Thin films (thickness of ∼2 to 4 μm) comprised of SiO2 particles (300–400 nm) were deposited on glass slides using a screen-printing approach. After annealing, films were functionalized with polyethyleneimine (PEI) to facilitate the adsorption of 3 to 5 nm Au nanoparticles (Au NPs) on the SiO2 particle films, and the resulting Au@SiO2 substrates were used to evaluate their potential for Surface-Enhanced Raman Spectroscopy (SERS) applications. Two types of supported silica particles, monodisperse solid silica (s-SiO2) and textured silica (t-SiO2), were used to investigate the impact of the SiO2 surface properties on the formation of Au NPs hotspots. Monodisperse s-SiO2 particles were synthesized using the Stöber method, while textured surface materials were obtained by incorporating cetrimonium bromide (CTAB) into the solution at a concentration below the critical micelle concentration.SERS studies were conducted using Rhodamine 6G (R6G) as a target molecule. For Au@s-SiO2, the SERS signal results showed a reasonable spatial reproducibility with relative standard deviation values (RSD ∼5 to 16% for studies on six different samples), in agreement with literature values (RSD ∼10 %). In the case of t-SiO2, the spatial reproducibility was also good in some slides with low% RSD between 7 and 12 %, but in other substrates with values as high as 20 % to 50 % RSD. Also, a lower slide-to-slide %RSD value, 26 % vs. 34 %, was found for the s-SiO2 sample.For the Au@t-SiO2 substrates, the estimated calibration sensitivity and the calculated limit of quantification (LoQ) for R6G were 0.392 (log (peak area) vs. log ([R6G]) and 0.381 μM, respectively, close to the visual practical limit. For Au@s-SiO2, the sensitivity was lower, 0.305 vs 0.392, and the LoQ was 0.016 μM, which is lower than the visual practical limit 0.1 μM. To put these results in context, SiO2 and Au@SiO2 films were characterized using different methods, including profilometry, high resolution scanning and transmission electron microscopy (HR-SEM and TEM), UV–visible and Raman spectroscopy.In addition, finite-difference time domain (FDTD) simulations were performed to understand the relationship between substrate morphology and SERS enhancement. Image processing and 3D modelling software were used to convert the HR-SEM images of the SERS substrates into FDTD simulation-compatible models. The simulations showed that the Au@s-SiO2 substrate had a greater average plasmon resonance enhancement over the Au@t-SiO2 substrate, in good agreement with the LoQ differences between substrates, ∼0.02 μM for Au@ s-SiO2 vs. ∼0.4 μM for Au@t-SiO2.