Pulsed laser deposition (PLD) is one the most successful thin film growth methods for oxides and other materials with very high melting points. The main merit of PLD is the ability to grow thin films of almost any material for which stoichiometric bulk evaporation sources are available. This universal nature of PLD stems from the near-stoichiometric transfer of material from the source to the film in the laser ablation process. However, despite this universal nature of PLD, it is also generally understood that the highest crystalline quality in thin films can be achieved by molecular beam epitaxy (MBE), rather than PLD. The inferior quality of PLD-grown films often manifests itself in the form of higher defect density, lower carrier mobility in semiconductors, inferior dielectric permittivity compared to bulk crystals, and in systematic shifts of film lattice parameters compared to known bulk reference values. These functional property differences in PLD films can be traced back to stoichiometry shifts in the deposition process and mechanical damage to the film due to the high-energy plasma in the laser ablation plume.

The stoichiometry shift may be related to nonstoichiometric evaporation of different elements from an ablation target, to a difference in the scattering spread of atoms with different weight in the gas phase as the ablation plume expands and interacts with the ambient background gas, or to selective re-evaporation of some atomic species from a film surface at the growth temperature. Depending on which mechanism is responsible, the target composition, ablation laser pulse power density, or the growth temperature and growth rate can be adjusted to correct for the stoichiometry errors.

A more difficult question is how to handle the formation of point defects in the film due to the high kinetic energy of atoms that impinge on the film surface. Time-of-flight measurements have shown that the kinetic energy of atomic species in a laser ablation plume can reach 100 eV, depending on the energy transfer from laser light to the expanding plasma. This energy is sufficient to cause sputtering damage on the film surface and thus leads to the formation of point defects that are not related to any stoichiometry errors. For oxide thin films, such point defect damage can be detected as an expansion of the film lattice compared to known bulk reference value. An effective way to reduce such sputter damage is to reduce the kinetic energy of the plume by introducing a dense background gas. As illustrated in Fig. 1, the position of a SrTiO₃ film x-ray diffraction (XRD) peak changes as a function of the ambient oxygen pressure. An overlap of the homoepitaxial film peak with the SrTiO₃ substrate peak was obtained at 100 mTorr oxygen pressure. Using very high oxygen pressure during film growth is not a good solution because it reduces the deposition rate, may cause gas-phase stoichiometry errors, and degrades the surface flatness of oxide films.

Fig. 1. X-ray diffraction patterns of homoepitaxial SrTiO₃ films grown by PLD at several oxygen pressures. The film peaks separate from the substrate position 46.4° due to lattice expansion caused by point defects. An overlap with the substrate is achieved only at very high oxygen pressure (100 mTorr).

In this work, plasma kinetic energy moderation was therefore done by introducing He gas into the process chamber to a pressure of 600 mTorr [1]. Despite the high pressure, the presence of He does not reduce the deposition rate significantly and allows for independent adjustment of the reactive oxygen pressure while the kinetic energy of the plume can be reduced by an order of magnitude. As the maximum energy of plasma species drops below about 20 eV, sputter damage of the thin film is nearly eliminated. A dramatic change in the film structure can be seen in the reciprocal space maps in Fig. 2. The map shown in panel (a) is for a homoepitaxial SrTiO₃ film grown at an oxygen pressure of 10^-5 Torr. The film diffraction peak is clearly shifted away from the substrate, indicating that the film lattice is strongly expanded. Separate composition analysis showed that the film was very close to stoichiometric and a Sr/Ti ratio error cannot be the cause of the lattice expansion. In contrast, panel (b) shows a map for a film grown in a mixture of 10^-5 Torr of oxygen and 600 mTorr of He. No separate film peak is visible in this case since the homoepitaxial film peak overlaps exactly with the substrate peak.

Fig. 2. Reciprocal space maps of homoepitaxial SrTiO₃ films grown in (a) 10^-5 Torr of O₂ and (b) mixture of 10^-5 Torr O₂ / 600 mTorr He. The film peak exactly overlaps with the substrate peak when the kinetic energy of the plasma plume is moderated by the presence of He in the film growth chamber.

High-pressure He in the PLD chamber thus effectively moderates the kinetic energy of the ablation plume and eliminates the sputtering damage of the oxide film. The kinetic energy is a function of the laser peak power and thus depends not only on the pulse energy but also on the pulse length. The kinetic energy moderation is thus an especially important technique for improving thin film quality when a Nd:YAG lasers with a pulse length of close to 4 ns are used for PLD growth of thin films.

References

• [1] R. Takahashi, T. Yamamoto, and M. Lippmaa, Cryst. Growth. Des. 21, 5017 (2021).

Authors

• R. Takahashi^a, T. Yamamoto^b, and M. Lippmaa
• ^aNihon University
• ^bNagoya University