The effective large signal longitudinal piezoelectric coefficient (d33,f∗) of piezoelectric thin films on rigid substrates has been widely investigated. Unclamped piezoelectric thin films are predicted to have a higher d33,f∗ coefficient due to reduced constraints on piezoelectric strain, domain reorientation, and domain wall motion, but quantitative measurements of this coefficient have been limited. This study uses microfabrication techniques along with double-beam laser interferometry (DBLI) to accurately determine the longitudinal piezoelectric coefficient of Pb(Zr,Ti)O3 thin films in partially released piezomicroelectromechanical structures. A two-step backside release process was used: first, deep reactive ion etching to create backside vias and second, wet etching of the ZnO sacrificial layer to release the area beneath the Pb(Zr,Ti)O3 thin films. Post wet etching, optical profilometry showed concavely deformed diaphragms resulting from asymmetrical stress profiles through the diaphragm thickness. DBLI was then used to examine diaphragm deflection under an applied unipolar voltage ranging from 0 to 10 V. Devices with 50% and 75% of the area beneath the top electrode released exhibited large signal d33,f∗ values of 410 ± 6 and 420 ± 8 pm/V, respectively, more than three times higher than the d33,f∗ value of the clamped samples: 126 ± 13 pm/V. The reasons contributing to the large d33,f∗ include (i) the change in stress levels due to the release process, (ii) the elimination of mechanical constraints from substrate clamping, and (iii) enhanced domain reorientation. These findings confirm that substrate declamping significantly boosts the piezoelectric coefficient, bringing d33,f∗ closer to the bulk longitudinal piezoelectric coefficient (d33).
Read full abstract