A comparative study of the corrosion stability of dental amalgams with electrochemical impedance measurements

The corrosion susceptibility of a selection of amalgams used in dentistry has been examined with the electrochemical impedance method. The results are compared with data derived from cyclic voltammetry performed with these materials before. Most examined materials including a conventional amalgam show similar corrosion resistance; however, only one product shows a significantly higher corrosion resistance.

this in vitro corrosion presents a serious drawback. [1] The situation is further complicated by the growing number of materials used in dentistry (e.g., cements, adhesives, and filling materials) as well as food constituents previously not encountered. Amalgams have been frequently used in restorative dentistry for more than a century. Concerns regarding the health effects of the metallic components, in particular mercury, have been raised sometimes; however, no convincing evidence strongly discouraging the use of amalgams has emerged. The availability of other materials as possible replacements has not changed the situation much because these materials come with their own drawbacks. The American Dental Association Council has thus confirmed in 2009 that amalgams are a safe material, [2] and this opinion has been reaffirmed in 2016. [3] We have reported earlier about the corrosion stability of a selection of amalgams (both of high-copper dispersant [HCD] and high-copper single composition [HCSC] alloy type [4] ) in artificial saliva. [5,6] Linear scan voltammetry has been applied to some amalgams. [7] Phase formation and conversion in high-copper amalgams have been discussed elsewhere. [8] As corrosion of metallic materials in the mouth is almost exclusively an electrochemical process, [9] we have employed an electrochemical method in previously reported studies of the corrosion of metallic materials used in dentistry. [5,6,[10][11][12][13][14][15][16][17][18] Cyclic voltammetry (CV) or similar techniques with an electrolyte solution quite similar to natural saliva proved to be particularly helpful and relatively easy to perform. [19][20][21][22] Although in the case of the studied alloys, well-defined breakthrough potentials E B could be identified, with amalgams no such characteristic potentials could be observed. We have instead taken the current observed at a selected electrode potential and its change as a function of potential cycle number as a measure of corrosion stability.
Unfortunately, the experimental results and thus the comparability (and compatibility) between various studies depend strongly on the experimental conditions, in particular, the composition of the electrolyte solution, [20,23] the sample pretreatment, and the scan rate. The latter parameter has been examined in investigations of the corrosion behavior of alloys of semiprecious metals or with low precious metal content with CV with widely varying scan rates; the results have been reviewed elsewhere. [24] Low scan rate CV has been shown to be a method giving results strongly correlated with those of other, in particular in vivo, studies of corrosion behavior (see e.g., Weber et al. [25] ); however, the sample under investigation is away from the electrochemical equilibrium, that is, at the spontaneously established rest potential.
In the present study, we have used a synthetic saliva solution composed of various salts as suggested by Meyer. [26] A very similar solution has been described first by Swartz et al. [27] and modified later by Fusayama et al. [28] ; it is now generally known as Fusayama saliva. Its properties are close to those of natural saliva. Ringer's solution (0.8 wt% NaCl, 0.02 wt% KCl, 0.02 wt% CaCl 2 , 0.1 wt% NaHCO 3 in water), frequently used in corrosion studies, shows an artificially enhanced chloride corrosion (see e.g., Finkelstein and Greener [29] ). Nevertheless, the corrosion products formed in vivo and in Ringer's solution are very similar to those formed at dental amalgams [30] ; there do not exist similar comparability studies for other materials. Engel has reviewed differences in results and provided further arguments in favor of Fusayama saliva. [24] The low scan rate of 1 mV/s (60 mV/min) chosen in our earlier studies resulted in a long duration of experiments, leaving this approach somewhat unattractive. In addition, the exposure of the samples to electrode potentials possibly far away from the electrode potentials commonly encountered in daily use in vivo (i.e., the spontaneously established corrosion or open circuit potential) may result in data possibly not characteristic of the corrosion behavior under actual conditions. Inspired by a recently provided review of experimental methods in corrosion research, [31] we have reexamined selected samples of our previous studies assumed to be representative with electrochemical impedance measurements with the aim of relating corrosion susceptibilities derived from breakthrough potential E B (corresponding to thermodynamic system properties) as reported in previous studies with corrosion currents obtained from impedance measurements (corresponding to kinetic system properties). [32] Impedance measurements are well established in corrosion research. [31,[33][34][35] In comparison to slow scan voltammetry aimed at determining breakthrough potentials E B or to get Tafel plots, they have a major advantage: They are executed at the open circuit potential (i.e., the corrosion potential) with the system in electrochemical equilibrium.
Nevertheless, they have been employed in studies of amalgams used in dentistry only infrequently: In a comparative study of various silver-containing materials used or at least suggested for use in dentistry, an amalgam with a particularly high silver content (Dispersalloy®) impedance measurements was employed. [36] To interpret results showing basically an incomplete and only slightly deformed semicircle in the Nyquist plot, an equivalent circuit with a double layer, rather unusually represented by a simple capacitive element (no constant phase element), and an adsorption impedance connected in series to the charge-transfer resistance with a constant phase element instead of a simple capacitive element as commonly used in an adsorption impedance [37] was employed. The latter discrepancy was not explained. A polarization resistance with some unclear relation to impedance results is used to calculate porosity data; the penetration depth is compared with that obtained from a polarization resistance presumably taken from slow scan CV showing a difference of about an order of magnitude. Polarization resistance of the amalgam was the smallest; in a very rough approximation, this might suggest that the amalgam was most prone to corrosion. Dispersalloy was again examined in different electrolyte solutions: natural saliva, artificial saliva, and aqueous 0.9 wt% NaCl solution. [38] The solution of 0.9 wt% NaCl turned out to be most aggressive, quite in agreement with earlier studies that have confirmed excessive corrosion in Ringer's solution containing 0.8 wt% NaCl. [24,29] Although natural saliva contains significantly more chloride and sodium ions, [38] it is less aggressive, presumably because of the presence of many constituents possibly passivating the surface by forming insoluble surface layers. The same argument applies to Fusayama artificial saliva. Somewhat surprisingly, the authors conclude the existence of three time constants from the rather featureless impedance plots. The corresponding equivalent circuit is not shown; it may be guessed only from an entry resembling a circuit description code as suggested by, for example, Boukamp. [39,40] The authors conclude that in vitro studies of amalgam corrosion are insufficient to predict in vivo corrosion.
Acciari et al. [8,41] have studied the corrosion of three metal phases commonly found with modern high-copper amalgams with impedance and noise measurements. The evaluation of the impedance measurements was based on the same equivalent used in our previous study [32] (see below also). Results of impedance and noise measurements were in good agreement; no attempt to compare with results of CV (Tafel plots) has been made. The silver-copper eutectic found to be most prone to corrosion plays no role in modern amalgams. In a similar study of three phases including the practically obsolete γ 2 -phase in an aqueous solution of 0.9 wt% NaCl using only impedance, open circuit, and polarization measurements, the observed corrosion resistance was found to be associated with the formation of a corrosion layer. [41] In a very similar study, a rating of corrosion stability of the phases was suggested, with the γ 2 -phase being most corrosion prone. [42] Except for data pertaining to the Ag-Cu alloy, the same results have been reported elsewhere. [43] The corrosion behavior of Tytin-Plus® and Dispersalloy (the former alloy been lathe cut, the latter is a blend) has been compared. [44] A further study of Dispersalloy corrosion remained inconclusive. [45] Corrosion current densities and polarization resistance were rather similar for both materials; only the corrosion potential of Dispersalloy is slightly more positive, possibly indicating a better corrosion stability. On this comparison of two alloys, a further report is available. [46] Again a solution of NaCl (1 wt%) turned out to be more aggressive than a Fusayama artificial saliva solution. Somewhat disturbingly impedance measurements are mentioned in the Abstract and Section 2; however, there is no mention of possibly obtained results.
Impedance measurements were applied in a comparative study of two amalgams. The author claims to compare a low-copper amalgam (Amalcap®) and a high-copper amalgam (Dispersalloy) without providing the composition data of their studied materials. As shown in Table 1, the copper content according to manufacturers' data is the same. [47] An equivalent circuit was not provided; instead, parameters in a provided equation were fitted. It can be concluded that a Randles circuit including a diffusion impedance was assumed. A lower chargetransfer resistance and accordingly a higher corrosion current were found for Amalcap. The claim that a simple equivalent circuit cannot be expected in a multiphase material has been assigned apparently in error to previous authors. [48] It has been shown frequently that a simple Randles circuit in the case of studies limited to high frequency data focused on charge-transfer resistance determination can be used successfully.
Lemaitre et al. have employed electrochemical impedance measurements in corrosion studies related to amalgams. In conventional amalgams, the γ 2 phase corrodes particularly easily. According to the authors' statement, corrosion of this phase is indicated by a current peak around E SCE = 250 mV (CV not shown). [48,49] Impedance data (not shown) showed a depressed semicircle at high frequencies and a sloped line at lower frequencies.
Assuming applicability of a Randles-type equivalent circuit (fitting was not performed), charge-transfer resistance values showing a wide scattering were extracted. Capacitance values presumably associated with the existence of an electrochemical double layer were surprisingly small. With a slightly larger set of different amalgams, measurements at E SCE = 250 mV were performed; the results are slightly confusing. [50] The charge-transfer resistance R ct was found to be inversely proportional to the corrosion current density -which hardly surprises, given the simple and straightforward mathematical relationship. As the authors neither explain how they extract R ct from the impedance data nor show a conceivable other source of corrosion current densities (e.g., Tafel plots), the report remains murky. The influence of silver content in the alloy (about 60 wt% or <50 wt%) has been examined with impedance measurements. [51] In a comparative study of a small selection of amalgams, the inapplicability of the Randles circuit was concluded, and no further results related to corrosion of the amalgams were provided. [52] A short report on a study of two non-γ 2 amalgams did not provide any information related to corrosion susceptibility. [53] Brett et al. [54] have employed impedance measurements in a comparative study of the influence of various electrolyte solutions on the corrosion behavior of a single amalgam. Displayed equivalent circuits did not show a charge-transfer resistance or any other entity related to the rate of corrosion. The further discussion remains open. The findings in Brett et al. [54] have been republished again by the same authors elsewhere. [55] In a corrosion study of tin and tin amalgams, enhanced corrosion of the amalgamated tin was observed [56] using the same equivalent used in our previous study [32] and presented below.
We have reexamined selected amalgams previously examined with slow scan CV. [5,6,13] The relative corrosion susceptibility expressed as anodic dissolution current is shown for all samples studied previously in Figure 1.
The use of electrochemical impedance measurements in studies of palladium-base and nonprecious metal alloys has been inspected previously. [32] 2 | EXPERIMENTAL Taking into account the results of an earlier study (see Figure 1 above), we have selected representative amalgams including a conventional one with significantly lower copper content, with their composition as provided by the manufacturer listed in Table 1.
Samples of amalgam fillings were triturated and condensed according to manufacturers' instructions into hollow Plexiglas® cylinders. A surface area of 1 cm 2 of amalgam filling was exposed to the electrolyte solution. Before electrochemical measurements, the samples were cleaned ultrasonically in ultrapure 18 MΩ water. Subsequently, they were connected to a sample holder fitted with a tapered joint. A conventional electrochemical H-cell with a corresponding tapered joint and cell compartments for a platinum sheet counter electrode and a saturated calomel reference electrode separated from the main compartment by glass frits was used.
A Fusayama synthetic saliva containing 0.4 g KCl (Acros Organics), 0.4 g NaCl (VEB Laborchemie Apolda), 0.69 g NaH 2 PO 4 ·H 2 O, 0.002 g Na 2 S (Grüssing GmbH), 0.684 g CaCl 2 (Grüssing GmbH), and 1 g urea (Fluka) per dm 3 dissolved in 18 MΩ water was used as electrolyte solution; its pH was 4.7-5.00. All chemicals were of analytical purity, and they were used as received. The solution was saturated with pressurized air filtered with activated carbon instead F I G U R E 1 Corrosion current density at E SCE = 300 mV for tempered and polished samples, grouped according to the alloy type. [5,6] HCD, high-copper dispersant; HCSC, high-copper single composition  of nitrogen or argon before every measurement to keep the composition of the electrolyte solution as similar as possible to the in vivo situation. Electrochemical impedance measurements were performed on a potentiostat IVIUMSTAT Electrochemical Interface in the frequency range 5 kHz to 1 Hz, with 7 data points per frequency decade at a sine wave amplitude of 5 mV. Measurements were repeated five times per sample. They were repeated five times after repeated polishing. This handling was repeated again; in total, 15 data sets were obtained for every amalgam. For data handling, in particular for fitting of the obtained impedance data, Boukamp software (version 2.4) was used. A Randles-type circuit ( Figure 2) was assumed as has been employed in our previous study [32] with a constant phase element Q instead of a simple double-layer capacitance. All parameters of all elements have been fitted; only the values of R ct relevant for this comparative study are discussed below.

| RESULTS AND DISCUSSION
Impedance measurements were started soon after sample immersion after the potential fluctuations immediately after immersion had abated. To examine reproducibility, the following sequence was performed five times with every sample: abrasive treatment-cleaning-immersion -impedance measurement. Results are displayed in Figure 3. Figure 4 shows the obtained values of charge-transfer resistance R ct ; they correspond to the rate of corrosion with smaller resistances indicating faster corrosion. A direct transformation of R ct into j corr is basically possible according to: The obtained values of R ct do not vary significantly during repetitions. Although comparisons between various products rarely go beyond three materials in most studies discussed above, similarities of data are also found without significant differences in corrosion stability. During repeated measurements in the procedure described above, the results did not show a significant change. As any possibly formed passivation layer during a run was removed subsequently after polishing, this was expected. Data are also rather similar for the studied amalgams except for SDI Permite C. This amalgam appears to be most corrosion stable according to obtained impedance data. A comparison with the stability rating reported before, based on slow scan CV (see Figure 1 and References [5,6] ), does not confirm the overall lower stability of HCD amalgams with these new impedance data. The amalgam appearing most stable according to impedance measurements showed only average stability in our previous studies. In the absence of further studies, as briefly reviewed above, and of evidence from other methods, a definitive identification of a most stable product cannot be provided here. As indicated above, impedance measurements performed in vitro under spontaneously established electrochemical conditions without forcing possibly unrealistically high electrode potentials as in slow scan CV on the electrode may provide more appropriate corrosion susceptibility information.
In a comparative study, Marek noticed similar corrosion rates with chronocoulometry with an externally applied electrode potential for eight high-copper amalgams, with only one product showing a significantly higher rate. [57] 4 | CONCLUSIONS Electrochemical impedance measurements performed with selected amalgams have yielded corrosion resistance values rather invariable during repeated measurements. They are rather similar except for one product; the previously observed lower corrosion stability of HCD amalgams could not be confirmed.