Dielectric properties of food (dielectric constant, ε', and loss factor, ε'') determine how electromagnetic energy is absorbed and dissipated, and are influenced by frequency, temperature, moisture, composition and structure. Their characterization is crucial for microwave/radio-frequency drying, cooking, thawing and pasteurization equipment design, and for online quality control.
Unlike most previous studies, which report single-frequency data, this work introduces a low-cost, non-destructive, broadband (100 MHz-4 GHz) measurement system, based on a Keysight FieldFox vector network analyzer and an open-ended coaxial probe with custom MATLAB calibration, validated against nine independent literature sources (2-6% relative error for water, apple cortex/puree and sucrose solutions; 10-14% for potato).
Sucrose solutions (0–60% w/w) served both as dielectric reference standards for calibration/validation and as a controlled water-sugar matrix, since water and soluble sugars dominate the dielectric response of fruit tissue at these frequencies. Potato, banana and apple (slices and puree) were then measured, as a first approximation, to evaluate how closely these foods behave as a simple water-sugar mixture, over the analyzed frequencies.
For the sucrose–water system, second-order polynomial models were developed by multiple regression to describe the dependence of dielectric constant and loss factor on frequency and sucrose concentration. The resulting correlations showed a very good fit to the experimental data, with R2 = 0.9923 and 0.9290 for ε' and ε'', respectively; concentration affected the dielectric constant more than frequency, since sucrose replaces polar water molecules with larger, less polar sugar molecules that also restrict remaining water's mobility.
The resulting sucrose-based correlations were further used as a first approximation to estimate the dielectric properties of the fruit and potato matrices using representative total soluble solids (°Brix) values reported in the literature. This water-sugar model reasonably reproduced homogeneous, low-ionic samples (e.g., apple purée), but underestimated banana and potato loss factor and overestimated intact-tissue dielectric constant (apple cortex), reflecting their starch/ionic content and structural/moisture heterogeneity. This first approximation provides a basis for further models incorporating additional compositional variables that should provide important information about cultivar, ripeness and tissue structure.