J. Am. Ceram. Soc., 90 [12] 3800–3803 (2007) DOI: 10.1111/j.1551-2916.2007.02056.x r 2007 The American Ceramic Society Journal Surface Composition and Imprint in CSD-Based PZT Films Aleksey Etin,z Gennady E. Shter,z Reuven Brener,y Sioma Baltianski,z and Gideon S. Graderw,z z Chemical Engineering Department, Technion, Haifa 32000, Israel y Solid State Institute, Technion, Haifa 32000, Israel Reports on PZT films often suggest the contradicting presence of Pb-deficient pyrochlore (Py) and a Pb-rich layer on the surface. We show that standard Ar1 ion sputtering X-ray photoelectron spectroscopy (XPS) depth profiles of PZT films artificially exhibit a Pb-rich surface, independent of actual Pb content of the chemical solution deposition solution. However angle-resolved XPS measurements reveal that films derived from solutions with 10% Pb excess, which give rise to Py surface grains, actually have the expected Pb-deficient surface layer. Alternatively, films derived from solutions with 30% Pb excess are Py free and have Pb-rich surface layer. The Pb-rich films show an increased imprint effect with increasing Pb content. top and bottom interfaces of the film. For example, it has been well recognized that the top surface is susceptible to lead oxide volatility, leading to formation of the Pb-deficient Py.9 In order to compensate for the PbO evaporation, an excess Pb of B10–30 mol% is usually used in the CSD precursor solutions. Evaporation of the excess PbO is believed to provide a ‘‘self-controlling’’ mechanism for the correct final PZT stoichiometry.1 This occurs because the PbO vapor pressure is two orders of magnitude higher over PbO than over the final PZT perovskite (Per) phase. Thus, it was suggested that after Per crystallization, residual PbO is quickly evaporated if a sufficiently high temperature is used.1 However, as suggested by Watts et al.,7,10 the ambipolar diffusion of Pb21 and O2 in the film during evaporation of PbO excess can cause imprint effect. As the secondary phases degrade the film properties, the homogeneity of the perovskite is very important.3,4,8 However, common elemental microanalysis such as X-ray photoelectron spectroscopy (XPS) and Auger electron spectroscopy usually reveal that PZT films are not homogeneous.3,10–15 A Pb-rich layer is consistently found on the films’ surface in all these studies. This is particularly odd since the Py, which develops under Pb-deficient conditions, is usually observed on the surface. Currently, there is no explanation for this discrepancy. Moreover, the effect of Pb-rich surface layer on the film properties is not well understood. It was proposed by Watts et al.7 that a built-in field can be created at the top interface due to a faster diffusion of Pb21 ions to the surface as compared to the O2 ions. However, to our best knowledge there are no experimental measurements that correlate the preparation conditions with the Pb inhomogeneity in the films and with the resulting ferroelectric properties. The main goal of this paper is therefore to study the Pb content in the films in relation to different processing conditions and to correlate the Pb content with the ferroelectric properties. The obtained results are important for understanding the origin of the imprint effect, thus pointing to the processing conditions that avoid it, and improving the PZT films’ performance. I. Introduction I N the last two decades there has been much interest in PZT (PbZrxTi1 xO3) films due to their special properties and numerous potential applications.1 One of the potentially attractive ways of producing these films is the chemical solution deposition (CSD) method.2 However, this method is still not widely accepted due to insufficient homogeneity of the resulting films on the nano-3,4 and macroscale5 as we have recently described. As a result of these inhomogeneities, the ferroelectric properties are significantly degraded. Specifically, one of the main problems in the Ferroelectric Random Access Memory (FeRAM) applications is an imprint effect, which gives rise to an asymmetric hysteresis loop.6 This effect complicates the design and functioning of FeRAM and reduces its reliability. Currently, the origin of this effect is not well understood; however, it is quite plausible that the imprint is related to the inhomogeneity in the films, including the nanoscale nonstoichiometry.7 The elemental stoichiometry in the film has significant impact on the ferroelectricity and therefore it is crucial for the optimization of the PZT film properties. The Zr/Ti stoichiometry is usually maintained close to the morphotropic phase boundary at 0.52/0.48, since the ferroelectric and piezoelectric properties are maximized at this ratio.8 The Pb/(Ti1Zr) ratio should be close to unity since the perovskite (Per) structure of PZT does not tolerate large excess or deficiency of the Pb content.9 Deviations from unity would favor formation of the pyrochlore (Py) phase that has a wider stoichiometry range.9 During film processing, temperatures as high as 6001–7001C are usually used for complete decomposition of metalorganic precursors and crystallization of the PZT film. At these temperatures, local stoichiometry variations are potentially more prone to occur at the II. Experimental Procedure The CSD precursors were prepared from the following starting materials: lead acetate trihydrate Pb(O2C2H3)2
3H2O (99.5%, Fluka, Seelze, Germany), titanium iso-propoxide Ti(OC3H7)4 (97%, Aldrich, St. Louis, MO) and zirconium acetyl acetonate Zr(O2C5H8)4 (98%, Sigma, St. Louis, MO). The solvent was 1,3propanediol (99%, Fluka) while acetylacetone (98%, Fluka) was employed as a stabilizer of Ti iso-propoxide. The synthesis procedure was reported in detail previously.3 Precursors were diluted with n-octanol5 and PZT thin films were deposited by spin coating on Pt/Ti/SiO2/Si substrates in class 100 clean room conditions. One cycle consisted of three coatings with intermediate heating for 1 min at 3001C and final annealing at 6501C for 10 min. Two series of films were prepared: with 10% and 30% Pb excess. In each series, samples resulting from one, two, and three coating cycles with corresponding film thicknesses of B0.35, B0.7, and B1 mm were studied. At least two samples were prepared at similar processing conditions. Gold electrodes with a 500-mm pad diameter were deposited on the films using a M. Kuwabara—contributing editor Manuscript No. 23107. Received April 19, 2007; approved July 26, 2007. This investigation was supported by the MOEMS Consortium of the Israel Ministry of Industry and Trade and Technion’s fund for the promotion of research. Additional support is from the Center for Adsorption in Science of the Ministry of Immigrant Absorption State of Israel and in part by a joint grant from the Committee for Planning and Budgeting of the Council for Higher Education under the framework of the KAMEA and GILADI Programs. w Author to whom correspondence should be addressed. e-mail: grader@tx.technion.ac.il 3800 December 2007 3801 Surface Composition and Imprint shadow mask technique. The CV characteristics were measured by a Precision Impedance Analyzer, Agilent 4294A (Santa Clara, CA). The applied field for this measurement consisted of a DC bias in the 710 V range, a scanning time of 5 s, and a low signal oscillator with 50 mV rms and frequency of 1 kHz. The film surface and cross-sectional microstructure were studied with a high-resolution field emission gun (FEG)-equipped scanning electron microscope (LEO 982, Cambridge, U.K.). Conditions included 4-kV accelerating voltage and 3–4-mm working distance, using the in-lens detector of secondary electrons. XPS was used to determine the elemental composition of films. Measurements were carried out with a Thermo VG Scientific Sigma Probe (East Grinstead, U.K.) using a monochromatized AlKa B(1486.5 eV) X-ray beam of 400 mm for photoelectron excitation. The spectrometer was calibrated to position the 3d5/2 line of sputter-cleaned Ag at a binding energy of 368.3 eV. A 2 kV rastered Ar1 ion beam was used for PZT film sputtering. The Pb4f, Zr3d, Ti2p, and O1s spectra were used for calculation of film composition as a function of depth. Atomic concentration was calculated using a standard bulk PbZr0.5Ti0.5O3 sample (verified with X-ray diffraction) sputtered at the same Ar1 ion bombardment conditions. The depth profiles were measured in the standard mode of operation, collecting the photoelectrons at a wide angle of 537301, relative to the normal. Etching rate calibration was carried out with a 20-nm SiO2 standard. The angle-resolved measurements did not involve sputtering of the samples (unlike the depth profile). The as-received samples were measured in the bulk- and surface-sensitive modes of operation acquiring the spectra at take-off angles of 30.577.51 and 75.577.51, respectively. In the latter experiment, atomic concentration was calculated using a scratched bulk PbZr0.5Ti0.5O3 as a standard. The typical error of the elemental concentration was below 5%. Each measurement except the depth profiles was reproduced on the duplicated samples. III. Results and Discussion The typical microstructure of films prepared with 10% and 30% Pb excess is shown in Fig. 1. It can be seen that two types of grains are found on the surface of 10% films in Fig. 1(a): the small grains with B10-nm diameter above the underlying layer having grains with B100-nm diameter. The small grains are ascribed to the Py phase while the large grains are the perovskite phase (Per).4,9 As explained in the introduction, the current consensus is that the Py phase forms due to Pb loss during hightemperature processing. The Pb loss ultimately leads to Pb deficiency that is beyond the Per stoichiometry limits and therefore nonstoichiometric Py is formed.9 Because the Pb loss is compensated by 30% Pb excess, no Py grains are observed at any thickness as shown in Figs. 1(b) and (d). We have previously shown4 that the largest Pb loss occurs in the first layer due to lead diffusion into the substrate. Therefore there is a decrease of the Py fraction with increasing film thickness as can be seen when Figs. 1(a) and (c) are compared. By analogy, the lead can accumulate in the 30% Pb excess films with increasing film thickness. After deposition of three layers, particles with a characteristic length of several 100 nm and width of B10–20 nm are observed on the surface, as highlighted in Fig. 1(d). These particles consistently appear in the 30% films only after deposition of at least three layers and therefore we believe that these particles are formed due to segregation of Pb excess. Because the goal of this paper is to elucidate whether there is Pb excess or deficiency relative to the stoichiometric Per, we Fig. 1. Typical morphology of PZT films with different Pb excess and thickness: (a) one layer, 10% Pb excess; (b) one layer, 30% Pb excess; (c) three layers, 10% Pb excess; (d) three layers, 30% Pb excess, with highlighted particles on the surface 3802 Vol. 90, No. 12 Journal of the American Ceramic Society—Etin et al. 2 10 1.6 30% Pb, 1 layer 8 10% Pb, 3 layers Pb/(Zr+Ti) Atomic ratio Pb /(Ti+Zr) 30% Pb, 3 layers 10% Pb, 1 layer 6 bulk Bulk 1.2 0.8 4 0.4 2 0 10% Pb excess 0 20 40 Depth (nm) 60 80 Fig. 2. Depth profile of Pb/(Ti1Zr) ratio in bulk PZT and films with different Pb excess and thickness. concentrate on the Pb/(Ti1Zr) atomic ratio. We expected that the sputtering XPS depth profile will reveal the differences in the Pb content between the samples. However, as shown in Fig. 2, there is almost no distinction in the Pb/(Ti1Zr) profile between the various samples, including bulk PZT. Both 10% and 30% films showed decrease in the Pb content as a function of thickness. The results indicate that there is a huge Pb excess up to 700% on the surface of each film regardless of the starting composition of the CSD solution. After etching of B30 nm, the Pb/ (Ti1Zr) ratio decreases to unity as expected for stoichiometric Per phase. This result is consistent with the depth profiles elsewhere3,10–15 that show Pb excess on the films surface. However, this result is inconsistent with HRSEM analysis that showed Py particles on the 10% films surface. We associate this contradiction with preferential etching that frequently occurs in Pb containing ceramics.11,13,16 As a result of the preferential etching, the analyzed and actual composition are different. In order to avoid this phenomenon, XPS analysis without etching was performed in the present work. The depth of XPS analysis decreases with the take-off angle (relative to the normal). Thus, information from deeper parts of the film is measured at smaller take-off angles.16 Therefore two modes of measurement were used: at 30.51 and 75.51 that are characteristic of ‘‘bulk’’ and ‘‘surface’’ compositions, respectively. It should be noted that even measurement in the ‘‘bulk’’ mode characterizes composition of a layer that is thinner than B10 nm. The region with Pb4f and Zr3d peaks in the ‘‘surface’’ and ‘‘bulk’’ modes of the one-layer PZT film prepared with 10% Pb excess and the three-layer 30% PZT film is shown in Fig. 3. In order to compare the cation ratios in the different measurement modes, the Zr3d peak was normalized in all the spectra. The comparison of ‘‘surface’’ and ‘‘bulk’’ spectra in the 10% Pb excess film clearly shows that the Pb peak is lower in the ‘‘sur- 10% Pb Pb4f 30% Pb Pb4f Zr3d Intensity (a.u.) Zr3d Surface Surface Bulk (a) 160 130 Bulk (b) 100 190 160 130 30% Pb excess Fig. 4. Average Pb/(Zr1Ti) atomic ratio of six samples in the ‘‘surface’’ and in the ‘‘bulk’’ in films with 10% and 30% Pb excess. 0 190 Surface 100 Binding energy (eV) Fig. 3. X-ray photoelectron spectra in ‘‘bulk’’ and ‘‘surface’’ modes of PZT films with 10% and 30% Pb excess. face’’ mode than in the ‘‘bulk’’ mode. Therefore it can be concluded that this film has a Pb-deficient layer on the surface. This observation is consistent with the Py phase on the surface of this film as was shown in Fig. 1(a). The main difference in the spectrum of 30% Pb excess film is the increase in the Pb signal in both ‘‘surface’’ and ‘‘bulk’’ modes as compared to the 10% film. This is expected since a larger Pb excess was used during film preparation. In addition, the Pb peak in the 30% film is higher in the ‘‘surface’’ mode than in the ‘‘bulk’’ mode. Therefore, unlike the 10% film that showed a Pb-deficient surface, the 30% film shows a Pb-rich surface. The quantitative Pb/(Ti1Zr) ratios in films prepared with 10% and 30% Pb excess are shown in Fig. 4. The atomic ratios in Fig. 4 were obtained using Pb4f, Zr3d, and Ti2p spectra in different measurement modes and represent an average of the six samples in each case. The measurements do not involve etching of the samples (unlike the depth profile measurement). The difference between the surface and bulk modes are in the take-off angle of 30.577.51 and 75.577.51, respectively, in which the data are collected. The dashed line in Fig. 4 represents an ideal Per phase where the ratio of Pb/(Ti1Zr) equals unity. The Pb/(Ti1Zr) ratio indicates the phase content on the surface: PbO segregation in indicated by a ratio 41, while a significant amount of Py is indicated by a ratio o1. It can be seen that in contrast to the sputtering XPS experiment, there is a significant difference between the films prepared with 10% and 30% of Pb excess in angle-resolved measurement. The Pb/ (Ti1Zr) ratio in 10% films with surface Py is below unity indicating Pb deficiency, while 30% films show Pb excess. In the films with 10% Pb excess, the average ‘‘bulk’’ ratio is consistently higher than the ‘‘surface’’ ratio. This effect is reversed in the films with 30% Pb excess. Thus we can conclude that the surface of the 10% films is Pb deficient while the surface of the 30% films is Pb rich. It can be noted from Fig. 2 that the asymptotic cation ratio in all films is unity, indicating that the nonstoichiometry is limited to the surface layer. It should be emphasized that the difference of the Pb/(Ti1Zr) atomic ratio at 0-nm depth in Fig. 2 and in the ‘‘surface’’ mode in Fig. 4 is due to the preferential lead etching in the depth profile case, which gives rise to an apparently high and incorrect Pb content on the top 40 nm. The average Pb/(Ti1Zr) ratio on the surface of the 10% film is 0.8. The Per phase is not stable at this stoichiometry9 and therefore the Py phase is expected on the surface of these films. This is consistent with the Py particles in the 10% films shown in Figs 1(a) and (c). The average Pb/(Ti1Zr) ratio of 1.7 on the surface of the 30% film is also beyond the Per stability limits. Thus, segregation of PbO is expected in these films, which correlates with the appearance of the particles on the surface of the three-layer-thick film with 30% Pb excess in Fig. 1(d). The CV characteristics of the films prepared with 10% and 30% Pb excess are presented in Fig. 5. It can be seen that in both cases the dielectric constant increases with thickness, consistent with previous results.4 The 10% film, Fig. 5(a), showed symmetric CV curves, as usually observed for classical ferroelectric films.8 However, the 30% films, Fig. 5(b), exhibited a shift of the December 2007 3803 Surface Composition and Imprint 2000 1500 10% Pb 30 % Pb 1600 1200 1200 k k 3 layers 900 3 layers 800 600 2 layers 2 layers 400 300 1 layer (a) 0 0 −20 −10 0 Bias (V) 1 layer (b) 10 20 −20 −10 0 Bias (V) 10 20 Fig. 5. Dielectric constant as a function of bias voltage at different film thicknesses: (a) 10% Pb excesses, (b) 30% Pb excess. curves’ intersection point to a negative bias as shown by a dashed arrow. This effect can be explained by a permanent dipole that is ‘‘imprinted’’ into the surface region due to the higher mobility of the Pb12 ions to the surface relative to that of the O 2 ions as originally suggested by Watts et al.7 The ‘‘imprinted’’ dipole causes the shift of the signal in the observed direction because the domains in these films switch more easily when a positive bias is applied. The results were well reproducible with repetitive measurements of the fresh samples as well as on samples that we stored for 1–2 months. It is interesting that the negative shift increases with layer number in the case of 30% films. This is another evidence of Pb excess accumulation that increases with each layer, consistent with the HRSEM image in Fig. 1(d). The presence of excess lead is consistent with our previous results that showed a decreased grain size in 30% films as compared with the 10% films, due to inhibition of grain growth in these films by the excess PbO.4 Thus it can be concluded that as a result of Pb accumulation in the film the imprint effect is amplified. On the other hand, 30/10/10 films with 30% Pb excess in the first layer (to compensate for the higher Pb loss into the substrate) and 10% in the following layers4 showed similar CV characteristics (no imprint effect) and grain size4 to the 10% films, indicating that there was no Pb accumulation in this case. IV. Summary This paper suggests that the surface of the sol–gel-derived PZT films is not always Pb rich as often reported in the literature.3,10– 15 Preferential etching during depth profile measurement appears to mask the distinction between the surfaces with different compositions, while angle-resolved XPS measurements enable a distinction between them. The presence of the Py grains on the surface is accompanied with the Pb/(Ti1Zr) ratio below one. On the other hand, the Py-free films prepared with 30% Pb excess showed a Pb/(Ti1Zr) ratio above one indicating Pb segregation on the surface. The CV characteristics showed symmetric curves in 10% films while curves of 30% films were highly asymmetric due to an imprint effect. The curves intersection shifted to the negative bias in the 30% films, indicating a positive built-in field in the film. This effect can be explained by the faster diffusion to the film surface of Pb21 compared with O2 ions, as was previously suggested by Watts.7 Following the CV pattern of the film allows fine-tuning of the Pb surface content. This allows avoiding imprint effect and therefore improves the quality of PZT films in FeRAM and other industrial applications. Acknowledgments We would like to thank Mr. M. Kabla from RAFAEL for preparation of Pt electrodes. References 1 P. Muralt, ‘‘PZT Thin Films for Microsensors and Actuators, Where do We Stand?,’’ IEEE Transact. Ultras., Ferroel., Freq. Contr., 47, 903–14 (2000). 2 R. W. Schwartz, T. Schneller, and R. Waser, ‘‘Chemical Solution Deposition of Electronic Oxide Films,’’ J.C.R. Chimie, 7, 433–61 (2004). 3 A. Etin, G. E. Shter, G. M. Reisner, S. Baltianski, and G. S. Grader, ‘‘Controlled Elemental Depth Profile in Sol–Gel Derived PZT Films,’’ J. Am. Ceram. Soc., 89, 2387–93 (2006). 4 A. Etin, G. E. Shter, G. M. Reisner, and G. S. Grader, ‘‘Interrelation of Ferroelectricity, Morphology and Thickness in Sol–Gel Derived PZT Films,’’ J. Am. Ceram. Soc., 90, 77–83 (2007). 5 A. Etin, G. E. Shter, V. Gelman, and G. S. Grader, ‘‘Uniformity, Composition and Surface Tension in Solution Deposited PZT Films,’’ J. Mater. Res., 22, 103–12 (2007). 6 J. Lee, C. H. Choi, B. H. Park, T. W. Noh, and J. K. Lee, ‘‘Built-in Voltages and Asymmetric Polarization Switching in Pb(Zr,Ti)O3 Thin Film Capacitors,’’ Appl. Phys. Lett., 72, 3380–2 (1998). 7 B. E. Watts, F. Leccabue, G. Tallarida, S. Ferrari, M. Fanciulli, and G. Padeleti, ‘‘Surface Segregation Mechanisms in Ferroelectric Thin Films,’’ J. Electroceram., 11, 139–47 (2004). 8 D. Damjanovic, ‘‘Ferroelectric, Dielectric and Piezoelectric Properties of Ferroelectric Thin Films and Ceramics,’’ Rep. Prog. Phys., 61, 1267–324 (1998). 9 C. G. Levi, ‘‘Metastability and Microstructure Evolution in Synthesis of Inorganics from Precursors,’’ Acta Mater., 46, 787–800 (1998). 10 B. E. Watts, F. Leccabue, G. Bocelli, G. Padeletti, S. Kaciulis, and L. Pandolfi, ‘‘Lead Enrichment at the Surface of Lead Zirconate Titanate Thin Films,’’ J. Eur. Ceram. Soc., 25, 2495–8 (2005). 11 S. M. Mukhopadhyay and T. C. S. Chen, ‘‘Surface Properties of Perovskites and their Response to Ion Bombardment,’’ J. Appl. Phys., 74, 872–6 (1993). 12 O. Sugiyama, K. Murakami, and S. Kaneko, ‘‘XPS Analysis of Surface Layer of Sol-Gel-Derived PZT Thin Films,’’ J. Eur. Ceram. Soc., 24, 1157–60 (2004). 13 J. N. Kim, K. S. Shin, D. H. Kim, B. O. Park, N. K. Kim, and S. H. Cho, ‘‘Changes in Chemical Behavior of Thin Film Lead Zirconate Titanate During 1 Ar -Ion Bombardment Using XPS,’’ Appl. Surf. Sci., 206, 119–28 (2003). 14 C. R. Cho, ‘‘Surface Chemical Bonding States and Ferroelectricity of Pb(Zr0.52Ti0.48)O3 Thin Films,’’ Cryst. Res. Technol., 35, 77–86 (2000). 15 Z. Qian, D. Xiao, J. Zhu, Z. Li, and C. Zuo, ‘‘X-ray Photoelectron Spectroscopy and Auger Electron Spectroscopy Studies of Ferroelectric (Pb,La)TiO3 Thin Films Prepared by Multi-Ion-Beam Reactive Cosputtering Technique,’’ J. Appl. Phys., 227, 224–7 (1993). 16 J. C. Vickerman, Surface Analysis—The Principal Techniques, 1st edition, John Wiley & Sons Ltd, Chichester, U.K., 1997. &