Статьи на перевод PVDF_P(VDF-TrFE) / Piezoelectric Composites for Sensor and Actuator Applications

dissipation factor in the figure of merit, the PVDF copolymer gave the highest dhgh/ tan δ of 2000 × 10−13 m2/N.

Recently Chen et al. [121] studied incorporation of nanosized PT powder, which was synthesized by a solgel technique into a vinylidene fluoride/trifluoroethylene (VDF-TrFE) copolymer matrix to form 0.2-mm thick films of (0-3) nanocomposites with ceramic volume fractions ranging from 0.11 to 0.41. The observed permittivity and dielectric loss of the composites were found to agree with the predictions of the Bruggeman model. The piezoelectric and pyroelectric coe cients of the poled composites can be reasonably described by modified linear mixture rules. The pyroelectric current, voltage, and detectivity figure of merit of the composites at 0.1 volume fraction are higher than those of the copolymer [121].

B. Composites with (1-3) Connectivity

In composites with (1-3) connectivity, the ceramic phase is continuously connected in one dimension, and the polymer phase is continuous in all three dimensions (Fig. 2). Harrison [108] obtained the first composites with (1-3) connectivity, which was followed by Savakus et al. [52], and Kawai [122]. A more common type of (1-3) composite was developed by Klicker [49], and Klicker et al. [50], [83]. The composite consisted of sintered, extruded rods aligned and embedded in the polymer. In this composite, the polymer phase is far more compliant than the PZT rods, allowing e cient stress transfer to PZT rods. The enhanced stress transfer, combined with the lower permittivity of the composite, has resulted in a much enhanced gh coe cient.

The internal stress created in (1-3) composite opposes the applied stress, which in turn reduces the stress amplification due to the high Poisson’s ratio of the polymer [123]. To reduce the Poisson’s ratio and increase the piezoelectric properties, Lynn [123] introduced porosity into the polymer by incorporating either a foaming agent or glass spheres. Both additives increased the hydrostatic properties; however, the composite with foaming agent in the polymer exhibited a pressure dependence on the piezoelectric properties. This connectivity pattern of this type of composites usually is designated as (1-3-0), in which the third number represents the connectivity of the voids or micro-glass spheres.

One way to increase the stress amplification in a (1-3) composite is to introduce transverse reinforcement [34]. In this design, the PZT rods are parallel to the poling direction, and sti glass fibers are placed in the transverse directions. The glass fibers carry the stress in the transverse directions, thereby decreasing the d31 coe cient without a ecting the d33 coe cient.

Another design developed to counteract the Poisson’s ratio problems consisted of a (1-3) structure, but without any lateral contact between the polymer matrix and the PZT rods, in which no charge is generated in the lateral mode [124]. The stress transfer is carried out by two metallic armature plates used as electrodes, which also provide a reinforcement of the polymer matrix in the transverse

directions. To prepare these composites, a polyurethane matrix is cut to the desired size and holes are drilled into it in a square or hexagonal arrangement, tailored to the required volume ratio. Armature plates of steel are cut to various thicknesses. The PZT rods are first aligned and bonded to one of the armature plates with a conductive adhesive (epoxy and silver). The polymer with holes is placed on top and bonded to the armature with the rods such that the PZT rods are standing without any contact with the surrounding polymer. The rods then are ground to the matrix thickness height, and the second armature is bonded to the PZT rods and the matrix. Some samples are further reinforced with an epoxy outer shell and all composites then are embedded fully in polyurethane. This type of composite is named a (1-3-1) in which the third number refers to the air element introduced to the composites. Typical piezoelectric d33 coe cients range from 250 to 400 × 10−12 C/N. The figure of merit (dh gh) is about 30 × 10−12 m2/N. The dh gh of composites with thicker armature plates generally are more stable, as are those with a small lateral epoxy shell. Both the armature plate thickness and the epoxy shell lower the stress supported by the lateral edges of the matrix and, therefore, limit its lateral strain and the related e ects. The dhgh of composites with a thinner armature plate decreases with pressure.

Over the past several years, many attempts have been made to simplify the assembly process for (1-3) design with the intention of improving manufacturing viability, lowering the material cost, and preparing large area composites. The early attempts involved dicing solid blocks and backfilling with polymer [37]–[42], [47], [125], [126].

Recently, Kim et al. (see [127]) devised a method to fabricate composites, which consisted of enclosing PZT rods and soft epoxy in the compartments of a honeycomb structure. The design of the process involved the selection of a sti corrugated preform made from a sti epoxy such as polycarbonate. The preforms were made in the shape of strips with trapezoidal grooves and ridges that would form a honeycomb when bonded on top of each other with epoxy. The PZT rods used in this study were PZT-5H with diameters of 0.08–0.13 cm. The volume percent PZT was varied by changing the rod diameter, introducing dummy rods, or by using a combination of rods with di erent diameters. The latter two approaches resulted in a nonperiodic arrangement of rods, which was beneficial in attenuating the spurious transverse resonances. The preform honeycomb polymer, in this composite structure, provided mechanical reinforcement to minimize the transverse contribution d31, and the soft epoxy reduced the Poisson’s ratio, allowing the ceramic rods to expand and contract. The prototype composite had a rod diameter of 0.08 cm, resulting in 17% ceramic, 70% polycarbonate (honeycomb preform), and 13% Rho-C soft epoxy, all by volume. Optimum values of piezoelectric properties of the composites were as follows: K = 406, d33 = 564 × 10−12 C/N, g33 = 157 × 10−3 Vm/N. dh, gh, and dh gh measured without a cover plate were 135 × 10−12 C/N, 45 × 10−3 Vm/N, and 6040 × 10−15 m2/N, respectively.

Fig. 11. Typical microstructure obtained by netshape molding. The piezoelectric phase (bright lines) has approximately 30 µm line width and has (1-3) connectivity. (Courtesy of L. J. Bowen)

Bowen and French [55] used an injection molding method to produce large area (1-3) ceramic/polymer composites. By using this technique, it is possible to make finescale (<50 µm) piezoelectric fiber-polymer composites for high frequency medical ultrasound and nondestructive applications as well as for underwater hydrophones. The injection molding process overcomes the di culty of assembling oriented ceramic fibers into composite transducers by net-shape preforming ceramic fiber arrays. Furthermore, the process makes possible the construction of composite transducers with more complex ceramic element geometries than those previously made.

The process consists of injecting a hot thermoplastic mixture of ceramic powder and organic binder into a cooled mold. Precautions such as hard-facing the metal contact surfaces are important to minimize metallic contamination from the compounding and molding machinery. After binder removal, firing and backfilling with polymer follow in typical composite manufacturing fashion. The PZT elements with 0, 1, and 2 degree taper are made in both 0.5 and 1 mm diameters. The preform size is maintained at 50 × 50 mm to accommodate molding shrinkage. The piezoelectric coe cient (d33) was 755 × 10−12 C/N, and the low-frequency relative dielectric constant 3588; respectively [55]. The capability of net-shape molding for fabricating very fine scale preforms has been demonstrated by making 30-µm wide PZT elements as shown in Fig. 11.

Polar glass ceramics, whose connectivity could be considered as (1-3), also have been investigated for use in hydrophones [45]. These glass ceramics can be thought of as diphasic composites composed of a glassy phase, which is continuous in three dimensions, and one or more crystalline phases, which are continuous in only one dimension. The gh and dhgh of these composites are comparable to those of PVDF. However, glass ceramics are favorable

especially because they involve no problems with aging or depoling as they are nonferroelectric and exhibit no pressure dependence [45].

Panda [82] fabricated staggered square rod (1-3) composites using solid freeform fabrication in which the negative replica mold of the desired structure was first built, then infiltrated with PZT slurry, followed by sintering. Defect-free structures with hundreds of uniform rods and with easy reproducibility were obtained. Composites with 350-µm square rods, separated from each other by typically 450 µm, could be reproducibly obtained. The structures had a ceramic content of 23 vol.%, and a dielectric constant of 425. The d33, kt, and kp of the composite were typically around 290 pC/N, 68% and 29%, respectively [82].

C. Composites with (3-0) Connectivity

The PZT ceramic is self-connected in three dimensions, and the second phase (polymer, voids) is not connected in any dimension in composites with (3-0) connectivity. Kahn et al. [128] and Kahn [129] developed ceramic-air composites with (3-0) connectivity by using a tape casting and multilayer ceramic technology. In these studies, a slip consisting of PZT ceramic powder, organic binder, and solvent were tape cast and dried to form thick-film tapes (34 µm). Fugitive ink then was screened onto the tapes in a computer-generated pattern. One hundred layers of tape then were stacked in a chosen configuration and laminated at 55◦C and 2 MPa. The samples then were heat treated to burn out the organic binders and carbon, leaving ordered voids. To prevent defects in the ceramic due to the rapid gas evolution from exothermic reactions during burnout, a well-controlled lower partial oxygen pressure was used. Sintering of the samples then was achieved in a covered alumina crucible. The total porosity of the final samples ranged from 15–22% depending on the configuration and size of the ink pattern. The highest hydrostatic piezoelectric values were reported for samples with a crossed-bar void configuration. The d33 of these composites was 350 × 10−12 C/N with a dielectric constant of 500. The dh and dhgh of these samples were 230 × 10−12 C/N and 13300 × 10−15 m2/N, respectively. Another type of (3- 0) composite was fabricated by hot pressing a mixture of large polymer spheres of polyethylene and PZT powder, thus yielding an irregular array of low K grains of polymer surrounded by high K (PZT) boundaries. Exceptionally low acoustic impedance is obtainable using this method [128], [129].

D. Composites with (3-1) and (3-2) Connectivity

The connectivity in (3-1) and (3-2) composites is such that the ceramic phase is continuous in all three dimensions, and the polymer phase is continuous in one and two dimensions, respectively. An early composite in this category is the perforated composite, the fabrication of which involves drilling holes in sintered PZT blocks, then back

filling with an epoxy polymer [35]. As an alternative to polymer impregnation of the composite, the perforation could be left empty, and an alumina plate then could be used to cap it o , followed by encapsulation in epoxy resin. Actually, composites as the one just described also are referred to as perforated (3-1-0), or (3-2-0) composites. It was found that the size of holes, as well as the volume fraction of piezoelectric filler (PZT) had a substantial impact on the hydrostatic sensitivity of the perforated composites. Also, an increase in the hydrostatic coe cients was observed, which was attributed to the decoupling of the d31 and d33 coe cients [35]. Large arrays (400 × 400) prepared using (3-1) composites showed comparable gh with respect to a single element, indicating the scale-up had no adverse e ect on properties [130]–[132].

The other type of (3-1) composite is the so-called honeycomb configuration (see Fig. 2). This particular process has the added flexibility of producing composite microstructures in which the polymer phase can be continuous in either poling direction (3-1P connectivity), or perpendicular to the poling direction (3-1S connectivity). A comparative study on (3-1P) and (3-1S) composites has revealed that the stress distribution in the (3-1P) configuration was very e ective in lowering the d31 coe cient, thereby imparting great anisotropy to the (d33/d31) ratio [82].

E. Composites with (2-2) Connectivity

The most common type of (2-2) composites uses sheets of piezoelectric ceramic material separated by epoxy, as shown in Fig. 2. These types of composites have shown good properties for ultrasonic transducers [133]–[135]. Similar to the (1-3) design, it is critical to maintain a high kt in these (2-2) composites for high-energy transfer through the thickness mode. The variation of the thickness coupling coe cient kt is related to the interaction between thickness and lateral modes. For kt to attain a maximum value, the thickness mode has to be completely decoupled from the lateral resonant mode. Shui and Xue [136] have computed the kt2 for (2-2) composites as a function of the ceramic volume percent and the aspect ratio [137]. Fig. 12 shows that the thickness-coupling coe cient falls drastically for a smaller aspect ratio (t/w) on decreasing volume percent ceramic. For t/w > 4, the drop would be negligible.

As schematically shown in Fig. 13, Schae er et al. [87] used sintered tape-cast sheets of PZT-5H to fabricate composites with (2-2) and (2-0-2) connectivity. The third phase in the form of powder included zirconia, alumina, or hollow and solid glass spheres. The (2-0-2) composites showed a ring-down time up to 50% faster than (2-2) composites. The dh value of the (2-0-2) composites were typically 20% higher than those of the (2-2) composites at equivalent PZT volume fractions. At 40% PZT loading, (2-0-2) composites exhibited a dh of about 90×10−12 C/N.

Huebner et al. [88] developed fine-scale piezocomposites of (2-2) connectivity by bonding thin (≤20 µm) sintered PZT plates and sheets of a thermoplastic polymer film us-

Fig. 12. The variation of kt2 as a function of ceramic volume percent for (2-2) composites for four di erent aspect ratios (t/w): 166.66 (dashed line), 4.16 (solid line with a dot), 1.04 (solid line), and 0.79

(dashed line with a dot). (Courtesy of R. K. Panda)

Fig. 13. Method for making fine-structured 2-2 and 2-0-2 piezoelectric ceramic/polymer composites by tape casting.

ing thermal processing. The piezoelectric plates were obtained by stack-sintering, tape-cast PZT, and tape-cast polymers were used to control the thermoplastic layers. Composite blocks were cut to required dimensions to fabricate linear arrays as shown in Fig. 14. It was demonstrated that the process is capable of producing fine-scale composites that were impossible to obtain by the dice and fill method. Typical thickness coupling coe cients for such composites were about 0.65 with a mechanical quality factors 5 for PZT and polymer thickness ranging from 5– 55 µm and 9-20 µm, respectively.

The traditional dicing technique poses severe di culties to introduce a volume fraction gradient in the elevation direction of a (2-2) composite as the so-called kerf width goes on increasing from the center to the edges while a cutting blade with a single thickness is used in the dicing machine. Furthermore, the resolution of the traditional techniques cannot be pushed under 50 µm. To overcome these limitations, Panda [82] studied volume fraction

Fig. 14. Fabrication of (2-2) connectivity PZT/thermoplastic composites for high-frequency linear arrays by tape casting and dicing.

gradient (VFG) (2-2) composites that were fabricated by the indirect lost mold route using Sanders prototyping—a CAD-based prototyping technology. In that study, many mathematical functions—including regular, Gaussian, linear, and exponential gradients—were designed in order to study the e ect of introducing di erent types of gradients along the elevation direction. The electromechanical and acoustic properties of such VFG composites were compared to those having no VFGs.

In Panda’s work [82], all the VFG distributions were designed to have 60 vol.% ceramic in that center region of the composite to obtain a high acoustic pressure output at the center. However, no further increase in the ceramic content at the center was pursued because of the acoustic impedance di erence between the central region and the edges. As a consequence, very complex matching layers would be needed for good energy transfer across the entire composite. The ceramic content in such VFG composites was gradually decreased to approximately 20% at the edges to ensure that the edges would act as a good receiver of ultrasonic waves due to the low dielectric constant of those regions. It should be noted that a further decrease in the total ceramic amount at the edges could further enhance the receiving sensitivity. However, this would concomitantly cause a fall in the thickness-coupling coe cient due to interference from the lateral modes for the ceramic element width and spacing of interest.

Figs. 15(a) and (b) show the sacrificial molds for two di erent functional distributions, Gaussian, and linear VFG, as obtained using the Sanders prototyping technique. Throughout the structure, an air gap of 135 µm between the walls was created. This enables one to ascertain that ceramic walls remain of uniform width after slurry infiltration, as well as after subsequent heat treatment steps. In that particular design, the mold wall thickness was increased from 95 µm at the center to 575 µm at the edges, following the respective mathematical function. The positioning and thickness of the mold walls were very accurate with an X-Y deposition error of only ±5 µm. After infiltrating PZT slurry into the molds and subsequent heat treatment, the sintered structures were embedded in Spurr epoxy (Polysciences, Inc., Warrington, PA).

Fig. 15. (a) SEM photograph showing the Sanders build wax sacrificial polymer molds for the Gaussian, and (b) linear volume fraction gradient distribution. (Courtesy of R. K. Panda)

The electromechanical properties of the Gaussian, linear, exponential, and regular (2-2) composites are shown in Table IV. The dielectric constant was as high as 950 for the Gaussian and 430 for the exponential design, following the total volume percent of PZT in the composite. All structures had well-discernible thickness mode resonances with kt’s on the order of 66%, and moderately low values of the kp. The high kt’s in such composites were attributed to the high aspect ratio of the ceramic elements (>6), and to the soundness of the VFG design by Sanders prototyping. The longitudinal piezocharge coe cient (d33) of all these composites were found to be greater than 400 pC/N.

The vibration profiles of the (2-2) composites with and without VFG’s are shown in Fig. 16 as a function of the distance from the center. The pressure outputs of the VFG designs were compared with a regular 2-2 composite, which was a diced PZT-5H ceramic structure having a wall width of 220 µm and a kerf of 440 µm. The regular design had33 vol.% ceramic with a d33 and kt of 340 pC/N and

Fig. 16. Graph showing the vibration amplitude profiles for 2-2 sheet composites with regular, Gaussian, linear, and exponential VFG distributions as a function of the distance from the center of the structure. (Courtesy of R. K. Panda)

63%, respectively. As expected, the amplitude of the vibration for this composite did not vary much with the distance from the center, as this structure was the same everywhere. The small fall in the amplitude at the edges could be due to the edge-clamping e ect. However, all the VFG distributions showed a decrease in the vibration amplitude as distance from the center increased. The Gaussian VFG was designed to have ceramic volume percent fall very slowly with distance from the center. The vibration amplitude observed in Fig. 16 followed a similar trend, initially remaining constant and dropping by only −4 dB at a distance of 5 mm. Thus, this distribution was not very different from the regular 2-2 composite out to 5 mm from the center. The linear gradient showed a very smooth decline in the vibration output with distance, decreasing linearly to about −8 dB at a distance of 5 mm. The exponential distribution had a vibration profile that was slightly lower in amplitude than the linear distribution. It should be noted that, although the ratio of the total ceramic content in the center and the edges is constant for all VFG distributions,

Fig. 17. Predicted far field beam pattern plots for a regular, Gaussian, and linear VFG distribution. (Courtesy of R. K. Panda)

the pressure output ratios are not similar. This could be because the 2.3-mm diameter photonic sensor probe shows an average reading over that area. Hence, the readings encompass not only the edge, instead a slightly larger region. The precise manipulation of the volume fraction within the same composite allows the control of the vibration amplitude profile to obtain the desired pressure output as seen in Fig. 17. The Gaussian vibration amplitude profile did not fall down fast and was very similar to the regular distribution. The linear or the exponential gradient showed a very smooth decrease in the vibration from the center to the edges.

Turcu [138] and Turcu et al. [139] studied oriented (2- 2) soft PZT-epoxy composites, which were fabricated by FDC as shown in Fig. 18. This study was inspired by the work of Nan et al. [140], [141] who investigated the effects of orientation of the ceramic phase relative to the poling directions were modeled in 0-3 and 1-3 piezocomposites. The modeling results showed that the dielectric and piezoelectric properties of the piezocomposites with (1-3) connectivity decreased when the poling direction deviated from the orientation of the ceramic phase. In Turcu and co-workers’ studies, however, the orientation angle of the ceramic phase relative to the poling direction varied in the range of 0◦ to 75◦ with 15◦ increments (Fig. 18)

[138], [139]. The volume fraction of the ceramic phase in both composites was fixed at 30 vol.%. It was found that the dielectric constant and piezoelectric charge coe cients of both composites decreased with increasing orientation angle. However, the piezoelectric voltage coe cient, g33, showed a modest (about 10%) improvement in (2-2) composites. The relevant piezocharge coe cients and the FOM of the oriented composites were calculated as shown in Figs. 19 and 20, respectively [138], [139]. The data exhibit dhgh maxima at an orientation angle of 45◦. It also is interesting to note that d31 goes to zero at an orientation angle of 38◦, which does not coincide with the orientation angle in which the maximum is observed [see Fig. 19(b)].

F. Composites with (3-3) Connectivity

A composite with (3-3) connectivity is comprised of two phases that mutually penetrate, and thereby form two three-dimensionally, self-connected networks in immediate contact. A series of (3-3) composites have been fabricated out of PZT in a number of di erent processes, which include replamine composites by Newnham et al. [142], burned-out plastic sphere composites (BURPS) by Rittenmyer et al. [84] and Shrout et al. [131], ladder composites by Creedon and co-workers [78], [79] and Miyashita et al. [143], and sandwich composites as shown in Fig. 2. These types of composites are relatively easy to pole as the piezoelectric phase is continuous between the electrodes.

Hikita et al. [144] have implemented a process to impart interconnected porosity to BURPS composites with (3-3) connectivity. Another (3-3) connectivity pattern was developed by Zhuang [145], which consisted of creating a sandwich composite out of one PZT/polymer layer surrounded by two solid PZT surface layers. Yet, another completely di erent fabrication method was developed to make (3-3) composites, termed the fired composite [113], [146], also was developed. This method involves mixing of PZT powder with an organic binder, pressing the mixture into a pellet, and removal of the binder by burnout. The result was a porous PZT compact that could be back filled polymer if needed. These were found to exhibit high d33 (×10−12 C/N), and kt (34%). However, it should be noted that fired composites are not of (3-3) connectiv-

Fig. 19. Piezoelectric properties of oriented (2-2) and (3-3) composites fabricated by FDC: (a) longitudinal piezostrain response, and

(b) transverse piezostrain response. (Courtesy of S. Turcu)

Fig. 20. E ect of piezoelectric phase orientation on the figure of merit (dh gh) in (2-2) and (3-3) oriented composites by FDC. (Courtesy of S. Turcu)

ity per se because there always is a certain amount of particle-particle interaction due to the heat treatment step involved. Nevertheless, there is some degree of (3-3) connectivity present within the composite [45].

Turcu and co-workers [138], [139] studied oriented (3-3) soft PZT-epoxy composites, which were fabricated by FDC as shown in Fig. 18. In this process, piezoelectric powder that is dispersed in a thermoplastic matrix and extruded into a filament is deposited layer by layer under full computer control so as to obtain a three-dimensional structure (see Fig. 18). The thermoplastic fugitive matrix then is eliminated by binder burn out, and the remaining PZT preform then is sintered. This CAD-based approach allows one to design the piezoelectric “road” thickness, line spacing, composite thickness, and orientation of the “road”. Testing revealed that the piezoelectric voltage coe cient (g33) showed a significant increase (about 85%) in (3-3) composites at an orientation angle of 45◦. The FOM of the oriented (3-3) composite reached an extremely high value of 50, 000 × 10−15 m2/N (refer to Fig. 4 for comparison with other composites), but a more modest increase was seen for the (2-2) composites (14, 000 × 10−15 m2/N) as shown in Figs. 19 and 20, respectively.

The feasibility of the FDC process in making piezoelectric composites was demonstrated in the fabrication of PZT/polymer composites with (3-3) connectivity (threedimensional honeycomb and ladder structures) [82]. In the direct processing route, FDC was used with a PZT powder loaded polymer filament as the feed material for a directlayered manufacturing of the three-dimensional green ceramic structure. In the indirect process, StratasysTM (Stratasys, Inc., Eden Prairie, MN) commercialized polymer filament was used to fabricate a mold via FDMTM. The final structure then was created using the lost mold method. In both methods, after embedding the ceramic structure in epoxy, electroding and poling resulted in the desired composite.

Two scanning electron micrographs (SEM) of a typical heat-treated ceramic ladder structure prepared by the direct FDC method prior to epoxy infiltration are shown in Fig. 21. The ladder structures were built by using a rasterfill strategy with a fixed inter-road spacing. The consecutive layers were built 90◦ to one other. The volume fraction of the ceramic phase in the structure shown in Fig. 21 is approximately 70%, and can be varied by varying the width and spacing between the ceramic roads. In Fig. 21, the ceramic roads are 300-µm wide with 800 µm center- to-center spacing. The mesoscopic structures created by FDC, as depicted in Fig. 21, are very uniform in structure with excellent unit cell repeatability, clearly demonstrating the superiority of the FDC process over traditional methods of making composites [82].

The indirect method was used to fabricate threedimensional honeycomb composites with (3-3) connectivity. The SEM in Fig. 22 depict the various crystallographic orientations of a typical heat-treated ceramic structure prior to epoxy infiltration. Fig. 22(a) shows the top, front, and side faces, and Fig. 22(b) shows the top and front

Fig. 21. Ladder structure with (3-3) connectivity as obtained by FDC. Upon impregnation with polymer, a (3-3) piezoelectric ceramic/polymer is formed. (Courtesy of R. K. Panda)

faces. The structure consists of a three-dimensional lattice of “air pipes” defining the three-dimensional piezoelectric skeleton (matrix). The “air pipes” were created by burning out the sacrificial polymer. By virtue of the flexibility o ered by the FDC process, the diameter of the holes, the spacing, and the volume fraction of the ceramic phase can be varied between the holes. In the example, given in Fig. 22, the diameter of the holes is 200 µm, with350 µm center-to-center uniform spacing, and the PZT content is 25% by volume [82], [147], [148].

Several three-dimensional honeycomb composites, whose relative permittivity (K33) and piezocharge coe cient (d33) are shown in Fig. 23, were fabricated by the indirect method with 10 to 35 vol.% PZT loading. An increase in K as the volume percent of PZT increased was observed, which is in conformity with mixing laws to a first approximation. It also is seen that there is an increase in d33 as the volume percent of PZT increases from 10 to 30. The d33 value then levels o at about 35 vol.% PZT, in which pseudo-single phase behavior is reached [147].

Recently Robocasting, which is another SFF process, has been used successfully to fabricate piezoelectric composites [149]. As shown in Fig. 9, composites with (3-3) connectivity have been built out of PZT, and were found to exhibit reasonable hydrostatic FOM (see Fig. 24). It was also shown that the incorporation of a faceplate in these (3-3) composites increased the FOM by approximately one order of magnitude, reaching a value of approximately 8000 × 10−15 m2/N.

G. Fine-Scale (1-3), (2-3), and (3-3) Piezoelectric

Ceramic Fiber Composites

The PZT rod-polymer composites with (1-3) connectivity have proven to be very e ective transducers in ultrasonic applications in which water is the acoustic load [45].

It was speculated that the increase in performance in such composites was due to the strong interaction between the PZT and the epoxy. The said interaction occurred at frequencies in which the transverse wavelength in the epoxy was much larger than the periodicity of the (1-3) lattice. Experimental observations suggested that the resolution of such composites could be improved if it were possible to operate them at higher frequency. That, of course, could be achieved only by a substantial scaling of the composite structure. In other words, the rod diameter and the periodicity of the lattice in the composite structure had to be reduced.

Card and co-workers [150], [151] and Waller and coworkers [66], [67], who used the so-called relic process, were the first to obtain fine-scale woven PZT fiber/polymer composites. These composites have essentially (2-3) connectivity. This process, which is believed to be adaptable for mass production, involves the following steps: preparation of an alkoxide PZT stock solution, impregnation of a carbon template with the stock solution, burn out of the carbon, and sintering of the PZT relic, and backfilling of the PZT relic with a polymer. In these studies, activated carbon fabric was used.

The series of SEM photomicrographs in Fig. 25 illustrate the relic process. Fig. 25(a) depicts the woven carbon fiber template material. Fig. 25(b) shows how the woven form is preserved in the PZT relic after the sintering stage. Individual fibers of PZT can be seen that correspond to the individual carbon fibers in the original template. Woven composites were poled in two orientations (i.e., parallel and perpendicular to fiber directions), and properties were

Fig. 24. E ect of solids loading on the figure of merit (dh gh) PZT composites obtained by Robocasting. (Courtesy of D. Dimos)

evaluated for both directions. The dielectric properties of composites with PZT fibers oriented perpendicular to the poling direction are K = 100, d33 = 150 × 10−12 C/N and g33 = 60 × 10−3 Vm/N.

The relic processing method has been adapted to produce large-area composites. The scaling-up of the procedure has included studying of variables such as the firing and poling conditions of increasingly larger samples beginning with 1 in. by 1 in., 2.5 in. by 2.5 in. and ending with 4 in. by 4 in. The electromechanical properties of the largest composites are comparable to those of 1 in. by 1 in. samples. Improved processing methods have enabled properties of K = 140, d33 = 180 × 10−12 C/N, and g33 = 65 × 10−3 Vm/N [152].

Fine-scale piezoelectric ceramic-polymer composites also were developed at Siemens Inc. (Munich, Germany) using a new ceramic processing method called the jet-

Fig. 25. SEM micrographs of (a) woven carbon template, (b) sintered PZT relic obtained from the template. (Courtesy of B. Jadidian)

machining technique [37]. In this method, photoresist was deposited in a unique pattern on the surface of the green PZT pellets using a standard lithographic technique. Ceramic powder of the uncovered area was removed by jet spraying. The jet-machining process cut deep, fine scale, concentric circular grooves with a coupler pattern, which are not possible to create with a dicing-saw. Firing at high temperature sintered the green pattern into a dense ceramic. The important feature of jet-machining is the simplification of fabrication and the creation of complex shapes, which are normally made using the dicing technique [37]. This same team devised a lost mold method for net-shape casting of piezoceramics in which the groovewidths and ceramic-widths can be below 10 µ, well beyond the limit of present diamond-wheel machining technology [37]. The ceramic slurry was cast into a plastic mold, dried, and fired. The fired ceramic compact then was flipped over and the space occupied by the lost mold was refilled with a polymer. Grinding, electroding, and poling completed the fabrication of a fine-scale piezocomposite. Composites con-

Fig. 26. Cross-sectional view of the microstructure of bundled composites with (1-3) connectivity. (Courtesy of B. Jadidian)

sisting of PZT rods with a diameter of 100 µm and a thickness of 0.5 mm were prepared by this method. This technique also can form solid-ceramic transducer structures with complex shapes not readily made with a dicing saw. Large-scale production of fine-scale structures is feasible; the key is forming the plastic molds, which are lost in the firing process. To make molds with very fine scale structures, a deep X-ray synchrotron lithographic technique was used to first form a reusable metal mold with the desired shape of the final ceramic part; the sacrificial plastic mold then was cast from this metal master. Making the plastic molds is a low-cost process, but producing the metal master is costly when very fine scale features are involved.

Jadidian [153] developed a series of fine-scale composites using fibers supplied from Advanced Cerametrics, Inc. (Lambertville, NJ). Sized and unsized green PZT fibers were incorporated into epoxy polymer to develop novel piezoelectric ceramic/polymer composites by three di erent methods. Sized bundles of PZT fibers were sintered to fabricate PZT rods. Rods were bundled and embedded in polymer to form (1-3) type composites as shown in Fig. 26. Collimated bundles of unsized fibers were sintered and backfilled with polymer to form (1-(3-3)-3) structures. Stacks of woven fabrics were sintered with and without applied pressure during heat treatment; the resultant microstructures as well as a representative preform is shown in Fig. 27. Composites with (3-3) connectivity were formed after imbedding sintered stacks in a polymer matrix. Strips of plane weave fabrics also were rolled tightly to form spirals as shown in Fig. 28. Bundled (1-3) composites embedded in Eccogel and Spurr epoxy had comparable thickness coupling coe cients. The kt value for these composites varied in the range of 60 to 63%. Collimated (1-3) composites also had comparable kt values to bundled composites. Fast

Fig. 27. Top view SEM micrograph of (a) green plane weave fabric made using sized PZT fibers, (b) and (c) sintered laminate of plane weave fabric sintered (at low and high magnifications, respectively), and (d) plane weave fabric made using sized ends and unsized picks. (Courtesy of B. Jadidian)

sintering the collimated bundles at 1240, 1275, and 1285◦C for 2.5 to 5 minutes, respectively, collimated the composites. The dielectric constant and piezoelectric charge coefficient of fast-fired composites were increased by increasing the soak time at all temperatures, then decreased due to lead loss. In composites with (3-3) connectivity, the dielectric constant increased linearly with the applied pressure. This was attributed to an increase in the sintering contact between the fabric sheets in the stack. The d33 value of these composites increased with pressure and saturated at 588 Pa pressure. This was in agreement with the variation of composite density, saturating at 588 Pa thresholds. The type of polymer matrix had significant e ect on the electromechanical properties of these composites. Composites with Eccogel epoxy had a d33 three times higher than the composites with Spurr epoxy. The poling direction had significant e ect on the dielectric constant of (3-3) composites. The value of parallel poled samples was twice as much as the K values of perpendicular poled samples. The spiral composites had the highest thickness-coupling coefficient compared to the (1-3) composites. The d33 values of this structure and (1-3) bundled composites were comparable (see Fig. 26 for microstructure). In Tables V–VII, the electromechanical properties of the composites mentioned above are summarized [153].

Bent et al. [154] developed piezoelectric fiber composites (1-3) connectivity as an alternative to monolithic piezoceramic wafers for structural actuation applications. In their work [154], interdigitated electrodes were used to enhance the in-plane strain output of the composite actuators. The PZT fiber composites, with fiber volume fractions in the range 7–58%, were tested. Composites with

Fig. 28. (a) The orientation of fibers in spiral and (b) green spiral. (Courtesy of B. Jadidian)

TABLE V

Elastodielectric Properties of Eccogel-Based, Spurr

Epoxy-Based, and Collimated Fine-Scale Composites

Fabricated from PZT Filament.1

1Courtesy of B. Jadidian.