The rate of mineral loss during dissolution of enamel indirect contact with an acidic demineralizing solution is governed primarily by the slowest event in the process. Enamel demineralization in vitro involves a variety of reaction processes at the dissolving surfaces, as well as diffusion of reactants into and out of the developing lesion. There are many factors that might determine the rate-limiting step; these include porosity of the enamel, connectivity of the pores, the presence or otherwise of surface dissolution inhibitors, surface phase changes, solubility of the mineral, and the available surface area for the dissolution events to take place. Although the rate of loss of mineral during in vitro enamel demineralization has been the subject of many studies, different changes in demineralization rate with time have been reported, suggesting different mechanisms are rate-limiting. For example, Featherstone et al. (1979), Featherstone and Mellberg (1981) and Poole et al. (1981) reported that lesion depth was proportional to the square root of demineralization time in both human enamel and bovine enamel. Christoffersen and Arends (1982) demonstrated that the combined data of Featherstone et al. (1979) and Groeneveld and Arends (1975) gave a plot of lesion depth that varied with the cube root of time. A loss of mineral that varies with the square root of time suggests that diffusive transport through the enamel body is rate-limiting. More complex relationships with time would be expected if other factors are taken into account, e.g. gradients with depth in solubility, porosity, and available reaction surface area, and the fact that the process is not one-dimensional.

However, others, for example Holly and Gray (1968), Theuns et al. (1985), de Josselin de Jong et al. (1987), de Josselin de Jong et al. (1988), Chow and Takagi (1989), Hafström-Björkman et al. (1992) and Gao et al. (1993), found that under conditions of constant composition of the demineralizing solution, enamel demineralization was essentially linear with time. This suggests that processes at the dissolving surfaces, i.e., at the dissolving front, are rate-limiting.

Clearly, there is uncertainty concerning the fundamental mechanisms of lesion formation in vitro, which can make the task of understanding mechanisms in vivo even harder. One possible reason for this is that, in many of the studies described above, measurements were recorded only a few times during the entire demineralization process, so subtle changes in rate would be overlooked. Although a few contact microradiographic methods have been described to monitor some of the intermediate steps in the process, these techniques involve considerable disruption of the specimen, so repeated measurements cannot be taken sufficiently frequently to observe lesion formation in detail. Scanning microradiography enables monitoring of local mineral loss in thin sections of enamel during demineralization without disrupting the specimen. Solid-state photon-counting X-ray detectors are used with good counting statistics, so small changes in mineral content can be detected over a wide range of specimen thicknesses. The number of incident and transmitted photons from the characteristic radiation of the target selected by the counting system are recorded in fixed times. From these, the mineral mass can be calculated, knowing the appropriate mass attenuation coefficient. In addition, the statistical accuracy of any measurement can be calculated directly from the number of X-ray photons counted.

For many scanning (and contact) microradiographic studies, thin sections are used in which the direction of acid attack is perpendicular to the direction of the X-ray beam. Quantitative measurements of mineral loss or gain per unit exposed area require accurate determinations of section thickness, which are difficult and likely to introduce errors. The rate of mineral loss can be determined by subtraction of successive mineral content profiles at known times, but there will be large errors as the profiles are poorly defined. However, if the specimen is oriented parallel with the direction of acid attack, these problems are avoided. de Josselin de Jong et al. (1987) have described such a photographic microradiographic method (longitudinal microradiography) in which the integrated mineral loss per unit exposed area is determined from photographic density measurements (calibrated with an aluminium step-wedge).

Fig. 1. Scanning microradiographic cell in perpendicular profiling configuration with a tooth section varnished all over except for the natural surface. Direction of acid attack is perpendicular to the direction of X-ray beam.

Fig. 2. Scanning microradiographic cell in parallel integrating configuration with an enamel block varnished all over except for the natural surface. Direction of acid attack is parallel to the direction of X-ray beam.

Our aim now was to measure accurately and at frequent intervals the total mineral loss over small areas of the natural surface of human and bovine enamel during in vitro demineralization by combining the idea of integrated measurements of mineral loss, as in longitudinal microradiography, with the advantages of scanning microradiography, particularly those that arise from photon counting. This methodology could have been called longitudinal scanning microradiography, in which longitudinal would have referred to the orientation of the sections originally used by de Josselin de Jong et al. (1987). However other section orientations can be used, so we prefer to use the term parallel integrating scanning microradiography, to emphasize the more fundamental aspect that the direction of acid attack is parallel to the X-ray beam. Conventional scanning microradiography is then termed perpendicular profiling scanning microradiography. For ease of subsequent use, the participles integrating and profiling are often omitted. Perpendicular scanning microradiography was used to confirm the presence of a surface layer in adjacent sections for the human samples.

Blocks (5×5 mm) with a maximum thickness of about 2 mm were cut from four unerupted caries-free human third molars. All the cut surfaces of the blocks were varnished leaving the natural surface exposed. In addition, approx. 500-?m thick sections were cut adjacent to the blocks so that an enamel block and an enamel section were cut from the same tooth. The sections were varnished except for the natural surface.

Approx. 6-mm dia. discs were cut from the mid-labial aspect of each of four permanent bovine lower incisors using a trephine. Visual inspection revealed no indication of any lesions already present in any of the teeth used. The discs were then thinned, with retention of the natural enamel surface, to approx. 2-mm in depth with approx. 1-mm depth of enamel. All surfaces except the natural surface were coated with nail varnish.

Lactic acid (0.1 mol/l) buffered to pH 4.0 with NaOH was pumped from a 5-l reservoir through the scanning microradiography cells at 4.0 ml/min. Solutions were made up in deionized water, but no further attempt was made to remove fluoride. All enamel samples were exposed to identical conditions.

Scanning microradiography is an X-ray absorption method in which a 15-?m beam from a microfocus X-ray generator (Mo target, 40 kV, 1.0 mA tube current) passes through the sample. As the tooth sample did not need to be in contact with a photographic film, it could be mounted in a scanning microradiography cell (approx. vol. 0.5 ml) through which demineralizing solutions were circulated. The scanning microradiographic cells were mounted on an X–Y stage, which was moved with micrometers under computer control to scan different parts of the sample. The number of photons in preset times at each scan position, and outside the specimen, were counted with a high-purity germanium detector with a pulse height analyser (EG & G Ortec, Oak Ridge, TN, USA) set to pass MoK radiation.