The validation of TDA-AAS was performed via evaluations of the limit of detection (LOD), the limit of quantification (LOQ), trueness, and precision. The LOD of TDA-AAS was calculated from the average background signal obtained from 10 parallel analyses of 100 µL ultrapure water and raised by three times the standard deviation (SD). The LOQ of TDA-AAS was calculated from the average background signal obtained from 10 parallel analyses of 100 µL ultrapure water and raised by ten times the SD. Trueness and precision were evaluated by analyses of ERM-BB422, a certified reference material. The long-term stability of the TDA-AAS performance was evaluated through analysing the certified reference material ERM-CC580 every day for the duration of the experimental period leading up to sample analysis and a control chart was generated from these measurements. Additionally, TDA-AAS performance was further validated through participation in an interlaboratory trial (PT/CHA/2/2018), in which samples of wastewater were analysed.
The validation of DGT was performed both using a model solution and a diluted fish sauce solution. The linearity of Hg accumulation in the binding gel over time, the value of the effective diffusive coefficient (De), the thickness of the diffusive boundary layer (DBL, δ), and the performance of DGT in solutions with different pH and NaCl concentrations were all evaluated, along with the LOD, LOQ, trueness, and precision. LOD was calculated from the average mass of Hg in binding gel blanks raised by three times the SD and by taking into account the determined effective diffusion coefficient, the standard parameters of the DGT unit, and the exposure time of 24 h, according to Equation (1), where M = mass of analyte; Δg = diffusive layer thickness; De = effective diffusion coefficient of analyte in the diffusive gel; t = deployment time; A = exposure area.
The LOQ was calculated in a similar way using the average mass of Hg in the binding gel blanks raised by ten times the SD. The trueness of DGT was evaluated using the recovery for a 5-fold diluted fish sauce sample spiked with a known amount of Hg. The precision of DGT was assessed by calculating the relative standard deviation (RSD) of DGT concentrations calculated from Equation (1) in the analysis of 6 parallel samples of binding gel peeled off from DGT pistons at the end of the deployment time in 5-fold diluted fish sauce.
The effective diffusion coefficient, De, was calculated after performing tests on the linear accumulation of Hg during the exposure time. Ten DGT pistons were immersed in the model solution and two units were removed from the solution after 2, 4, 6, 8, and 24 h. The DGT pistons were then dismantled and the binding gel was analysed using TDA-AAS. By plotting the dependence of M.c−1 on time, the slope k of linear regression was obtained and De was calculated using Equation (2) [16], where ts is the conversion factor of hours to seconds.
This procedure was also used to calculate the De of Hg in diluted fish sauce. The De can be affected by the uptake efficiency of the analyte on the binding gel; therefore, it is essential to verify uptake efficiency, especially in complex matrices. In this work, we adopted our procedure for determining the uptake efficiency of Hg on the binding gel from the work of Abdulbur-Alfakhoury et al. [24].
The thickness of the DBL was calculated from the simultaneous deployment of eight DGT pistons with different thicknesses of diffusive layer (diffusive gel and membrane filters 0.039, 0.064, 0.089, and 0.114 cm, tested in duplicate) in the model solution. After plotting M−1 against Δg, the DBL was calculated from the slope k and intercept q obtained from the regression equation of dependence of M−1 to Δg [25]. The diffusion coefficient of Hg in water, Dw, was taken from the work of Docekalova and Divis [16].
All experiments were performed at the laboratory temperature of 20 °C and were repeated three times.
The effect of pH of the model solution of DGT performance was evaluated by immersing four pistons into the model solution with a pH of 3 for 4 h. This test was then repeated in model solutions adjusted to pH 4, 5, and 6. The concentration of Hg in the model solution was monitored for the duration of the test by analysis of TDA-AAS and recorded as cSOL. This concentration was then compared with the concentration determined by DGT (cDGT) and recovery R was calculated as R = (cDGT/cSOL) × 100. To evaluate the effects of NaCl concentration of the solution on DGT performance, DGT was used on model solutions with NaCl concentration, ranging from 2–50 g·L−1, corresponding to the salt concentration in five-fold diluted fish sauce. The test arrangement was similar to that used in the testing of pH influence on DGT performance, and the R value was calculated again.
2.4. Analysis of Fish Sauce Samples
Samples of fish sauce were purchased at Asian markets in Brno, Czech Republic, and in Brussels, Belgium. Fish sauces were diluted with ultrapure water to a volume ratio of 1:4 and stirred for a total of 1 h in a 2 L glass beaker. Subsequently, four DGT pistons were inserted into the solution for a total time of 24 h. After the exposition, DGT units were dismantled, and the binding gels were directly analysed using TDA-AAS. The DGT concentration was calculated using Equation (1). The concentration of Hg in all undiluted samples was measured directly at the same time in triplicate by TDA-AAS. Concentrations were obtained in µg·L−1 (from both methods) and recalculated to the average concentrations (µg·L−1) and then recalculated to mass concentrations in mg·kg−1 using the average fish sauce density of 1.2 g·mL−1 [1]. All experimental data were statistically processed using XLstat software (Addinsoft, New York, NY, USA).