Joseph F. Shepard

The wafer flexure technique for the determination of the transverse piezoelectric coefficient (d31) of PZT thin films

Abstract This paper describes a simple and inexpensive method for evaluating the transverse piezoelectric coefficient ( d 31 ) of piezoelectric thin films. The technique is based upon the flexure of a coated substrate which imparts an ac two-dimensional stress to the piezoelectric film. The surface charge generated via the mechanical loading is converted to a voltage by an active integrator. Plate theory and elastic stress analyses are used to calculate the principal stresses applied to the film. The d 31 coefficient can then be determined from knowledge of the electric charge produced and the calculated mechanical stress. For 52/48 sol-gel lead zirconate titanate (PZT) thin films, the d 31 coefficient was found to range from − 5 to − 59 pC/N and is dependent on poling field.

Size Effects and Domains in Ferroelectric Thin Film Actuators

Ferroelectric thin films typically differ from bulk ceramics in terms of both the average grain size and the degree of stress imposed on the film by the substrate. Studies on bulk ceramics have demonstrated that the number of domain variants within grains depends on the grain size for sizes 3 films are ferroelectric at 77K down to thicknesses as low as ˜ 60nm. Data on the low and high field electrical properties are reported as a function of temperature, the film crystallinity, and film thickness for representative perovskite films.

The Impact of Domains on the Dielectric and Electromechanical Properties of Ferroelectric Thin Films

In ferroelectric thin films for capacitive and piezoelectric applications, it is important to understand which mechanisms contribute to the observed dielectric constant and piezoelectricity. In soft PZT (PbZr 1−x Ti x O 3 ) ceramics, over half the room temperature response is associated with domain wall contributions to the properties. However, recent studies on bulk ceramics have demonstrated that the number of domain variants within grains, and the mobility of the twin walls depend on the grain size. This leads to a degradation in the dielectric and piezoelectric properties for grain sizes below a micron. This has significant consequences for thin films since a lateral grain size of 1 μm is often the upper limit on the observed grain size. In addition, since the pertinent domain walls are ferroelastic, the stress imposed on the film by the substrate could also clamp the piezoelectric response. To investigate these factors, controlled stress levels were imposed on PZT films of different thickness while the dielectric and electromechanical properties were measured. It was found that for undoped sol-gel PZT 40/60, 52/48, and 60/40 thin films under a micron in thickness, the extrinsic contributions to the dielectric and electromechanical properties make very modest contributions to the film response. No significant enhancement in the properties was observed even when the film was brought through the zero global stress condition. Comparable results were obtained from laser ablated films grown from hard and soft PZT targets. Finally, little twin wall mobility was observed in AFM experiments. The consequences of this in terms of the achievable properties in PZT films will be presented. Work on circumventing these limitations via utilization of antiferroelectric phase switching films and relaxor ferroelectric single crystal films will also be discussed.

High-resolution dry etch patterning of PZT for piezoelectric MEMS devices

Dry etch patterning of lead zirconate titanate with self-aligned top and bottom platinum electrodes was demonstrated using a combination of reactive ion etching of PZT and argon ion milling of Pt. Monochlorotetrafluoroethane (HC/sub 2/ClF/sub 4/) etch gas and a Pt etch mask were employed for PZT patterning. PZT etch rates in the range 13-110 nm/min were measured as a function of rf discharge power for alumina, graphite, and ardel electrode shields in a conventional parallel plate reactor. The top and bottom platinum films were patterned by argon ion milling with a photoresist etch mask which was applied at the outset of the process and left in place throughout patterning. Etched profiles of Pt(200 nm)/PZT(500 nm)/Pt(150 nm) multilayer films were characterized by SEM and exhibited submicron definition.

Measurement of Effective Longitudinal Piezoelectric Coefficient of thin Films by Direct Piezoelectric Effect

This paper presents a new method for the measurement of the longitudinal piezoelectric coefficient of piezoelectric thin films using the direct piezoelectric effect. A uniform uniaxial stress was applied to the piezoelectric thin film by high-pressure gas and the induced charge was collected and measured by a charge integrator. The effective longitudinal piezoelectric coefficient of lead zirconate titanate (PZT) 52/48 thin films made by sol-gel processing was measured by this method. Undoped films typically have d 33 values of ∼ 5 pC/N, while poled films have values up to 220 pC/N.

Dry Etching of Sol-Gel Pzt

Two techniques for dry etching of sol-gel lead zirconate titanate (PZT 52/48) thin films were investigated: reactive ion etching and argon ion milling. Etched profiles were characterized by scanning electron microscopy. For reactive ion etching, a parallel plate etcher was used with HC 2 ClF 4 , an environmentally safe etch gas, in a process described by other researchers. Etch rates were measured and compared as a function of electrode shield material (ardel, graphite, alumina) and RF input power (100 to 500 W). These etch rates varied from 10 to 100 nm/min. Reactive ion etched sidewall angles 12° off normal were consistently produced over a wide range of RF powers and etch times, but overetching was required to produce a clean sidewall. For argon ion milling, a 300 mA/500 V beam 40° off normal to the substrate operating in a 72 mPa argon pressure was used. These ion milling conditions produced an etch rate of 250 nm/min with a sidewall slope angle of about 70°. The ion milling etch rate for sol-gel PZT was significantly faster than rates reported for bulk PZT. The 500 nm thick PZT films used in this study were prepared by the sol-gel process that used methoxyethanol solvent, spin coating on t/Ti/SiO 2 silicon substrates, and rapid thermal annealing for 30 s at 650 °C for crystallization of the perovskite phase.

Recent advances in piezoelectric materials

Recent developments in piezoelectric materials include submicron grain size ceramics and single crystals. Pb(Zr,Ti)O3 (PZT) ceramics with submicron grain sizes (approximately 0.5 micrometer) have been produced with properties comparable to conventional, coarse grained (approximately 3 to 5 micrometer) ceramics. The fine grain ceramics exhibit improved machinability over conventional materials. Ultrasonic transducer arrays with post widths less than 15 micrometers have been fabricated as well as thin plates with thicknesses as low as 10 micrometers. The yields and performance of such operations are expected to be much greater with fine grain ceramic than with conventional material. The single crystal piezoelectrics developed offer field induced strain an order of magnitude higher than what can be achieved in piezoelectric ceramic actuators (greater than 1%). Furthermore, the strain electric field hysteresis and dielectric losses are very low for these materials and electromechanical coupling factors (k33) are greater than 90%. Applications which may benefit from the recent developments include smart materials and structures and MEMS.

A Technique for the Measurement of d 31 Coefficient of Piezoelectric Thin Films

This paper describes a new technique by which the d 31 coefficient of piezoelectric thin films can be characterized. Silicon substrates coated with lead-zirconate titanate (PZT) are flexed while clamped in a uniform load rig. When stressed, the PZT film produces an electric charge which is monitored together with the change in applied load. The mechanical stress and thus the transverse piezoelectric coefficient can then be calculated. Experiments were conducted as a function of poling field strength and poling time. Results are dependent upon the value of applied stress, which itself is dependent upon the mechanical properties of the silicon substrate. Because the substrate is anisotropie, limiting d 31 values were calculated. In general, d 31 was found to be ∼20 pC/N for field strengths above 130 kV/cm and poling times of less than 1 minute, d 31 was increased more than a factor of three, to ∼77 pC/N, when poled at 200 kV/cm for ∼21 hours.