The deposition rate is thus proportional to the electric field E and the particle concentration C, so that nonnuniformities in E or C can lead to nonuniform deposition. For substrates with a trench geometry, like those illustrated in Figure 2, the electric field is concentrated at the exterior corners at the top of the trench and minimized at the interior corners at the bottom of the trench. As a result, the deposit will tend to be thicker on the exterior corners than on the interior corners, resulting in the nonuniform film shown at the left in Figure 2. The resistivity of the electrophoretically deposited film is usually higher than the resistivity of the suspension, however, so that after the initial preferential deposition at the exterior corners occurs, the thicker film there has higher resistance than the rest of the film, decreasing the potential difference across the suspension phase available to drive electrophoretic deposition. This decrease in the potential gradient near areas that had high deposition rates initially allows the other areas of the film to “catch up”. If the resistivity of the deposited film is much larger than the resistivity of the suspension, then the uniform film illustrated in Figure 2 may be obtained.

Figure 3 shows a simple one-dimensional deposition model used to represent the two-dimensional trench geometry shown in Figure 2. Our goal is to produce the same thickness of film at location (a) and location (b) of Figure 2. Initially, there is a higher deposition rate at (a) than at (b), due to the higher electric field there initially. If the geometry of the trench and the positions of (a) and (b) are such that the initial electric field at location (a) is 10 times the initial electric field at location (b), then we may obtain an approximate solution by modeling locations (a) and (b) as parallel plate electrodes where the factor of 10 difference in initial electric field is produced by spacing the electrodes 10 times further apart in the model for location (b) compared to the model for location (a).

With the model depicted in Figure 3, we seek to calculate deposit thicknesses at location (a) and location (b) under various conditions to see what conditions produce conformal films. Graphs of deposit thickness as a function of time at locations (a) and

(b) are provided in Figure 4. For the graph on the left, the film resistivity and suspension resistivity are the same (1:1 resistivity ratio), resulting in constant deposition rates with time and nonconformal deposition. For the graph on the right the film resistivity is 104 times more than the suspension resistivity (104:1 resistivity ratio). Such high resitivity films tend to deposit in conformal, thin layers. Table I lists assumed values of parameters used to generate the Figure 4 plots.

EXPERIMENTAL – SUSPENSION APPROACH TO EPD

The deposition bath for electrophoretic deposition consists of a suspension of particles dispersed in a liquid media. The particles are typically dispersed in the liquid by ultrasonic agitation (sonication). Chemical interactions with the liquid cause the particles to acquire a surface charge, and this surface charge sets up repulsive interactions between particles so that they remain dispersed (unagglomerated). The dispersion media is chosen so that the chemical interaction produces the maximum amount of surface charge, thereby making the suspension more stable.

Another approach to electrophoretic deposition, the use of a liquid media that actually solvates the particles rather than simply acting to keep them dispersed, is discussed later.

Procedure

The standard electrode material (for both the deposition substrate and counter electrode) is lightly doped silicon, cut into a 25 mm by 30 mm rectangular dies. (The high resistivity of the nonaqueous dispersion media (isopropanol) allows lightly doped silicon to be used as an electrode without causing potential variations across the electrode surface.) The particles used in these depositions are 0.2 m diameter P(VDF-TrFE) particles, with (VDF/TrFE) ratio of (75/25), obtained from Measurement Specialties, Inc. (Fairfield, NJ, USA). Figure 5 shows a scanning electron micrograph (SEM) of these particles on silicon. The standard procedure for electrophoretic deposition with the suspension approach follows.

1. Clean substrate. For Si substrates, contamination is removed in hot “piranha” (H2SO4 & H202) bath and native oxide is removed in HF, followed by multiple DI water rinses.

2. Prepare suspension. Standard concentration of P(VDF-TrFE) powder in isopropanol dispersion media is 10 g/l.

3. Sonicate suspension. Place ultrasonic horn in suspension and use Sonifier (Branson Ultrasonics, Danbury, CT) to apply ultrasonic energy for 2 minutes, adjusting intensity to keep output power at 50 W, and allowing suspension to cool for 1 minute in middle of the 2 minute period. Suspension should be allowed to cool for 5-10 minutes before beginning deposition, because isopropanol evaporates too fast after deposition if the suspension temperature is too high, and because the deposition rate depends on temperature.

4. Deposit P(VDF-TrFE). Position electrodes (the deposition substrate and counter electrode) in deposition jig parallel to each other and separated by 6 mm. Immerse electrodes 25 mm into suspension (immersed area = 6.25 cm2) by raising suspension beaker up to electrodes with labjack. Deposit at 70 A constant current (11A/cm2 current density) for desired duration. Voltage varies during deposition, but 50 – 200 V is a typical range.

5. Post-deposition procedure. Remove suspension from substrate quickly and smoothly with labjack, and then rapidly raise beaker of isopropanol onto substrate (raise rapidly before film has time to dry out). Let substrate sit in isopropanol beaker for 1-2 minutes, remove quickly and smoothly, and allow film to dry in air.

6. Anneal P(VDF-TrFE). Put substrate in oven stabilized at temperature of 150200°C for 1-10 minutes. Remove substrate and place on aluminum cooling block.

This procedure was used to deposit most of the suspension-based EPD films described in this section. Important deviations from this procedure are noted where they occur. In addition to these six steps, two more should be carried out to make P(VDFTrFE) films piezoelectric, namely a crystallization anneal and electrical poling. The crystallization anneal (120-140ºC for 12-18 hours (14)) is carried out to increase the degree of crystallinity of the P(VDF-TrFE) polymer. Higher crystallinity gives better piezoelectric and ferroelectric performance. Poling imparts a preferred direction to the polarization of the film by applying high voltages at elevated temperatures. Without being poled, ferroelectric materials are not strongly piezoelectric.

Process Details

When a potential difference is applied between two electrodes immersed in a suspension of P(VDF-TrFE) in isopropanol, a porous, white P(VDF-TrFE) film deposits on the positive electrode. By this we know that P(VDF-TrFE) particles are negatively charged in isopropanol. Deposition voltages are typically in the range of 50 to 200 V for the current density given in the standard procedure (11 A/cm2). Deposition voltages are so high because the suspension is highly resistive. Compared to metal electroplating, EPD is a low-current and high-voltage electrodeposition procedure. Electrophoretic deposition at lower voltages is possible, albeit with proportionally lower deposition rates, but caution should be exercised, because we have observed that deposition at low electric fields (e.g. 80 V across a gap of 2.5 cm, a field of 32 V/cm) may produce poorly adhered films, consistent with the “minimum deposition field for coagulation” referred to in (1).

Further details on other aspects of the deposition process are not included here due to space constraints, although they could be useful to someone attempting to carry out an electrophoretic deposition. Contact the authors for more information.

Deposition Rates

Deposition rates of films deposited by the EPD suspension approach can be very high. For example, a film of 12 m thickness (when dense) was deposited in 80 seconds of deposition time (deposition rate 9 m/min). The deposition time does not include the time of drying the substrate or annealing the substrate, just the time that current was flowing. The deposition rate can be controlled by varying the parameters specified in Equation [1], particularly the applied electric field E and the suspension concentration C. Higher deposition rates are obtained with higher voltages and higher concentrations. Suspension concentration is limited by the fact that there is a trade-off between the concentration of a suspension and its stability.

Uncontrolled variability in E and C can also affect deposition rates, either because electric field variations can cause the deposition rate to vary during a deposition, or changes in concentration can cause the deposition rate to vary between depositions (see Table II for an example of this).

The high deposition rates of suspension-based electrophoretic deposition allow very thick films to be deposited. Figure 6 shows SEMs of a porous ~100 m-thick P(VDF-TrFE) film deposited simultaneously on both sides of a silicon substrate from a suspension of P(VDF-TrFE) particles in isopropanol. Conventional spin coating is not capable of forming a film like this, because spin coating is not capable of depositing films with such high porosity (estimated at ~50%). For applications where high porosity on a very fine scale (on the scale of the 0.2 m P(VDF-TrFE) particles) is important, such as fluid filters and piezoelectric composites, suspension-based EPD is well suited to the task. For applications where thick, dense films are desired instead, the porous 100 m film would be annealed to attempt to make a dense 50 m film. The production of thick, dense films has been problematic, however.

Annealing Problems

Figure 7 shows cracks in two P(VDF-TrFE) films after annealing. The samples were annealed in an oven at 235°C for 4 minutes. The melting point of this variety of

P(VDF-TrFE) copolymer is 141°C according to differential scanning calorimetry (DSC) measurements, so the polymer is definitely molten at the 235°C anneal temperature. We might expect a viscous liquid (like the molten polymer) to be tougher and more resistant to crack growth compared to other materials (such as brittle ceramic). This fluidity did not prevent cracks from developing, however. The mechanism of crack formation in suspension-based EPD films of P(VDF-TrFE) has not been determined, but cracks could form during the drying of isopropanol from the porous film, during shrinkage of the film during the densification anneal, or perhaps due to poor wetting of the polymer to the silicon.

Comparing the two Figure 7 images shows the effect of film thickness on crack formation. The thick film shown at the left has pronounced cracks widely separated, while the thinner film at the right has less pronounced cracks more closely spaced. This leads us to try thinner films to see whether we can avoid cracks entirely, although thin films tend to have problems with pinholes. Trials comparing films of different thickness and annealing temperature have produced a variety of results: in summary, it may be said that the process window for producing a film without cracks or a significant number of pinholes is not large. So far, films made in a single deposition thin enough to avoid cracking have still had enough pinholes to prevent their use in broad area piezoelectric sensors and actuators.

Multiple Deposit/Anneal Cycles and Ferroelectric Testing

There is a way out of this dual problem of cracking and pinholes. On top of an EPD film that has already been annealed, we perform another electrophoretic deposition. This second deposition tends to occur preferentially at the defects (cracks, pinholes, etc.) of the first film. Subsequent annealing melts the second deposit so it fuses into the cracks and pinholes, partially or completely healing them. By repeating this process of deposition and annealing, it is possible to decrease the defect density until an acceptably low value has been reached. This is one of the strengths of electrophoretic deposition, that defects that would otherwise cause electrical shorts can be engineered out of the film by multiple deposition/annealing cycles. This approach has previously been used with the electrophoretic deposition of PZT (9,11). The progress made in eliminating defects can be tracked by monitoring the increase in resistance during electrodeposition.

By using the multiple deposition/annealing process, films suitable for electrical testing have been formed. Figure 8 shows a ferroelectric hysteresis loop obtained from a 12 m thick P(VDF-TrFE) film formed by five cycles of deposition and annealing. In the making of this film, the resistance of the deposition setup (film and suspension) increased 38% by the end of the 5th deposition compared to the start of the 1st deposition, a larger increase than obtained for any suspension-based single-cycle deposition/anneal process. Poling was not conducted beforehand, instead, ferroelectric hysteresis loops of gradually increasing amplitude were applied, up to the maximum output of the high voltage amplifier of 535 V. The 12 m film did not experience electrical breakdown over the area of the 0.9 mm diameter top electrode at all voltages up to and including 535 V (a field of 45 V/ m), showing that electrical breakdown of EPD films is, at a minimum, within the range specified by the literature (30 – 120 V/ m, (15) ). That the film did not short-circuit over the area of the top electrode lends credence to the efficacy of multiple deposition/annealing cycles for reducing the density of cracks and pinholes. Ferroelectric properties measured from a hysteresis loop of 530 V amplitude are as follows: the coercive field Ec = 30.1 V/ m, and the remanent polarization Pr = 4.1