Pulsed laser ablation-deposition of La0.5Sr0.5CoO3 ...

Journal of

MATERIALS RESEARCH

Welcome

Comments

Help

Pulsed laser ablation-deposition of La0.5 Sr0.5 CoO3 for use as electrodes in nonvolatile ferroelectric memories R. Dat, O. Auciello,a) D. J. Lichtenwalner, and A. I. Kingon Department of Materials Science & Engineering, North Carolina State University, Raleigh, North Carolina 27695-7919 (Received 29 October 1994; accepted 21 February 1996)

La0.5 Sr0.5 CoO3 (LSCO) thin films have been deposited on (100) MgO substrates using pulsed laser ablation-deposition (PLAD). The crystallographic orientation of LSCO was found to be dependent on the surface treatment of (100) MgO prior to deposition. PLAD deposition parameters were optimized to yield LSCO films with an ˚ A smooth surface morphology was reproduced as RMS surface roughness of 40–50 A. long as the oxygen content of the LSCO target was preserved. Otherwise, “splashing” occurred which resulted in the transfer of condensed particles from molten spherical globules of LSCO from the target to the substrate. Splashing was subsequently eliminated and smooth surface quality was restored after annealing the LSCO target at 550 ±C in oxygen for 3 h. Optical emission spectroscopy (OES) of the LSCO’s plume identified excited atomic cobalt neutrals, excited singly ionized strontium and lanthanum, and excited molecular LaO species. Oxygen interaction with the plume produced no new species. Furthermore, the OES data suggest that the observed LaO molecules were not created by the chemical reaction between La and O2 during ablation, but were ejected directly from the target during the PLAD process.

I. INTRODUCTION

The wide range of electrical resistivities possible in thin film oxides makes these materials very attractive for various device applications. At one extreme, they can be used as superconductors, while at the other extreme their insulating properties can be employed in charge storage devices. Between these two ranges of applications, thin film oxides can also be utilized for their metallic characteristics. In nonvolatile memory applications, both the conducting and insulating properties of oxides are utilized. The ferroelectric material (insulator) is sandwiched between two conducting electrodes. Recent research has shown that conducting oxide electrodes are essential to the reliability of ferroelectric capacitors.1–7 PZT-based ferroelectric capacitors with metallic top and/or bottom electrodes show a dramatic decrease of switched polarization (fatigue) after about 106 bipolar voltage pulses. The reason for this decrease is not known; however, it is speculated that this behavior is related to the quality of the ferroelectric/electrode interface, including the presence of oxygen vacancies, as well as other charged defects in the ferroelectric layer. While oxide electrodes are essential in controlling the long-term properties of ferroelectric capacitors, their successful utilization in memory devices will ultimately a)

Also, MCNC, Electronics Technology Division, Research Triangle Park, North Carolina 27709-2889.

1514

J. Mater. Res., Vol. 11, No. 6, Jun 1996

depend on factors such as their chemical reaction with the substrate and ferroelectric material at the deposition temperature, their ability to maintain the desired compositional and electrical properties after the deposition cycle, their structural compatibility with the perovskite phase of the ferroelectric layer so as to minimize interface states, their surface roughness, and their processing temperature. Some of these factors prompted the use of YBa2 Cu3 O72x (YBCO), prepared by pulsed laser ablation-deposition (PLAD), as an appropriate electrode material for nonvolatile memory capacitors.5 While YBCO thin films proved successful in alleviating some of the fatigue problems in PZT-based capacitors, they presented various problems; namely, they required accurate control of the deposition and postprocessing parameter (especially oxygen partial pressures) in order to achieve good electrical properties, they required a high deposition temperature (800 ±C versus 600 ±C for PbZr0.53 Ti0.47 O3 (PZT), they exhibited rough surface morphologies due to the generation of “boulders” and other smaller sized particles during the PLAD process, and they needed to be c-axis oriented since the resistivity of YBCO is lowest along the (001) direction.8 More recently, other conducting oxides such as SrRuO3 ,9 Sr12x Cax RuO3 ,10 and La12x Srx CoO3 2,11–14 have emerged as electrode materials. While SrRuO3 has a resistivity that is comparable to that of LSCO, its deposition temperature is approximately 150 –200 ±C higher than that of LSCO. On the other hand, Sr12x Cax RuO3  1996 Materials Research Society

Downloaded from https://www.cambridge.org/core. Eugene McDermott Library, University of Texas at Dallas, on 03 Mar 2018 at 16:11:32, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1557/JMR.1996.0189

R. Dat et al.: Pulsed laser ablation-deposition of La0.5 Sr0.5 CoO3

has a resistivity and a deposition temperature that are higher than that of LSCO. LSCO is one of the most conductive oxide-perovskites with a room temperature bulk resistivity of 90 mV-cm when x 0.5.8 Because of its high electrical conductivity, LSCO is also a good thermal conductor. This property is important during the PLAD process since it is one of the factors that determines the amount of debris that is ejected from the target’s surface as a result of thermal shock caused by short and intense laser pulses. The focus of this paper is to investigate the growth and characterization of LSCO layers prepared by the PLAD process on MgO substrates. The reason for using MgO substrates as opposed to others more technologically relevant, such as Si, is because a large number of studies have been performed on MgO substrates, which can be used for comparison, and because it is easier to grow highly oriented films on single crystal MgO than on Si substrates. Attention is given to the conditions necessary to deposit thin LSCO layers with minimal surface roughness. In addition, the effect of MgO substrate preparation on the resulting orientation of the LSCO layer is discussed. Finally, the characteristics of the laser ablated flux of LSCO is investigated as a function of target-substrate distance and oxygen partial pressure. II. EXPERIMENTAL

The experimental setup used for this investigation is briefly discussed here. A more detailed description of the PLAD system is published elsewhere.15,16 A pulsed laser beam was generated by a KrF excimer laser operating at a wavelength of 248 nm. The pulsed output power was 400 mJ (1.1 Jycm2 ), at a repetition frequency of 3 Hz, and a pulse width of 25 ns. The laser beam impacted a La0.5 Sr0.5 CoO3 target that was mounted on a rotating holder assembly. The substrate and target holder assemblies were each attached to linear motion drives for positioning during deposition. Substrates were radiatively heated to approximately 600 ±C in an ambient of 300 mTorr of oxygen during deposition. The acquisition of the emission spectra of the excited species in the ablated plume was achieved by a vacuum feedthrough fiber optic cable (0.4 mm diameter) positioned perpendicularly to the plume’s axis and mounted on a linear motion drive. The optical fiber was recessed in a stainless steel tube to prevent deposition on the tip of the fiber. In this arrangement, the tube’s aperture was used as an acceptance slit. The other end of the fiber optic cable was connected to a computer-controlled optical multichannel analyzer (OMA) equipped with a 512-element photodiode array detector at the focal plane of a 0.25 m monochromator with a 150 grooveymm diffraction grating (0.5 nm spectral resolution).

The LSCO films were examined by x-ray diffraction (XRD) for crystallographic orientation. Scanning electron microscopy (SEM) and atomic force microscopy (AFM) were used to evaluate the surface morphology of the deposited films. Resistivity measurements were made using the four-point probe technique. III. RESULTS AND DISCUSSION

In Fig. 1(a), the XRD spectrum of an LSCO film grown on an as-purchased (100) MgO substrate (without any pretreatment prior to deposition) is compared with two other cases where (b) the MgO substrate was ion beam etched (IBE) with 500 eV Ar1 ions for 5 min, and (c) the substrate was ion beam etched (IBE) and then annealed at 1140 ±C in oxygen for 14 h prior to LSCO deposition. Procedures for the preparation of MgO substrates with smooth surfaces have been developed by several groups, including ours.12,17,18 If the substrate is annealed and/or IBE, a highly oriented LSCO layer is produced on (100) MgO, as indicated by phi-scan and rocking curve analysis, the latter done on the (001) peak, which showed a FWHM of ,1±. The high orientation of the LSCO layers produced with our PLAD method has also been confirmed by TEM studies that showed electron diffraction patterns characteristic of epitaxial films (see Fig. 7 of Ref. 19). If the substrate is not annealed or IBE, a polycrystalline LSCO film results with preferred nucleation along (100) and (110) crystallographic directions. The reason for achieving highly oriented LSCO on pretreated (100) MgO substrate may be due to the removal of defects, absorbed gas species, k110l surface steps, and work damage (from sawing and polishing) during the anneal/IBE process.17,18 For nonvolatile ferroelectric memories, it is desirable to have highly oriented LSCO films so that the ferroelectric layer may also be highly oriented and, hence, produce a larger remanent polarization when compared to polycrystalline films. However, with respect to the long-term properties of ferroelectric capacitors (such as fatigue, retention, and imprint), it is not critically important to have oriented films.14 As discussed elsewhere,3 the presence of LSCO as the top and bottom electrodes plays a significant role in controlling the long-term properties of ferroelectric capacitors. Topographical features such as particulates, outgrowth of crystal grains, and globules are commonly found on films produced by PLAD. Some of these morphological features are detrimental to device yield and reliability. In general, the density of these defects can be controlled by optimizing certain deposition parameters (temperature, substrate-target distance, oxygen pressure, laser fluence, etc.). For LSCO thin films, we have observed that the surface morphology can be drastically altered depending on the characteristics of the target


1515



FIG. 1. XRD spectrum of LSCO grown on (a) as-purchased (100) MgO substrate, (b) ion beam etched (100) MgO, and (c) annealed and ion beam etched (100) MgO. FIG. 2. LSC surface topography by atomic force microscopy for PLAD without splashing.

material. As mentioned above, the electrical and thermal conductivities of the target material are factors that determine the amount of debris that is ejected from the target during ablation. The surface morphology of an LSCO layer ˚ thick) grown on a (100) MgO substrate (,1000 A (annealed at 1140 ±C in oxygen for 14 h) is given in Fig. 2. The RMS surface roughness is approximately ˚ as determined by AFM. The deposition 40 –50 A conditions are temperature at 600 ±C, oxygen pressure of 300 mTorr, laser energy density of 1.1 Jycm2 , laser frequency of 3 Hz, and target-substrate distance of 4.5 cm. It should be noted that the surface of the LSCO target was not polished after each deposition and this did not seem to affect the surface quality of the deposited film. The above conditions were routinely used to produce LSCO films with smooth surface morphology. After an extended period of ablating the same LSCO target, the above deposition conditions no longer produced smooth surfaces. Instead, a condition known as “splashing” developed during LSCO deposition by PLAD, which resulted in a significant increase in surface roughness of the deposited layers. Figure 3 shows some of the spherically shaped globules of LSCO (10–15 mm in diameter) that are characteristic ˚ of the splashing process. The LSCO layer (,1000 A thick) shown in Fig. 3 was grown on an MgO substrate prepared under the same conditions as those used for producing the LSCO layer shown in Fig. 2. The LSCO morphology between the globules was about a factor of 2–3 rougher than that observed in Fig. 2. Ferroelectric capacitors fabricated on LSCO electrodes with globules similar to those shown in Fig. 3 resulted in short-circuited devices. The results presented in Figs. 2 and 3 suggest that the properties of the LSCO target changed as a function of ablation time. It is believed that with increased 1516

FIG. 3. SEM micrograph of an LSCO layer showing the splashing phenomenon.

ablation time, the thermal and electrical conductivities of the LSCO target decrease due to the loss of some of its oxygen content. Splashing ensued as a result of an increased thermal shock on the surface of the target. The physical effect of thermal shock on the target is shown in Fig. 4. During normal ablation, the surface of the LSCO target is relatively smooth with the subsurface cracks that accommodate the normal thermal stress caused by intense laser pulses [see Fig. 4(a)]. The surface morphology of the LSCO target after splashing is shown in Fig. 4(b) where extensive recrystallization suggests that the subsurface temperature was greater than that of Fig. 4(a) due to poor thermal conduction. While the effect of oxygen loss on the thermal conductivity




(a)

(b) FIG. 4. (a) Surface of LSCO target after normal ablation and (b) surface of LSCO target after splashing. For both cases, ablation was done at 400 mJ, 3 Hz, and 300 mTorr oxygen.

of the LSCO target has not been quantified, support to the possible explanation for splashing presented above was given by the experiments involving the LSCO target annealing in oxygen at 550 ±C for 3 h. PLAD of LSCO after annealing the target material showed no splashing, and smooth surface morphology was again achieved using the deposition conditions quoted above. However, deterioration of the target characteristics reappeared after subsequent prolonged ablation, indicating the target deterioration and annealing-induced recovery are repeatable. The resistivity of LSCO films on (100) MgO was approximately 200 mV-cm at room temperature, as determined by the four-point probe technique. Optical emission spectroscopy of the LSCO plasma plume was investigated to categorize the type and state of the species in the plume, as well as reveal effects caused by interaction with oxygen. The 350 –650 nm emission

spectra from La0.5 Sr0.5 CoO3 using a laser energy of 400 mJ, pressures ranging from vacuum up to 900 mTorr oxygen, and a repetition rate of 10 Hz is given in Fig. 5. The identification of the spectral lines of neutral and ionized atomic species and of molecular species is made using standard tables.20,21 The fiber optic cable was placed at 0.5 cm from the LSCO’s surface and positioned so that its line of sight was parallel to the target’s surface. The LSCO spectra reveal (a) an abundance of excited species in the blue region; (b) all of the excited species are cobalt and strontium atomic neutrals and singly ionized, respectively; (c) all of the excited lanthanum species are either singly ionized or molecular; (d) no new excited species are created as oxygen is introduced during the PLAD process; and (e) the intensity of the emission lines is not strongly dependent on the oxygen pressure used in this study. The predominance of excited neutral cobalt may be due to Co having a higher ionization energy (7.88 eV) than La (5.57 eV) and Sr (5.69 eV). The abundance of excited ionized La and Sr may be important for the PLAD process. As mentioned earlier, the electrical conductivity of La0.5 Sr0.5 CoO3 is dependent on the Sr content (conductivity is maximum when x 0.5). For LSCO films deficient in Sr or La, it is conceivable that techniques (such as the application of an electric field between substrate and target) may be employed to change the energy and direction of these charged species in the plume and, hence, alter the composition (and resistivity) of the LSCO film. The small dependence of the emission intensity on the oxygen pressure is indicative of minimal reaction between the excited species and atomic/molecular oxygen. The existence of atomic oxygen is not verified in this experiment since their emission wavelength at approximately 777 nm is beyond our detection window.

FIG. 5. Plasma emission spectra resulting from ablating an LSCO target in vacuum, 100 mTorr, 300 mTorr, and 900 mTorr of oxygen. Excited neutrals are identified as (I), while excited singly ionized species are denoted as (II). The ablation of LSC was done at 400 mJ and 10 Hz.


1517



The visual examination of the plume as oxygen was introduced also verified minimal reaction between the species and oxygen. The color of the plume was whiteblue in vacuum and remained the same as oxygen was admitted into the PLAD chamber. This is in contrast to the case of PbZr0.53 Ti0.47 O3 (PZT) where the color of the plume is white-blue in vacuum but changes at its outer periphery to a yellowish-red as oxygen is added.13 Emission bands of diatomic molecules are observed and identified as LaO at 362.2, 365.17, 371.72, and 539.77 nm, as shown in Fig. 5. In light of the results presented in Fig. 5, it is suggested that LaO molecules are ejected from the target during ablation instead of being formed by oxidation of La species within the plume, since the LaO molecules are present in the plume when the ablation is done in vacuum. If on the other hand, it is assumed that the LaO molecules are formed by the combination of La and O atoms during the flight from the target to the substrate, then the peak corresponding to ionized La would be almost nonexistent, since the La species would be paired with oxygen from the target (for every La atom in La0.5 Sr0.5 CoO3 there are six oxygen atoms). Also, when oxygen is intentionally introduced during the ablation process, the relative O : La ratio will be significantly larger than when the ablation is done in vacuum. This would cause the peak intensity from the excited ionized La species to decrease. In reality, this is not the case since the data in Fig. 5 do not support the suggestion that La and O combine during their flight to the substrate. The qualitative data from the optical emission spectroscopy (OES) studies presented above suggest that LaO molecules are mainly ejected from the target during the ablation process. As the distance between the optical probe and the target is increased, no new species are identified (see Fig. 6). At a distance greater than 4 cm, the intensity of the emission spectra is significantly reduced. It should be noted that the emission spectra from the species in the plume are produced by either photon excitation due to interaction with the laser beam or by impact excitation from high velocity electrons emitted from the target during the ablation process. As the target-to-substrate distance is increased, the electrons lose energy due to collisions in the plume and are no longer able to excite species at distances greater than 4 cm in 300 mTorr of oxygen. IV. CONCLUSION

In summary, the experiments discussed in this section indicate that the depletion of oxygen from the LSCO target, during the ablation process, can seriously degrade the surface quality of LSCO films produced by PLAD. Molten globules can be ejected from the LSCO target due to thermal shock caused by the degradation of its electrical and thermal conductivities. This undesirable 1518

FIG. 6. Plasma emission spectra of LSCO as a function of distance from the target’s surface. Ablation done at 400 mJ and 10 Hz in 300 mTorr of oxygen.

effect can be minimized or eliminated by annealing the LSCO target in oxygen for 3 h at 550 ±C. LSCO films grown on (100) MgO substrates show an RMS ˚ The orientation of these films roughness of 40–50 A. depends on the treatment of (100) MgO substrates prior to deposition. Annealing and/or IBE of (100) MgO result in highly oriented LSCO films; otherwise, the films are polycrystalline. OES of the LSCO plume revealed that molecular LaO species originate directly from the target rather than being formed by the reaction between lanthanum and oxygen in the gas phase. Other species detected in the plume are excited atomic cobalt neutrals and excited singly ionized strontium and lanthanum. The existence of ionized Sr and La species suggests that their incorporation into the growing film may be controlled (with an external electric field) to alter the properties of the LSCO film. No new species are created as oxygen gas is allowed to interact with the plume. REFERENCES 1. H. N. Al-Shareef, K. R. Bellur, O. Auciello, and A. I. Kingon, Proc. 5th Int. Symp. on Integrated Ferroelectrics, Colorado Springs, CO (1994). 2. R. Ramesh, H. Gilchrist, T. Sands, V. G. Keramidas, R. Haakensaasen, and D. K. Fork, Appl. Phys. Lett. 63 (26), 3592 – 3594 (1993). 3. D. J. Lichtenwalner, R. Dat, O. Auciello, and A. I. Kingon, Proc. 8th Int. Meeting on Ferroelectricity, Gaithersburg, MD (1994). 4. T. Mihara, H. Watanabe, C. A. Araujo, J. D. Cuchiaro, M. C. Scott, and L. D. McMillan, Proc. 4th Int. Symp. on Integrated Ferroelectrics, Colorado Springs, CO (1992), pp. 137 – 157. 5. R. Ramesh, W. K. Chan, B. Wilkens, H. Gilchrist, T. Sands, J. M. Tarascon, V. G. Keramidas, D. K. Fork, J. Lee, and A. Safari, Appl. Phys. Lett. 61 (13), 1537 – 1539 (1992). 6. S. D. Bernstein, T. Y. Wong, Y. Kisler, and R. W. Tustison, J. Mater. Res. 8, 12 –13 (1993). 7. I. K. Yoo and S. B. Desu, Proc. Int. Symp. on Applications of Ferroelectrics (1992), pp. 225 – 228.




8. J. T. Cheung, P. E. D. Morgan, D. H. Lowndes, X-Y. Zheng, and J. Breen, Appl. Phys. Lett. 62, (17), 2045 – 2047 (1993). 9. X. D. Wu, S. R. Foltyn, R. C. Dye, Y. Coulter, and R. E. Muenchausen, Appl. Phys. Lett. 62 (19), 2434 – 2436 (1993). 10. C. B. Eom, R. J. Cava, R. M. Fleming, J. M. Phillips, R. B. van Dover, J. H. Marshall, T. W. P. Hsu, J. J. Krajewski, and W. F. Peck, Jr. Science 258, 1766 –1769 (1992). 11. It should be noted that some of the original research on LSCO began thirty years ago in relation to the application of LSCO as electrodes for high temperature fuel cells. See, for example, J. B. Goodenough and R. C. Raccah, J. Appl. Phys. 36, 1031 (1963); D. B. Meadowcroft, Nature (London) 226, 847–848 (1970). 12. J. T. Cheung, P. E. D. Morgan, and R. Neurgaonkar, Proc. 4th Int. Symp. on Integrated Ferroelectrics (University of Colorado Press, 1992), p. 158. 13. J. F. M. Cillessen, R. M. Wolf, and A. E. M. De Veirman, Appl. Surf. Sci. 69, 212 –215 (1993). 14. R. Dat, D. J. Lichtenwalner, O. Auciello, and A. I. Kingon, Appl. Phys. Lett. 64, 2673 – 2675 (1994).

15. O. Auciello, L. Mantese, J. Duarte, X. Chen, S. H. Rou, A. I. Kingon, A. F. Schreiner, and A. R. Krauss, J. Appl. Phys. 73, 5197 – 5207 (1993). 16. D. J. Lichtenwalner, O. Auciello, R. Dat, and A. I. Kingon, J. Appl. Phys. 74, 7497– 7504 (1993). 17. S. H. Rou, Thesis (North Carolina State University Library, 1994). 18. R. A. McKee, F. J. Walker, E. D. Specht, and K. B. Alexander, in Epitaxial Oxide Thin Films and Heterostructures, edited by D. K. Fork, J. M. Phillips, R. Ramesh, and R. M. Wolf (Mater. Res. Soc. Symp. Proc. 341, Pittsburgh, PA, 1994), pp. 309 – 314. 19. O. Auciello, N. N. Al-Shareef, K. D. Gifford, D. J. Lichtenwalner, R. Dat, K. R. Bellur, A. I. Kingon, and R. Ramesh, in Epitaxial Oxide Thin Films and Heterostructures, edited by D. K. Fork, J. M. Phillips, R. Ramesh, and R. M. Wolf (Mater. Res. Soc. Symp. Proc. 341, Pittsburgh, PA, 1994), pp. 341–363. 20. A. R. Stringanov and N. S. Sventitiskii, Tables of Spectral Lines of Neutral and Ionized Atoms (Plenum, New York, 1968). 21. R. W. B. Pearse and A. G. Gaydon, The Identification of Molecular Spectra, 3rd ed. (John Wiley & Sons, New York, 1963).


1519
