Supplementary Information Computational details Structural model

FE structural models of a SAW device based on LGS substrate, with dimensions 1600 μm length x 1600 μm depth x 200 μm height, were developed. Figure 1a in the main manuscript shows a schematic diagram of the simulated SAW device with orthogonal IDT configuration integrated with delay path modification, in the form of micro-cavities. IDT finger pairs with periodicity 40 μm, were defined on the surface for each port along (0, 22, 0) and (0, 22, 90) Euler directions. The piezoelectric domain was meshed using coupled field tetragonal solid elements with four degrees of freedom accounting for the three translations and voltage. High mesh densities were ensured near the surface. Impulse response analysis was performed during 190 ns by applying an impulse voltage of 100 V at the input IDT and employing a time step of 0.95 ns^1. Subsequently, an ac analysis was carried out using a peak voltage of 2.5 V with the device frequency obtained from impulse response analysis for each of the propagation directions.

The fluid region, with 50 μm height, was modeled as an infinite reservoir of Newtonian, incompressible fluid on top of the piezoelectric domain, subject to stress free boundary condition at the free surface (i.e. P =0). The piezoelectric domain was meshed using tetragonal solid elements with four DOF to account for the three translations and the voltage. The fluid domain was meshed using eight node fluid elements and discretized employing an Arbitrary Lagrangian Eulerian (ALE) frame for the kinematical description. The FSI model contained millions of nodes and elements: 2, 217, 889 nodes and 2, 085, 525 elements. Numerical solutions were obtained by sequential solution of the generalized Navier-Stokes equation for the fluid domain and acoustoelectric equations for the piezoelectric motion ^2, employing the iterative sequential coupling algorithm^3. In this algorithm, the governing equations for the two domains are solved separately and the solver iteration between the two domains continues till convergence of the load transferred across the interface is achieved.

To obtain the best micro-cavity configuration for sensing, 20 micro-cavities of λ/2 square cross sections with varying depths were etched on the device surface along the delay path with IDTs along the (0, 22, 90) direction. These micro-cavities were filled with a waveguide material polystyrene (PS) to provide increased entrapment of acoustic energy near the device surface^4. Based on our simulated node translation data, we find that Love waves incident on a shallow groove or cavity on the surface of a piezoelectric substrate filled with another elastically isotropic material such as polystyrene are weakly scattered with a part of the energy contained in the reflected and the transmitted Love waves and the other part converted into the bulk shear waves that propagate into the substrate. Our simulations indicate that PS filled micro-cavities with dimensions λ/2 x λ/2 x λ/2 exhibit the highest energy transmission and smallest insertion loss (lesser by 2 dB compared to the next best PS filled λ/2 x λ/2 x λ/8 micro-cavities) (Fig. 1S) Our analyses of displacement profiles for each of the simulated micro-cavity designs indicate that these improvements are brought about by a larger coherent reflection of the incident Love wave and subsequent reduced conversions into bulk shear modes which radiate into the substrate (Fig. 1b). Details of the pass-bands are shown in Fig. IS where a 21.8 dB increase in energy transmission is clearly shown for the best case micro-cavity (λ/2 x λ/2 x λ/2) over the SAW sensor with unidirectional IDTs along the (0, 22, 90) Euler direction and no delay path modifications (standard SAW). Therefore, 20 PS-filled micro-cavities with dimensions λ/2 x λ/2 x λ/2 were incorporated in the delay path of a SAW device with orthogonally-oriented, mutually-interacting IDTs along the (0, 22, 0) and (0, 22, 90) Euler directions.

Bio-fouling elimination in the simulated SAW devices due to mixed mode waves propagating along (0, 22, 0) Euler direction is dependent on the magnitude of the induced acoustic streaming forces computed as^{5, 6}:

Here,  and k_i refer to the attenuation constant and leaky SAW wave number, respectively, and are obtained using a perturbation approach^{7, 8},  refers to the angular frequency and A refers to the acoustic wave displacement, which are obtained using FE simulations.

The device sensitivity was computed using the perturbation theory utilizing the mass sensitivity equation applicable to all types of surface acoustic devices independent of the type of acoustic modes ^9:

(2)

To calculate the mass sensitivity, the values of mode center frequency f₀, angular frequency , average area density of wave energy U, and displacements u_x, u_y, and u_z were derived from the finite element simulations.