Molecularly imprinted polymers (MIP) represent a novel area of biomaterials that mimic the behaviour of natural antibodies exhibiting far greater stability than their natural counterparts. Even though their powerful interest has no more to be demonstrated for biological applications such as immunoassays or biosensors1, the difficulty for commercial development of MIP may be due to their tough compatibility with standard immunoassay formats. This is all the more evident when considering the necessity of labelling the template molecules for biodetection purposes (using fluorescent or radioactive labels), thus decreasing the experiments ease. Coupling silicon-based microfabricated structures with MIPs used as sensitive layer could overcome the main drawbacks of the latter ones by taking advantage of the high sensitivity and high resolution of the former ones.2 In this study, a MIP-based label-free immunoassay for the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) is validated using piezoelectric circular micro-membranes working in dynamic mode. 4x4 matrices of piezoelectric membranes have been fabricated by standard micromachining techniques (figure 1(a)). The membranes are circular shaped with a total radius R2 equal to 100 µm or 150µm. Each membrane can be individually actuated by a piezoelectric PbZrxTi1-xO3 (PZT) thin film with its two platinum electrodes. This active part is circular shaped with a radius R1 designed so that three ratios R1/R2 co-exist in the same matrix (for each R2), respectively equal to 0.3, 0.5 and 0.7 (figure 1(b)). The piezoelectric layer allows simultaneous actuation and detection of the resonant frequency (and corresponding quality factor) of the membranes using a HP4294A impedance analyzer. First, after the assembly step (figure 2), MIP with template molecule 2,4-D and corresponding negative imprinted polymer (NIP) for control purpose were deposited on the membranes using a specific cantilever array-based microspotting technique3. The second step consisted in the polymerization of the MIP under UV light (365 nm) and nitrogen stream. For the first time to the authors' knowledge, real time MIP polymerization has been monitored by following the membrane's resonant frequency shift thus determining a minimum time for a complete polymerization and allowing the study of viscoelastic properties of the MIP during this phase (figure 3). Indeed, two main parameters induce variations of the resonant frequency: the effective stiffness keff traducing changes in the stress of the stacked layers (and thus, viscoelastic properties of the added layer) and the equivalent mass Meff due to mass variation on the surface (f0 = (keff/Meff)0.5). In our case, MIP was deposited on the clamped part of the membrane, allowing a much higher sensitivity for stiffness changes than for mass changes. Thus, the increase of f0 during polymerization traduces a higher rigidity of the MIP due to its progressive cross-linking on the membrane's surface. Following to the polymerization, dip-and-dry experiments were performed in order to validate the MIP's functionality. A first washing step of the template resulted in a high decrease of the resonant frequency and the quality factor traducing a decrease of MIP's stiffness (figure 4). This is mainly confirmed by the quality factor evolution as its decrease is relevant of the MIP softening (figure 5). These results demonstrated the extraction of the template molecule 2,4-D. On the contrary, after incubation, frequency and quality factor levels are close to the initial ones showing that the MIP becomes a “rigid” film again as its pores are progressively occupied by the template molecules. Relevance of the obtained results is confirmed by the low frequency and Q-factor shifts obtained on the NIP control. Works are currently under progress on the precise determination of the sensor performances and the extension of the method to other MIP-template molecule couples.