This in vitro study was approved by the institutional review board at the Dental Hospital of Wonkwang University (No. WKDIRB201610-01). Seventy-five human premolars, extracted from young adults (age range, 20–29 years) for orthodontic purposes, were used in this study. Teeth with caries, erosion and/or other damage, or restorations were excluded. These teeth were obtained from various private practices and from clinics at the Wonkwang University Dental Hospital (Iksan, Korea). All teeth were cleansed and stored in normal saline solution. The roots of the teeth were cut off with a water-cooled, low-speed diamond saw for ease of use. The teeth were cleansed with nonfluoridated pumice and a rubber cup, thoroughly washed, and then air-dried. The left side of the buccal surface of each tooth was incisogingivally covered with melted acid-resistant wax (Sticky Wax; Kerr Co., Orange, CA, USA), and after cooling, the area was isolated from the test solution or artificial saliva during the experiment (Figure 1A and 1D). This established the Unexposed enamel subgroups prior to immersion of the teeth in the test solutions, and allowed comparison of the exposed surfaces. The right side of the buccal surface, i.e., the unwaxed part of each tooth, was etched with 37% phosphoric acid (DB-Etching 37; Denbio, Gwangju, Korea) for 20 seconds, followed by cleansing for 20 seconds, and drying for 10 seconds (Figure 1B). When the white, chalky surface appeared, Transbond XT Light Cure Adhesive Primer (3M Unitek, Monrovia, CA, USA) was applied on the cervical half of the etched surface, and this was then cured with an light emitting diode (LED) curing light (Ortholux; 3M Unitek) for 20 seconds (Figure 1C). Thus, the occlusal part of the etched surface was designated as the Etched enamel subgroup of each tooth, and the cervical part of the etched surface was designated as the Sealed enamel subgroup of each tooth (Figure 1D).

A soda carbonator with a 3-level LED fizz indicator (Source Sparkling Water Maker, SodaStream, NJ, USA), which is a popular commercial apparatus for manufacturing carbonated water, was used in this study. The LED fizz indicator shows the 3 stages of carbonation levels, by changes in LED lights when the carbonated level reaches a certain range. This was used to prepare carbonated water with different carbonation levels according to the manufacturer's instructions.

The tooth specimens were then randomly divided into 5 groups, in accordance with the carbonation level and the presence of calcium ions in the test solutions (i.e., a control group and 4 experimental groups):

Control group: Ultrapure deionized water

Group 1: Low-level carbonated water

Group 2: High-level carbonated water

Group 3: Low-level carbonated water with calcium ions

Group 4: High-level carbonated water with calcium ions

The product name, manufacturing method, and pH value of each test solution are listed in Table 1.

Over a period of 7 days, all the specimens were submerged in each test solution for 15 minutes, 3 times a day, with 2-hour intervals. When the teeth were not submerged in test solutions, they were stored in artificial saliva at a constant temperature of 37℃. Before being submerged in the test solution, all the specimens were washed with ultrapure deionized water. The artificial saliva was prepared with 0.4 g NaCl, 1.21 g KCl, 0.78 g NaH₂PO₄·2H₂O, 0.005 g Na₂S·9H₂O, 1 g CO(NH₂)₂, and 1,000 mL of distilled and deionized water in the laboratory, as described previously.4 Then, sodium hydroxide was added to this solution until the pH value was measured electrometrically as 6.75.4 The test solutions and artificial saliva were remade and refreshed at the beginning of every experimental session.

After the immersion procedure, the specimens were washed with ultrapure deionized water and the wax was removed mechanically as a whole block without any effect on the enamel surface.

After the immersion procedure, 10 specimens from each group were cross-sectioned mesiodistally to divide the tooth into occlusal and cervical specimens. The cross-section was made at the level of the occlusal third (Figure 1E). Each occlusal specimen was embedded in acrylic resin, such that the cut surface was exposed for the microhardness test (Figure 1F). The surface of each specimen was ground to achieve a favorable flat enamel surface, using 600, 1,000, 1,200 grit silicon carbide abrasive paper sequentially, followed by polishing with 0.05-mm alumina slurry. All the specimens were kept in ultrapure deionized water prior to the microhardness test.

Microhardness tests on the Unexposed enamel subgroup and the Etched enamel subgroup were performed using a Vickers indentor attached to a microhardness tester (MXT-70 Microhardness tester; Matsuzawa, Tokyo, Japan). Microhardness tests were conducted under a 100-g load for 20 seconds dwell time. Three indentations per test were made on the outer enamel surfaces which were 10, 55, or 100 µm apart, making rows toward the dentin. Three tests were performed on each subgroup (Figure 1F). Three rows of indentations were made with a distance of at least 300 µm. The length and width of each indentation were measured and the Vickers hardness number was calculated from the obtained values by the indentation apparatus of a measuring microscope. The microhardness was determined using the average value of 3 indentations per test and 3 rows were assessed for each subgroup.

After the immersion procedure, 5 specimens from each group were cross-sectioned incisogingivally to divide the tooth into buccal and palatal specimens (Figure 1G). Each buccal specimen was kept in a desiccator containing CaCl₂ for 3 days, affixed to scanning electron microscopy (SEM) stubs, coated with platinum twice with a 108 Auto Sputter Coater (Cressington, Watford, UK), and was then examined with a JSM-6360 SEM (JEOL, Tokyo, Japan) at 15 kV, at different magnifications.

The microhardness of each Unexposed enamel subgroup was also compared (Table 2). The Shapiro-Wilks test found a lack of normality of the distribution of the microhardness of each Unexposed enamel subgroup. Thus, the nonparametric Kruskal-Wallis test was used for analysis, and showed no significant difference (p > 0.05).

Descriptive statistics were calculated for the microhardness of each subgroup. Statistical analyses were performed using the Predictive Analytics Software (version 18.0; SPSS Inc., Chicago, USA).

To assess microhardness changes of each group after the exposure to test solutions, microhardness of the Unexposed enamel subgroup and the Etched enamel subgroup were compared (Table 2). Difference values between subgroups in the Control group and Groups 1, 3, and 4 were normally distributed. Paired t-tests were used for the Control group, and Groups 1, 3, and 4, while the Wilcoxon signed rank test was used for Group 2. The nonparametric Kruskal-Wallis test and Mann-Whitney U-test with Bonferroni correction were used as post-hoc comparisons to compare difference values between subgroups (Tables 3 and 4).

Descriptive data and statistical comparisons of microhardness between the Unexposed enamel subgroup and Etched enamel subgroup in each group are given in Table 2, and microhardness changes in specimens after exposure to test solutions are given in Tables 3 and 4. The results showed statistically significant differences between groups (p = 0.000).

All experimental groups, except Group 3, showed greater statistically significant change than the Control group. High-level carbonated water (Group 2, 90.18 ± 33.16) resulted in greater statistically significant changes than low-level carbonated water, without or with the presence of calcium ions (Group 1, 49.57 ± 26.13; Group 3, 32.60 ± 19.83). High-level carbonated water with calcium ions (Group 4, 71.33 ± 38.28) resulted in a greater statistically significant change than low-level carbonated water with calcium ions (Group 3, 32.60 ± 19.83). Low-level carbonated water with calcium ions (Group 3, 32.60 ± 19.83) showed lesser statistically significant change than low-level carbonated water without calcium ion (Group 1, 49.57 ± 26.13). There was no significant difference in microhardness changes between high-level carbonated water with calcium ions (Group 4, 71.33 ± 38.28) and low/high-level carbonated water without calcium ions (Group 1, 49.57 ± 26.13; Group 2, 90.18 ± 33.16).

The enamel surface of the Unexposed enamel subgroup in each group was observed. All groups showed a smooth enamel surface (Figure 2).

Demineralized areas were observed in each group that had been etched and exposed to test solutions. The etched enamel surfaces in Groups 1–4 lost more tissue than the Control group (Figure 3). Dissolution of the peripheral prism was observed in the Control group (Figure 3A). Apparent dissolution of the prism core and a slightly flattened honeycomb-like appearance were observed in Group 1 (Figure 3B), and a flattened and agglomerated honeycomb-like structure was observed in Group 2 (Figure 3C). Dissolution of the prism core and a well-defined honeycomb-like appearance were observed in Group 3 (Figure 3D), while a fuzzy and worn honeycomb-like structure was observed in Group 4 (Figure 3E).

Different amounts of adhesive primer removal were observed on the sealed enamel surfaces (Figure 4). Intact adhesive primer was observed in the Control group (Figure 4A). Adhesive primer with a roughened surface remained in Group 1 (Figure 4B) and some adhesive primer was stripped off (Figure 4C) in Group 2. The border of the adhesive primer was partially removed in Group 3 (Figure 4D) and the border of the adhesive primer was irregular and some part of it had disappeared in Group 4 (Figure 4E).