Texture and phase variation of ALD PbTiO₃ films crystallized by rapid thermal anneal

PbTiO₃ (PTO), or lead titanate, is a material of technological interest primarily due to its ferroelectricity in both bulk and thin film varieties. PTO is a perovskite ceramic with large remanent polarization and coercive field.¹ Thin film PTO has been deposited using a variety of techniques, including sputtering, sol-gel, pulsed-laser deposition, metal organic chemical vapor deposition, and more recently, atomic layer deposition (ALD).^2–9 PTO film growth by ALD is primarily investigated as a prerequisite for the deposition of PbZr_xTi_1-xO₃ (PZT), or lead zirconate titanate, which exhibits several important properties, including large piezoelectric coefficients, high dielectric constant near the morphotropic phase boundary, high Curie temperature, and ferroelectric polarization hysteresis.¹⁰ PZT is an especially useful material for microelectromechanical systems and has enabled technologies such as actuators, switches, resonators, and gyroscopes.^11–13 PTO may serve as a growth template for PZT enabling oriented grain growth which can help to increase the ε₃₁.^14,15 Hiboux and Muralt demonstrated that (100)-oriented sputtered thin film PTO may be deposited on a Pt-coated Si substrate with the use of a sputtered PbO-rich nucleation layer 1–10 nm in thickness.¹⁶ (111)-oriented PTO was grown using a crystallized TiO₂ nucleation layer 1–2 nm thick prior to PTO deposition by sol-gel and sputtering.¹⁷ However, literature regarding PTO grown by ALD is relatively sparse, and thus far, no attempt to control ALD PTO film texture by use of seed layers has been reported.

PTO previously has been deposited by ALD using a variety of precursors. ALD PTO is typically grown using a sequence of binary oxide processes which are alternated to form a supercycle that yields the desired stoichiometry. All previous attempts report that the ALD PTO films are initially grown amorphous and require a postdeposition anneal for perovskite crystallization. Harjuoja et al. demonstrated ALD PTO growth on (100) Si at substrate temperatures of 250 °C and 300 °C using tetraphenyl lead (PbPh₄) and titanium tetraisopropoxide (TTIP) as cation precursors with O₃ and H₂O serving as oxidizers for the PbPh₄ and TTIP, respectively.⁶ The films were shown to have crystallized primarily into the perovskite phase with randomly oriented grains following rapid thermal anneal (RTA) in either N₂ or O₂ in the temperature range of 600–900 °C and held steady for 10 min. The PTO films were not electrically characterized. Hwang et al. demonstrated ALD PTO growth at 200 °C using lead bis(3-N,N-dimethyl-2-methyl-2-propanoxide) [Pb(DMAMP)₂] and titanium tert-butoxide [Ti(OtBu)₄] as cation precursors with H₂O as the oxidizer.⁷ Perovskite PTO with a preferred orientation of (001)/(100) was achieved following rapid thermal anneal when the films were deposited on sputtered Ir/IrO₂ coated Si. Polarization loops obtained using metal-insulator-metal capacitors with 37 and 78 nm PTO did not saturate with increasing electric field magnitude. The crystallized PTO films displayed a maximum dielectric constant of 280 with a positive remanent polarization (P_r) of 11.2 μC/cm². Watanabe et al. demonstrated ALD PTO deposited at 240 °C substrate temperature with excess lead content that crystallized into randomly oriented perovskite following anneal when deposited on sputtered Pt-coated Si substrate.⁸ Ti(OC₃H₇)₂(C₁₁H₁₉O₂)₂ [Ti(Oi-Pr)₂(DPM)₂] and Pb(C₁₁H₁₉O₂)₂ [Pb(DPM)₂], dissolved in ethylcyclohexane, were used as the titanium and lead cation precursors, respectively, with H₂O as the oxidizer. The ALD PTO films showed conformal coating of sidewalls, though the Pb(DPM)₂ dose was not self-limiting and the films were not electrically characterized. Recently, Sbrockey et al. demonstrated ALD PTO using tetraethyl-lead [Pb(Et)₄] and tetrakis dimethylamino titanium (TDMAT).⁹ Ozone was used as the oxidizer for Pb(Et)₄ while ozone or water was used as the oxidizer for TDMAT. Lead excess and stoichiometric PTO was achieved at a growth temperature of 250 °C. ALD PTO films deposited on platinum-coated Si crystallized into the perovskite phase following anneal. Trenches 5 μm wide by 20 μm deep were coated with ALD PTO but were not annealed or crystallized. A polarization versus electric field curve was provided for a stoichiometric film annealed at 600 °C in O₂ for 1 min which showed a P_r of 13.5 μC/cm², and maximum polarization (P_max) above 20 μC/cm² at 800 kV/cm, though leakage current was not reported.

Only reports by Hwang et al.⁷ and Sbrockey et al.⁹ of PTO grown by ALD contained electrical characterization of the crystallized films. While they displayed promising ferroelectric figures of merit, neither reported the leakage current and only Hwang et al.⁷ reported the dissipation factor which was greater than 0.2 at relatively low voltages (±2 V). Leakage current and dissipation factor are relevant figures of merit for device integration and can be strongly influenced by extrinsic material properties. Surprisingly, none of the previous reports of PTO grown by ALD have included high-resolution cross-section images of annealed samples which would have helped to connect the microstructure to the measured electrical properties. The objectives of this work are to analyze the effect of PbO_x and TiO_x nucleation layers deposited by ALD on the crystallization of ALD PTO grown using a new combination of the commercial precursors Pb(DMAMP)₂ and TDMAT and to provide detailed electronic characterization along with scanning transmission electron microscope (STEM) cross-section images to inform the nucleation properties.

PTO thin films deposited by ALD were grown in a Kurt J. Lesker Company ALD-150LX reactor. Laminar purge flow was constantly supplied using mass flow controller (MFC)-regulated ultrahigh purity argon supplied by a cryogenic liquid argon dewar. The purge flow was used to provide a diffusion barrier to prevent deleterious chamber wall deposition and to serve as a carrier gas for the precursors. The process pressure was held at approximately 1.6 Torr. Pb(DMAMP)₂ heated to 90 °C and TDMAT heated to 85 °C in stainless steel ampoules were used as the lead and titanium cation precursors, respectively. TDMAT was selected due to its high vapor pressure and high reactivity at 250 °C, which is near the top of the ALD window for the binary TiO_x oxide process.¹⁸ Pb(DMAMP)₂ was selected as the lead precursor due to its reasonable vapor pressure and the quality of the electrical properties demonstrated previously. The vapor pressure of the Pb(DMAMP)₂ was increased by briefly pulsing argon into the ampoule prior to dosing into the reactor. Demineralized H₂O at ambient temperature and ozone were both used as oxidizers. Ozone was supplied via an Absolute Ozone® Nano Ozone Generator. Ozone flow was controlled via MFC to 200 sccm, and the ozone concentration was measured to be approximately 10% by volume. The ozone generator was continuously running during the deposition process and was collected in a 1 l stainless steel reservoir which was evacuated into the reactor using an ALD valve during the ozone dose step. TDMAT was pulsed into the reactor for 0.5 s followed by a 20-s purge step prior to oxidation. H₂O was dosed using three sequential 1-s pulses to ensure saturation, followed by a 5-s ozone dose, and a 20-s purge. Pb(DMAMP)₂ was pulsed into the reactor using six-sequential 0.25-s pulses followed by a 20-s purge and was oxidized using eight sequential 1-s H₂O pulses, followed by a 5-s ozone dose and a 20-s purge. No thickness change was observed using a Film Sense (Lincoln, NE, USA) FS-1™ multiwavelength ellipsometer following the first pulse of either the H₂O or the Pb(DMAMP)₂; however, the additional pulses improved the wafer nonuniformity. The substrate temperature was held at 250 °C for all depositions. The substrates were “platinum lower electrode film stacks,” each consisting of Si (150 mm substrate)/500 nm SiO₂ (elastic layer, thermal wet oxide)/40 nm TiO₂ (adhesion layer, sputtered Ti followed by furnace oxidation)/100 nm Pt (bottom electrode, sputtered at 500 °C). Further details regarding the Pt film stack may be found elsewhere.¹⁹ 

PTO growth was achieved using a combination of binary oxide processes for PbO_x and TiO_x with the relative number of PbO_x:TiO_x cycles varied over the range from 1:1 to 4:1. The cycle ratio refers to the relative number of PbO_x to TiO_x cycles in one supercycle, which is repeated to achieve the desired thickness. For example, PTO films grown with a 3:1 PbO_x:TiO_x cycle ratio indicates that the supercycle consists of three PbO_x ALD cycles performed in sequence followed by a single TiO_x cycle. The films grown with fractional cycle ratios such as 3:2 and 5:2 are grown with the constituent binary ALD cycles occurring in back-to-back sequence as follows: (PbO_x) × 3-(TiO_x) × 2 for 3:2 and (PbO_x) × 5-(TiO_x) × 2 for 5:2.

The PTO films were annealed by RTA at 700 °C for 1 min with a 90 °C /s standard ramp rate in O₂ atmosphere using an AG Associates 610 system for crystallization prior to electrical characterization. The sample temperature was measured by a thermocouple in contact with the backside of the substrate near the center of the RTA. Each sample selected for electrical characterization received at least one additional deposition layer and anneal to help avoid electrical shorting due to pinholes. A 50-nm Pt thin film, sputtered at 500 °C to promote adhesion, was used as the top electrode. Capacitors with 4.92 × 10⁻⁴ cm² area were patterned using photolithography, and the electrode area was defined using UV-stabilized resist and etched via ion milling. The capacitor array was evenly spaced over a 100-mm diameter working area, which left a 25 mm edge exclusion on the 150-mm substrates. Additional piece-part samples were annealed for varying times, temperatures, and ramp rates in an Allwin 21 810 RTA using a carrier wafer to evaluate a variety of thermal treatment recipes.

Capacitance and tan(δ) were measured using an HP/Agilent 4192A LF Impedance Analyzer with a probe station. Polarization versus voltage and leakage current measurements were performed using a Radiant Technologies Inc. RT66-A Ferroelectric Test System. Thickness and roughness measurements were primarily carried out ex situ using a J.A. Woollam M-2000 spectroscopic ellipsometer, while elemental composition was quantified by Rutherford backscattering (RBS). Crystal phase and texture were identified by x-ray diffraction (XRD) using a Rigaku MiniFlex benchtop x-ray diffractometer with Cu-kα radiation in the θ-2θ configuration and a D/tex Ultra silicon strip detector. The x-ray tube was held at a constant 40 kV and 15 mA and underwent a start-up conditioning prior to measurement. The scan parameters were held at 20 °/min scan speed, 0.02° increment, with measurement range 10°–70°. All in situ ellipsometry measurements were performed using the Film Sense FS-1 ellipsometer referred to in Sec. II A.

The transmission electron microscopy (TEM) thin samples were prepared by focused ion beam (FIB) in an FEI Quanta 3D FIB/SEM DualBeam system. All samples received a protective Pt capping layer prior to FIB treatment. 250 nm of Pt was evaporated in the FIB by an electron beam followed by a 2 μm Pt deposition with the ion beam. The FIB was used to pattern the sample area. After the in situ lift-out, the samples were transferred into a probe aberration corrected FEI Titan G2 microscope for imaging and chemical analysis. High-angle annular dark-field (HAADF) STEM images and energy dispersive spectroscopy analysis were performed on the Titan microscope, operated at 200 kV, and equipped with Super-X silicon drift detectors.

High-speed, dynamic in situ thickness measurements were performed on each of the films reported in this work. The data were collected throughout the duration of the film growths and later reanalyzed following ex situ characterization. Final thickness and roughness measurements were performed on each ALD PTO sample using ex situ spectroscopic ellipsometry. The PTO films were measured using a Tauc-Lorentz parametrized model confined to the wavelength range of 500–1000 nm. The optical constants and roughness, determined by the ex situ model, were then inserted into the in situ fixed n and k model, while the thickness was allowed to vary over the duration of the film growth. Quantitative growth per cycle (GPC) characteristics were evaluated versus cycle number from the in situ ellipsometry data. One such example is shown in Fig. 1.

The ALD deposition was carried out using a 3:1 PbO_x:TiO_x cycle ratio for a total of 1401 ALD cycles or 350 supercycles with a single TiO_x ALD cycle in the beginning for nucleation promotion on the sputtered platinum substrate surface. Unless otherwise stated, the ALD films presented in this work were all initiated with a single TiO_x ALD cycle followed by a sequence of identical supercycles. The end of the growth is highlighted in the inset to show the final ALD film thickness and to exemplify a single 3:1 PTO supercycle. As is shown in Fig. 1, the supercycle consists of three binary PbO_x ALD cycles followed by one ALD TiO_x cycle. Although the apparent thickness increase for the PbO_x cycles is far smaller than the increase after the TiO_x cycle, the 3:1 sample was measured to be near-stoichiometric PTO by RBS and crystallized into the perovskite phase upon postdeposition anneal as indicated by XRD. The details of the structural and chemical analyses are discussed in Secs. III B and III C. The disparity between the apparent thickness increase and Pb atomic concentration implies that the PbO_x incorporation is not fully represented by the in situ thickness measurements until after the TiO_x ALD cycle. The details of the mechanism by which the surface-bound PbO_x contributes to the optical properties of the growing film have not been explored.

The in situ thickness characterization shows that the film displays nearly perfectly linear growth from the start with no nucleation delay. ALD PTO films with 1:1, 3:2, 2:1, 5:2, 3:1, and 4:1 PbO_x/TiO_x cycle ratios all showed similar characteristics, including nearly perfectly linear growth throughout the deposition, no nucleation delay, and growth-per-cycle measured to be 0.08, 0.08, 0.08, 0.07, 0.07, and 0.06 nm per cycle, respectively. The GPC was calculated using the final thickness in the center of the samples, corresponding to the center of the ALD reactor. The highly linear thickness versus number of cycles profiles are strongly suggestive of a deposition process within the ALD growth regime. The thermal stability of the process will be explored through rigorous precursor dose uptake curves in future work. Following RTA, the films were remeasured by ex situ ellipsometry and a comparison to preanneal ex situ ellipsometry measurements indicated that the thickness decreased approximately 10% due to densification.

Crystal structure was analyzed by XRD for samples with binary cycle ratios ranging from 1:1 to 4:1 PbO_x:TiO_x. 1401 ALD cycles were deposited for each sample on the standard sputtered Pt/sputtered and annealed TiO₂/Thermal SiO₂/Si lower electrode film stack. Piece parts were cleaved near the center of the 150 mm substrates for post-treatment and analysis. The films were crystallized by RTA. The phases and corresponding orientations are summarized in Fig. 2.

The RTA process was initially optimized for sol-gel films with Pb-excess ranging from 8% to 15% as reported by Sanchez et al. and consisted of a 90 °C/s ramp to 700 °C, with a 60-s hold under flowing O₂ atmosphere of approximately 5 SLPM.²⁰ As shown in Fig. 2, no obvious XRD peaks from the ALD film are observed from the samples that were not annealed, though a small hump is evident near 29° for films with 5:2 PbO_x:TiO_x cycle ratios or greater. The hump is consistent with the (101) reflection from the PbO litharge structure (PDF#5-0561) and (111) reflection from the massicot structure (PDF#38-477) and may be due to randomly oriented PbO_x grains; detailed phase identification of the small hump would require a higher resolution scan. The small hump near 29° is distinct from the peak observed near 29.5° in the postannealed 1:1 and 3:2 samples which corresponds to the PbTi₃O₇ (−102) reflection. Phase-pure perovskite was observed only in the samples with 3:1 and 4:1 PbO_x:TiO_x cycle ratios. The perovskite grains appear to be randomly oriented, though the (111) peak is not clearly observed in this sample set.

The crystallization of near-stoichiometric PTO was investigated with respect to variation in the ramp rate from ambient oven temperature to 700 °C within the range of 10–200 °C/s. Four separate RTA strategies were tested including 700 °C, 1 min hold and 600 °C, 5 min hold with an optional 400 °C, 5 min hold pyrolysis immediately prior to final crystallization. The pyrolysis preanneal was added to mimic a treatment that is often performed on PTO and PZT films deposited by sol-gel.²⁰ TiO_x and PbO_x seed layers were implemented to investigate their effect on crystallization and preferred orientation. XRD was performed as the primary method of characterization.

Figure 3(a) indicates that the ramp rate did not affect the phase formation or the orientation of the perovskite grains. Each of the annealed films in Fig. 3(a) appear to be phase-pure perovskite with randomly oriented grains, although the (111) peak is not observed. Figure 3(b) shows that the pyrolysis step by itself did not completely crystallize the sample, though a small unidentified hump is apparent near 30° that is not present in the unannealed sample. The 30° hump may be a result of PbO oxidation in O₂ to monoclinic Pb₂O₃ (#00-036-0725) that exhibits strong powder diffraction peaks at 29.5° and 30.2°. The variation of final crystallization temperature and time did not seem to strongly affect the phase or orientation of the crystallized PTO films.

For samples with a TiO_x nucleation layer, 20 cycles of ALD TiO_x were deposited as the nucleation layer using the same conditions as the TiO_x cycle used in the PTO supercycle. The nucleation layer deposition was followed by an RTA at 700 °C for 60 s in O₂. The TiO_x thickness was determined to be 1 nm by in situ ellipsometry using a tabulated TiO₂ optical constants. 1401 cycles of ALD PTO were deposited on top of the annealed TiO_x layer. Piece parts were cleaved from the sample and heat-treated by RTA using a variety of temperatures and times and then characterized by XRD [Fig. 3(c)]. Contrary to our expectation, the TiO_x seed layer did not cause a preferred (111) perovskite crystallite orientation. The grains appear to be nearly randomly oriented over the range of high-temperature anneals; however, the (001) and (111) peaks appear to be suppressed. No secondary phases are observed; however, the (101) and (110) peaks appear to overlap for all RTA conditions except for the 700 °C, 1 min hold.

For samples with 250 cycles of ALD PbO_x deposited as an alternative nucleation layer to TiO_x, the expectation was that the ALD PTO films deposited on top would display a preferred {100} orientation following postanneal resulting from the Pb-rich interface. The PbO_x seed layer was not annealed prior to the PTO growth, and the thickness was found to be 8 nm by in situ ellipsometry using a Cauchy model. The XRD results of the PbO_x seed layer with varied RTA conditions are presented in Fig. 3(d). Unlike the other samples, the as-deposited PTO-on-PbO_x films display peaks near 39° consistent with (121) reflection from massicot (#38-1477) at 39.5° and near 43° consistent with (102) reflection from litharge (#5-0561) at 42.5°. The peaks are likely a result of lead-oxide crystallization during the seed layer growth; this is further evidenced by the fact that these peaks disappear following the pyrolysis anneal. It is probable that the PbO_x crystallites are absorbed into the PTO matrix upon postdeposition anneal. The samples that received a final crystallization RTA in Fig. 3(d) appear to have stronger (001) and (111) peaks compared to the samples with no seed layer or samples with a TiO_x seed layer. While the ALD PbO_x seed layer did have an effect on the crystallization, the expected primary {100} texture was not observed.

The addition of O₃ as a second oxidizer was required to mitigate macroscopic pinhole defects that likely resulted from residual carbon from the cation precursors. The defects were visible pinholes tens of microns across that were observable by microscope and also caused a visible haze across the wafer. Devices for films that did not receive the additional O₃ dose were electrically shorted. The O₃ dose every half-cycle served as an in situ clean, but it also reduced the PbO_x incorporation and necessitated additional PbO_x ALD cycles in the supercycle to reach near-ideal stoichiometry. However, use of O₃ exclusively resulted in a reduction of the GPC by over 50% compared to the GPC observed in films oxidized with H₂O followed by O₃.

PTO film stoichiometry was measured directly for a range of PTO samples with varied PbO_x:TiO_x cycle ratios by RBS. The low-field dielectric constant or relative permittivity was also measured for films with a range of PbO_x:TiO_x cycle ratios which enabled the direct comparison between film composition, crystal structure (see Sec. III B), and electronic properties. The results of the aggregate data are presented in Fig. 4. There are two points plotted for each PbO_x:TiO_x cycle ratio because piece parts were cleaved from each whole-wafer sample wafer in different locations. The divergence results from slightly non-optimized precursor dose across the area of the wafer which also affected the thickness nonuniformity.

Phase identification from the XRD scans was used to subdivide the data points in Fig. 4. The points plotted with the triangles were samples that received two sets of coats and anneals, each of 1401 ALD cycles. The maximum permittivity of the samples that received two coats and anneals occurs at PbO_x:TiO_x 5:2 (plotted as 2.5); this was surprising since the maximum was expected to occur at 3:1, i.e., near the ideal PTO stoichiometry. Since the samples were quite thin, the dielectric constant measurements were repeated with select samples receiving four coats and anneals of 1401 cycles each. The thicker films led to an overall increase in dielectric constant for both samples and caused the samples to show the expected trend with the maximum now occurring near the ideal PTO stoichiometry. The dielectric constant decreased significantly for the PbO_x:TiO_x 4:1 samples despite having near-ideal stoichiometry (as measured by RBS near the center of the wafer) and having crystallized into phase-pure perovskite. The reason for the steep drop has not been thoroughly investigated, though there is some evidence that electrode adhesion and surface morphology contribute significantly. The electrical measurements for the PbO_x:TiO_x 4:1 sample were taken only from capacitors within a 1.5-in. radius of the center of the wafer because the capacitors near the edge of the mask were shorted due to pinholes observed by optical microscope. The data plotted in Fig. 4 were recorded from virgin samples that did not receive a recovery anneal following top electrode processing. After the PbO_x:TiO_x 4:1 sample received an additional 600 °C/30 min recovery anneal, the low-field dielectric constant increased by 20% and the polarization/voltage hysteresis loops dramatically improved.

The PTO with PbO_x:TiO_x of 4:1 that received four coatings and four rapid thermal anneals was selected for detailed capacitance and loss measurements. The results are presented in Fig. 5. The sample underwent a 600 °C, 30 min recovery anneal in O₂ following top Pt electrode processing. The sample displayed a maximum dielectric constant of 290 at 85 kV/cm and loss tangent below 4% throughout the tested electric field range of ±275 kV/cm (±10 v). Nested polarization/electric field loops for the same sample and test point were measured and are reported in Fig. 6. The sample displays a maximum polarization (P_max) 48 μC/cm² when tested at 415 kV/cm (15 V), respectively, remanent polarization (2P_r) = 60 μC/cm², E_c1 = −73 kV/cm, and E_c2 = 125 kV/cm. The polarization saturation is less sharp in the negative field direction which implies higher leakage current compared to the positive direction. Leakage current measurements were performed immediately following polarization/field testing and indicated a leakage current of 15 μA/cm² at 138 kV/cm (5 V) and 211 μA/cm² at −138 kV/cm (−5 V), confirming higher leakage current density in the negative direction. The test probes were positioned such that the bottom electrode served as the drive and the top electrode served as the ground.

STEM images were obtained for PTO samples deposited with a PbO_x:TiO_x ratio of 3:1 and are detailed in Fig. 7. The sample is shown as-deposited (no anneal) in Fig. 7(a). The as-deposited ALD PTO film is smooth as-deposited with the surface roughness nearly matching the underlying substrate. Grain boundaries are not apparent in the PTO layer at the stated resolution, which implies that the film is not fully crystallized as-deposited. Unlike the Pt, TiO₂, and SiO₂ substrate layers, the PTO layer has finely clustered regions of differing contrast. The HAADF detector provides z-contrast that shows heavy elements brighter and lighter elements darker. The difference in contrast within the PTO layer implies that there is a segregation of the Pb and Ti cations on the nm-scale. The white nm-scale clusters approximately 2–5 nm in diameter likely correspond to the PbO_x phases previously identified by XRD in section B2 because Pb is far heavier than Ti. Metallic Pb can be ruled out because the RBS analysis showed that the as-deposited films contained 62 (±3) at. % O₂, which indicates that the films were fully oxidized before the anneal. PbO_x nanoclusters were also observed by Watanabe et al. in as-deposited PZT films deposited using ALD precursors.²¹ In that study, high-resolution transmission electron microscopy images showed that the PbO_x nanoclusters were crystalline, although the specific phase could not be identified because the measured d-spacing was shared by several PbO_x phases. The PbO_x nanoclusters were shown to be surrounded by an amorphous Ti-Zr-O_x matrix which indicated that the as-deposited films were a composite structure, rather than a homogeneous solid-solution. The results of the present study are consistent with Watanabe et al.²¹ and suggest that the as-deposited film consists of PbO_x nanocrystals embedded in an amorphous matrix.

Figure 7(b) shows the microstructure of the annealed ALD PTO sample. The protective cap was deposited in the FIB after the PTO film was annealed and therefore did not influence the crystallization. The apparent delamination of the protective cap is due to either poor adhesion as-deposited or FIB damage during sample preparation. Grain boundaries are readily observed in the PTO film and indicate that the PTO film crystallized upon RTA treatment, which is consistent with the XRD patterns from section B2. The grains appear to be distributed at random with diameters ranging from approximately 10–50 nm. None of the grains in contact with the lower Pt electrode appear to extend through the entire film thickness, which suggests that the grains nucleated throughout the film and that the nucleation was not localized to the Pt-PTO interface. This is in distinct contrast to the columnar growth observed for sputtered and sol-gel PZT films where grains nucleate primarily at the lower electrode-PZT interface and extend to the top electrode.²² The difference in nucleation mode explains why the TiO_x and PbO_x nucleation layers did not strongly influence the texture of the crystallized films. Only the minority fraction of grains that nucleated at the Pt/PTO interface could be directly affected by the nucleation layers; therefore, the orientation of the PTO grains was not strongly influenced by the substrate.

Defects were observed in various locations throughout the film and a representative STEM-HAADF image is presented in Fig. 8. The top-right section of the PTO film is quite dark in contrast compared to the rest of the film which suggests the presence of a void. Since the void appears to be a surface defect, it is better described as a pinhole roughly 20 nm wide and 25 nm deep. Other pinholes (not pictured) of similar size were observed throughout the film, and none of the pinholes appeared to penetrate the entire film thickness. Another region of dark contrast shown in Fig. 8 is observed at the Pt-PTO interface and is most likely a void 15 nm in diameter. Other circular patches of darker contrast are observed throughout the film that are not as dark as the pinhole or the void at the Pt-PTO interface. Secondary phases as well as voids could show contrast differences in the STEM-HAADF image; however, since no secondary phases were observed in the XRD pattern for the corresponding growth conditions (see Fig. 2, top pattern), the circular patches are attributed to voids. The presence of pinholes and voids throughout the film suggests that the film is not fully densified after a single 700 °C/60 s anneal. Each of the films that were processed for electrical characterization received the same 700 °C/60 s initial crystallization anneal with 90 °C /s ramp rate. An additional 600 °C/30 min recovery anneal in O₂ had a dramatic improvement on the electrical quality of the films, mainly in the reduction of leakage current. While STEM images were not captured for the ALD PTO films following the recovery anneal, it is likely that the films were further densified in addition to any improvement to electrode adhesion. A nearly identical recovery anneal was performed by Hwang et al.⁷ and similarly led to an improvement of the electrical properties.

In this work, we demonstrate a PTO ALD process using a new combination of the commercial precursors Pb(DMAMP)₂ and TDMAT characterized primarily by RBS, XRD, STEM, and dielectric measurements. The films were amorphous as-deposited and near-stoichiometric films crystallized into phase-pure, randomly oriented perovskite grains when grown directly on Pt-coated Si substrates. PTO films with an ALD-grown unannealed PbO_x seed layer displayed mostly randomly oriented grains with slightly increased (001) and (111) perovskite peaks compared to samples with no seed layer but were still mostly randomly oriented. PTO films grown on a 20 cycle ALD TiO_x annealed seed layer did not show a clear orientation preference. The seed layers did not have a significant effect because STEM images indicated that the PTO grains nucleated randomly throughout the film which minimized any effect of seed layers at the Pt/PTO interface. ALD-grown PTO films with a 4:1 PbO_x:TiO_x cycle ratio and a thickness of 360 nm were electrically characterized and showed polarization hysteresis loops typical of ferroelectric behavior, with P_max = 48 μC/cm² at 415 kV/cm, remnant polarization (2P_r) = 60 μC/cm², E_c1 = −73 kV/cm, and E_c2 = 125 kV/cm. The same sample exhibited a leakage current density of 15 μA/cm² at 138 kV/cm. The dielectric constant/electric field butterfly loops were measured and showed a loss tangent of less than 4% tested between ±275 kV/cm, with a maximum relative permittivity of 290 at 85 kV/cm when tested at 10 kHz.

The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for the United States Government purposes. DOD will provide public access to these results of federally sponsored research in accordance with the DOD Public Access Plan (http://www.dtic.mil/dtic/pdf/dod_public_access_plan_feb2015.pdf).

The authors wish to acknowledge Glen R. Fox of Fox Materials Consulting, LLC for timely feedback related to XRD patterns and ferroelectric hysteresis loops and Noel O'Toole and G. Bruce Rayner of the Kurt J. Lesker Company, Inc. for equipment support and technical expertise. 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).