Controlled graphene oxide assembly on silver nanocube monolayers for SERS detection: dependence on nanocube packing procedure

Hybrid graphene oxide/silver nanocubes (GO/AgNCs) arrays for surface-enhanced Raman spectroscopy (SERS) applications were prepared by means of two procedures differing for the method used in the assembly of the silver nanocubes onto the surface: Langmuir–Blodgett (LB) transfer and direct sequential physisorption of silver nanocubes (AgNCs). Adsorption of graphene oxide (GO) flakes on the AgNC assemblies obtained with both procedures was monitored by quartz crystal microbalance (QCM) technique as a function of GO bulk concentration. The experiment provided values of the adsorbed GO mass on the AgNC array and the GO saturation limit as well as the thickness and the viscoelastic properties of the GO film. Atomic force microscopy (AFM) measurements of the resulting samples revealed that a similar surface coverage was achieved with both procedures but with a different distribution of silver nanoparticles. In the GO covered LB film, the AgNC distribution is characterized by densely packed regions alternating with empty surface areas. On the other hand, AgNCs are more homogeneously dispersed over the entire sensor surface when the nanocubes spontaneously adsorb from solution. In this case, the assembly results in less-packed silver nanostructures with higher inter-cube distance. For the two assembled substrates, AFM of silver nanocubes layers fully covered with GO revealed the presence of a homogeneous, flexible and smooth GO sheet folding over the silver nanocubes and extending onto the bare surface. Preliminary SERS experiments on adenine showed a higher SERS enhancement factor for GO on Langmuir–Blodgett films of AgNCs with respect to bare AgNC systems. Conversely, poor SERS enhancement for adenine resulted for GO-covered AgNCs obtained by spontaneous adsorption. This indicated that the assembly and packing of AgNCs obtained in this way, although more homogeneous over the substrate surface, is not as effective for SERS analysis.