Pulsed Laser Deposition of Epitaxial Non-Doped PbTiO₃ Thin Films from PbO–TiO₂ Mosaic Targets

Abstract

PbTiO₃ (PTO) suffers from difficulty in preparing high-density robust bulk ceramics, which in turn has been a bottleneck in thin films growth with physical vapor deposition (PVD) methods. In the present work, we prepared non-doped PTO thin films by a pulsed laser deposition (PLD) method with either a single PTO target or a mosaic target consisting of PbO and TiO₂ pie-shaped pieces. On the PTO single target, laser irradiation caused selective ablation of Pb, resulting in Ti-rich cone-shaped pillar structure on the surface, whereas the irradiated surface of PbO and TiO₂ pieces was smoother. Epitaxial PTO films deposited on SrTiO₃ (001) substrates from the pie-chart targets with PbO:TiO₂ areal ratio from 3:5 to 5:3 resulted in composition, crystallinity, flatness, and ferroelectric properties almost independent of the areal ratio. The averaged composition of each film was close to stoichiometric, suggesting a compositional self-control mechanism. For growing epitaxial and high-quality non-doped PTO films, a PbO–TiO₂ pie-chart target is advantageous in easiness of handling and stable surface structure.

1. Introduction

PbTiO₃ (PTO) is a pioneering and still important ferroelectric material with a perovskite-type structure [1,2]. The crystalline structure of PTO is cubic at temperatures above 490 °C, whereas it is tetragonal at lower temperatures with a large c/a ratio of 1.06 at room temperature (RT) [1]. The significant structural phase transition often causes cracks in the whole volume of PTO bulk ceramics during cooling after sintering at a higher temperature. This makes the PTO bulk ceramics extremely fragile, which in turn is a difficulty for using it as a target material typically needed in thin films growth processes by physical vapor deposition (PVD) techniques. To overcome this fragility, one popular solution is using sintering aids or doping with other elements to form “modified PTO” [3,4].

Another factor that could make the growth of PTO thin films difficult is the high vapor pressure of lead oxide. In general, PLD method allows thin films growth with cation composition close to that of the target. However, when the material contains a volatile element, it is often observed that film composition progressively deviates from stoichiometry because of the volatile element depletion when depositing at elevated temperatures [5]. Many groups growing PTO films by PVD techniques with a single PTO target attempt to prevent the Pb deficiency with strategies that can be categorized into two, namely, to mitigate the re-evaporation of Pb from the growing film or to supply species with an excess Pb/Ti ratio onto the substrate to compensate for the Pb loss. The practical deposition conditions based on the former strategy include (a) low substrate temperature (550 °C or lower) [6,7,8], and (b) high oxygen ambient pressure (100 mTorr or higher) [7,9,10,11,12,13,14]. A common strategy for the latter category is (c) to use a Pb-rich target [6,7,10,15], although the exact amount of Pb in excess will depend on the deposition conditions.

Supplying species from two targets has been reported to effectively function for preparing high-quality thin films containing a high vapor-pressure element. The alternative ablation of two (or more) targets has been regularly employed since the beginning of combinatorial PLD researches [16,17]. The species sequentially ablated from multiple targets are expected to mix on the growing surface to form a uniform film provided that the individual layer thickness for each cycle is at the monolayer level to allow for intermixing. In that way, the ratio of species supplied from multiple targets can be controlled by modifying the number of laser pulses on each target. However, this method typically results in a larger setup complexity due to the need for constant exchange of target positions (or laser redirecting to the different targets) and result in an overall reduced growth rate. This can be regarded as an advanced technique belonging to category (c). In this way, Maruyama et al. have succeeded in obtaining epitaxial LiCoO₂ thin films by alternative laser ablation of Li₂CO₃ and CoO targets [18]. A similar approach by using the simultaneous ablation of two targets (with two independent lasers or a split laser beam) can also be a powerful tool for this purpose, in this case avoiding continuous target exchange [19,20]. In this manner, Hussey and Gupta have prepared good quality YBa₂Cu₃O_7−d (YBCO) films by simultaneously shedding laser beams to individual YBa₂O_x and CuO targets [21].

A much simpler approach for sequential ablation is based on using a mosaic target consisting of pieces of multiple materials, which requires only one rotating target [5]. Since PLD film growth rates produce sub-monolayer thickness per pulse (typically below 1 Å/pulse depending on the deposition conditions, namely laser fluence, gas pressure and target-to-substrate distance), the number of pulses in each part of the target at standard target rotation speeds produce a sequence of very short pulses allowing for homogenous cross plane composition. Besides, the possible composition inhomogeneities caused by slight variations in the sequence, or when ablating at the boundary regions between both compounds in the target, are averaged during film growth and expected to produce negligible effects. In this way, Lee et al. have demonstrated to compensate the typical Zn deficiency in the PLD growth of ZnGa₂O₄ (ZGO) films at high temperature by using a mosaic target consisting of ZGO and ZnO [22]. The mosaic target strategy has been successfully adopted to deposit PTO films, since both PbO and TiO₂ are robust ceramics unlike PTO. Funakoshi et al. have already performed PLD of PTO polycrystalline films on silicon (Si) substrates using a PbO–TiO₂ mosaic target [23]. Nothing precludes the use of this method for growing PTO epitaxial films, except for the larger difficulty due to the need of higher deposition temperatures to achieve crystalline order. However, to our knowledge, there has been no previous report on the growth of epitaxial PTO films using a mosaic target.

In the present study, we prepared epitaxial PTO films on perovskite-type oxide single-crystal substrates with PLD. Either a mosaic target consisting of PbO and TiO₂ pie-shaped pieces or, for comparison, a single PTO target was employed. A cone-shaped pillar structure, accompanied by Pb-deficient composition, was observed on the surface of the single PTO target after laser irradiation and supposed to impede the reproducible Pb/Ti ablation during deposition. PTO epitaxial films were grown from a series of pie-chart targets resulting in composition, crystallinity, flatness, and ferroelectric properties comparable with the film obtained from a single PTO target.

2. Experiments

2.1. Fabrication and Characterization of Targets

To prepare the PbTiO₃ target, mixed powders of PbO and TiO₂ were pressed in 30 mm-diameter dies, and calcinated in air at 800 °C for 1 h. No sintering aids were used in this case. The product was ground and re-pressed and sintered in air at 1160 °C for 2 h. The bulk was extremely fragile so that it easily breaks when picked up with tweezers. On the other hand, individual PbO and TiO₂ targets were prepared by pressing the powder in the 30 mm-diameter dies and sintering in air at 750 °C for 1 h, and at 1200 °C for 6 h, respectively. Both PbO and TiO₂ disks thus obtained were robust. The 3 mm-thick disks were then cut into quarter circle or one-eight circle shape with a diamond saw. The pie-shaped pieces were fixed on an aluminum disk with a double-sided carbon tape in various PbO:TiO₂ areal ratios of 3:5, 4:4, and 5:3 (Figure 1a).

A commercial (Nb,Mg)-modified Pb_1.1TiO₃, much denser, was also used from comparison (Nb and Mg are sintering aids). The laser-induced evolution of surface morphology and composition was studied for the homemade PbTiO₃ target, as well as for the commercial Pb_1.1TiO₃, and the homemade PbO/TiO₂ targets. For this, the target surface was irradiated by a focused excimer laser (COMPex 102, KrF gas, wavelength of 248 nm, Lambda Physik, Göttingen, Germany) beam at different fluence in 100 mTorr oxygen pressure. The morphology and the composition of the target surface was observed with a scanning electron microscopy (SEM, Quanta 650 FEG, FEI, Hillsboro, OR, USA) and an energy-dispersive X-ray (EDX) spectroscopy (X-Max, Oxford Instruments, High Wycombe, UK), respectively.

2.2. Thin Films Growth

PTO thin films were grown mainly on single-crystal SrTiO₃ (STO; 001) substrates using either the PTO single target or the PbO–TiO₂ pie-chart target. The STO substrates were chemically and thermally pre-treated to form atomically flat surface with a step (~4 Å height) and terrace structure [24]. The deposition conditions for PTO films were as follos: substrate temperature of 550 °C, ambient oxygen pressure of 100 mTorr, laser repetition rate of 2 Hz, total pulse number of 1760, energy on the target surface of 50 mJ, fluence of 1.3 J/cm², and target–substrate distance of 50 mm. As a bottom electrode layer for electrical characterization, a 15 nm-thick La_0.6Sr_0.4MnO₃ (LSMO) layer was deposited with deposition conditions of 650 °C, 100 mTorr, 2 Hz, 740 pulses, 50 mJ, 2.0 J/cm², and 50 mm, prior to the PTO layer growth. The target rotates with a speed of 3 rpm during laser irradiation. Rastering, as well as rotation, were used on the single PTO and LSMO targets for uniform ablation [25], whereas only rotation (no rastering) was used on the mosaic targets. After the deposition, the sample was cooled to room temperature at a rate of −10 °C/min in O₂ ambient of 10 Torr. Hereafter, we label the PTO/LSMO/STO samples prepared with the pie-chart target with the PbO:TiO₂ areal ratios of 3:5, 4:4, and 5:3 as samples PC35, PC44, and PC53, respectively, and the one with the single PTO target as sample SP. The actual pulse numbers fed to PbO and TiO₂ pieces during one cycle of rotation are shown in

Table 1. Deposition parameters and characteristics of four samples.

Sample Name	PbO:TiO2 Areal Ratio	Number of Pulses Irradiated on PbO:TiO2 per 1 Cycle	PTO Growth Rate (10−3 nm/pulse)	PTO out-of-Plane Lattice Spacing (Å)	PTO 002 Rocking Curve FWHM (deg)
SP	-	-	28	4.158	0.080
PC35	3:5	12:30	15	4.146	0.081
PC44	4:4	18:24	14	4.146	0.087
PC53	5:3	24:18	8.6	4.141	0.090

. The deviation of pulse numbers ratio from the central angle ratio between PbO and TiO₂, which should theoretically be identical (Figure 1b), is due to a slight offset between the target center and the rotation axis.

2.3. Characterization of Films

Crystallinity of PTO films was evaluated with 2θ–ω scans, rocking curves and reciprocal space mapping (RSM) by X-ray diffraction (XRD) using a diffractometer with a four-circle goniometer (X’pert PRO-MRD, Malvern-PANalytical, Almelo, The Netherlands). Surface morphology was observed with atomic force microscopy (AFM). X-ray photoemission spectroscopy (XPS) and Rutherford backscattering spectroscopy (RBS) were employed to characterize the composition of the PTO films. XPS measurements were performed at RT with a hemispherical analyzer (PHOIBOS 150, SPECS, Berlin, Germany). Here, calculation of the Pb ratio, defined as Pb/(Pb + Ti), was based on three different Pb lines, namely 4d (binding energies 413 and 435 eV), 4f (138 and 143 eV), and 5d (20 and 22 eV), and one Ti line 2p (455 and 461 eV). Due to the different binding energies, the three Pb lines give the signals originating from surface layers of different thicknesses. This was used to account for possible Pb composition gradients at the surface. The RBS measurement was performed with a tandem accelerator at Centro de Micro-Análisis de Materiales, Madrid, Spain. He⁺ ions of an energy of 2 MeV were directed to the sample at an angle of 2.8° from the normal direction, and the scattered particles were detected at an angle of 165° with respect to the incident angle. The ferroelectric properties of the films were evaluated through phase–bias curves and write–erase tests using a piezoresponse force microscopy (PFM) system. Both AFM and PFM were carried out using a scanning probe microscopy system (MFP-3D Classic, Asylum Research, Santa Barbara, CA, USA). See [20] for the detail of XRD, XPS, and PFM procedures.

3. Results

3.1. Morphology and Composition of the Target Surface

Surface SEM images of a commercial (Nb,Mg)-doped Pb_1.1TiO₃ target are shown in Figure 2a–c. The non-irradiated pristine region (Figure 2a) consists of random-shaped grains of several tens’ micrometers. Adjacent grains are weakly connected, resulting in an overall porous structure. Cone-shaped pillars appeared on the surface of the target immediately after irradiation of a few laser shots. The pillars pointed to the direction of laser incidence (Figure 2b). The surface after being used for several tens of depositions with fluence of 1 to 2 J/cm² shows significant cone-shaped pillars at the level of several tens of micrometers (Figure 2c). EDX analysis revealed Ti-rich (Pb-poor) composition on the surface with such a pillar structure. The off-stoichiometry suggests that selective ablation of Pb has occurred due to its high vapor pressure, leaving Ti component on the surface. This cone-shaped pillar structure looks similar to the one generated on a YBCO target [26]. The pillars on the YBCO bulk were reported to be Cu-poor and Y-rich, and was generated when the laser fluence was below a threshold of 1 J/cm² [26]. In the case of PTO, unlike YBCO, the off-stoichiometry remained in any fluence examined in the experiment, from 0.3 to 4.7 J/cm².

Figure 2d,e shows SEM images of the homemade PTO target. The pristine surface, even though flatter than the commercial Pb_1.1TiO₃ target, is filled with microcracks that explain its fragility (Figure 2d). Appearance of cone-shaped pillars due to laser irradiation (Figure 2e) was not as remarkable as that on the commercial target shown in Figure 2b. Still, this target was not free from a compositional shift towards Pb-poor values. Figure 2f shows the relationship between the laser fluence and the Pb ratio in the commercial (Nb,Mg)-doped Pb_1.1TiO₃ and the homemade PbTiO₃ targets. In both cases, large deviations from the initial composition were observed down to about Pb/(Pb + Ti) = 0.4. Although deviation tends to reduce as the fluence increases after a minimum point, it is not completely suppressed. Similar surface morphology observations were carried out for the bulk targets of the binary oxides PbO and TiO₂. The PbO surface was quite flat and crack-free either at pristine state (Figure 3a) or after laser irradiation (Figure 3b). TiO₂, initially consisting of ~10 µm-sized grains (Figure 3c), was deformed by the laser to show a 3d structure, wavy rather than pillarlike (Figure 3d).

3.2. Properties of PTO Films

XRD 2θ–ω scans around STO 002 reflection of samples SP, PC35, PC44, and PC53 are shown in Figure 4a. In all the samples, the PTO 002 diffraction peak was accompanied by Laue fringes, which indicate high crystallinity of the PTO layer. Thickness of the PTO layer was estimated from the fringe interval to be 32.7, 26.8, 24.1, and 15.2 nm for samples SP, PC35, PC44, and PC53, respectively. The growth rate of the PTO layer (Table 1) was highest in sample SP and lowest in sample PC53. The slight difference in the cell parameters (Table 1) can qualitatively be explained through the difference of degree of strain originating in the thickness variation. The rocking curves around the PTO 002 reflection in the four samples showed FWHM smaller than 0.1°, the ones of SP and PC35 being slightly smaller than the others (Figure 4b, Table 1). The epitaxial relationship between films and substrate was confirmed with RSM, as shown in Figure 4c. The spots corresponding with PTO, LSMO, and STO are aligned vertically, indicating the in-plane lattice of the films fully coherent to the substrate.

AFM topographic images of the four samples are shown in Figure 5. All the images show lines corresponding to the terraces on the STO substrates. The peak-to-valley values were within 1 nm in the 5 × 5 µm² area for all the samples.

The results of XPS analysis of four samples are reported in Figure 6. Unexpectedly, the Pb ratio in the PTO layers was found to be quasi-consistent among the four samples. Meanwhile, the calculated Pb composition strongly depended on the Pb lines used for the evaluation. The Pb lines with higher binding energy resulted in the Pb ratio closer to stoichiometry. Generally, the higher the binding energy, the thicker the surface layer probed. Therefore, the dependence of Pb ratio suggests the possibility of Pb segregation in PTO layers towards the surface. The Pb ratio larger than 50% even for the highest binding energy line (Pb 4d) implies that the thickness of Pb segregation was larger than the XPS observation depth (a few nanometers). The composition difference among the samples was negligible compared with the evolution along the depth.

We also carried out RBS characterization for two PTO films grown for this purpose, one directly on a STO substrate using the single PTO target and the other on a CeO₂-buffered r-plane sapphire (Al₂O₃ (1$\overline{1}$02)) substrate using the pie-chart target of PbO:TiO₂ = 4:4. The Al₂O₃ substrate was chosen so that the Ti signals from the film would be completely separated from signals from the substrate. The film on CeO₂/Al₂O₃ (1$\overline{1}$02) was regarded to be quasi-epitaxial since the PTO layer was confirmed to be (h00) oriented. The RBS analysis suggested Pb ratio of 49% and 50% for the whole thickness of the PTO layers in PTO/STO and PTO/CeO₂/Al₂O₃ samples, respectively (Figure 7). As the composition analyzed with XPS was consistent among the four samples grown on LSMO/STO, we speculate that the averaged Pb ratio of ~50% is adopted for all the samples.

In order to confirm the ferroelectric properties of the four samples, write–erase test in a few micrometers’ square region and phase–bias measurement was performed using PFM. As shown in Figure 8, all samples showed good write and erase characteristics and typical phase–bias curves, which prove good ferroelectricity for all the PTO films.

4. Discussion

In a binary oxide (AO_x) target the only possible composition deviation among the ablated species is the ratio x of O to A, which can be compensated through the oxygen ambient pressure. Furthermore, deposition using PbO and TiO₂ bulks, free from cone-shaped pillars, is desirable in terms of reproducibility in film growth. The appearance of cone-shaped pillars could perturb the distribution of local laser fluence, which could drastically affect not only the ablation rate, and therefore the film growth rate, but also the quality of the film.

The composition of the PTO layers, independent of the areal ratio of the pie-type PbO and TiO₂ pieces, indicates that a compositional self-control (CSC) mechanism occurs during film growth. The CSC mechanism has been reported in PZT thin films grown on Pt/Ti/SiO₂/Si substrates with a sputtering method [27]. Provided that the deposition rate is not too high, as in the present case, the adatoms are allowed to migrate over the substrate or growing film surface. They will be crystallized to form a stoichiometric crystal with Pb:(Ti + Zr) equal to 1:1 or, if they are in excess, will be re-evaporated. The present result confirmed that this should have occurred for the A cations excess (Pb). Not incorporated into the perovskite structure, the excess Pb segregated towards the surface and ultimately was re-evaporated. Most probably, composition of the adatoms fed to the substrate was the closest to stoichiometric in PC35, resulting in the highest growth rate and the smallest FWHM, whereas it was the farthest in PC53 with the lowest growth rate. The growth rate variation among samples is quite consistent with the TiO₂ supply rate (ratio in growth rate between PC35 and PC53 was 1.7, exactly the same as the ratio in number of TiO₂ pulses/target turn between them; see Table 1). This suggests that Pb was always in excess, and is expelled from the film in the CSC mechanism. Similarly, It was also the CSC mechanism that resulted into stoichiometric films when deposited from the single PTO target on which Pb is selectively ablated (see Section 3.1).

It would be informative to compare the present results with those by Funakoshi et al. [23]. In their PTO films grown on Si substrates, Pb/Ti was 2.59 (i.e., Pb ratio 72%) when PbO and TiO₂ pieces were irradiated with the same number of laser pulses, whereas the pulse numbers ratio of 29:71 to PbO:TiO₂ resulted in a stoichiometric PTO film. The Ti deficiency in the former film could be due to the difference of the deposition rate between PbO and TiO₂ pieces. In the present study on epitaxial films, stoichiometric films were obtained when the pulse number ratio of PbO:TiO₂ was at least in the range of 12:30 (=29:71) to 24:18 (=57:43). The difference clearly originated in the difference between perovskite-type single crystal substrates that promote epitaxial growth and Si substrates that do not. The best film in terms of crystallinity (PC35) was obtained with a pulse numbers ratio of 29:71, which was in agreement with the result of Funakoshi et al. It should be noted that PTO films deposited on Si substrates in the present study, from either PTO or pie-chart target during the same runs as those on STO, have always resulted in Pb-rich composition in EDX analysis (not shown). We attribute this difference to whether the CSC mechanism functioned or not depending on STO or Si substrates, rather than to the difference of the analysis techniques.

The composition of a PTO film prepared by PLD would be affected by the balance among multiple factors such as:

(i)

selective ablation at the target (which would drive the film composition to Pb-rich);

(ii)

re-evaporation from the substrate (to Pb-poor);

(iii)

the CSC mechanism (to stoichiometry).

In the case of epitaxial films grown in O₂ of 100 mTorr at 550 °C, the effect of (iii) was revealed to be more significant than (i) and (ii), meaning that the laser-induced off-stoichiometry on the single PTO target is not a severe issue in the composition viewpoint. However, the pie-chart target, which can compensate for the composition deviation due to the effect of (i), is useful when one attempts to deposit PTO films in a non-epitaxial manner. The robustness and pillar-free surface are also the advantages of the pie-chart target over the PTO target. The pie-chart target strategy would be generally applicable to materials containing elements with high vapor pressure, and probably to other physical vapor deposition techniques such as sputtering.

5. Conclusions

Irradiation of a PTO target with a pulsed laser beam of any fluence causes selective ablation of Pb, which induces a Ti-rich cone-shaped pillar structure, affecting the stability and composition control of the thin film deposition process. On the other hand, the PbO and TiO₂ targets are robust and free from cone-shaped pillars after laser irradiation. This allows for a better control of the deposition process.

We prepared PTO thin films by the PLD method using either a PbO–TiO₂ pie-chart targets or PTO single target. On STO (001) substrates, which allow the epitaxial growth, the PbO–TiO₂ pie-chart target with the areal ratio from 3:5 to 5:3 provided films with composition, crystallinity, flatness, and ferroelectric properties almost independent of the areal ratio. Even the PTO single target produced films with properties similar to those from pie-chart targets, despite the surface off-stoichiometry caused by the laser irradiation history. To explain this coincidence a compositional self-control (CSC) mechanism was suggested to occur during the growth of epitaxial PTO films.

In conclusion, the use of PbO–TiO₂ pie-chart target produces high-quality epitaxial films comparable to the ones from a PTO single target. The advantages of the pie-chart target include the robustness of the target and the long-term controllability of ablated compositions. In the preparation of films containing volatile species (Pb, Bi, Zn, Sn, Li, …) by PVD methods, the mosaic-target scheme is a realistic solution.