such that pairs of data at each time point did not differ significantly (‘t’ test for null hypothesis, p < 0.01). The errors in data were smaller for the results for bovine tooth specimens (cv 10–15%).

To compensate for the variability in the erosion rates of dental specimens under controlled conditions, the effects of protective treatments have been assessed in vitro by comparing measurements made before and after their application. Dissolution by citric acid (10 g/L, pH 3.75; 37 °C) was monitored by sampling at intervals over 30 min. The method was evaluated by using it to measure the effects of treating specimens with fluoride ion. Dose-response plots (n = 4 for each experiment) were constructed for HA discs that had been immersed (30 s or 120 s) in water or in aqueous sodium fluoride (fluoride ion concentrations 250-2,000 μg/g).

Treatment with fluoride ion (300 μg/g, 120 s) reduced the rate of erosion by 40 ± 10% over 0 to 5 min, increasing to 70 ± 20% over 20 to 30 min, Figure 5(a), reflecting the considerably reduced rate of erosion of the treated sample. Inhibition was increased by the use of higher concentrations of fluoride ion but the effect of using the longer treatment exposure time was much greater (Figure 5(b)). The proportional reductions in sampled phosphorus concentrations at each time point show lower variability (cv 8–22%) than the absolute values (cv 20–32% both before and after treatment).

Poly(alkyl methacrylate)s were selected as putative erosion-protection coatings because of their established good film-forming characteristics and of their affinity for the tooth surface [36]. For systematic investigation, the study employed a homologous series of polymers with glass transition temperatures in the range −100 to +65 ° C and surface energies in the range 17–34 mJ/m², Table 1. For each specimen the inhibition of erosion was assessed from the relative initial rates of dissolution of phosphorus (average over 0–5 min) before and after coating. Relative rates of dissolution over 20–30 min were also obtained.

For HA specimens, erosion was slightly enhanced by pre-immersion in water, Table 2. Relative to this, significant inhibition of erosion was shown by specimens coated with PBMA (32% initially, decreasing to 22%) and PHxMA (24% initially, decreasing to 20%), Figure 6(a,b) and Table 2. Erosion also appeared to be inhibited by coating with PHMA (ca 9% initially) but the standard deviations were too large for the effect to be significant. The polymers PLMA and POMA formed coatings with very slight inhibiting effects while any effect of PEMA coatings was undetectable. For comparison, the inhibition by fluoride ion (300 μg/g) was 43% over 0–5 min increasing to 73% over 20–30 min, Figure 5(a).

In erosion tests using bovine tooth sections prior to treatment, Figure 6(c,d), dissolved phosphorus concentrations initially increased rapidly, reaching 40 mg/L after 10–15 min, but dissolution progressively slowed thereafter. Following measurements to establish baseline erosion rates [37], immersion of the specimens in water (120 s) appeared to inhibit erosion considerably, Table 2. The inhibition (fractional reduction in initial erosion rate) due to PBMA was similar to that given by fluoride ion (Table 2) and to that found using HA discs (F⁻ 250 μg/g, 30 s; Figure 5(b)). Treatment with PHxMA also apparently inhibited erosion but the effect was not significantly greater than the inhibition by water. The effects of combined or sequential treatments with fluoride and PBMA or PHxMA were very close to those for single treatments with fluoride, Table 2. Longer term effects (10–30 min) could not be assessed from the measurements made.

The precision, consistency, the scatter of experimental data and the lower detection limits have been compared for the two analytical methods. In both methods samples were taken using 100 μL micropipettes. In the phosphovanadomolybdate method micropipettes were also used for adding the complexing reagent, whereas in ICP-MS each sample was diluted with a precise, relatively large, amount of aqueous HCl. Absorbances were measured to ±0.001, giving precision in the range 0.2–2%. In ICP-MS, counting rates were reproducible to <2%; other contributions to error arising from procedures are likely to be relatively small. The observed errors in the HA disc erosion experiments (cv, 16–26%; similar for the two methods) were much greater than 2%, while the differences (5–11%) between comparable mean analytical values were not significant. Therefore up to 8% of the variability may be attributed to the HA discs. For erosion experiments using bovine tooth specimens, the errors were smaller–indicating that these specimens were less variable than the HA discs. As used for these experiments, there was little to choose between the two analytical methods, which have different potential problems of interference. Samples for ICP-MS were, however, diluted by 1:140 whereas for the phosphovanadomolybdate method the dilution factor was 1:2. Therefore, allowing also for the minimum detection levels obtained, the ICP-MS method would be more sensitive by a conservative factor of ca. 40. This method could be used for measuring much lower levels of dissolved phosphate, for example from erosion experiments either over shorter time scales or using smaller dental samples.

New agents or technologies incorporated into dentifrices and/or mouth-rinses in future may play an important role in enhancing the anti-erosion efficacy of the formulation. Ideally, these technologies should be fluoride-compatible since anti-caries activity is a key benefit of modern dentifrices and rinses, and the available in vitro and in situ studies suggest fluoride may be clinically relevant as an anti-erosion active. Although the polymeric materials investigated here are intended to form protective barrier coatings that protect the underlying mineral from acid attack, the established in vitro and in situ anti-erosion efficacy of fluoride effectively dictates its adoption as the positive/benchmark control in anti-erosion studies. The homologous series of poly(alkyl methacrylate)s investigated in the present study are characterised by a range of glass transition temperatures, and surface energies of deposited films; both exhibit decreasing trends with increasing pendent-chain length. With the exception of PEMA, the relative efficacy of each coating as an inhibitor of acid erosion of HA discs was also lower for the polymers with longer pendent chains, as indicated in Table 2. The behaviour of PEMA may be due to the poor film-forming characteristics of this polymer. Its Tg is higher than the temperature used for film deposition and its latex particle sizes exceeded 100 nm (Table 1), which is the maximum for the deposition of good quality films [36].

Data from the citric acid-mediated dissolution of untreated HA discs showed a progressive general decrease in erosion rate of 10–20% over each 30 min assay as shown in Figure 5(a). Since erosion might have been expected to accelerate due to a progressive increase in accessible area as the external surface became pitted, this decrease probably reflected the reducing aggressiveness of the challenge as the citric acid was consumed and calcium and phosphate were released. Throughout each assay the dissolution of HA discs after their treatment with PBMA or PHxMA was significantly less (p < 0.01) than that of the corresponding controls (before treatment). These consistent and statistically significant results identify PBMA or by PHxMA as potentially useful anti-erosion agents that may offer additional efficacy benefits to fluoride-based formulations, or new fluoride-free formulations.

That the initial rate of erosion relative to dimensions was approximately three times higher for bovine tooth sections than for HA discs is consistent with the tooth sections' greater exposed surface area, which was associated with the corrugated surface of a cut dentine surface together with the porosity associated with the tubules (Figure 2). Pre-etching of the enamel is also likely to have increased its surface area. Additional factors are possible, however, since the dissolution of biological HA is dependent on the molar ratio of calcium to phosphorus in the sample [38] and also because biological apatite is both more soluble and susceptible to partial transformation to other phosphate phases [39]. Further, the size and perfection of HA crystals may influence the rate of erosion, since smaller or less perfect crystals are more susceptible to dissolution [40]. Finally, the presence in biological apatite of carbonate ions and of other impurities may impact upon solubility [41,42]. The greater susceptibility of dentine to erosive demineralisation is well known and has been attributed to a combination of its porosity and its relatively small crystallites of HA [29]. Although the experiments showed that ca. 11% of the available citric acid had been consumed over the course of each assay (30 min), the decrease in the rate of erosion during the control measurements appeared to exceed what might have been expected from the changes in composition of the liquid phase. The rapidly declining rate of erosion of dentine (Figure 6(c,d)) may be due to the gradual build up of a collagen mat on the surface. This would explain the apparent inhibiting effect of water: after the control measurements the specimens would have become partially protected. The suitability of bovine- tooth specimens to act as their own controls in these experiments therefore becomes questionable. As a consequence, data obtained using these substrates may only be treated as indicative of trends in dissolution behaviour. Nevertheless, significant inhibition by PBMA of citric acid mediated erosion is confirmed. Treatments with both fluoride and a polymer coating produced similar inhibition to that with fluoride alone. Therefore although enhanced inhibition was not shown, neither was there any indication of incompatibility between the treatments.