All the pre-treatment scans of the artificial demineralized enamel lesions show that each tooth had a subsurface demineralized lesion with a surface mineralized layer (Figures 2A,D, 3A,D, and 4A,D). With no SDF treatment, the lesion did not have any observable changes in LAC after immersion in remineralization solution for 120 h (Figures 2B,D). After subsequent demineralization for 72 h, the subsurface lesion extended further into enamel by about 0.06 mm (Figure 2C) and the line profile shows a second mineralized layer on the surface of the advancing front (Figure 2D). This feature could also be seen in the SEM image, which shows substantial loss of mineral (Figure 2E). With the application of SDF, there was a layer of high LAC (∼3.2 cm⁻¹) material on the demineralized enamel surface, and into the lesion after remineralization (Figures 3B,D). After subsequent demineralization, the line profile (Figures 3C,D) shows that there was some loss of mineral in the depth of the lesion (indicated by the grey line shift to the right of the orange line). The radiopaque layer could be seen clearly in the BSE-SEM image covering the enamel, but it did not completely seal the surface. Therefore, a secondary demineralized layer could be detected (B in Figure 3E). With the application of SDF+KI, the initial surface mineralized layer was lost after remineralization (Figures 4B,D) and the normal enamel LAC was increased slightly from 2.6 to 2.8 cm⁻¹. After subsequent demineralization, there was a small loss of mineral in sound enamel (∼0.02 mm).

In all the natural dentin carious lesions, there was no surface mineralized layer as observed in the artificial enamel lesions (Figures 5A,D, 6A,D, 7A,D). In the control sample without SDF application, remineralization did not seem to have an effect (Figures 5B,D). After subsequent demineralization, there was further loss of minerals in the sound dentin (Figures 5C,D). This could also be observed in the BSE-SEM image (B in Figure 5E). With the application of SDF, the DS sample had high LAC at the surface, and a speckled appearance with high LAC islands after remineralization inside the lesion (Figures 6B,D). After subsequent demineralization, no further loss of mineral was observed within the lesion or into the sound dentin (Figures 6C,E). In the BSE-SEM image, high density particles were observed inside the lesion (Figure 6E). No dentinal tubule structures were observed inside the lesion, but was apparent in the sound dentin. With the application of SDF+KI, the DSK sample had a thick layer of radiopaque material (LAC > 3 cm⁻¹) from the carious dentin surface into the into the depth of the lesion (Figures 7B,D), after remineralisation. Even in the deeper part of the lesion, the LAC was increased to around 1.25 cm⁻¹, similar to that of sound dentin. After subsequent demineralization, there was no observable change of LACs in the lesion (Figures 7C,D). The BSE-SEM image (Figure 7E) showed that the lesion maintained some dentinal tubular structure and the high-density materials were deposited along the direction of the dentinal tubules.

Artificial enamel lesions were used because it is difficult to standardise natural enamel carious lesions. Also, in natural early enamel carious lesion, there is usually a surface mineralised layer, possibly with high fluoride content that may prevent penetration of SDF into the lesion. It is confirmed in this study that even with the thinner surface mineralized layer, the penetration of the SDF was limited on the surface (Figure 2B). As the outer surface layer was smooth, the Ag particles did not retain well. This was shown in the ESK sample (Figure 4) which showed total loss of the surface after SDF+KI remineralization (Figure 4B). This might be due to the surface layer was very thin and was lost during the application of SDF+KI.

In the enamel control sample (EC), no increase of LAC (i.e., no remineralization) was observed after immersion in remineralization solution for 120 h (Figure 2B). This might be due to the immersion time was too short; and/or the surface mineralized layer prevented any further exchange of ions in the deeper layer. However, when this tooth was re-immersed in demineralization solution for 72 h, there was further mineral loss in the body of the lesion, with an increase in depth (Figure 2C). It was noted the surface mineralized layer was intact and retained its LAC. Furthermore, there was a 2nd mineralized layer with the second wave of demineralization in the depth of enamel. This implies that clinically when enamel is subjected to cyclical de- and re-mineralization, the dynamics of ion exchange will create a complex enamel structure.

In the ES sample, after SDF application and remineralization, there was a very radiopaque layer on the enamel surface (Figure 3B). When SDF was in contact with the remineralization solution, AgCl is formed (4). From the high LAC value, this layer is due to the Ag content of AgCl. In this sample, the AgCl was retained, covering the surface of enamel. From the line profile (Figure 3D), the AgCl might have penetrated inside the lesion and increased LAC of the lesion. Although remineralization might also occur due to the high F content of SDF, the immersion time is too short for this to happen, as shown in the EC sample. Also, after demineralization, there is some mineral loss beyond the original lesion into normal enamel, showing that SDF does not prevent demineralization in enamel.

After application of SDF+KI and after remineralization, the surface layer of the ESK seemed to be destroyed (Figure 4B). As there was no acidic component in SDF+KI, the destruction was likely caused by the mechanical brushing during the application since the surface was thin and weak. However, there was a small increase in LAC in the normal enamel, indicating that a small amount of AgI (a product formed when SDF is mixed with KI) managed to be deposited on the surface and partly into the enamel. Once again, this layer did not protect the enamel from subsequent demineralization indicating no formation of acid resistant fluorapatite or fluoride substituted hydroxyapatite (FA/FSHA).

These results suggest that the SDF will penetrate into the enamel only if some porosities exist. Li et al. (10), reported the movement of silver ions through the pellicle along the prism boundaries. However, this current study shows that the penetration of SDF is limited to the surface of the demineralized lesion. In natural enamel carious lesion, the porous structure could be different from the artificial lesion. Clinically, when the lesion becomes black after SDF application, it indicates that the lesion is porous enough to retain the SDF byproduct that is photo reduced to metallic Ag, which might offer protection from bacterial invasion.

In natural dentin carious lesion (DC sample), no change of LAC was observed after immersion in remineralization solution for 120 h (Figure 5B). This implies that dentin caries is difficult to remineralize without the presence of fluoride. Hence, there was no protection when the tooth was subsequently immersed in demineralization solution for 72 h, resulting in a loss of mineral and decrease in LAC (Figure 5D). When SDF was applied to the dentin carious lesion (DS sample), after remineralization, a layer of high LAC (up to 3.4 cm⁻³ material was deposited on the surface of the lesion. This material was likely to be AgCl. Some of this material also penetrated into the lesion and randomly deposited in the dentin lesion. After subsequent demineralization, the LAC was unchanged. This implies that unlike enamel, FA or FSHA might have been formed inside the dentin lesions but was masked by the high LAC of precipitated AgCl. This might explain the multiple peaks in the line profile (Figure 6D). As FA/FSHA are more acid resistant and harder, subsequent demineralization did not result in further loss of mineral (11).

Several studies have reported occlusion of dentin tubules with the application of silver ions (2, 10, 12–15). Seto et al. (13) reported the formation of microwires of silver within the dentinal tubules but Kiesow et al. (14) and Menzel et al. (15) found particles of Ag occluding the dentinal tubules. However, if the dentinal tubule structure is destroyed, as shown in the BSE-SEM image (Figure 6E), Ag particles could only be retained within space in the matrix. This might explain the random high LAC island appearance in Figure 6B.

In the DSK sample which had SDF+KI application, there was a thick layer of high LAC material on the surface. Unlike the DS sample, the penetrated material had a structural pattern (Figure 7B). The BSE-SEM image (Figure 7E) showed that the dentinal tubule structure within the lesion was not destroyed. Hence, this radiopaque material, most likely to be AgI, was deposited inside the dentinal tubules, as shown by others (14, 15). Hence, it does not indicate that SDF+KI application results in deeper penetration than SDF, it depends on the structure of the natural dentin carious lesion. Similar to SDF, SDF+KI application would have formed FA/FSHA to harden the dentin and prevent mineral loss in subsequent demineralization (Figure 7D).

Regarding the remineralization effect of SDF, Heukamp et al. (16) found that SDF to be less effective than Cervitec F varnish on artificial enamel caries lesion, which agrees with the results of present study. |With regard to SDF on dentin, Cifuentes-Jiménez et al. (3) found that SDF and NaF resulted in an increase in mineral content, the formation of a crystalline precipitate with high flexural strength. They found no effect of acid challenge on dentin blocks treated with SDF under pH cycling. This agrees with this current study which shows application of SDF with or without KI protects dentin from acid attack.