PLD脉冲激光沉积简介

英文原文：CHAPTER 1Pulsed Laser Deposition of Complex Materials: Progress TowardsApplicationsDAVID P. NORTONUniversity of Florida, Department of Materials Science and Engineering, Gainesville, Florida1.1 INTRODUCTIONIn experimental science, it is a rare thing for a newly discovered (or rediscovered) synthesis technique to immediately deliver both enhanced performance and simplicity in use in a field of accelerating interest. Nevertheless, such was the case with the rediscovery of pulsed laser deposition(PLD) in the late 1980s. The use of a pulsed laser as a directed energy source for evaporative film growth has been explored since the discovery of lasers [Hass and Ramsey, 1969; Smith and Turner, 1965]. Initial activities were limited in scope and involved both continuous-wave (cw) and pulsed lasers. The first experiments in pulsed laser deposition were carried out in the 1960s; limited efforts continued into the 1970s and 1980s. Then, in the late 1980s, pulsed laser deposition was popularized as a fast and reproducible oxide film growth technique through its success in growing in situ epitaxial high-temperature superconducting films [Inam et al., 1988]. The challenges for in situ growth of high-temperature superconducting oxide thin films were obvious. The compounds required multiple cations with diverse evaporative properties that had to be delivered in the correct stoichiometry in order to realize a superconducting film. Simultaneously, the material was an oxide, requiring an oxidizing ambient during growth. Pulsed laser deposition had several characteristics that made it remarkably competitive in the complex oxide thin-film research arena as compared to other film growth techniques. These principle attractive features were stoichiometric transfer, excited oxidizing species, and simplicity in initial setup and in the investigation of arbitratry oxide compounds. One could rapidly investigate thin-film deposition of nearly any oxide compound regardless of the complexity of the crystal chemistry. Significant development of pulsed laser deposition has continued and over the past 15 years, PLD has evolved from an academic curiousity into a broadly applicable technique for thin-film deposition research [Saenger, 1993; Kaczmarek, 1997; Willmott and Huber, 2000; Dubowski, 1988; Dieleman et al., 1992]. Today, PLD is used in the deposition of insulators, semiconductors, metals, polymers, and even biological materials. Few material synthesis techniques have enjoyed such rapid and widespread penetration into research and application venues.Pulsed Laser Deposition of Thin Films: Applications-Led Growth of Functional MaterialsEdited by Robert Eason Copyright # 2007 John Wiley & Sons, Inc.34 PULSED LASER DEPOSITION OF COMPLEX MATERIALS1.2 WHAT IS PLD?The applicability and acceptance of pulsed laser deposition in thin-film research rests largely in its simplicity in implementation. Pulsed laser deposition is a physical vapor deposition process, carried out in a vacuum system,that shares some process characteristics common with molecular beam epitaxy and some with sputter deposition. In PLD, shown schematically in Figure 1.1, a pulsed laser is focused onto a target of the material to be deposited. For sufficiently high laser energy density, each laser pulse vaporizes or ablates a small amount of the material creating a plasma plume. The ablated material is ejected from the target in a highly forward-directed plume. The ablation plume provides the material flux for film growth. For multicomponent inorganics, PLD has proven remarkably effective at yielding epitaxial films. In this case, ablation conditions are chosen such that the ablation plume consists primarily of atomic, diatomic, and other low-mass species. This is typically achieved by selecting an ultraviolet (UV) laser wavelength and nanosecond pulse width that is strongly absorbed by a small volume of the target material. Laser absorption by the ejected material creates a plasma. For the deposition of macromolecular organic materials, conditions can be chosen whereby absorption is over a larger volume with little laser absorption in the plume. This permits a large fraction of the molecular material to be ablated intact. For polymeric materials,transfer of intact polymer chains has been demonstrated. For even ‘‘softer’’materials in which the direct absorption by the laser would be destructive to molecular functionality, the formation of composite ablation targets consisting of the soft component embedded in an optically absorbing matrix has been investigated (see, e.g., Chapter 3).Several features make PLD particularly attractive for complex material film growth. These include stoichiometric transfer of material from the target, generation of energetic species,hyperthermal reaction between the ablated cations and the background gas in the ablation plasma,and compatibility with background pressures ranging from ultrahigh vacuum (UHV) to 1 Torr. Multication films can be deposited with PLD using single, stoichiometric targets of the material of interest, or with multiple targets for each element. With PLD, the thickness distribution from aFigure 1.1 Schematic of the PLD process.WHAT IS PLD? 55 stationary plume is quite nonuniform due to the highly forward-directed nature of the ablation plume. To first order, the distribution of material deposited from the ablation plume is symmetric with respect to the target surface normal and can be described in terms of a cos nðyÞ distribution,where n can vary from ～4–30. However, raster scanning of the ablation beam over the target and/or rotating the substrate can produce uniform film coverage over large areas, and this topic is covered in Chapter 9.One of the most important and enabling characteristics in PLD is the ability to realize stoichiometric transfer of ablated material from multication targets for many materials. This arises from the nonequilibrium nature of the ablation process itself due to absorption of high laser energy density by a small volume of material. For low laser fluence and/or low absorption at the laser wavelength, the laser pulse would simply heat the target, with ejected flux due to thermal evaporation of target species. In this case, the evaporative flux from a multicomponent target would be determined by the vapor pressures of the constituents. As the laser fluence is increased, an ablation threshold is reached where laser energy absorption is higher than that needed for evaporation. The ablation threshold is dependent on the absorption coefficient of the material and is thus wavelength dependent.At still higher fluences, absorption by the ablated species occurs, resulting in the formation of a plasma at the target surface. With appropriate choice of ablation wavelength and absorbing target material, high-energy densities are absorbed by a small volume of material, resulting in vaporization that is not dependent on the vapor pressures of the constituent cations.In pulsed-laser deposition, a background gas is often introduced that serves two purposes. First, the formation of multication thin-film materials often requires a reactive species (e.g., molecular oxygen for oxides) as a component of the flux. The amount of reactant gas required for phase formation will depend on the thermodynamic stability of the desired phase. Interaction of ablated species with the background gas often produces molecular species in the ablation plume. These species facilitate multication phase formation. In addition to actively participating in the chemistry of film growth, the background gas can also be used to reduce the kinetic energies of the ablated species. Time-resolved spectroscopy studies of ablation plume expansion have shown that kinetic energies on the order of several hundred electron volts can be observed [Chen et al., 1996]. A background gas can moderate the plume energies to much less than 1 eV. The vapor formed by laser ablation compresses the surrounding background gas resulting in the formation of a shock wave.Interaction with the ambient gas slows the ablation plume expansion.For the deposition of multication materials, target selection can have significant impact on film growth properties, including particulate density, epitaxy, phase formation, and deposition rate. As a minimum requirement, ablation requires a target material possessing a high optical absorption coefficient at the selected laser wavelength. In general, the phase of the target does not need to be the same as that of the desired film. Only the cation stoichiometry need be identical to that of the films, assuming stoichiometric transfer and negligible evaporation from the film surface. For ceramic targets, one prefers target materials that are highly dense, as this will reduce particulate formation during the ablation process. As an alternative to polycrystalline ceramics, the use of single crystals as ablation targets has been investigated and shown to be effective in further reduction of droplet densities [Li et al., 1998]. The exception to this is wide bandgap insulators, such as Al2O3, where insufficient optical absorption makes single crystals unattractive as ablation targets. For soft materials, including biological materials, the target might be the material of interest or the material embedded in a matrix of an optically absorbing substance that does not deposit but yields an efficient ablation process.An alternative to ceramic or single-crystal targets is reactive PLD where the targets consist of the constituent cations, while the anion is supplied by the background gas. In general, the ablation process is less efficient for metal cations due to higher reflectivity and thermal conductivity. In addition, films deposited via ablation of metal targets can exhibit high particulate densities due to the ejection of molten droplets: for some systems, this problem can be addressed by using liquid metal targets. For some specific multication systems, metal targets have useful advantages. For the growth of multication films in which cation purity is an important issue, metals are often available with the6 PULSED LASER DEPOSITION OF COMPLEX MATERIALShighest purity. In addition, for insulators that possess particularly wide optical bandgaps, such as MgO, the ablation efficiency from ceramic or single-crystal targets is low for commercially available pulsed laser wavelengths.One also needs to consider the laser wavelength used for ablation. Efficient ablation of the target material requires the nonequilibrium excitation of the ablated volume to temperatures well above that required for evaporation. This generally requires the laser pulse to be short in duration, high in energy density, and highly absorbed by the target material. For ceramic targets, this is most easily achieved via the use of short wavelength lasers operating in the ultraviolet. High-energy ultraviolet laser pulses can be readily provided via excimer lasers or frequency-tripled or quadrupled Nd : YAG solid-state lasers. In some cases, a more efficient source is an infrared laser whose energy corresponds to a vibrational mode of the ablation target material [Bubb et al., 2002].In laser ablation, each ablation pulse will typically provide material sufficient for the deposition of only a submonolayer of the desired phase. The amount of film growth per laser pulse will depend on multiple factors, including target –substrate separation, background gas pressure and laser spot size, and laser energy density. Under typical conditions, the deposition rate per laser pulse can range from 0.001 to 1 A˚ per pulse. As such, PLD enables laser shot-to-shot control of the deposition process that is ideal for multilayer and interface formation where submonolayer control is needed. This degree of control can be seen from the in situ surface studies using reflection high-energy electron diffraction (RHEED), as discussed in detail in Chapter 8[Bozovic and Eckstein, 1995; Foxon, 1991]. RHEED provides a means of determining the crystallinity and smoothness of a surface, and oscillations in the intensity of diffraction spots during film growth correlate to the atomic layer-by-layer growth of the material. Figure1.2 shows the specular intensity of RHEED data for an epitaxial oxide film being deposited by PLD [Rijnders et al., 2000]. Two types of time-0 350 Time (s)0 40 Time (s) Figure 1.2 The specular RHEED intensity during PLD at 1 Hz (T ¼ 750o C, p O 2 ¼ 3 Pa). The insets give enlarged intensity after one laser pulse at 0.9 and 0.95 unit-cell layer coverage y. Also shown is (a) intensity variations of the specular reflection during PLD at 1 Hz and (b) interval deposition using the (b) t=0.45 s t=0.25 s(a)I n t e n s i t y (a r b i t r a r y u n i t s )WHAT IS PLD? 75 laser repetition rateof 10 Hz (T ¼ 800o C, p O2 ¼ 10 Pa) [Rijnders et al., 2000].8 PULSED LASER DEPOSITION OF COMPLEX MATERIALSdependent structure can be observed in the RHEED intensity plot. First, the oscillations observed in the intensity in Figure 1.2a represent the deposition of single unit cells of the oxide film. Specular RHEED intensity is dependent on the spatial coherence of the surface atoms. As layer-by-layer growth cycles through submonolayer coverage of the surface, RHEED intensity decreases, while for completed layers, the intensity is high. The oscillations seen in Figure 1.2a indicate that unit cell by unit cell growth on an atomically flat surface is occurring. The superimposed time-dependent substructure in the RHEED intensity seen in Figure 1.2b corresponds to surface redistribution of ablation plume species that have condensed on the surface from an individual ablation pulse. The time dependence of this structure yields insight into the nucleation and growth of the film at the submonolayer level for the arrival of each ablation plume.For multicomponent film growth, most of the limitations identified early in the development of PLD have been allieviated. A key development for the utilization of pulsed laser deposition for applications in industry has been the realization of schemes by which large area substrates can be effectively coated. The dynamics of the laser ablation process result in a highly focused plume of material ejected from the target. While this leads to a deposition efficiency on the order of 70%, it also results in a significant variation in deposition rate over distances on the order of a few centimeters. For uniform film thickness over large areas, manipulation of the plume –substrate positioning is required. Several approaches have been implemented to overcome this limitation, the most straightforward being to combine substrate rotation with rastering of the ablation beam over a large ablation target. This will, to first order, provide a means for covering large area substrates. However, one must take into account the decrease in plume energies and change in plume stoichiometry as one moves to the edge of the plume region.In pulsed laser deposition, the kinetic energies of ions and neutral species in the ablation plume can range from a few tenths to as high as several hundred electron volts. These energies are sufficient to modify the stress state of films through defect formation as has been documented for ion-beam- assisted approaches. The most common consequence of allowing deposition from an unabated energetic plume is the introduction of compressive stress. The origin of compressive stress due to energetic bombardment is associated with subsurface damage from the impinging energetic species, as schematically illustrated in Figure 1.3, leading to interstitial defects [Norton et al., 1999]. In this500Energetic atoms from ablation plume impinge on surface400300200100660 nm CeO 2 on 110 µm Si Curvature due to plume- induced compressive stress Collisions induce subsurface implantation to interstitials 00 200 400 600 800 1000 Scan distance (µm)Interstitials induce compressivestress in growing filmFigure 1.3 Schematic of plume-induced stress in PLD-deposited films.Z (n m )WHAT IS PLD? 95 case, the energetic depositing atoms displace underlying atoms in the film, resulting in atoms displaced to interstitial sites. Stress on the order of gigapascals has been observed. For thin substrates, this compressive stress can be sufficient to induce bowing of the structure as indicated in Figure 1.3 for CeO2 on a thin Si wafer. The kinetic energy necessary for the onset of recoil implantation of surface atoms into the film interior through bombardment is material dependent but is often observed for ion bombarding energies of a few electron volts or greater [Muller, 1989]. For oxides, the energetic bombarding cations can also preferentially sputter oxygen atoms from the surface, resulting in films that are oxygen deficient. The kinetic energy of ablated species is largely dependent on laser energy and gas-phase collisions. Fortunately, the use of a background gas to thermalize the plume is usually effective in eliminating this problem.Another potential issue with PLD is the ejection of micron-size particles in the ablation process. This is often observed when the penetration depth of the laser pulse into the target material is large.If these particles are deposited onto the substrate, they present obvious problems in the formation of multilayer device structures. The use of highly dense ablation targets and ablation wavelengths that are strongly absorbed by the target tends to reduce or eliminate particle formation. Mechanical techniques have been developed to reduce particle density in the event that target density and/or laser wavelength optimization fails to eliminate particulates. These include velocity filters [Pechen et al.,1995], off-axis laser deposition [Holzapfel et al., 1992], and line-of-sight shadow masks [Trajanovic et al., 1997]. Cross-beam techniques have also been considered as described in Chapter 6.In addition to particles ejected from the ablation targets, one can also observe nanoparticles that form in the gas phase when the background pressure is sufficiently high for heterogeneous particle nucleation. These particles can become embedded in a depositing film. Figure 1.4 shows a cross- section transmission electron microscopy image of a CeO2 film with the initial growth occurring at high pressure while the remaining film was grown at low pressure. CeO2 nanoparticles are clearly evident in the epitaxial CeO2 thin-film matrix [Norton et al., 1998]. In particular, the nanoparticles,with diameters ranging from 10 to 40 nm, are seen in the layer formed by ablating a CeO2 target in a hydrogen–argon background gas at a pressure of 200 mTorr. In contrast, the upper part of the CeO2film is devoid of nanoparticles and was deposited in a background pressure of 10-5 Torr where gas-phase collisions are minimal. While the formation of nanoparticles is generally undesirable for thin-film growth, this heterogeneous gas-phase nucleation has been intentionally exploited in the synthesis and study of nanomaterials, including oxides and semiconductors.While stoichiometric transfer of target composition is readily achieved for nearly every material,this does not ensure stoichiometric film growth at elevated temperature if any of the cation species possess high vapor pressures. Specific cations for which the sticking coefficient is an issue include K, Li, Na, Tl, Mg, Pb, Cd, and Zn. We consider as an example ZnGa2O4 [Lee et al., 1999], which is a wide bandgap semiconducting spinel that is of interest as a phosphor material. Films deposited using a stoichiometric single ZnGa2O4 target will show significant Zn deficiency for elevated deposition temperatures due to the higher vapor pressure of Zn relative to that of Ga. Figure 1.5 shows theFigure 1.4 Cross-section TEM image of a CeO2 film grown at high and low pressure, with CeO2 nanoparticles forming at the high background pressure.10 PULSED LASER DEPOSITION OF COMPLEX MATERIALSTarget PO2 P tot(mTorr)Single Mosaic Target400 450 500 550 600 650 700 750Substrate temperature (C)Figure 1.5 Plot of the Zn/Ga ratio for ZnGa2O4 films deposited using mosaic ablation targets.Zn/Ga ratio for films deposited at several different background pressures. Note that stoichiometric ZnGa2O4 corresponds to a Zn/Ga ratio of 0.5. Using a stoichiometric target, a Zn/Ga ratio as low as0.12 is observed for films deposited at 500o C. The optimal growth temperature, on the order of 600 –700o C, yields even greater deficiency in Zn content. One means of compensating for the loss of the volatile species in the films is to use a mosaic target consisting of the desired material and an additional region rich in the volatile component. Figure 1.5 compares the results for ZnGa2O4 films deposited with a single ZnGa2O4 target versus mosaic ZnGa2O4/ZnO targets. Stoichiometric ZnGa2O4 films required a target mosaic of 50% ZnGa2O4, 50% ZnO. The additional ZnO flux provides a means of overcoming the disparity in cation vapor pressure. Note that the stoichiometry also depended on oxygen partial pressure, which likely reflects the difference in vapor pressure for Zn as compared to ZnO.1.3 WHERE IS PULSED LASER DEPOSITION BEING APPLIED?Given the attractive characteristics of pulsed laser deposition in the synthesis of multicomponent thin-film materials, a number of applications are being actively pursued using this technique. In some cases, the application focuses on the synthesis of a thin-film material or structure. In other cases, the research has targeted the development of specific devices. It is interesting to consider specific structures, devices, and applications for which PLD has been successfully applied. Many of these topics are discussed in more detail in later chapters in Parts 3 and 4.1.3.1 Complex Oxide Film GrowthIn the growth of crystalline oxides, PLD has proven to be most effective. The growth of complex oxides requires the delivery of a growth flux with the correct stoichiometry in an oxidizing ambient that is favorable for the desired phase formation. The utility of pulsed laser deposition in reproducing target stoichiometry has been demonstrated for a number of multication oxides. Early success in realizing stoichiometric YBa2Cu3O7 clearly delineated this advantage for pulsed laser deposition. In recent years, even more complex crystal structures have been successfully grown using this approach. Consider, for example, the growth of the Y-type magnetoplumbite Ba2Co2Fe12O22compound. This material is a ferromagnetic oxide of potential interest in thin-film magnetic device applications [Sudakar et al., 2003]. The epitaxial growth of Ba2Co2Fe12O22 is challenging as it possesses a remarkably complex crystal structure as illustrated in Figure 1.6a. The unit cell possesses a huge lattice parameter of 43.5 A˚. Despite the complexity of the crystal structure, the epitaxial growth of Ba2Co2Fe12O22 has been realized via pulsed laser deposition [Ohkubo et al., 2003]. Figure 1.6b and 1.6c show the X-ray diffraction data for an epitaxial film, along with theI IIZnGa2O4ZnOZn/GaSingle 24 60Mosaic I 24 60Mosaic II 24 60Mosaic II 45 60Mosaic II 60 60Mosaic II 75 100WHERE IS PULSED LASER DEPOSITION BEING APPLIED?9Figure 1.6 Crystal structure, X-ray diffraction data, and TEM image of a Ba2Co2Fe12O22 film grown by PLD [Ohkubo et al., 2003].cross-sectional transmission electron microscopy (TEM) image. It should be noted that this result was obtained using a combinatorial synthesis approach in which multiple process parameters(composition, temperature) are explored in parallel through the use of combinatorial arrays or compositional spread techniques. Pulsed laser deposition has proven to be easily adaptable to combinatorial techniques for thin-film research. The ability to grow epitaxial, multication complex inorganic thin films has been, and continues to be, one of the enabling strengths of PLD.1.3.2 Epitaxial Interface and Superlattice FormationDevelopments in oxide PLD film growth have provided remarkable opportunities in the synthesis of epitaxial heterostructures and superlattices [Yilmaz et al., 1991; Fernandez et al., 1998; Chang et al., 1999; Smolenskii et al., 1984; Xu et al., 2000]. Superlattices of oxides, such as the (001) KNbO3/KTaO3 perovskite structure shown in Figure 1.7, have been realized for a number of material systems, with individual layers as thin as a single unit cell [Christen et al., 1996, 1998; Specht et al.,1998]. Excellent film flatness and crystallinity are evidenced in these films, and the interfaces can beFigure 1.7 Cross-section Z-contrast STEM image of a KTaO3/KNbO3 superlattice structure grown by pulsed laser deposition.10 PULSED LASER DEPOSITION OF COMPLEX MATERIALSFigure 1.8 Cross-section Z-contract STEM image of a CeO2/Ge epitaxial interface fabricated using PLD. compositionally sharp on an atomic scale as seen in Figure 1.7. In the formation of atomically abrupt interfaces and superlattice structures, PLD is competitive with other film growth techniques,including molecular orbital chemical vapor deposition (MOCVD) and molecular beam epitaxy(MBE). The formation of epitaxial oxide superlattices has been used to investigate the effects of reduced dimensionality on a number of phenomena. Superconductivity in single unit cell YBa2Cu3O7 layers was first demonstrated using PLD. Low dimensionality behavior has been investigated for ferroelectric and magnetic oxides.Pulsed laser deposition also yields the opportunity to create atomically abrupt interfaces between materials that are chemically dissimilar, including epitaxial metal–oxide and semiconductor–oxide structures. A key factor in the formation of atomically abrupt interfaces between oxide films and nonoxide surfaces is to identify conditions where undesirable interfacial reactions are minimal yet compatible with oxide epitaxy. Laser ablation film growth is particularly well suited for nucleation in a reactive environment since it is compatible with a large range of background pressures. For example, a hydrogen-assisted nucleation approach has been used to grow atomically abrupt,epitaxial CeO2/Ge interfaces [Norton et al., 2000]. During the initial nucleation of CeO2 on Ge, the hydrogen partial pressure and substrate temperature are chosen such that the native oxide, GeO2, is thermodynamically unstable, and under these conditions, the epitaxial growth of an oxide can be achieved on the Ge surface without interference from native oxides. The deposited oxide material must itself be thermodynamically stable under the conditions used during nucleation, and epitaxy will be determined by the chemistry and structure of the two materials. For CeO2 nucleated on (001) Ge in hydrogen gas using pulsed laser deposition, the film can be epitaxial with an interface that is atomically abrupt as is evident in the cross-section Z-contrast scanning transmission electron microscopy image shown in Figure 1.8. A similar approach has been used to form epitaxial interfaces between various oxides and metals (e.g., Ni and Ni alloys), as well as compound semiconductors, including InP. The latter is of interest as it provides a means of integrating electronic oxides with photonic and microwave electronics.。