This work investigates the role of microstructure on radiation-induced changes to the functional response of ferroelectric thin films. Chemical solution-deposited lead zirconate titanate thin films with columnar and equiaxed grain morphologies are exposed to a range of gamma radiation doses up to 10 Mrad and the resulting trends in functional response degradation are quantified using a previously developed phenomenological model. The observed trends of global degradation as well as local rates of defect saturation suggest strong coupling between ferroelectric thin film microstructure and material radiation hardness. Radiation-induced degradation of domain wall motion is thought to be the major contributor to the reduction in ferroelectric response. Lower rates of defect saturation are noted in samples with columnar grains, due to increased grain boundary density offering more sites to act as defect sinks, thus reducing the interaction of defects with functional material volume within the grain interior. Response trends for measurements at low electric field show substantial degradation of polarization and piezoelectric properties (up to 80% reduction in remanent piezoelectric response), while such effects are largely diminished at increased electric fields, indicating that the defects created/activated are primarily of low pinning energy. The correlation of film microstructure to radiation-induced changes to the functional response of ferroelectric thin films can be leveraged to tune and tailor the eventual properties of devices relying on these materials.

Chemical solution deposition (CSD) of perovskite ferroelectric thin films has proven to be a cost-effective, large-area processing method for a variety of applications since it was first implemented by Budd et al. in the mid-1980s.1,2 The fine microstructural control inherent to CSD has led to efficient processing of ferroelectric materials that demonstrate exceptional multifunctional dielectric, polarization, and piezoelectric responses without the need for expensive and size-limiting substrates and/or processing equipment.3,4 Modification of CSD processing parameters can influence grain and domain morphology, chemical heterogeneity, defect concentrations, stress states, etc., and thus leads to a wide range of functional material responses. Accordingly, many CSD-processed ferroelectrics, especially lead zirconate titanate (PZT), have found use in ferroelectric random access memories (FeRAM), multilayer ceramic capacitors, infrared (IR) detectors, acoustic and ultrasound devices, and microelectromechanical systems (MEMS) sensors and actuators.5–12 However, new generations of novel microelectronics devices require operation in increasingly demanding environments, including radiation-hostile settings like space or facilities managing radioisotopes. While many studies on the microstructure of CSD ferroelectric thin films have centered around (multi-) functional response enhancement, the coupling between microstructural variations and radiation tolerance is still unclear. In this work, we investigate the effects of grain morphology on radiation tolerance in gamma-irradiated ferroelectric PZT thin films.

Many ferroelectric thin films, such as PbZr0.60Ti0.40O3, PbZr0.52Ti0.48O3, and BaTiO3 with columnar grain structure exhibit greater dielectric and electromechanical responses compared to similar compositions with equiaxed grains.8,13,14 Furthermore, changes to grain morphology can modify the energy landscape of the material, affecting the motion of internal interfaces, e.g., domain walls and phase boundaries.15–18 The nonlinear and hysteretic motion of these internal interfaces is largely responsible for the large magnitude of the functional response of ferroelectric thin films. Hence, any restriction to the motion of these interfaces can potentially result in degradation of the functional properties.15,19 Interactions of lattice, vacancy, and point defects with internal interfaces like domain walls, can pin domain wall motion and result in degradation of the ferroelectric response.17,18 Exposure to radiation can further aggravate these effects, resulting in additional dependencies of the functional response on microstructural and morphological factors.

Generally, the effects of irradiation on the functional response are strongly correlated with the grain size (and thus, grain boundary density) in the material. Greater grain boundary density potentially results in increased interaction of radiation-induced perturbations (electron-hole pairs, defect dipoles, and vacancies) with defective interfaces, thus yielding greater radiation-induced degradation (RID).20 However, grain boundaries can act as effective defect “sinks,” and counteract to a large degree the deleterious effects of RID in the material.21 In a ferroelectric material, radiation-induced ionization and displacement events can increase stable defect concentrations and cause changes in the defect energy landscape. Specifically, trapped charges generated by X-ray, gamma, and proton irradiations, as well as defect dipoles activated by neutron irradiation, have been shown to cause degradation of polarization, dielectric, and electromechanical responses.22–26 Work by Leray et al. suggested that irradiation of PZT thin films mimics and/or exacerbates the effects of ferroelectric aging and fatigue, and hypothesized that gamma radiation-induced space charges are likely proportional to grain boundary density across the thickness of interrogated samples.27 

While prior work on irradiated ferroelectric materials has shown considerable amounts of RID of functional properties, very little research has been undertaken to specifically correlate such variations in ferroelectric response to grain morphology and microstructure.24,28–34 Variations in grain morphology and microstructure will inevitably alter the fundamental interaction between radiation and the ferroelectric material. Modifying parameters such as grain size, orientation, and degree of anisotropy can potentially alter the effects of radiation interaction, defect creation and mobility, and even the extent of degradation that defects can impart. In this work, we investigate the role of grain morphology on radiation-induced changes to the functional response of ferroelectric materials by studying gamma-irradiated PZT thin films with columnar and equiaxed grain structures.

Two separate PbZr0.52Ti0.48O3 (PZT) precursor solutions were prepared, one via a 2-methoxyethanol-based (2-MOE) route, and the other using a methanol-based inverted mixing order (IMO) process.35–39 Both 2-MOE and IMO PZT solutions were deposited by spin coating on 150 mm-diameter 100-silicon wafers consisting of 100 nm Pt, 35 nm TiO2, 2035 nm SiO2, Si, and resulting in PZT films with thicknesses of 500 ± 14 nm.35,38,40 A seed layer of PbTiO3 was deposited for the 2-MOE films to induce 100-texture.35 2-MOE-PZT films were deposited with a target thickness of 500 nm using 0.4 M solutions, pyrolysis temperatures of 365 °C for 60 s for each spin layer, and a crystallization anneal at 700 °C for 60 s after every 2 layers in a rapid thermal anneal furnace. IMO-PZT films were deposited to a target thickness of 500 nm using 0.35 M solutions. IMO-PZT films were pyrolyzed at a temperature of 350 °C for 60 s for each spin layer and subsequently crystallized in a preheated furnace at 700 °C for 10 min; the crystallization step was repeated halfway through the film thickness and again at 500 nm. IrO2 top electrodes were selected for continuity with prior work.24 100 nm-thick electrodes were sputter-deposited onto both the 2-MOE- and IMO-films at 500 °C and processed with a post-deposition anneal at 650 °C in flowing O2 for 30 min. The top electrode and PZT layer were patterned using argon ion milling and a series of additional metallization steps to create interconnects to the device structures. This general process is outlined elsewhere.23 Figure 1 shows the baseline X-ray diffraction (XRD) phase analysis of the films deposited using 2-MOE- and IMO-prepared solutions, and the resulting columnar and equiaxed grain morphologies, respectively. Additionally, scanning electron microscopy (SEM) was performed to observe grain size and porosity in the films deposited with each precursor solution (Fig. 2).

FIG. 1.

X-ray diffraction (XRD) crystallographic phase analysis comparing representative samples with columnar and equiaxed grain structures that were prepared using 2-MOE and IMO PZT solutions, respectively. Noteworthy is the large 100-peak in samples with columnar grains, compared to the large presence of 110-texture in samples with equiaxed grains. “Signal” indicates artifacts of rapid increases to XRD signal intensity.

FIG. 1.

X-ray diffraction (XRD) crystallographic phase analysis comparing representative samples with columnar and equiaxed grain structures that were prepared using 2-MOE and IMO PZT solutions, respectively. Noteworthy is the large 100-peak in samples with columnar grains, compared to the large presence of 110-texture in samples with equiaxed grains. “Signal” indicates artifacts of rapid increases to XRD signal intensity.

Close modal
FIG. 2.

Scanning electron microscopy (SEM) images of samples with columnar and equiaxed grain structures. Notable is the greater degree of porosity in the samples with equiaxed grains (b) (see also Table III).

FIG. 2.

Scanning electron microscopy (SEM) images of samples with columnar and equiaxed grain structures. Notable is the greater degree of porosity in the samples with equiaxed grains (b) (see also Table III).

Close modal

The fabricated samples were exposed to radiation from a 60Co gamma source at doses ranging from 0.2 to 10 Mrad (equivalent Si dose) at a dose rate of approximately 600 rad(Si)/s at the U.S. Naval Research Laboratory (NRL). The geometry of the 60Co gamma source surrounds the sample, resulting in an isotropic exposure, and thus eliminating any effects of radiation directionality. All electrodes were left floating during radiation exposure. Dielectric, polarization, and piezoelectric responses of the samples were fully characterized both before and after irradiation, including (in order) measurements of low-field dielectric permittivity, polarization response, nonlinear AC dielectric response, DC electric field-dependent permittivity response, and DC electric field-dependent piezoelectric response. All measurements were performed on the same sample/electrode both before and after irradiation in order to accurately monitor changes in response. A summary of these measurements for samples with both columnar and equiaxed grain structures as a function of radiation dose is shown in Tables I and II. A 600-second poling step at 10 V (approximately five times the coercive voltage, VC) was performed directly before the piezoelectric measurements in both pre- and post-irradiation measurement sets in order to maximize polarization alignment in the out-of-plane direction. Low-field dielectric permittivity (εr) measurements were conducted at 100 mV and 1 kHz using an Agilent 4284 A precision LCR meter. Polarization-electric field (P-E) hysteresis loops were performed up to fields of 250 kV/cm at 100 Hz, using a Radiant P-PM2 ferroelectric test system. The nonlinear AC dielectric permittivity (εr-EAC) was measured up to approximately 150 kV/cm AC at 1 kHz. DC-dependent dielectric permittivity (εr-EDC) measurements were performed up to 250 kV/cm DC bias with an overlapping small-signal 100 mV AC bias at 1 kHz. Measurements of the converse, effective longitudinal piezoelectric response (d33,f) were performed on an aixACCT double beam laser interferometer (DBLI) measurement system up to 250 kV/cm DC bias with an overlapping AC signal VAC≈ 0.5Vc. All measurements reported are subject to experimental error up to 3%–5% due to sample variability.

TABLE I.

Measured dielectric permittivity, loss tangent, polarization, and effective longitudinal piezoelectric responses at increasing radiation doses for PZT thin films with columnar grain structures. Percent change (where applicable in the text) is calculated from measurements before and after the given radiation dose for the set of samples and electrodes exposed to that dose. Uncertainties expressed represent standard error to one significant figure. Measurement values are reported to the same decimal place as the uncertainties.45 

Columnar0 Mrad0.20.51.02.05.010.0
Low-field dielectric         
εr Virgin 1229 ± 1 1247 ± 3 1230 ± 1 1227 ± 2 1237 ± 1 1233 ± 1 1248 ± 1 
Irradiated 1222 ± 6 1297 ± 10 1229 ± 4 1208 ± 4 1188 ± 3 1127 ± 4 1083 ± 2 
tan(δ) (%) Virgin 1.3 ± 0.1 1.3 ± 0.1 1.3 ± 0.1 1.3 ± 0.1 1.5 ± 0.1 1.4 ± 0.1 1.2 ± 0.1 
Irradiated 1.6 ± 0.1 2.4 ± 0.2 1.5 ± 0.1 1.6 ± 0.1 1.3 ± 0.1 1.2 ± 0.1 1.0 ± 0.1 
Rayleigh analysis         
εinit Virgin 1219 ± 4 1263 ± 8 1230 ± 10 1253 ± 8 1223 ± 4 1221 ± 8 1237 ± 8 
Irradiated 1146 ± 3 1137 ± 6 1136 ± 9 1157 ± 6 1169 ± 10 1142 ± 10 1156 ± 27 
α (cm/kV) Virgin 26.9 ± 0.4 27.2 ± 0.4 36.8 ± 0.9 30.1 ± 0.4 34.7 ± 0.2 29 ± 2 29.1 ± 0.4 
Irradiated 29 ± 4 25.2 ± 0.3 23.6 ± 0.2 19 ± 1 20.1 ± 0.4 21 ± 4 15 ± 1 
α/εinit × 103 (cm/kV) Virgin 22.1 ± 0.3 21.5 ± 0.4 30 ± 1 24.1 ± 0.5 28.3 ± 0.1 24 ± 1 23.5 ± 0.5 
Irradiated 26 ± 4 22.2 ± 0.3 20.8 ± 0.3 16 ± 1 17 ± 1 18 ± 4 13 ± 1 
Polarization         
Psaturation (μC/cm2Virgin 35.4 ± 0.1 35.6 ± 0.1 35.7 ± 0.1 35.4 ± 0.1 35.3 ± 0.1 35.4 ± 0.1 35.5 ± 0.1 
Irradiated 35 ± 0.1 35 ± 0.2 35 ± 0.2 35 ± 0.1 35 ± 0.1 35 ± 0.1 35 ± 0.1 
Premanent (μC/cm2Virgin 12.2 ± 0.1 10.4 ± 0.2 10.7 ± 0.1 11.1 ± 0.2 11.1 ± 0.1 10.4 ± 0.1 11.1 ± 0.6 
Irradiated 11.2 ± 0.4 10.1 ± 0.3 9.8 ± 0.1 9.2 ± 0.1 9.6 ± 0.3 7.9 ± 0.1 8.3 ± 0.4 
εr-EDC         
% Diel. tunability Virgin 74.3 ± 0.2 74.7 ± 0.2 74.3 ± 0.1 74.5 ± 0.1 74.6 ± 0.2 74.2 ± 0.1 74.8 ± 0.1 
Irradiated 74.0 ± 0.4 74.2 ± 0.2 74.0 ± 0.3 73.1 ± 0.1 73.0 ± 0.2 71.8 ± 0.3 71.3 ± 0.3 
εDC,low-field Virgin 1402 ± 7 1354 ± 2 1400 ± 5 1359 ± 5 1390 ± 20 1400 ± 20 1382 ± 8 
Irradiated 1350 ± 20 1324 ± 2 1348 ± 2 1307 ± 1 1320 ± 20 1320 ± 20 1247 ± 1 
Piezoelectric         
d33,f,saturation (pm/V) Virgin 79 ± 3 76.2 ± 0.5 77 ± 5 78 ± 3 79 ± 1 79 ± 3 80.8 ± 0.6 
Irradiated 79 ± 2 76 ± 1 72 ± 2 77 ± 2 78 ± 1 76 ± 1 77 ± 2 
d33,f,remanent (pm/V) Virgin 49 ± 6 38 ± 2 43 ± 1 10 ± 1 43 ± 1 47 ± 5 41.5 ± 0.7 
Irradiated 34 ± 4 28 ± 1 25 ± 1 22 ± 1 23 ± 2 24 ± 6 7 ± 2 
Columnar0 Mrad0.20.51.02.05.010.0
Low-field dielectric         
εr Virgin 1229 ± 1 1247 ± 3 1230 ± 1 1227 ± 2 1237 ± 1 1233 ± 1 1248 ± 1 
Irradiated 1222 ± 6 1297 ± 10 1229 ± 4 1208 ± 4 1188 ± 3 1127 ± 4 1083 ± 2 
tan(δ) (%) Virgin 1.3 ± 0.1 1.3 ± 0.1 1.3 ± 0.1 1.3 ± 0.1 1.5 ± 0.1 1.4 ± 0.1 1.2 ± 0.1 
Irradiated 1.6 ± 0.1 2.4 ± 0.2 1.5 ± 0.1 1.6 ± 0.1 1.3 ± 0.1 1.2 ± 0.1 1.0 ± 0.1 
Rayleigh analysis         
εinit Virgin 1219 ± 4 1263 ± 8 1230 ± 10 1253 ± 8 1223 ± 4 1221 ± 8 1237 ± 8 
Irradiated 1146 ± 3 1137 ± 6 1136 ± 9 1157 ± 6 1169 ± 10 1142 ± 10 1156 ± 27 
α (cm/kV) Virgin 26.9 ± 0.4 27.2 ± 0.4 36.8 ± 0.9 30.1 ± 0.4 34.7 ± 0.2 29 ± 2 29.1 ± 0.4 
Irradiated 29 ± 4 25.2 ± 0.3 23.6 ± 0.2 19 ± 1 20.1 ± 0.4 21 ± 4 15 ± 1 
α/εinit × 103 (cm/kV) Virgin 22.1 ± 0.3 21.5 ± 0.4 30 ± 1 24.1 ± 0.5 28.3 ± 0.1 24 ± 1 23.5 ± 0.5 
Irradiated 26 ± 4 22.2 ± 0.3 20.8 ± 0.3 16 ± 1 17 ± 1 18 ± 4 13 ± 1 
Polarization         
Psaturation (μC/cm2Virgin 35.4 ± 0.1 35.6 ± 0.1 35.7 ± 0.1 35.4 ± 0.1 35.3 ± 0.1 35.4 ± 0.1 35.5 ± 0.1 
Irradiated 35 ± 0.1 35 ± 0.2 35 ± 0.2 35 ± 0.1 35 ± 0.1 35 ± 0.1 35 ± 0.1 
Premanent (μC/cm2Virgin 12.2 ± 0.1 10.4 ± 0.2 10.7 ± 0.1 11.1 ± 0.2 11.1 ± 0.1 10.4 ± 0.1 11.1 ± 0.6 
Irradiated 11.2 ± 0.4 10.1 ± 0.3 9.8 ± 0.1 9.2 ± 0.1 9.6 ± 0.3 7.9 ± 0.1 8.3 ± 0.4 
εr-EDC         
% Diel. tunability Virgin 74.3 ± 0.2 74.7 ± 0.2 74.3 ± 0.1 74.5 ± 0.1 74.6 ± 0.2 74.2 ± 0.1 74.8 ± 0.1 
Irradiated 74.0 ± 0.4 74.2 ± 0.2 74.0 ± 0.3 73.1 ± 0.1 73.0 ± 0.2 71.8 ± 0.3 71.3 ± 0.3 
εDC,low-field Virgin 1402 ± 7 1354 ± 2 1400 ± 5 1359 ± 5 1390 ± 20 1400 ± 20 1382 ± 8 
Irradiated 1350 ± 20 1324 ± 2 1348 ± 2 1307 ± 1 1320 ± 20 1320 ± 20 1247 ± 1 
Piezoelectric         
d33,f,saturation (pm/V) Virgin 79 ± 3 76.2 ± 0.5 77 ± 5 78 ± 3 79 ± 1 79 ± 3 80.8 ± 0.6 
Irradiated 79 ± 2 76 ± 1 72 ± 2 77 ± 2 78 ± 1 76 ± 1 77 ± 2 
d33,f,remanent (pm/V) Virgin 49 ± 6 38 ± 2 43 ± 1 10 ± 1 43 ± 1 47 ± 5 41.5 ± 0.7 
Irradiated 34 ± 4 28 ± 1 25 ± 1 22 ± 1 23 ± 2 24 ± 6 7 ± 2 
TABLE II.

Measured dielectric permittivity, loss tangent, polarization, and effective longitudinal piezoelectric responses at increasing radiation doses for PZT thin films with equiaxed grain structures. Percent change (where applicable in the text) is calculated from measurements before and after the given radiation dose for the set of samples and electrodes exposed to that dose. Uncertainties expressed represent standard error to one significant figure. Measurement values are reported to the same decimal place as the uncertainties.45 

Equiaxed0 Mrad0.20.51.02.05.010.0
Low-field dielectric         
εr Virgin 1405 ± 1 1481 ± 1 1366 ± 6 1512 ± 2 1445 ± 3 1454 ± 1 1496 ± 2 
Irradiated 1357 ± 4 1323 ± 2 1215 ± 6 1284 ± 4 1284 ± 5 1262 ± 6 1175 ± 2 
tan(δ) (%) Virgin 3.5 ± 0.1 3.3 ± 0.1 4.0 ± 0.1 3.8 ± 0.1 2.8 ± 0.1 2.5 ± 0.1 3.7 ± 0.1 
Irradiated 2.7 ± 0.1 1.7 ± 0.1 1.9 ± 0.1 1.5 ± 0.1 1.4 ± 0.1 1.3 ± 0.1 1.3 ± 0.1 
Rayleigh analysis         
εinit Virgin 1290 ± 40 1251 ± 3 1251 ± 5 1330 ± 10 1300 ± 3 1224 ± 3 1280 ± 10 
Irradiated 1200 ± 30 1161 ± 6 1100 ± 10 1138 ± 5 1263 ± 7 1227 ± 4 1112 ± 7 
α (cm/kV) Virgin 40 ± 2 57.7 ± 0.2 35.0 ± 0.5 51 ± 2 38.9 ± 0.1 52.2 ± 0.3 39 ± 1 
Irradiated 33 ± 1 37 ± 1 29 ± 1 34.5 ± 0.2 26.4 ± 0.2 23.2 ± 0.3 20.4 ± 0.4 
α/εinit × 103 (cm/kV) Virgin 30.9 ± 0.9 46.1 ± 0.3 27.9 ± 0.5 39 ± 2 29.9 ± 0.2 42.6 ± 0.3 30 ± 1 
Irradiated 28 ± 1 32 ± 1 26.1 ± 0.2 30.3 ± 0.3 20.9 ± 0.1 18.9 ± 0.2 18.4 ± 0.5 
Polarization         
Psaturation (μC/cm2Virgin 43.8 ± 0.1 43.3 ± 0.5 42.4 ± 0.1 41.9 ± 0.2 42.6 ± 0.3 41.7 ± 0.1 41.9 ± 0.1 
Irradiated 43 ± 1 42.5 ± 0.1 41.8 ± 0.2 41 ± 1 41 ± 1 40.9 ± 0.1 42.0 ± 0.1 
Premanent (μC/cm2Virgin 14.8 ± 0.2 17 ± 1 19.2 ± 0.1 16 ± 1 15.3 ± 0.1 16.6 ± 0.4 14.7 ± 0.3 
Irradiated 15.4 ± 0.4 16 ± 1 17.2 ± 0.3 15 ± 1 13.9 ± 0.3 13.5 ± 0.3 12.8 ± 0.5 
εr-EDC         
% Diel. tunability Virgin 77.9 ± 0.1 77.6 ± 0.1 77.4 ± 0.2 79.2 ± 0.1 78.1 ± 0.1 78.0 ± 0.1 79.5 ± 0.2 
Irradiated 77.9 ± 0.2 79.1 ± 0.1 76.7 ± 0.1 78.2 ± 0.3 77.4 ± 0.1 77.0 ± 0.2 76.4 ± 0.2 
εDC,low-field Virgin 1520 ± 9 1657 ± 8 1480 ± 20 1675 ± 6 1644 ± 5 1631 ± 4 1712 ± 1 
Irradiated 1443 ± 7 1489 ± 4 1360 ± 10 1440 ± 2 1455 ± 3 1410 ± 2 1297 ± 5 
Piezoelectric         
d33,f,saturation (pm/V) Virgin 66.8 ± 0.6 68.2 ± 0.7 67.7 ± 0.5 64 ± 4 60 ± 2 62.5 ± 0.4 64 ± 1 
Irradiated 68 ± 3 61 ± 3 63 ± 1 64 ± 1 59 ± 2 64 ± 1 64 ± 6 
d33,f,remanent (pm/V) Virgin 39 ± 2 76 ± 8 48 ± 3 57 ± 1 47 ± 2 54 ± 5 69 ± 6 
Irradiated 43 ± 3 48 ± 6 25 ± 3 33 ± 3 30 ± 4 30 ± 2 19 ± 5 
Equiaxed0 Mrad0.20.51.02.05.010.0
Low-field dielectric         
εr Virgin 1405 ± 1 1481 ± 1 1366 ± 6 1512 ± 2 1445 ± 3 1454 ± 1 1496 ± 2 
Irradiated 1357 ± 4 1323 ± 2 1215 ± 6 1284 ± 4 1284 ± 5 1262 ± 6 1175 ± 2 
tan(δ) (%) Virgin 3.5 ± 0.1 3.3 ± 0.1 4.0 ± 0.1 3.8 ± 0.1 2.8 ± 0.1 2.5 ± 0.1 3.7 ± 0.1 
Irradiated 2.7 ± 0.1 1.7 ± 0.1 1.9 ± 0.1 1.5 ± 0.1 1.4 ± 0.1 1.3 ± 0.1 1.3 ± 0.1 
Rayleigh analysis         
εinit Virgin 1290 ± 40 1251 ± 3 1251 ± 5 1330 ± 10 1300 ± 3 1224 ± 3 1280 ± 10 
Irradiated 1200 ± 30 1161 ± 6 1100 ± 10 1138 ± 5 1263 ± 7 1227 ± 4 1112 ± 7 
α (cm/kV) Virgin 40 ± 2 57.7 ± 0.2 35.0 ± 0.5 51 ± 2 38.9 ± 0.1 52.2 ± 0.3 39 ± 1 
Irradiated 33 ± 1 37 ± 1 29 ± 1 34.5 ± 0.2 26.4 ± 0.2 23.2 ± 0.3 20.4 ± 0.4 
α/εinit × 103 (cm/kV) Virgin 30.9 ± 0.9 46.1 ± 0.3 27.9 ± 0.5 39 ± 2 29.9 ± 0.2 42.6 ± 0.3 30 ± 1 
Irradiated 28 ± 1 32 ± 1 26.1 ± 0.2 30.3 ± 0.3 20.9 ± 0.1 18.9 ± 0.2 18.4 ± 0.5 
Polarization         
Psaturation (μC/cm2Virgin 43.8 ± 0.1 43.3 ± 0.5 42.4 ± 0.1 41.9 ± 0.2 42.6 ± 0.3 41.7 ± 0.1 41.9 ± 0.1 
Irradiated 43 ± 1 42.5 ± 0.1 41.8 ± 0.2 41 ± 1 41 ± 1 40.9 ± 0.1 42.0 ± 0.1 
Premanent (μC/cm2Virgin 14.8 ± 0.2 17 ± 1 19.2 ± 0.1 16 ± 1 15.3 ± 0.1 16.6 ± 0.4 14.7 ± 0.3 
Irradiated 15.4 ± 0.4 16 ± 1 17.2 ± 0.3 15 ± 1 13.9 ± 0.3 13.5 ± 0.3 12.8 ± 0.5 
εr-EDC         
% Diel. tunability Virgin 77.9 ± 0.1 77.6 ± 0.1 77.4 ± 0.2 79.2 ± 0.1 78.1 ± 0.1 78.0 ± 0.1 79.5 ± 0.2 
Irradiated 77.9 ± 0.2 79.1 ± 0.1 76.7 ± 0.1 78.2 ± 0.3 77.4 ± 0.1 77.0 ± 0.2 76.4 ± 0.2 
εDC,low-field Virgin 1520 ± 9 1657 ± 8 1480 ± 20 1675 ± 6 1644 ± 5 1631 ± 4 1712 ± 1 
Irradiated 1443 ± 7 1489 ± 4 1360 ± 10 1440 ± 2 1455 ± 3 1410 ± 2 1297 ± 5 
Piezoelectric         
d33,f,saturation (pm/V) Virgin 66.8 ± 0.6 68.2 ± 0.7 67.7 ± 0.5 64 ± 4 60 ± 2 62.5 ± 0.4 64 ± 1 
Irradiated 68 ± 3 61 ± 3 63 ± 1 64 ± 1 59 ± 2 64 ± 1 64 ± 6 
d33,f,remanent (pm/V) Virgin 39 ± 2 76 ± 8 48 ± 3 57 ± 1 47 ± 2 54 ± 5 69 ± 6 
Irradiated 43 ± 3 48 ± 6 25 ± 3 33 ± 3 30 ± 4 30 ± 2 19 ± 5 

The nonlinear dielectric response was analyzed through the Rayleigh approach to quantify changes in intrinsic and extrinsic dielectric response.41–43 This analysis yields the reversible Rayleigh parameter, εinit, which is the intercept of the linear AC field-dependent relative dielectric permittivity; and the irreversible Rayleigh parameter, α, which is the slope of the AC field-dependent relative permittivity. εinit describes intrinsic contributions to the dielectric response such as lattice vibration as well as low field, reversible extrinsic contributions, while α designates extrinsic contributions to the dielectric response, e.g., from irreversible domain wall and phase boundary motion. Accordingly, the ratio α/εinit is commonly employed as a measure of extrinsic contributions to the functional dielectric response.44 

Cross-sectional scanning electron microscopy (SEM) was used to visually observe the grain microstructure through the thickness of the films and measure porosity. Samples were sputter-coated with a 20 nm-thick Au layer to increase sample conductivity during SEM imaging. Cross sectional characterization was performed using an FEI Verios field-emission scanning electron microscope in secondary lens mode. Thickness measurements of distinct layers in the samples and porosity calculations were performed using ImageJ (Table III).

TABLE III.

Statistical data for measurements of in-plane grain size, grain height, and various geometry-related values for samples with columnar and equiaxed grain structures. Also calculated are the number of electron-hole pairs (ehp) per grain.

Grain structure% PorosityMean grain height (nm)Mean in-plane grain size (nm)Mean grain surface area (cm2)Mean grain volume (cm3)Surface area per volumeNumber of electron-hole pairs (ehp) per grain·Mrad
Columnar (cylindrical) 1.10 391 ± 220 66 ± 27 8.8 × 10−10 1.3 × 10−15 6.8 × 105 1.2 × 105 
Equiaxed (ellipsoidal) 1.35 215 ± 109 177 ± 160 11.3 × 10−10 3.6 × 10−15 3.1 × 105 3.4 × 105 
Grain structure% PorosityMean grain height (nm)Mean in-plane grain size (nm)Mean grain surface area (cm2)Mean grain volume (cm3)Surface area per volumeNumber of electron-hole pairs (ehp) per grain·Mrad
Columnar (cylindrical) 1.10 391 ± 220 66 ± 27 8.8 × 10−10 1.3 × 10−15 6.8 × 105 1.2 × 105 
Equiaxed (ellipsoidal) 1.35 215 ± 109 177 ± 160 11.3 × 10−10 3.6 × 10−15 3.1 × 105 3.4 × 105 

In order to quantify the microstructure and grain orientations in the film, transmission Kikuchi diffraction (TKD) was performed on the samples (Fig. 3). Focused ion beam (FIB) milling was used to prepare thin specimens for TKD, after which they were sputter-coated with a 20 nm-thick Au layer to increase sample conductivity and to mitigate drift during the data acquisition. A 3 μm-thick Pt layer was also deposited on the sample surface before FIB milling under FEI Quanta 3D FEG to protect the sample surface from potential electron and ion damage. Cross-sectional TKD samples were prepared using FEI Quanta 3D FEG which employs both electron and ion beam guns (SEM/FIB). TKD imaging was performed on the FIB samples using an Oxford Instruments NordlysNano Electron Backscatter Diffraction (EBSD) detector. Kikuchi diffraction patterns were acquired at 30 kV with the Oxford Instruments Aztec software. TKD data was processed using the HKL CHANNEL5 program Tango. Kikuchi images for the PZT were indexed to the cubic phase of the perovskite structure, which results in a description of the grain orientations in the PZT film. The PZT composition is in close proximity to a morphotropic phase boundary (MPB) and the small structural distortions from cubic cannot be reliably indexed.

FIG. 3.

Representative transmission Kikuchi diffraction (TKD) orientation maps and Kikuchi band contrast images for samples with (a) columnar and (b) equiaxed grain structures images from the x-, y-, and z-axes. (c) shows the inverse pole figures for the cubic indexing of both PZT and Pt, and the coordinate system used for this figure. Note that in the highly 100-textured columnar samples, the x- and y-axis TKD images are very similar, due to the 4-fold rotational symmetry of the cubic lattice used for indexing. In the samples with equiaxed grains and weak texture, the corresponding images from the x- and y-axes do not show a similar symmetry, confirming their more randomly oriented nature.

FIG. 3.

Representative transmission Kikuchi diffraction (TKD) orientation maps and Kikuchi band contrast images for samples with (a) columnar and (b) equiaxed grain structures images from the x-, y-, and z-axes. (c) shows the inverse pole figures for the cubic indexing of both PZT and Pt, and the coordinate system used for this figure. Note that in the highly 100-textured columnar samples, the x- and y-axis TKD images are very similar, due to the 4-fold rotational symmetry of the cubic lattice used for indexing. In the samples with equiaxed grains and weak texture, the corresponding images from the x- and y-axes do not show a similar symmetry, confirming their more randomly oriented nature.

Close modal

Fabrication of PZT thin films using 2-MOE- and IMO-CSD methods resulted in films with large differences in the grain orientation and microstructure. XRD crystallographic orientation analysis (Fig. 1) shows highly 100-textured films processed with 2-MOE-based CSD, while the IMO CSD-processed samples demonstrate a relatively strong 110-peak. Minor tetragonal peak splitting is observable in the samples processed with 2-MOE-based solutions, due to proximity of the precursor solutions to the morphotropic phase boundary and accompanying changes to the Curie temperature (TC). Differences in microstructure, porosity, and crystallographic texture are further elucidated via SEM imaging (Fig. 2) and TKD analysis (Fig. 3). SEM images show highly columnar grain structures with low porosity concentrated at the crystallization interfaces in the samples processed via 2-MOE-based solutions. The films processed with IMO solutions show randomly oriented grains with a greater degree of porosity (1.35% for equiaxed and 1.10% for columnar) distributed through the thickness of the film (Table III). TKD analysis further confirms the grain orientation and texture, showing large regions of 100-oriented columnar grains for the samples processed with 2-MOE-based CSD [Fig. 3(a)] compared with a more random distribution of crystallographic orientations for those processed via IMO CSD [Fig. 3(b)]. For simplicity throughout the results and discussion, the films processed via 2-MOE-based and IMO CSD methods will be referred to as “columnar” and “equiaxed,” respectively.

Shown in Fig. 4 are statistical measurements of the distribution of in-plane grain sizes, out-of-plane grain heights, and size of multi-grain regions of similar orientation, as calculated using ASTM E112–12 average grain intercept methods from multiple TKD band contrast images.46 Samples with columnar grains show a smaller mean in-plane grain size [Fig. 4(b)] and a larger mean grain height [Fig. 4(c)], with a greater frequency of multi-grain regions [Fig. 4(d)]. Equiaxed samples have a greater mean grain surface area, but due to their much greater mean volume, a smaller surface-area-to-volume ratio. Quantified statistical results are available in Table III.

FIG. 4.

Statistical analysis of grain characteristics for PZT thin films with columnar (blue) and equiaxed (orange) grains. Schematic for statistical measurements shown in (a). Measurements include (b) in-plane grain size, (c) out-of-plane grain height, and (d) size of multi-grain regions of similar orientation. The normal distribution curve is shown as a dotted line to suggest trends in data.

FIG. 4.

Statistical analysis of grain characteristics for PZT thin films with columnar (blue) and equiaxed (orange) grains. Schematic for statistical measurements shown in (a). Measurements include (b) in-plane grain size, (c) out-of-plane grain height, and (d) size of multi-grain regions of similar orientation. The normal distribution curve is shown as a dotted line to suggest trends in data.

Close modal

Figure 5 shows trends in the characterization of dielectric response at the low AC electric field (εr) and the intermediate AC electric field (α). Samples with equiaxed grains show greater degradation of relative dielectric permittivity, εr, as a function of radiation dose, as well as greater reductions in dielectric losses, tan(δ) (Tables I and II). Similarly, at increasing AC field, degradation of the extrinsic contributions to the dielectric response, α, is slightly greater for samples with equiaxed grains (Tables I and II, Fig. 5), potentially indicating that contributions from irreversible motion of internal interfaces are greater in those samples than in samples with columnar grains.

FIG. 5.

Degradation trends for dielectric responses at low [(a) and (b)] and increasing (c) AC electric field. Notable are the large differences in degradation trends at low field when comparing samples with different grain structures, but similar trends of degradation of the Rayleigh extrinsic to intrinsic ratio are observed for both sets of samples.

FIG. 5.

Degradation trends for dielectric responses at low [(a) and (b)] and increasing (c) AC electric field. Notable are the large differences in degradation trends at low field when comparing samples with different grain structures, but similar trends of degradation of the Rayleigh extrinsic to intrinsic ratio are observed for both sets of samples.

Close modal

The degradation data for remanent polarization response and coercive field (negative potential) are plotted in Fig. 6 [with sample loops shown in Figs. 7(a) and 7(b)]. Samples with columnar grains show more severe degradation for both remanent and saturated polarization (Prem and Psat, respectively) compared to samples with equiaxed grains. Decrease in the negative coercive field (EC-) is closely matched for both sets of samples (Tables I and II).

FIG. 6.

Degradation trends for polarization and coercive field. Note the severe degradation of remanent polarization, compared to the mild degradation of saturation polarization.

FIG. 6.

Degradation trends for polarization and coercive field. Note the severe degradation of remanent polarization, compared to the mild degradation of saturation polarization.

Close modal
FIG. 7.

Comparison of (a) and (b) polarization-field (P-E) hysteresis loops; (c) and (d) permittivity-DC field (εr-EDC); and (e) and (f) piezoelectric-field loops (d33,f-EDC) for samples with (a), (c), and (e) columnar grains and (b), (d), and (f) equiaxed grains in virgin (0 Mrad) samples and after exposure to 10 Mrad gamma irradiation. Notable is the larger shift in the positive direction of the P-E loop at 10 Mrad for the sample with columnar grains (a), as well as slight pinching of the P-E loops for both samples (a) and (b), indicating changes to the defect energy landscape of the material. The internal bias is also visible in piezoelectric plots (e) and (f). The formation of new peaks in the εr-EDC loops (c) and (d) potentially indicates changes to the defect energy landscape.

FIG. 7.

Comparison of (a) and (b) polarization-field (P-E) hysteresis loops; (c) and (d) permittivity-DC field (εr-EDC); and (e) and (f) piezoelectric-field loops (d33,f-EDC) for samples with (a), (c), and (e) columnar grains and (b), (d), and (f) equiaxed grains in virgin (0 Mrad) samples and after exposure to 10 Mrad gamma irradiation. Notable is the larger shift in the positive direction of the P-E loop at 10 Mrad for the sample with columnar grains (a), as well as slight pinching of the P-E loops for both samples (a) and (b), indicating changes to the defect energy landscape of the material. The internal bias is also visible in piezoelectric plots (e) and (f). The formation of new peaks in the εr-EDC loops (c) and (d) potentially indicates changes to the defect energy landscape.

Close modal

The summary of functional response characterization shown in Tables I and II allows for a variety of observations regarding the behavior of samples with both columnar and equiaxed grains as a function of radiation dose, particularly from a high-dose perspective. For example, focusing on dielectric response, PZT thin films with columnar and equiaxed grain structures show relatively low levels of degradation of low-field dielectric permittivity (εr) at 10 Mrad dose (13% and 22%, respectively), compared to much greater degradation of remanent effective piezoelectric coefficient (d33,f,rem) – up to 83% for samples with columnar grains, and 70% for samples with equiaxed grains (see Fig. 8). Additionally, large degradation (up to 50%) of α is noted for all samples at 10 Mrad. This suggests that gamma radiation negatively interacts with sources of extrinsic contributions to the dielectric response, such as the irreversible motion of domain walls. This result is consistent with prior studies at discrete gamma radiation doses.24 In order to more comprehensively quantify radiation-induced defect interactions and the resulting effects on the functional response in PZT thin films of different grain morphologies, we apply a phenomenological model developed in prior work to analyze trends as a function of radiation dose.47 The implementation thereof will be addressed in the following Discussion.

FIG. 8.

Degradation trends and fitted model for DC field-dependent piezoelectric responses (d33,f-EDC) and percent dielectric tunability. Notably, samples with columnar grains appear to be slightly more susceptible to radiation-induced degradation of DC field-dependent responses. However, the magnitude of degradation of d33,f,sat and % tunability (high DC field) is somewhat negligible, compared to d33,f,rem (low DC field).

FIG. 8.

Degradation trends and fitted model for DC field-dependent piezoelectric responses (d33,f-EDC) and percent dielectric tunability. Notably, samples with columnar grains appear to be slightly more susceptible to radiation-induced degradation of DC field-dependent responses. However, the magnitude of degradation of d33,f,sat and % tunability (high DC field) is somewhat negligible, compared to d33,f,rem (low DC field).

Close modal

The employed phenomenological model relies on the assumption that exposure of ferroelectric materials to gamma radiation creates or activates defects in the material.24 We note that radiation-induced defects (RIDs) may refer to both ionic/electronic defects, such as electron-hole pairs and trapped charges, or atomic displacements, e.g., vacancy/interstitial pairs, with the former being the more likely result of ionizing gamma irradiation. The newly created/activated defects interact with a given volume of ferroelectric material and, potentially, with existing defects therein, such as domain walls, grain boundaries, existing point defects, defect dipoles, etc., thereby modifying the defect-energy landscape of the material. The result is an effective “pinning” of the ferroelectric volume (normalized to the total probed volume)47 due to defect interactions, Vd, which can be directly correlated to the degradation of functional response parameters. The volume of the pinned ferroelectric material is related to the number of radiation-induced defects, N, (assumed proportional to the radiation dose) by the expression

Vd=1eφNN1k1k,
(1)

where two parameters determine the response to radiation: (1) φN, the normalized effective change in material volume pinned per new defect introduced; and (2) k, the effective rate of defect saturation. φN can be considered as the global susceptibility of the material to radiation-induced change, while k is useful for describing the nonlinear physical phenomena governing the effective rate of change and the more nuanced effects of defect concentration on degradation of response. Full details of the model are available elsewhere, including derivation and detailed discussion of the assumptions and parameters.47 It should be noted that the values of φN are normalized to an arbitrary total volume, thus comparisons between values of φN represent fractional comparisons, rather than the absolute value. Results of fitting the model to the functional characterization data (Table I) are given in Table IV.

TABLE IV.

Extracted parameters from applying the phenomenological model developed in prior work to TID study data of dielectric, polarization, and piezoelectric responses of the irradiated samples with columnar and equiaxed grains. Note that values of φN are multiplied by three orders of magnitude to make interpretation more manageable.

ColumnarEquiaxed
φN × 103kφN × 103k
Low-field dielectric     
εr 19 0.48 26 0.79 
tan(δ) 25 0.77 57 0.92 
Rayleigh analysis     
Α 85 0.76 96 0.77 
εinit 11 0.76 19 0.79 
α/εinit 86 0.68 97 0.68 
Polarization     
Prem 140 0.47 52 0.64 
Psat 0.65 0.85 
EC- 173 0.59 153 0.70 
εr-EDC response     
% Tunability 0.43 0.44 
εDC,low-field 15 0.63 34 0.72 
Piezoelectric     
d33,f,rem 185 0.66 104 0.82 
d33,f,sat 0.61 0.62 
ColumnarEquiaxed
φN × 103kφN × 103k
Low-field dielectric     
εr 19 0.48 26 0.79 
tan(δ) 25 0.77 57 0.92 
Rayleigh analysis     
Α 85 0.76 96 0.77 
εinit 11 0.76 19 0.79 
α/εinit 86 0.68 97 0.68 
Polarization     
Prem 140 0.47 52 0.64 
Psat 0.65 0.85 
EC- 173 0.59 153 0.70 
εr-EDC response     
% Tunability 0.43 0.44 
εDC,low-field 15 0.63 34 0.72 
Piezoelectric     
d33,f,rem 185 0.66 104 0.82 
d33,f,sat 0.61 0.62 

By comparing the fitting parameters in Table IV, we can now perform a more quantitative analysis of both the global and local response of PZT thin films and the effect of grain morphology on gamma radiation-induced defect interactions. Generally, gamma irradiation of PZT thin films with both columnar and equiaxed grains resulted in the degradation of dielectric, polarization, and piezoelectric functional properties. However, several notable trends are observed regarding the dependence of radiation-induced degradation on the grain structure, magnitude/type of electric field, and the type of functional response measured.

First and foremost, samples with equiaxed grains exhibit consistently greater values of k throughout all measurements in Table IV. The distinct trend of larger k across all functional response parameters in equiaxed-grained samples suggests a strong correlation between grain morphology and defect saturation rates in gamma irradiated PZT thin films. The coefficient k accounts for the effects of material anisotropies, e.g., grain boundaries, pre-existing defects, chemical heterogeneity, etc. Elevated values of k potentially signal either enhanced rates of defect creation/activation, increased damage by newly introduced defects, or a combination of both. A variety of reports have suggested that the harmful effects of radiation-induced defects are potentially alleviated by decreasing grain size in functional materials, and thus increasing the density of grain boundary sinks for defect accumulation.21,48–52 Defects either created or accumulated at grain boundaries potentially result in less severe and shorter-range degradation, due to the relative absence of functional material volume at grain boundaries. The columnar grains studied in this work are both smaller than the equiaxed grains studied and have a surface-area-to-volume ratio that is more than twice as large (Table IV). The result is a slower rate of defect saturation and resulting degradation of response in samples with columnar grains, signaled by consistently smaller values of k compared to samples with equiaxed grains.

Trends of global susceptibility (φN) to radiation-induced degradation for both columnar- and equiaxed-grained samples are observed through measurements of dielectric properties. Focusing on the Rayleigh analysis of nonlinear AC dielectric response, values of φN are similar for measurements of εinit (φN = 11 × 10−3 for columnar samples, φN= 19 × 10−3 for equiaxed samples) and α (φN = 85 × 10−3 for columnar samples, φN = 97 × 10−3) for samples with both grain orientations. Two important observations can be gleaned from this data. First, the relative values of φN for α are four to five times greater than the corresponding values for εinit (see also the ratio, α/εinit). This result indicates that the impact of radiation on the irreversible motion of domain walls is larger than the effects on intrinsic response and reversible motion of domain walls, consistent with multiple reports on gamma, X-ray, proton, and neutron irradiation of PZT thin films.22,24,25 Prior work has suggested that ionizing radiation can excite electrons and holes, leading to potential trapped charges in the material.22,24,27 Such trapped charges can modify the charge state of existing defects and lead to pinning of domain wall motion, similar to mechanisms driving ferroelectric fatigue.24,27,53–56 Proie et al. and Brewer et al. have previously reported the nature of charge accumulation due to gamma irradiation and the subsequent creation of charged defects through TID and bias studies in PZT thin films.23,24

Second, the values of φN for both α and εinit are slightly greater for samples with equiaxed grains. It is noteworthy that the grain volume in samples with the equiaxed structure is larger than columnar grains, as noted in Table III, as well by visual inspection of the TKD band contrast maps in Fig. 3. Numerous reports have shown that domain size (and thus, wall mobility) in ferroelectric materials scales with grain size.57,58 In grains with larger domains, domain walls are expected to be more mobile, due to lower chances of interaction with other domain walls, potentially resulting in greater extrinsic contributions to the dielectric response. Indeed, comparing the actual values of α, in Tables I and II, the samples with equiaxed grains consistently show greater magnitude of extrinsic contributions to dielectric response across the entire range of radiation doses. In other words, samples with larger, equiaxed grains have higher extrinsic contributions in virgin samples. On the other hand, in samples with smaller columnar grains, the smaller domains and proximity to grain boundaries may result in to lower extrinsic contributions, even prior to irradiation.59 A higher value of α in the equiaxed samples might also be attributable to greater chemical homogeneity in those samples. Prior research compared cation distributions through the thickness of films prepared with 2-MOE-based and IMO PZT solutions, showing greater chemical homogeneity for IMO-based solutions prepared under conditions typically used for PZT thin film preparation.60 Additionally, it was demonstrated that films with increased chemical homogeneity can produce enhanced dielectric properties due to their consistent proximity to the morphotropic phase boundary (MPB), which is associated with enhanced extrinsic contributions to the dielectric response, e.g., greater domain wall mobility.61 The consequence is greater observed radiation-induced degradation of response in equiaxed grains as the larger, more mobile domains are pinned and result in greater percent degradation of response, while maintaining larger measured values of extrinsic response. This phenomenon is potentially supported by the previously mentioned greater values of k in samples with equiaxed grains, as defects are created/activated at an increased rate relative to columnar-grained samples.

It is also useful to compare the values of φN for polarization response at low and high field, i.e., remanent and saturated polarization response (Prem and Psat). At a low electric field, the effects of gamma radiation are quite severe for Prem, with φN values of 140 × 10−3 and 52 × 10−3 for columnar- and equiaxed-grained samples, respectively. However, at the higher fields (250 kV/cm) required to reach polarization saturation, the effects of radiation-induced defect interactions are substantially reduced (values of φN for saturated polarization of columnar and equiaxed samples are 6 × 10−3 and 2 × 10−3, respectively). These results suggest that the defects generated or activated with gamma irradiation are of relatively low energy, and the effects of their presence are almost fully overcome at higher electric fields.

As shown in Figs. 7(a) and 7(b), internal bias and the associated horizontal shift of the polarization-field (P-E) loops in the positive electric field direction are potentially responsible for some degree of degradation of remanent polarization response. However, reductions in the magnitude of negative coercive field, EC-, are very similar for both sets of samples [Fig. 6(b)], indicating that internal bias for samples with columnar grains may not be the sole contributor to the greater observed rates of degradation. Indeed, pinching of the hysteresis loops for both sets of samples is visible [Figs. 7(a) and 7(b)], consistent with work on single-dose radiation studies, and indicating the presence of charged defects and/or defect dipoles, and potential pinning of internal interfaces.24,25 Such pinching, coupled with the observed asymmetric internal bias, makes exact interpretation of the Prem and EC trends difficult, as deconvolution of these effects is not trivial.

Meanwhile, piezoelectric and dielectric measurements as a function of DC electric field show severe degradation at low-fields (d33,f,rem, φN = 185 × 10−3 for columnar, φN = 103 × 10−3 for equiaxed) and the relative abatement thereof at high-field measurements (d33,f,sat, φN = 7 × 10−3 for columnar, φN = −1 × 10−3 for equiaxed; and % dielectric tunability) (Table IV and Fig. 8). Li et al.62 and Bassiri-Gharb et al.41 have previously reported that increasing applied DC electric field significantly raises the threshold at which nonlinear domain wall motion is initiated. At the low DC electric field used for d33,f,rem measurements, domain walls are largely mobile, and the previously discussed radiation-induced reduction of domain wall motion dominates the large degradation observed for both sets of samples. At a substantially higher DC electric field (such as those used to measure d33,f,sat and percent dielectric tunability), domain wall motion is essentially restricted, and intrinsic factors, such as lattice strain and crystal anisotropy dominate the piezoelectric response. Prior work has shown that radiation-induced damage to the crystal lattice in gamma-irradiated PZT is virtually nonexistent, even at gamma radiation doses up to 10 Mrad.24 Accordingly, extremely mild trends of degradation are observed for measurements at a high DC electric field, reflected in the essentially negligible values of φN, and flat degradation trends as shown in Fig. 8.

Additionally, we note that the values of φN for remanent and saturated piezoelectric responses (d33,f,rem and d33,f,sat) are somewhat similar to those observed for Prem and Psat, respectively, for both set of samples. This similarity is consistent with the fact that the piezoelectric response often scales with the degree of coherent alignment of the polarization direction in the sample. In fact, piezoelectric response is often closely correlated to the polarization in ferroelectric materials, especially with decreasing film thickness below 1.5 μm.63 Consequently, the internal electrical bias and changes in the defect energy landscape causing pinching of the P-E loops are expected to also be present in the piezoelectric and DC field-dependent dielectric responses. Indeed, the presence of internal bias is noted in the piezoelectric response plots [Figs. 7(e) and 7(f)], as well as notable reductions in the irradiated positive remanent piezoelectric response (d33,f,rem+), indicating that defects created/activated prevent the alignment of dipoles to some degree.

A final factor to consider is the effect of defect creation at locations where gamma rays interact with grain boundaries. Work by Claeys and Simoen suggested that in ferroelectric materials exposed to ionizing radiation, rates of trapped charge creation and activation are highly dependent on the surface area of grain boundaries with which the radiation interacts.20 They further noted that such trapped charges at grain surfaces can screen local polarization.20 Leray et al. also suggested that potential trapped charges in gamma-irradiated PZT are activated proportionally to the exposed grain boundary area.27 The grain boundary density is greater in samples with columnar grains (Table III), and thus, incident gamma rays may create more trapped charges at grain boundary locations. While the previously discussed effects of grain boundaries, acting as defect sinks, likely neutralize these trapped charges and any extensive domain wall-pinning at those locations, the local charges potentially contribute further to the observed internal bias and previously discussed convolution of polarization/piezoelectric response trends. These phenomena are observed by slightly greater RID of Prem and d33,f,rem in columnar-grained samples [Figs. 6(a), 7(e), 7(f), and 8].

PZT thin films processed with 2-MOE-based and methanol-based IMO solutions were fabricated resulting in films with columnar and equiaxed grain structures, respectively, and irradiated with 60Co gamma radiation from 0.2 to 10 Mrad(Si). Degradation trend data was fitted using a modified sigmoid function to quantify defect interactions in irradiated ferroelectrics, and the resulting values describing (1) the ferroelectric volume pinned per new defect, and (2) the effective rate of defect saturation, were compared. For films with both types of grain structures, significant amounts of degradation were observed for dielectric, ferroelectric, and piezoelectric properties. Compared to samples with columnar grains, those with equiaxed grains consistently showed greater rates of defect saturation and nonlinear degradation of response as a function of radiation dose, due to their larger grains and relatively lower density of grain boundaries acting as defect sinks, the presence of which can reduce the nonlinear rates of defect saturation.

Trends in the Rayleigh analysis indicate that pinning of domain walls is the dominant mode of dielectric response degradation. This effect is marginally greater in samples with equiaxed grains than those with columnar grains, potentially due to the larger grain size and correspondingly larger domains and more labile domain walls in the former. Greater initial dielectric response is quickly degraded in such samples, and greater percent degradation of extrinsic contributions to the dielectric response is thus observed. On the other hand, samples with columnar grains show slightly increased degradation of both polarization and piezoelectric responses at the high radiation doses studied here. The pinning energy of the domain walls as a result of irradiation is, however, somewhat limited to lower energies: in fact, low-field measurements of the dielectric, ferroelectric, and piezoelectric response show substantial degradation, while high-field responses remain essentially unaffected by irradiation even at 10 Mrad.

The results presented in this study can be leveraged for the design of radiation-tolerant devices that rely on ferroelectric thin films, such as for use in satellite or nuclear energy applications. The distinct trends in functional response degradation are strongly correlated to the microstructure of the PZT thin films, and can be used to tailor eventual device properties to meet the needs of a variety of radiation-tolerant applications. Furthermore, the quantification of the effects of radiation dose allows for a nuanced assessment of material performance across device exposure and lifetime.

This work was supported by the Defense Threat Reduction Agency, Basic Research Award No. HDTRA1-15-1-0035 to the Georgia Institute of Technology. The contents do not necessarily reflect the position or the policy of the federal government, and no official endorsement should be inferred. IMO film synthesis was supported by the Laboratory Directed Research and Development program at Sandia National Laboratories, a multi-mission laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000. The authors acknowledge the contributions of Joel Martin and Brian Power of the Army Research Laboratory and Steven Isaacson of General Technical Services for their roles in the fabrication of the PZT thin films. This work was performed in part at the Analytical Instrumentation Facility (AIF) at North Carolina State University, which is supported by the State of North Carolina and the National Science Foundation (award No. ECCS-1542015). The AIF is a member of the North Carolina Research Triangle Nanotechnology Network (RTNN), a site in the National Nanotechnology Coordinated Infrastructure (NNCI).

1.
S. K.
Dey
,
K. D.
Budd
, and
D. A.
Payne
,
IEEE Trans. Ultrason. Ferroelectr. Freq. Control
35
,
80
(
1988
).
2.
K. D.
Budd
,
S. K.
Dey
, and
D. A.
Payne
,
Br. Ceram. Proc.
36
,
107
(
1985
).
3.
S.
Trolier-McKinstry
and
P.
Muralt
,
J. Electroceram.
12
,
7
(
2004
).
4.
N.
Setter
,
D.
Damjanovic
,
L.
Eng
,
G.
Fox
,
S.
Gevorgian
,
S.
Hong
,
A.
Kingon
,
H.
Kohlstedt
,
N.
Park
,
G.
Stephenson
,
I.
Stolitchnov
,
A.
Taganstev
,
D.
Taylor
,
T.
Yamada
, and
S.
Streiffer
,
J. Appl. Phys.
100
,
051606
(
2006
).
5.
C.
Chang
and
C.
Tang
,
Sens. Actuators, A
65
,
171
(
1998
).
6.
R.
Schwartz
,
T.
Schneller
, and
R.
Waser
,
C. R. Chim.
7
,
433
(
2004
).
7.
T.
Otsuki
and
K.
Arita
,
Integrated Ferroelectr.
17
,
31
(
1997
).
8.
R.
Schwartz
,
P.
Clem
,
J.
Voigt
,
E.
Byhoff
,
M.
Van Stry
,
T.
Headley
, and
N.
Missert
,
J. Am. Ceram. Soc.
82
,
2359
(
1999
).
9.
R.
Schwartz
,
T.
Reichert
,
P.
Clem
,
D.
Dimos
, and
D.
Liu
,
Integrated Ferroelectr.
18
,
275
(
1997
).
10.
P.
Clem
,
B.
Tuttle
,
J.
Ruffner
,
C.
Brinker
,
R.
Schwartz
,
M.
Rodriguez
,
W.
Warren
,
R.
Jones
,
S.
Summerfelt
, and
I.
Yoo
,
Mater. Res. Soc.
541
,
661
(
1999
).
11.
G.
Brennecka
,
J.
Ihlefeld
,
J.
Maria
,
B.
Tuttle
, and
P.
Clem
,
J. Am. Ceram. Soc.
93
,
3935
(
2010
).
12.
G.
Brennecka
,
C.
Parish
,
B.
Tuttle
, and
L.
Brewer
,
J. Mater. Res.
23
,
176
(
2008
).
13.
C.
Park
,
J.
Lee
,
S.
Lee
,
S.
Jun
, and
H.
Kim
,
J. Electroceram.
25
,
20
(
2010
).
14.
Z.
Song
and
C.
Lin
,
Appl. Surf. Sci.
158
,
21
(
2000
).
15.
D.
Damjanovic
,
Rep. Prog. Phys.
61
,
1267
(
1998
).
16.
T. M.
Shaw
,
S.
Trolier-McKinstry
, and
P. C.
McIntyre
,
Annu. Rev. Mater. Sci.
30
,
263
(
2000
).
17.
M.
Dawber
,
K. M.
Rabe
, and
J. F.
Scott
,
Rev. Mod. Phys.
77
,
1083
(
2005
).
18.
M.
Stengel
,
D.
Vanderbilt
, and
N. A.
Spaldin
,
Nat. Mater.
8
,
392
(
2009
).
19.
20.
C. L.
Claeys
and
E.
Simoen
,
Radiation Effects in Advanced Semiconductor Materials and Devices
(
Springer
,
Berlin; New York
,
2002
).
21.
T.
Shen
,
Nucl. Instrum. Methods Phys. Res. Sect. B
266
,
921
(
2008
).
22.
Y.
Bastani
,
A. Y.
Cortes-Pena
,
A. D.
Wilson
,
S.
Gerardin
,
M.
Bagatin
,
A.
Paccagnella
, and
N.
Bassiri-Gharb
,
Appl. Phys. Lett.
102
,
192906
(
2013
).
23.
R. M.
Proie
,
R. G.
Polcawich
,
C. D.
Cress
,
L. M.
Sanchez
,
A. D.
Grobicki
,
J. S.
Pulskamp
, and
N. J. H.
Roche
,
IEEE Trans. Nucl. Sci.
60
,
4505
(
2013
).
24.
S.
Brewer
,
C.
Deng
,
C.
Callaway
,
M.
Paul
,
K.
Fisher
,
J.
Guerrier
,
R.
Rudy
,
R.
Polcawich
,
J.
Jones
,
E.
Glaser
,
C.
Cress
, and
N.
Bassiri-Gharb
,
J. Appl. Phys.
120
,
024101
(
2016
).
25.
J. T.
Graham
,
G. L.
Brennecka
,
P.
Ferreira
,
L.
Small
,
D.
Duquette
,
C.
Apblett
,
S.
Landsberger
, and
J. F.
Ihlefeld
,
J. Appl. Phys.
113
,
124104
(
2013
).
26.
A.
Henriques
,
J. T.
Graham
,
S.
Landsberger
,
J. F.
Ihlefeld
,
G. L.
Brennecka
,
D. W.
Brown
,
J. S.
Forrester
, and
J. L.
Jones
,
AIP Adv.
4
,
117125
(
2014
).
27.
J.
Leray
,
O.
Musseau
,
P.
Paillet
,
J.
Autran
,
D.
Sodi
, and
Y.
Coic
,
J. Phys. III France
7
,
1227
(
1997
).
28.
D. D.
Glower
,
D. L.
Hester
, and
D. F.
Warnke
,
J. Am. Ceram. Soc.
48
,
417
(
1965
).
29.
J.
Gao
,
L.
Zheng
,
X.
Duo
,
J.
Huang
,
L.
Yang
,
C.
Lin
, and
R.
Yan
,
Thin Solid Films
340
,
132
(
1999
).
30.
J.
Gao
,
L.
Zheng
,
Z.
Song
,
L.
Wang
,
L.
Yang
,
D.
Zhu
, and
C.
Lin
,
Philos. Mag. Part B
79
,
829
(
1999
).
31.
S. A.
Yang
,
B. H.
Kim
,
M. K.
Lee
,
G. J.
Lee
,
N.-H.
Lee
, and
S. D.
Bu
,
Thin Solid Films
562
,
185
(
2014
).
32.
I.
Baturin
,
N.
Menou
,
V.
Shur
,
C.
Muller
,
D.
Kuznetsov
,
J. L.
Hodeau
, and
A.
Sternberg
,
Mater. Sci. Eng.: B
120
,
141
(
2005
).
33.
N.
Menou
,
A.-M.
Castagnos
,
C.
Muller
,
D.
Goguenheim
,
L.
Goux
,
D. J.
Wouters
,
J.-L.
Hodeau
,
E.
Dooryhee
, and
R.
Barrett
,
J. Appl. Phys.
97
,
044106
(
2005
).
34.
J. F.
Scott
,
C. A.
Araujo
,
H. B.
Meadows
,
L. D.
McMillan
, and
A.
Shawabkeh
,
J. Appl. Phys.
66
,
1444
(
1989
).
35.
L. M.
Sanchez
,
D. M.
Potrepka
,
G. R.
Fox
,
I.
Takeuchi
,
K.
Wang
,
L. A.
Bendersky
, and
R. G.
Polcawich
,
J. Mater. Res.
28
,
1920
(
2013
).
36.
K. D.
Budd
and
D. A.
Payne
,
Inst. Phys. Conf. Ser.
103
,
13
(
1989
).
37.
S.
Mhin
,
K.
Nittala
,
J.
Lee
,
D.
Robinson
,
J.
Ihlefeld
,
G.
Brennecka
,
L.
Sanchez
,
R.
Polcawich
, and
J.
Jones
,
J. Am. Ceram. Soc.
97
,
2973
(
2014
).
38.
R.
Schwartz
,
Chem. Mater.
9
,
2325
(
1997
).
39.
R. A.
Assink
and
R. W.
Schwartz
,
Chem. Mater.
5
,
511
(
1993
).
40.
D. M.
Potrepka
,
G. R.
Fox
,
L. M.
Sanchez
, and
R. G.
Polcawich
,
Mater. Res. Soc. Symp. Proc.
1299
(
2011
).
41.
N.
Bassiri-Gharb
,
I.
Fujii
,
E.
Hong
,
S.
Trolier-McKinstry
,
D. V.
Taylor
, and
D.
Damjanovic
,
J. Electroceram.
19
,
49
(
2007
).
42.
D.
Damjanovic
and
M.
Demartin
,
J. Phys. D: Appl. Phys.
29
,
2057
(
1996
).
43.
R.
Eitel
,
T.
Shrout
, and
C.
Randall
,
J. Appl. Phys.
99
,
124110
(
2006
).
44.
S. P.
Li
,
A. S.
Bhalla
,
R. E.
Newnham
, and
L. E.
Cross
,
Mater. Lett.
17
,
21
(
1993
).
45.
J. R.
Taylor
,
An Introduction to Error Analysis: The Study of Uncertainties in Physical Measurements
, 2nd ed. (
University Science Books
,
Sausalito, California
,
1997
).
46.
ASTM. in Designation: E112 − 12 (
2013
).
47.
S. J.
Brewer
,
C. D.
Cress
,
S. C.
Williams
,
H.
Zhou
,
M.
Rivas
,
R. Q.
Rudy
,
R. G.
Polcawich
,
E. R.
Glaser
,
J. L.
Jones
, and
N.
Bassiri-Gharb
,
Sci. Rep.
7
,
5071
(
2017
).
49.
50.
S.
Dey
,
J.
Mardinly
,
Y.
Wang
,
J.
Valdez
,
T.
Holesinger
,
B.
Uberuaga
,
J.
Ditto
,
J.
Drazin
, and
R.
Castro
,
Phys. Chem. Chem. Phys.
18
,
16921
(
2016
).
51.
B. N.
Singh
and
A. J. E.
Foreman
,
Philos. Mag.
29
,
847
(
1974
).
52.
I.
Ovid'ko
and
A.
Sheinerman
,
Appl. Phys. A
81
,
1083
(
2005
).
53.
A.
Tagantsev
,
I.
Stolichnov
,
E.
Colla
, and
N.
Setter
,
J. Appl. Phys.
90
,
1387
(
2001
).
54.
T.
Mihara
,
H.
Watanabe
, and
C.
de Araujo
,
Jpn. J. Appl. Phys., Part 1
33
,
3996
(
1994
).
55.
J.
Scott
,
C.
Araujo
,
B.
Melnick
,
L.
McMillan
, and
R.
Zuleeg
,
J. Appl. Phys.
70
,
382
(
1991
).
56.
X.
Du
and
I.
Chen
,
J. Appl. Phys.
83
,
7789
(
1998
).
57.
C.
Randall
,
N.
Kim
,
J.
Kucera
,
W.
Cao
, and
T.
Shrout
,
J. Am. Ceram. Soc.
81
,
677
(
1998
).
58.
J.
Ihlefeld
,
D.
Harris
,
R.
Keech
,
J.
Jones
,
J.
Maria
, and
S.
Trolier-McKinstry
,
J. Am. Ceram. Soc.
99
,
2537
(
2016
).
59.
D. M.
Marincel
,
H.
Zhang
,
S.
Jesse
,
A.
Belianinov
,
M. B.
Okatan
,
S. V.
Kalinin
,
W. M.
Rainforth
,
I. M.
Reaney
,
C. A.
Randall
, and
S.
Trolier-McKinstry
,
J. Am. Ceram. Soc.
98
,
1848
(
2015
).
60.
J.
Ihlefeld
,
P.
Kotula
,
B.
Gauntt
,
D.
Gough
,
G.
Brennecka
,
P.
Lu
, and
E.
Spoerke
,
J. Am. Ceram. Soc.
98
,
2028
(
2015
).
61.
J.
Ihlefeld
and
C.
Shelton
,
Appl. Phys. Lett.
101
,
052902
(
2012
).
62.
S. P.
Li
,
W. W.
Cao
, and
L. E.
Cross
,
J. Appl. Phys.
69
,
7219
(
1991
).
63.
F.
Xu
,
S.
Trolier-McKinstry
,
W.
Ren
,
B.
Xu
,
Z.
Xie
, and
K.
Hemker
,
J. Appl. Phys.
89
,
1336
(
2001
).