CeraNova
Corporation
Active Composites for Smart Structures
New: Transparent Ceramics
| Telephone: | 508 460-0300 |
| E-mail: | Info@CeraNova.com |
|
| Telephone: | 508 460-0300 |
| E-mail: | Info@CeraNova.com |
|
Active PZT Fibers, a Commercial Production Process 3. THE CERANOVA PROCESS CeraNova has developed a proprietary extrusion and firing method to make round, straight and contamination free PZT fibers having composition and piezoelectric performance suitable for AFC use. This is a practical process based on conventional ceramic technology that is capable of making large fiber quantities at low cost. The basic process steps are outlined in Table I. The improvements outlined below in each case increase productivity and improve product quality.
3.1
Raw materials and mixing PZT-5A (Morgan Matroc, Bedford, OH) powder is mixed under high-shear conditions with a proprietary binder formulation until a homogeneous blend is achieved. The PZT particle size is submicron. Binders are added sequentially during the mixing process. As only 100 kilograms of mix are required to produce sufficient fiber for 20,000 AFC packs per year, mixing capacity will not constrain future AFCC production requirements. Batch blending and remixing improves mix consistency and extrusion performance. Great care is taken to avoid contamination as this can result in extruder die blockage or unacceptable defects in fired fibers. 3.2
Extrusion A screw-feed extruder is used to extrude green fiber through a carbide die of 160 micrometer diameter circular opening. Multiple dies as well as a range of orifice sizes and cross-sections have been used. Rectangular and tubular products are easily produced in addition to the circular monofilament currently used in active fiber composites. Mix composition / rheology, homogeneity and cleanliness are key to a successful fiber extrusion process. Slow system time constants, for screw rotation rate and temperature inputs, limit the ability to control fiber diameter in real time. Consistent operating procedures have been established to achieve good fiber diameter control. The typically observed fiber diameter range is ±6%, i.e., 150mm ±10mm. Real-time fiber diameter control is possible via a subsequent fiber draw down process step . Table I: Active Fiber Processing Steps
Screw RPM or piston linear displacement for piston extruders is the primary determinant of extrusion pressure as such systems are capable of delivering material to the die at a rate much faster than it can pass through the orifice. A careful material feed / extrusion rate balance is required for stable long-term operation. Kilometers of continuous monofilament fiber are extruded per day. The extrusion process will not limit future active fiber composite production volumes. The mixes in standard production at CeraNova contain 40 to 60 volume percent PZT-5A ceramic powder in a proprietary binder formulation system. Binder systems are critical to process practicability and are carefully guarded trade secrets. Many aqueous and non-aqueous systems are used in the ceramics industry. The mix formulation used by CeraNova is a practical compromise between high solids loading and mix rheology. It meets the requirements of subsequent binder burnout and sintering process steps. In the formulation of any ceramic mix, the aim is to maximize solids loading while maintaining adequate mix flow and plastic behavior. The former is necessary for high sintered density and defect free fiber while the latter is important for extrudability. An example of defect free green fiber is shown in Figure 4.
Figure
4: Fiber cross section after binder burnout.
Porosity is not visible. Grain
size is submicron. Fiber diameter
is 150mm.
Mix flow is critically dependent on binder composition and ceramic volume fraction. Any inhomogeneity in either, through variations in binder formulation or solids loading, has a significant impact on extrusion rate and its variability over time. The dimensional scale of significant mix inhomogeneity is determined by the extrusion die orifice size. To illustrate the importance of mix homogeneity, recently implemented mixing improvements resulted in a 3x increase in average extrusion rate with corresponding reduction in the frequency of die blockages. Kilometers of fiber can be extruded over many days before a die blockage occurs. The use of a low fiber drawing force assures a relative insensitivity of extruded fiber diameter to any variation in extrusion rate.
(a)
(b) Figure
5: SEM micrographs of sintered fiber surfaces: a) large metallic inclusion,
“crater”; b) metallic detritus (fiber diameters »130mm) Homogeneous contamination free mixes and fiber extrudate currently achieve consistent high quality mechanical and electrical properties in fired active fiber product. Metal particulate and abrasive wear contamination observed early in this program, as illustrated in Figure 5, must be scrupulously avoided. Contaminants come from raw materials as well as from mixing and extrusion equipment. Particularly troubling metallic contaminants were introduced via metal brushes and metal fiber cleaning pads used to clean the extruder between mix composition changes. A mix granulator that experienced significant wear was another source of metal contamination. It introduced significant wear detritus. Large metal particles resulted in die blockage while those that passed through the die resulted in defects having a volcanic crater-like appearance as shown in Figure 5a. The consequence of the finer abrasive wear metallic detritus is shown in Figure 5b. Such large scale defects are easy to identify via SEM microscopy and in situ x-ray analysis. With current practice, they occur infrequently, e.g., once in 10,000 fibers. The very fine wear detritus from metal extruder parts, caused by the abrasive nature of the binder / PZT ceramic powder mix, is more difficult to detect. The localized discoloration of fired fibers is thought to be, at least in part, related to such contamination. Care has been taken to avoid such wear and its corresponding mix contamination. This is particularly important for the extruder screw, barrel and exit die. Hardened metal components are used in all mixing and extrusion equipment to minimize such effects. Proper mix rheology is important for stable process control and achieving fiber dimensional tolerances. Increasing mix temperature reduces its viscosity and increases its extensibility. Mix temperature / rheology and drawing force are important determinants of fiber diameter. The time lag to change extrusion die and extruder barrel temperature make it an impractical means for real time control of fiber diameter to tight tolerance levels. Diameter can be controlled at faster rates by controlling fiber drawing force. Such control at the extrusion die is expected to be unnecessarily complex. An off-line draw down process step is being investigated as an optional control technique. It is thought to be a more practical approach to both diameter control and the need for increased extrusion rates. The increase in linear extrusion rate due to draw down is in proportion to the ratio of the squares of corresponding fiber diameters. Higher drawing forces result in more elongation, i.e., cross-sectional area reduction, as a fiber is extruded. Nominally 160 micrometer fiber has been drawn to under 100 micrometers, increasing the linear extrusion rate by a factor of 2.56. Fiber draw down is practiced at very high speed in the monofilament fibers industry. This is a well understood technology that can be used to increase extrusion rate and / or control fiber diameter. Extrusion rate increases of 10x or more with diameter control in the ± 1 micrometer range are thought possible. Integration of draw down steps into commercial scale fiber processes producing 100 km of fiber per day should be straightforward. Any high speed process will require real time closed loop process control of fiber diameter. Laser micrometers capable of measuring fiber diameters in the 80 to 160 micrometer range are used in the fiber optics industry. Such instruments are unnecessarily expensive for current km / day active fiber production rates. Fraunhofer diffraction methods are a known alternative that is being explored (Pike3 (1998), Cielo4 (1988)). They offer the required speed and precision, 100 mm ±0.25 mm , with low cost for near term needs and real time control potential for future high speed process conditions. 3.3
Fiber handling Extruded green fiber is spooled in continuous lengths of one kilometer or more. Currently, fiber is cut to length while green. Placing individual fibers onto kiln furniture for subsequent firing is costly as it requires excessive manual labor. Techniques have been developed to handle up to 20 fibers simultaneously. This technology is more than adequate to produce the 12,000 fibers needed per week by the Consortium to produce 2,000 AFC packs per year. Mandrel unit handling will increase CeraNova’s preform production capacity to >10x above this requirement. Winding continuous green fiber directly onto handling / firing mandrels at the extruder has been demonstrated. Mandrel units containing up to 1000 fiber lengths are practical. For all intents and purposes, mandrel unit handling obviates the need for automated and complex individual fiber or small fiber group handling schemes. The challenge lies in firing the 1,500 m2 to 15,000 m2 of mandrel units needed for pre-commercial and commercial production, not in their fabrication and handling. 3.4
Firing / Sintering The primary technical challenge to produce quality piezoceramic PZT products is maintaining compositional stoichiometry, i.e., controlling lead loss during sintering. PZT and other lead containing ferroelectric materials are prone to lead loss. This is due to the high vapor pressure of lead oxide above its melting point, 880C, and the reactivity of its vapor species. Commercial firing techniques generally use sealed saggers built of well seasoned ceramic components into which additional sources of lead oxide are added to counter its volatilization from product parts. SEM micrographs of fired active fibers are shown in Figures 6a (cross section) and 6b (surface). Note the fired grain size is on the order of several micrometers. A fiber showing considerable lead loss is pictured in Figure 7 where the telltale surface exsolution of zirconia is apparent.
(a)
(b) Figure
6: SEM micrographs of sintered PZT 5A fibers
a) cross section; fiber diameter ~130mm b)fiber surface. Grain
size is about 2mm
(dimension marker is 5mm).
Figure
7: a) Secondary electron image of lead deficient PZT fiber surface, fired at
1225C. Grain size about 2mm.
b) Zr x-ray image of same fiber region. Light
gray spots are excess Zr x-ray intensity indicating exsolved zirconia.
The corresponding Pb map shows Pb deficiency at these locations. CeraNova has developed a proprietary micro-atmosphere control method that is particularly suited to firing fibers. It is critically important that this control technique be translated from the current small scale batch manufacturing methods suited to 5,000 fibers per week to the methods under development for 30x this production level. In particular, any firing scheme must be compatible with mandrel unit handling. This includes binder burnout as well as sintering temperature requirements. Preliminary results are very encouraging. Mandrel units of several hundred fibers have been fired with good lead control and sintered fiber straightness. However, problems with fiber-to-fiber overlap indicate that further modification to the binder burnout cycle is required. 3.5
Preform fabrication The active fiber composite packages, as shown in Figure 1, require a monolayer of aligned monofilament active PZT fibers to be sandwiched between electrode and protective polyimide layers. To achieve major AFC cost reduction in the future, it is essential that fiber preforms, analogous to structural composite prepregs, be developed. These will enable fiber layup to be carried out as a unit placement operation. The mandrel unit handling concept being developed by CeraNova is targeted at this requirement.
Figure
8: Mandrel Unit Preform. Sintered
PZT fiber glued to an adhesive substrate that is compatible with the AFC layup
process. 3.6
Quality control Significant data exists on the use of CeraNova fibers in composite actuators. This provides an averaged indication of the properties of the fibers, and a means for feedback in the fiber processing investigations. Characterization is in the form of metrics appropriate for actuators, namely low field (linear) piezoelectric constants (such as d33), and high electric field strain response. Low field piezoelectric constants are calculated from low voltage direct strain measurements. Observations are made at ± 50V, ± 100V and ± 200V and extrapolated to zero volts. Experience with bulk materials has shown good correlation between results obtained by low field resonance methods and direct strain methods at voltages below 5% of the coercive field. Low field measurement of the d33 constant for AFCs show values as high as 200 x 10-12 m/V. This value is about half the 370 value for bulk PZT 5A in the 33 direction. Note that d33 is aligned with the fibers for the AFC. Other measured low field properties include transverse d constants, dielectric constant, and elastic modulus. To first order, the modulus of the AFC device in the fiber direction is equal to the area weighted ratio of fiber and epoxy stiffness. Discussion within the Consortium continues as to whether or not sintered active fibers should possess the piezoelectric properties of their bulk counterparts. It is possible that the small diameter and high surface area of fibers will limit achievable properties. The possible role of interface effects and space charge build up such as seen with ferroelectric memory devices is cited in support of this position. To date, single fiber permittivity values are about half the bulk counterpart values. Considerable effort is being spent on ensuring the reproducibility of the capacitance measurements used to determine fiber relative permittivity. It is exceedingly difficult to measure permittivity and piezoelectric properties of individual 100 micrometer diameter fibers due to the small signals obtained. Figure 9 shows an example of remanent polarization and coercive field measurement from a polarization - electric field loop obtained with a modified Sawyer-Tower circuit.
Figure 9:
Electric Field - Polarization Loop from single active PZT fiber.
Remanent polarization of 25 micro Coulomb / cm2 is about 75%
of bulk PZT 5A values. Coercive field of 10 kV/cm is comparable to the bulk value. |
|
