Fabrication, Characterization, and Modulation of Functional Nanolayers

Abstract

Regions of a few nanometers at the surface or interface of a material exhibit various functional properties, which differ from those of the bulk because the electrons and/or ions receive different potentials due to the incoherent atomic arrangement. High-quality epitaxial films of functional materials called “nanolayers” are important to utilize such functional properties. However, fabrication of high-quality nanolayers of complex materials with complicated crystal structures is usually challenging due to the difference in the thermochemical properties of the constituents. In this chapter, epitaxial growth techniques, especially “reactive solid-phase epitaxy” of functional oxides and chalcogenides, are reviewed based on the authors’ efforts. Additionally, this chapter reviews several modulation methods of optical, electrical, and magnetic properties of functional oxide nanolayers.

1 Epitaxial Growth and Characterization of Functional Nanolayers

Regions of a few nanometers at the surface or interface of a material often exhibit various functional properties, which differ from those of the bulk due to the fact that the electrons and/or ions receive different potentials due to the incoherent atomic arrangement. High-quality epitaxial films of functional materials called “nanolayers” are important to utilize such functional properties.

In this chapter, epitaxial growth techniques, especially “reactive solid-phase epitaxy” of functional oxides and chalcogenides, are reviewed based on the authors’ efforts. Additionally, this chapter reviews several modulation methods of optical, electrical, and magnetic properties of functional oxide nanolayers.

2 Pulsed Laser Deposition

Pulsed laser deposition (PLD) is a physical vapor deposition technique [1]. By irradiating focused laser pulses of an excimer laser or a higher (3rd or 4th) harmonic of a Nd:YAG laser onto the target material (single crystals or ceramics or powder), which is located in an ultrahigh vacuum chamber, films can be deposited on the substrate as a result of vaporization of the target materials occur during laser irradiation (Fig. 10.1). PLD is one of the most powerful techniques for epitaxial film growth of inorganic solids, especially oxides. It has several advantages compared to the other deposition techniques, such as sputtering. In the case of PLD, the chemical composition of the resultant film is almost same as that of the target material, although generally it differs from the target because the chemical species show different sputtering yields in the case of sputtering. Moreover, the atmosphere in the PLD chamber can be widely controlled from an ultrahigh vacuum to ~10² Pa, allowing a thermodynamically nonequilibrium crystalline phase of a material to be fabricated.

Fig. 10.1

Schematic illustration of a PLD system

As an example, PLD growth and characterization of the SrTiO₃-SrNbO₃ solid solution system are explained. SrTiO₃ has attracted increasing attention as the next generation of oxide electronics [2]. Doping with the appropriate substituent, such as Nb⁵⁺ (Ti⁴⁺ site) or La³⁺ (Sr²⁺ site), easily varies the charge carrier concentration of SrTiO₃ from insulating to metallic (n _3D ~ 10²¹ cm⁻³). Electron-doped SrTiO₃ is one of the most extensively studied materials for thermoelectric applications [3, 4]. In 2001, Okuda et al. [5] synthesized Sr_1−x La _x TiO₃ (0 ≤ x ≤ 0.1) single crystals by the floating-zone method. They reported that the crystals exhibit a large power factor (S ² · σ) of 2.8–3.6 mW m⁻¹ K⁻² at room temperature. Later, Ohta et al. reported the carrier transport properties of Nb- and La-doped SrTiO₃ single crystals (carrier concentration, n ~ 10²⁰ cm⁻³) at high temperatures (~1000 K) to clarify the intrinsic thermoelectric properties of these materials [6].

The experimental discovery of unusually large thermopower outputs from superlattices and two-dimensional electron gases in SrTiO₃ [7, 8] spurred substantial research efforts into SrTiO₃ superlattices [9, 10] and heterostructures [11,12,13] for thermoelectric applications. For example, a superlattice composed of one unit cell (uc) of SrTi_0.8Nb_0.2O₃ and 10 uc of SrTiO₃ exhibits a giant thermopower, most likely due to an electron confinement effect. Although electron confinement is strongly correlated with the electronic structure [14, 15], a full understanding of the fundamental electronic phase behavior of the SrTi_1−x Nb _x O₃ solid solution system has yet to be developed.

Although high-quality single crystals of SrTi_1–x Nb _x O₃ species with x > 0.1 are not available due to the low solubility limit of Nb in the lattice [16], epitaxial films with these material compositions can be fabricated by PLD [17]. As summarized in Fig. 10.2, pure SrTiO₃ (space group $Pm\bar{3}m$, cubic perovskite structure, a = 3.905 Å) is an insulator with a bandgap of 3.2 eV. The bottom of the conduction band is composed of triply degenerate, empty Ti 3d−t _2g orbitals, while the top of the valence band is composed of fully occupied O 2p orbitals [18]. The valence state of Ti ions in crystalline SrTiO₃ is 4 + (Ti 3d⁰). On the other hand, pure SrNbO₃ (space group $Pm\bar{3}m$, cubic perovskite structure, a = 4.023 Å) is a metallic conductor [19,20,21]. The valence state of the Nb ion is 4 + (Nb 4d¹). In between SrTiO₃ and SrNbO₃ in the SrTi_1−x Nb _x O₃ ss, there are two possible types of valence state changes in the Ti and Nb ions, as shown in Fig. 10.2b and c. In the case of isovalent substitution (Fig. 10.2b), the mole fraction of Ti⁴⁺ proportionally decreases with increasing Nb⁴⁺ (x). On the other hand, heterovalent substitution, in which two Ti⁴⁺ or Nb⁴⁺ ions are substituted by adjacent (Ti³⁺/Nb⁵⁺) ions, can occur, as shown in Fig. 10.2c. Based on these considerations, we focused on the valence state changes of Ti and Nb ions in the SrTi_1−x Nb _x O₃ ss.

Fig. 10.2

Reprinted with permission from [22]. © 2017 AIP

Schematic of the crystal structure and possible valence state changes in the SrTiO₃-SrNbO₃ solid solution system. a Schematic of the crystal structure. Pure SrTiO₃ is an insulator with a bandgap of 3.2 eV, in which the valence state of the Ti ions (blue, TiO₆) is 4 + (Ti 3d⁰). In contrast, pure SrNbO₃ is a metal, in which the valence state of the Nb ions (Red, NbO₆) is 4 + (Nb 4d¹). b, c Possible valence state changes of the Ti and Nb ions in the SrTiO₃-SrNbO₃ solid solution system: b isovalent substitution, where Ti⁴⁺ is substituted by Nb⁴⁺ and c heterovalent substitution, where two Ti⁴⁺/Nb⁴⁺ ions are substituted by adjacent Ti³⁺/Nb⁵⁺ ions.

Zhang et al. [22] fabricated approximately 100 nm-thick SrTi_1−x Nb _x O₃ (x = 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.55, 0.6, 0.7, 0.8, 0.9, and 1.0) epitaxial films by PLD using dense ceramic disks of a SrTiO₃-SrNbO₃ mixture. Insulating (001) LaAlO₃ (pseudo-cubic perovskite, a = 3.79 Å) was used as the substrate. The growth conditions were precisely controlled with a substrate temperature of 850 ℃, an oxygen pressure of ~10⁻⁴ Pa, and a laser fluence of 0.5–1 J cm⁻² pulse⁻¹, yielding a growth rate of 0.3 pm pulse⁻¹.

Figure 10.3a summarizes the Xray reciprocal space mappings (RSMs) around the $(\overline{ 1} 0 3 )$ diffraction spot of LaAlO₃ (overlaid). Intense diffraction spots from $(\overline{ 1} 0 3 )$ SrTi_1−x Nb _x O₃ are seen together with those from the LaAlO₃ substrate, indicating that incoherent heteroepitaxial growth of the target materials occurs for all x compositions. The peak positions of the diffraction spots from each composition correspond well with the cubic line (q _z /q _x  = −3), suggesting that no epitaxial strain is induced in the films. It should be noted that a slight tetragonal distortion is observed in the x = 0.4 (c/a = 1.0057) and 0.5 (c/a = 1.0050) samples.

Fig. 10.3

Reprinted with permission from [22]. © 2017 AIP

Crystallographic characterization of the SrTi_1−x Nb _x O₃ epitaxial films on a (001) LaAlO₃ single-crystal substrate. a Xray reciprocal space mappings around the $(\bar{1}03)$ diffraction spot of the SrTi_1−x Nb _x O₃ epitaxial films. The location of the LaAlO₋₃ diffraction spot, (q _x /2π, q _z /2π) = (−2.64, 7.92), corresponds to the pseudo-cubic lattice parameter of LaAlO₃ (a = 0.379 nm). Red symbols (+) indicate the peak positions of the SrTi_1−x Nb _x O₃ epitaxial films. b Changes in the lattice parameters of the SrTi_1−x Nb _x O₃ films (circles, left axis) superimposed with isovalent/heterovalent substitution lines (black line: isovalent substitution, gray line: heterovalent substitution, right axis), calculated using Shannon’s ionic radii [23]. c Changes in the B-site occupation by [Ti⁴⁺/Nb⁴⁺] derived from the data in (b).

From the RSMs of the SrTi_1−x Nb _x O₃ films, we extracted the lattice parameters using the formula a = (2π/q _x  · 2π/q _x  · 6π/q _z )^1/3. Figure 10.3b plots the lattice parameters of the SrTi_1−x Nb _x O₃ film as a function of x. We observed an M-shaped trend along with a general increase in the lattice parameter with increasing x. In order to analyze the changes in the lattice parameter, we calculated the average ionic radii in the crystal structure and used Shannon’s ionic radii as a comparison [23]: Ti⁴⁺ (60.5 pm), Ti³⁺ (67.0 pm), Nb⁴⁺ (68.0 pm), and Nb⁵⁺ (64.0 pm). In the ranges of 0.05 ≤ x ≤ 0.3 and x ≥ 0.6, the observed lattice parameters closely follow the heterovalent substitution line, suggesting that two Ti⁴⁺ or Nb⁴⁺ ions are substituted by adjacent (Ti³⁺/Nb⁵⁺) ions [24]. On the other hand, at x = 0.4 and 0.5, the observed lattice parameter correspond well with the isovalent substitution line. Moreover, at x = 0.5, the B-site occupation of [Ti⁴⁺/Nb⁴⁺] is almost 100%, as shown in Fig. 10.3c.

Figure 10.4a shows a cross-sectional HAADF-STEM image of the SrTi_0.5Nb_0.5O₃ film. Periodical mismatch dislocations with intervals of ~8.5 nm are seen at the heterointerfaces. If the strain in the thin film is fully relaxed by such misfit dislocations, it is possible to calculate the spacing between dislocations (d) from $d = {\mathbf{b}}/\delta$, where b is the Burgers vector and δ is the lattice mismatch between thin film and substrate [25]. Using the lattice parameters obtained from XRD [δ = $(q_{{x\,{\text{sub}}}} - q_{{x\,{\text{film}}}} )/q_{{x\,{\text{film}}}} = + 0.0435$], the estimated dislocation spacing is 8.7 nm, suggesting that the dislocations fully relax the strain in the film. Although superspots originating from the (111) diffraction are often observed in AB_0.5B’_0.5O₃ compositions that crystallize in B-site-ordered double perovskite structures [26], they are not observed in the SrTi_0.5Nb_0.5O₃ film (Fig. 10.4b). This is most likely due to the slight tetragonal distortion of the crystal structure. Figure 10.4 shows the EELS spectra acquired around the Ti L (c) and O K edges (d). The reported EELS spectra of Ti³⁺/Ti⁴⁺ [27] and Nb⁴⁺/Nb⁵⁺ [28] are plotted for comparison. In the Ti L edge spectrum (c), t _2g and e. _g. peak splitting is clearly observed for Ti L ₃, indicating that the dominant valence state of Ti is 4+. In the O K edge spectra (d), two intense peaks (assigned as A and B) are clearly observed where peak B has a higher intensity than peak A. A previous study noted that this is a characteristic feature of Nb⁴⁺ [28]. The peak intensity ratio A/B is calculated to be 0.66, which roughly corresponds with the Nb⁴⁺ spectrum (0.66).

Fig. 10.4

Reprinted with permission from [22]. © 2017 AIP

Electron microscopy analyses of a SrTi_1−x Nb _x O₃ film with a composition of x = 0.5. a HAADF-STEM image acquired with the electron beam incident along the <100> direction. Periodic misfit dislocations (~8.5 nm interval) at the heterointerface are indicated by red lines. b Selected-area electron diffraction pattern acquired with the electron beam incident along the <110> direction. c, d EELS spectra acquired around the c Ti L edge and d O K edge. EELS spectra for Ti³⁺/Ti⁴⁺ [27] and Nb⁴⁺/Nb⁵⁺ [28] from previous studies are also plotted for comparison.

By using abovementioned films, Zhang and Ohta et al. clarified the thermoelectric phase diagram for the SrTi_1−x Nb _x O₃ (0.05 ≤ x ≤ 1) solid solution system (Fig. 10.5). They observed two thermoelectric phase boundaries in the system, which originate from the step-like decrease in the carrier effective mass at x ~ 0.3 and from the local minimum in the carrier relaxation time at x ~ 0.5. The origins of these phase boundaries are related to the isovalent/heterovalent B-site substitution. The parabolic Ti 3d orbitals dominate the electron conduction for compositions with x < 0.3, whereas the Nb 4d orbital dominates when x > 0.3. At x ~ 0.5, a tetragonal distortion of the lattice, in which the B-site is composed of Ti⁴⁺ and Nb⁴⁺ ions, leads to the formation of tail-like impurity bands, which maximize electron scattering. These results provide a foundation for further research to improve the thermoelectric performance of SrTi_1−x Nb _x O₃.

Fig. 10.5

Reprinted with permission from [22]. © 2017 AIP

Thermoelectric phase diagram for the SrTiO₃-SrNbO₃ solid solution system. The thermoelectric power factor (S ² · σ) of the SrTiO₃-SrNbO₃ solid solution system is plotted along with the previously reported values [7]. The x dependence of S ² · σ is shown in the inset. The system’s thermoelectric phase boundaries are clearly seen at x ~0.3 and ~0.5.

3 Reactive Solid-Phase Epitaxy

As explained above, PLD is a powerful technique for epitaxial film growth of metal oxides. Although there are many reports on epitaxial oxide film growth by the PLD method, it is still difficult to fabricate epitaxial films of complex oxides composed of different vapor pressure elements, especially alkali metals. Since complex oxides have high melting points (>1500 ℃), substrates must be heated at high temperature (>800 ℃ : ~60% of the melting point) in a vacuum during PLD. If the multiplex elements have different vapor pressures at the substrate temperature, the chemical composition of the resultant film completely differs from that of the target because re-vaporization of the high vapor pressure element occurs during deposition.

To overcome this issue, Ohta et al. developed the “Reactive Solid-Phase Epitaxy (R-SPE)” method in 2003 (Fig. 10.6) [29]. In this method, an epitaxial film of a monoxide, which is a component of a complex oxide, is fabricated by PLD. Then the film is heated at high temperatures with another member of the target oxide (thin film or powder). During the heat treatment, a solid-solid reaction between the monoxide film and the other member elements occurs while maintaining the crystallographic orientation. Using the R-SPE method, epitaxial films of InMO₃(ZnO) _m (M = Ga and In, m = integer) [29], ZnRh₂O₄ [30], LaCuOCh (Ch = S or Se) [31], and Na _x CoO₂ (x ~ 0.8) [32] have been fabricated. Thus, the R-SPE method effectively fabricates epitaxial films of complex oxides composed of high vapor pressure elements. In this section, recent progress of “reactive solid-phase epitaxy” of functional oxides and chalcogenides is reviewed.

Fig. 10.6

Schematic diagram of the “Reactive Solid-Phase Epitaxy” method. A bilayer laminate composed of a thin epitaxial layer of simple oxide (A–O or B–O or C–O) or metal grown on a substrate and a polycrystalline layer or powder source of target A _k B _m C _n O _x is thermally annealed at high temperatures (~1000 ℃). The solid-state reaction at high temperatures leads to the formation of a thin single-crystalline layer on the substrate, which may act as “an epitaxial template” for successive homoepitaxial SPE growth of the film

3.1 Na_≈2/3MnO₂ Epitaxial Film

Layered alkali ion-containing metal oxides (LAMO), A _x MO₂ (A: alkali metal and M: transition metal) have received considerable interest as candidate materials for energy storage and conversion applications. This is because their chemical potential can be readily controlled. Changing the concentration of A ⁺ in the interspace between adjacent MO₂ layers tunes the valence state of the M ion. In particular, there have been many studies on cobalt-based A _x CoO₂ because the A ⁺ concentration is easily controlled and these oxides possess a two-dimensional electronic structure. Li _x CoO₂ (0 ≤ x ≤ 1) is one of the best cathode active materials in commercial Li-ion batteries because the Li⁺ concentration can be controlled by an electrochemical process [33, 34]. Meanwhile, Na _x CoO₂ (x ~ 0.8) is a promising thermoelectric material, which can directly convert a temperature difference into electricity. Additionally, it exhibits a rather large thermopower even though it displays metallic conductivity due to the two-dimensional nature of the electronic structure [35, 36]. Furthermore, the two-dimensional electronic structure of the bilayer hydrated crystal Na_≈0.3CoO₂ · 1.3H₂O allows it to exhibit superconductivity at a critical temperature T _c of ~4 K [37, 38].

Unlike A _x CoO₂ systems, the physical properties of A _x MnO₂ have yet to be clarified, although Na _x MnO₂ has recently been proposed as a new candidate for the cathode active material in Na-ion batteries [39,40,41] because Mn (Clarke number: 0.09) is more abundant than Co (Clarke number: 0.004). At least two crystallographic phases of NaMnO₂ are known; low-temperature α-NaMnO₂ [39, 42] has an O3 layered structure with monoclinic symmetry and high-temperature β-NaMnO₂ [40] has a Pmnm structure with orthorhombic symmetry. It should be noted that the crystal symmetry of Na _x MnO₂ strongly depends on x. The low-temperature phase Na_0.67MnO₂ [41, 43] has a P2 layered structure with hexagonal symmetry. Recently, Billaud et al. [41] reported that Na_0.67MnO₂ with a P2 structure exhibits a high capacity of 175 mA h g⁻¹ with a good capacity retention.

However, there are a few studies on the electrical conductivity of Na _x MnO₂ [44]. The most rational reason is the lack of large single crystals. It is difficult to measure the intrinsic electrical property using powder compacts due to severe electron scattering. High-quality epitaxial films may be a solution to clarify the electrical conductivity of Na _x MnO₂.

In 2017, Katayama et al. fabricated Na _x MnO₂ epitaxial films by the R-SPE method using the following procedure (Fig. 10.7) [45]. First, a 70 nm-thick MnO _y thin film was heteroepitaxially grown on a (0001) α-Al₂O₃ substrate (10 × 10 × 0.5 mm) by PLD using a KrF excimer laser (λ = 248 nm, 20 ns, 10 Hz, ~1.5 J cm⁻² pulse⁻¹) to ablate the Mn₂O₃ ceramic disk. During the deposition, the substrate temperature and oxygen pressure were kept at 700 ℃ and ~10⁻² Pa, respectively. After deposition, the film was cooled to RT in the PLD chamber. The film’s top surface was completely covered with another sapphire plate, and the sandwiched specimen was subsequently preserved in Na₂CO₃ powder. Then the film was heated at 700 ℃ for 30 min in air to supply Na⁺ and O²⁻ into the MnO _y film. During heat treatment, the film color changed from light brown to dark brown.

Fig. 10.7

Reprinted with permission from [45]. © 2017 ACS

Schematic of the crystal structure change from MnO _y to Na _x MnO₂ during R-SPE [Gray: Na, blue: Mn, red: O]. First, a spinel-type MnO _y film is heteroepitaxially grown on a sapphire substrate. The MnO _y epitaxial film, covered with Na₂CO₃ powder, is heated at 700 ℃ in air. As a result, Na⁺ ions are supplied into the MnO _y film together with O²⁻ ions during the heating, forming the Na _x MnO₂ epitaxial film with a layered crystal structure.

Figure 10.8 shows a cross-sectional HAADF-STEM image of the R-SPE grown Na_≈2/3MnO₂ film around the interface observed from the direction of Na_≈2/3MnO₂||α-Al₂O₃. The stripe patterns correspond to the layered structure of Na_≈2/3MnO₂. It should be noted that an interfacial layer is not observed, confirming that the present sample has an atomically sharp interface between the film and substrate, contrary to the previously reported observation for a Na_≈0.8CoO₂ film.

Fig. 10.8

Reprinted with permission from [45]. © 2017 ACS

HAADF-STEM image of the Na_≈2/3MnO₂/α-Al₂O₃ substrate interface. Incident direction of the electron beam is $[ 1\overline{ 1} 0 0 ]$ Na_≈2/3MnO₂. The atomically sharp heterointerface is clearly seen.

Figure 10.9 summarizes the temperature (T) dependence of the electrical conductivity (σ) for the Na_≈2/3MnO₂ and hydrated Na_≈0.61MnO₂ ≈ 0.42H₂O epitaxial films. It should be noted that the σ – T curves for both films do not show a remarkable hysteresis in the heating–cooling cycles ranging from RT to 400 K, suggesting that the absorbed water does not significantly contribute to σ because the surface-adsorbed water should be released at 100–150 ℃, although a slight deviation from a straight line at ~100 ℃ is observed in hydrated Na_≈0.61MnO₂ ≈ 0.42H₂O film. At RT, σ of the Na_≈2/3MnO₂ epitaxial film is ~1 mS cm⁻¹, which is two orders of magnitude larger than that of an α-Na_0.70MnO_2.25 single crystal (~0.5 μS cm⁻¹). In contrast, σ of the Na_≈0.61MnO₂ ≈ 0.42H₂O film is ~0.1 mS cm⁻¹, which is comparable to that of the Na _x MnO₂ · nH₂O ceramic (~0.05 mS cm⁻¹). In both cases, σ increases exponentially with temperature because electron hopping becomes faster at higher temperatures. The activation energy for electron hopping (E _a) observed for both films in the 300–400 K range is 0.47 eV, which is comparable to those of other Mn³⁺/Mn⁴⁺-containing oxides, such as α-MnO_2−δ , Li _x MnO₂, and LiMn₂O₄. In contrast to Na_≈0.35CoO₂ · 1.3H₂O, the electron hopping conductivity of Na_≈2/3MnO₂ decreases by the hydration treatment. This decrease is most likely because the intercalated water molecules affect the ratio of Mn³⁺/Mn⁴⁺.

Fig. 10.9

Reprinted with permission from [45]. © 2017 ACS

Electrical conductivity. Temperature dependence of the electrical conductivity for the Na_≈2/3MnO₂ and the Na_≈0.61MnO₂ ≈ 0.42H₂O films measured by the AC impedance method in air. The resultant films show electrical conductivities of ~1 mS cm⁻¹ for Na_≈2/3MnO₂ film and ~0.1 mS cm⁻¹ for Na_≈0.61MnO₂ ≈ 0.42H₂O film at RT. Inset shows the σT–1/T plots for both films; the activation energy (E _a) for electron hopping conduction is 0.47 eV.

3.2 Li₄Ti₅O₁₂ Epitaxial Film

Li₄Ti₅O₁₂ (S.G. $Fd\overline{3} m$) is one of the most promising anode active materials of solid Li-batteries, [46, 47] due to its structural stability during charge/discharge reactions [48] with excellent reversibility [49, 50] and a long cycle life [51]. Although epitaxial film growth of Li₄Ti₅O₁₂ by PLD has been reported, [52, 53] the target ceramic containing excess Li species is required to fabricate stoichiometric Li₄Ti₅O₁₂ thin films. On the contrary, Li et al. fabricated an amorphous Li₄Ti₅O₁₂ film by PLD on a (001) SrTiO₃ single-crystal substrate at room temperature with a stoichiometric Li₄Ti₅O₁₂ target, and heated the amorphous film with molten LiNO₃ at 600 ℃ in air. As a result of this “solid–liquid phase epitaxy”, they successfully fabricated an epitaxial film of single-phase Li₄Ti₅O₁₂ [54].

The solid–liquid phase epitaxy procedure for the growth of Li₄Ti₅O₁₂ films is schematically illustrated in Fig. 10.10. Step 1(a): Amorphous Li-Ti-O films (100 nm thick) are deposited at RT on (001) SrTiO₃ single-crystal substrates (area: 10×10 mm², thickness: 0.5 mm) by PLD. The in-plane lattice mismatch is too large for Li₄Ti₅O₁₂ to coherently grow on (001) SrTiO₃ substrate, where the lattice mismatch between cubic Li₄Ti₅O₁₂ (the half of a-axis lattice parameter, a/2 = 0.4176 nm) and SrTiO₃ (a = 0.3905 nm) is estimated to be −6.9%. A KrF excimer laser with an energy fluence of ~2 J cm⁻² pulse⁻¹ and a repetition rate of 10 Hz is used to ablate a ceramic target of stoichiometric Li₄Ti₅O₁₂. The oxygen pressure during film deposition is kept at a low P _O2 of 1.0 × 10⁻³ Pa and the deposition rate is 3.3 nm min⁻¹. Step 2(b): The resultant film is covered with LiNO₃ powder. Then it is heated at 600 ℃ for 30 min in air at temperature increasing rate of 40 ℃/min in an Al₂O₃ crucible using an electric furnace. During the heating process, the LiNO₃ powder melts due to its low melting point of 261 ℃ and entirely covers the Li-Ti-O film at 600 ℃. The film is naturally cooled to RT in the furnace. Step 3(c): The resultant film is washed by distilled water since the film surface is covered with the remaining LiNO₃ film. The resultant film surface looks very clean, indicating that LiNO₃ is successfully removed.

Fig. 10.10

Reprinted from [54]. © 2016 The Japan Society of Applied Physics

Solid–liquid phase epitaxy of the Li₄Ti₅O₁₂ film. a Step 1: An amorphous Li₄Ti₅O₁₂ film is deposited at RT on a (001) SrTiO₃ single-crystal substrate by PLD using a dense Li₄Ti₅O₁₂ ceramic as the target. b Step 2: The resultant film is heated at 600 ℃ for 30 min in air with a LiNO₃ powder, which melts during the heating process, in an Al₂O₃ crucible using electric furnace. Then the film is naturally cooled down to RT in the furnace. c Step 3: The Li₄Ti₅O₁₂ epitaxial film is obtained after the resultant film is washed with distilled water since the film surface is covered with remaining LiNO₃ film.

Figure 10.11a–c shows the out-of-plane XRD patterns for Li₄Ti₅O₁₂ films [(a) as-deposited, (b) heated at 600 ℃ without LiNO₃, and (c) heated at 600 ℃ with LiNO₃]. Only the intense diffraction peaks of 00 l SrTiO₃ are observed in the as-deposited Li₄Ti₅O₁₂ film, indicating that the film is amorphous (Fig. 10.11a). After heating the amorphous Li₄Ti₅O₁₂ film without LiNO₃ powder at 600 ℃ in air, the 004 anatase-TiO₂ diffraction peak is observed, but the Li₄Ti₅O₁₂ diffraction peak is not seen in the out-of-plane XRD pattern (Fig. 10.11b). The c-axis lattice parameter (0.954 nm) for the TiO₂ phase is almost the same as 0.951 nm for the pure anatase-TiO₂ bulk [55].

Fig. 10.11

Reprinted from [54]. © 2016 The Japan Society of Applied Physics

Out-of-plane XRD patterns of the Li₄Ti₅O₁₂ films [a as-deposited, b heated at 600 ℃ without LiNO₃ (solid-phase epitaxy), c heated at 600 ℃ with LiNO₃ (solid–liquid phase epitaxy)]. Only intense diffraction peaks of 00 l SrTiO₃ are seen in the as-deposited Li₄Ti₅O₁₂ film (a), indicating that the as-deposited film is amorphous. The intense diffraction peak of 004 Li₄Ti₅O₁₂ is observed in (c), although 004 anatase TiO₂ is crystallized when the film is heated without LiNO₃ powder (b). FWHM of the Xray rocking curve (Δω) for the 004 Li₄Ti₅O₁₂ is ~0.8° (d). e In-plane XRD pattern of the Li₄Ti₅O₁₂ film grown by solid–liquid phase epitaxy. Only the intense diffraction peak of 400 Li₄Ti₅O₁₂ is seen together with h00 SrTiO₃. The ϕ scan of 400 Li₄Ti₅O₁₂ diffraction [(e) inset] shows a fourfold rotational symmetry with every 90° rotation originating from the cubic symmetry of Li₄Ti₅O₁₂ lattice.

In contrast, the intense diffraction peak of 004 Li₄Ti₅O₁₂ is observed after the film is heated with molten LiNO₃ at 600 ℃ (Fig. 10.11c). The full width at half maximum (FWHM) value of the out-of-plane rocking curve (Δω) for 004 Li₄Ti₅O₁₂ diffraction is ~0.8°, indicating that the Li₄Ti₅O₁₂ film is preferentially oriented perpendicular to the substrate surface (Fig. 10.11d). The chemical composition of the Li₄Ti₅O₁₂ film could not be accurately estimated when the Li/Ti ratio is an extremely large value of ~5.0, presumably due to the residual adhesive LiNO₃ and/or Li species incorporated into SrTiO₃ substrate. However, the film density characterized by the Xray reflectivity measurements for the Li₄Ti₅O₁₂ film is 3.5 g cm⁻³, which is consistent with the 3.48 g cm⁻³ of the Li₄Ti₅O₁₂ phase. In contrast, that of the Li₄Ti₅O₁₂ film heated without LiNO₃ is 4.1 g cm⁻³, approaching 3.90 g cm⁻³ of the anatase TiO₂ due to the decrease of in Li content in the film. At this stage, the actual vaporization temperature of Li in the Li₄Ti₅O₁₂ film has yet to be examined, but these results suggest that molten LiNO₃ plays an essential role in suppressing the vaporization of Li species and enables crystallization of the Li₄Ti₅O₁₂ phase at relatively low temperature (600 ℃) by solid–liquid phase epitaxy.

3.3 KFe₂As₂ Epitaxial Film

A novel attractive property has recently been theoretically predicted for KFe₂As₂. Pandey et al. [56] reported that KFe₂As₂, which is the end member of the 122-type iron-based superconductors (Ba_1–x K _x ) Fe₂As₂ (i.e., x = 1) with a critical temperature ≈ 3 K [57], may exhibit a large spin Hall conductivity (SHC), which is comparable to that of Pt [58]. That is, it exhibits 10⁴ times larger SHC (2 × 10⁴ Ω⁻¹ m⁻¹) than that of a semiconductor (0.5 Ω⁻¹ m⁻¹) [59]. Such a high SHC originates from the strong spin-orbit coupling of the Fe 3d states with Dirac cones below the Fermi level of heavily hole-doped KFe₂As₂. Indeed, high-resolution angle-resolved photoemission spectroscopy experiments show that the electron pocket at the M point of (Ba_1–x K _x )Fe₂As₂ completely disappear for KFe₂As₂ due to its heavily self-hole-doped nature (Ba²⁺ ↔ K⁺ + hole) [60].

Because KFe₂As₂ contains alkali metal K as its main constituent, it is very air sensitive. Therefore, the thin-film growth of KFe₂As₂ is difficult due to two intrinsic properties: its extremely hygroscopic nature and the high vapor pressure of potassium. Thin-film growth of KFe₂As₂ and electrical measurements with device patterning are challenging issues. These issues were solved by combining room-temperature pulsed laser deposition using K-rich KFe₂As₂ bulk targets with thermal crystallization in a KFe₂As₂ powder after encapsulation in an evacuated silica-glass tube. All of the setup processes must be conducted in a vacuum chamber and a dry Ar atmosphere in a glove box (Fig. 10.12) [61]. Optimized KFe₂As₂ films on (La, Sr)(Al, Ta)O₃ single-crystal substrates are obtained by crystallization at 700 ℃. These films are strongly c-axis oriented. Electrical measurements were performed with thin films protected by grease passivation to block reaction with the atmosphere. The KFe₂As₂ films exhibit a superconductivity transition at 3.7 K, which is the same as that of bulk KFe₂As₂. This result is the first demonstration of a superconducting KFe₂As₂ thin film.

Fig. 10.12

Reprinted from ref. [61]. Copyright © 2014 American Chemical Society

Solid-phase epitaxy for KFe₂As₂ film growth. a Set up before thermal annealing. b XRD patterns of the films annealed at T _a = 500–800 ℃ buried in KFe₂As₂ powder. b Out-of-plane rocking curve of the 002 diffraction of the KFe₂As₂ film annealed at T _a = 700 ℃. c Pole figure of the 103 diffraction of the KFe₂As₂ film annealed at T _a = 700 ℃.

The obtained KFe₂As₂ films are, however, not epitaxial films, but c-axis orientated ones without an in-plane orientation. This is attributed to the maximum thermal annealing temperature up to 700 ℃ in the conventional annealing method sealed in an evacuated silica-glass tube. When we raised the annealing temperature to >700 ℃, the films decompose into Fe₂As, FeAs, and Fe. It indicates that the gas-tightness of this synthesis condition is poor >700 ℃ for K and the alkali metal component K does not remain in the films.

Therefore, an improved solid-phase epitaxy technique using a custom-made alumina vessel, which realizes a high annealing temperature of 1000 ℃ without vaporization of K from the films, was developed, and high-quality heteroepitaxial KFe₂As₂ thin films on MgO single crystals were successfully obtained (Fig. 10.13) [62]. This result demonstrates that this solid-phase epitaxy technique is a powerful method for the complex compounds with extremely high vapor pressures, such as K.

Fig. 10.13

Reprinted from ref. [62]. Copyright © 2016 The Japan Society of Applied Physics

Improved solid-phase epitaxy for KFe₂As₂ film growth.

3.4 (Sn, Pb)Se Epitaxial Film

SnSe is usually a p-type semiconductor and has the orthorhombic GeS-type layered crystal structure composed of an alternating stack of (Sn²⁺Se^2–)₂ layers along the a-axis. In contrast, a simple binary selenide PbSe has a cubic rock-salt (RS-) type structure. Different from SnSe, the RS-type structure is thermodynamically stable for PbSe at room temperature. Comparing these crystal structures, one expects that a smaller hole effective mass and a higher hole mobility would be realized if the crystal structure of SnSe is changed from the thermal equilibrium GeS-type one to the RS-type one because the three-dimensional network of high-coordination number polyhedra [sixfold (PbSe₆) in the RS-type one] forms larger band dispersions than two-dimensional layered structures [threefold (SnSe₃) in the GeS-type one]. As found in the SnSe–PbSe phase diagram, isovalent Pb²⁺ ions can substitute for part of the Sn²⁺ sites in the orthorhombic GeS-type SnSe at thermal equilibrium [63]. For example, at 400 K, ~20% Pb²⁺ can occupy the Sn²⁺ sites in the orthorhombic GeS-type structure, while more than 60% Pb²⁺ substitution is necessary to stabilize the RS-type structure in (Sn, Pb)Se. On the other hand, in the intermediate Pb concentration region between 20 and 60%, single-phase (Sn, Pb)Se is not obtained. Only a mixture of the GeS-type and the RS-type phases is obtained at a thermal equilibrium at 400 K. However, the RS-type (Sn, Pb)Se with smaller Pb concentrations down to ~40% is stabilized at higher temperatures (e.g., ~1100 K).

Recently, RS-type SnSe and (Sn, Pb)Se have gathered renewed attention because they are expected to be new topological insulators. So far, RS-type (Sn, Pb)Se single crystals are grown using a self-selecting vapor-growth method by a large amount of Pb doping to SnSe (63 and 77% Pb doping) [i.e., the chemical compositions are very Pb-rich, (Sn_0.37Pb_0.63)Se and (Sn_0.23Pb_0.77)Se.] Although a higher Sn concentration would provide a higher topological insulator transition temperature, the maximum Sn concentration is limited to 37%, corresponding to a minimum Pb concentration as high as 63% to stabilize the cubic RS-type structure in SnSe. From the phase diagram, the RS-type (Sn, Pb)Se composition region by freezing the high-temperature RS-type (Sn, Pb)Se phase may be extended.

Thus, isovalent Pb doping to the orthorhombic GeS-type SnSe in order to stabilize the nonequilibrium RS-type (Sn, Pb)Se phase was examined. Reactive solid-phase epitaxy [29], in which a thin RS-type PbSe epitaxial template layer works as a sacrificial layer (Fig. 10.14), was employed [64]. Additionally, a quenching process from 600 ℃ to RT also effectively stabilizes the nonequilibrium RS-type epitaxial (Sn, Pb)Se. Using this technique, we succeeded in varying the Pb concentration from 0 to 100%. The minimum Pb concentration to stabilize the RS-type SnSe is 50%, which is the lowest minimum Pb content ever reported. A structural transition from the GeS-type to the RS-type drastically increases the hole mobility from 60 for SnSe to 290 cm² V⁻¹ s⁻¹ for 58% Pb-doped RS-type film, as expected. The p-type to n-type conversion is also observed upon further increasing the Pb doping up to 100% (i.e., the end member PbSe). A maximum electron mobility of 340 cm² V⁻¹ s⁻¹ is achieved by 61% Pb doping.

Fig. 10.14

Reprinted from ref. [64]. Copyright © 2016 American Chemical Society

Reactive solid-phase epitaxy for the (Sn_1–x Pb _x )S epitaxial films and their carrier transport properties at room temperature as a function of x. σ, N _h,e, and μ _Hall show the electrical conductivity, carrier concentration of hole or electron, and Hall mobility, respectively.

4 Modulation of Functional Nanolayers

State-of-the-art information storage devices such as USB flash drives are electronic data storage devices, which store digital information by electrical resistivity changes of semiconducting silicon using the electric field effect to process information into “words” consisting of various combinations of the numbers “0” and “1”. Since miniaturization technology has already reached its limit, epoch-making technology is strongly required to further improve the storage capacity.

We have demonstrated multi-information memory devices, which are composed of metal oxides showing both an electrical resistivity change and magnetism/color change simultaneously by a redox reaction of the metal oxides. Three-terminal thin-film transistor (TFT) structures were fabricated on a functional metal oxide using an insulating oxide, water-infiltrated calcium aluminate (C12A7) with mesoporous structure, as the gate insulator. We utilized H⁺/OH⁻ ions in the water to change the valence state of the metal oxides by applying a gate voltage since H⁺/OH⁻ ions are strong reducing/oxidizing agents for metal oxides. Upon changing the valence of the transition metal ion, the metal oxide changes from an insulator to a metal as well as from nonmagnetic to magnetic or from colorless transparent (invisible) to visible, as schematically illustrated in Fig. 10.15. Although the present device requires a relatively long storage time (a few seconds) because it utilizes mobile ion diffusion in the functional metal oxide, it has great merits. For example, it has nonvolatile operations, which mean no standby power is required after storing information. The present multi-information storage device should be useful for Internet of Things (IoT) technologies.

Fig. 10.15

Schematic concept of a reversible conversion of optical-, electrical-, and magnetic properties of functional metal oxides using an electrochemical redox reaction of metal oxides with H⁺ and OH⁻ of liquid water. For example, by combining the electrical properties and the optical properties, a novel electrochromic device, which can store A/B in addition to 0/1 for storing information, can be developed

As IoT technologies become more ubiquitous, the information gathered annually is rapidly increasing as various machines, as well as personal computers, are connected to the internet. State-of-the-art information storage devices such as USB flash drives are electronic data storage devices. They store digital information using an electrical resistivity change of semiconducting silicon and an electric field effect that process information into “words” consisting of various combinations of the numbers “0” and “1”. Although the storage capacity of such devices increased annually due to miniaturization techniques, the limit of such miniaturization techniques has already been reached. Consequently, complicated multi-levelization techniques such as “0”, “1”, “2”, and “3” are utilized to improve the storage capacity. To further improve the storage density, epoch-making technology is strongly required.

To overcome this obstacle, we proposed the following idea: utilizing functional materials whose optical transmittance or magnetism can be dramatically changed together with a change in the electrical resistivity. Such features would be very useful to improve the storage capacity. For example, multiple storing/reading of information becomes possible when vision and electrical signals are combined with displays. Such devices are appropriate for future IoT technologies. However, it is impossible to use optical transmittance or magnetic properties in case of semiconductor Si. In addition, it is impossible to use the electrical resistivity change in case of a magnetic metal.

Our research has focused on metal oxides because some metal oxides exhibit changes in their optical property or magnetic property together with the electrical property via an oxidation/reduction (redox) reaction. Generally, such redox reactions occur at a high temperature (several hundred degrees) heat treatment in an oxidizing or reducing atmosphere. However, this method is inappropriate for device operations. On the other hand, redox reactions using electrochemistry such chemical battery cells occur at room temperature. The latter technique is appropriate for practical applications, but the device must be sealed to prevent electrolyte leakage.

Unexpectedly, in 2010, Ohta et al. found that water, which is automatically absorbed into the mesoporous structure of an insulating oxide due to capillary action, can be a good electrolyte for this purpose [65]. Three-terminal thin-film transistor (TFT) structures were fabricated on a functional metal oxide using an insulating oxide, calcium aluminate (12CaO∙7Al₂O₃, C12A7) with mesoporous structure (namely CAN, calcium aluminate with nanopores) as the gate insulator (Fig. 10.16). C12A7 can be prepared by PLD at room temperature under a relatively high oxygen atmosphere of ~5 Pa. CAN films contain many mesopores (~10 nm in diameter) whose volume fraction is ~30% [65]. Temperature desorption spectra of the CAN film reveal that the mesopores are fully occupied with molecular water. The AC conductivity of the CAN film is ~10⁻⁹ S cm⁻¹ at room temperature, [66] which is comparable to that of ultrapure water. Thus, water moisture in air is automatically absorbed in the mesopores of CAN film by capillary action.

Fig. 10.16

Schematic illustration of CAN (calcium aluminate with nanopore) gated functional oxide thin-film transistor with a three-terminal electrodes geometry. Since 30 volume percent of the CAN film is occupied with liquid water, H⁺ and OH⁻ ions in the CAN film move with a gate voltage application. The percolation AC conductivity is ~10⁻⁹ S cm⁻¹, which is comparable to that of ultrapure water. Changing the valence of the transition metal ion by a gate voltage application changes the functional oxide from an insulator to a metal as well as from a nonmagnetic to a magnetic material or from colorless and transparent to a black color

In 2016, Katase and Ohta et al. utilized H⁺/OH⁻ ions in a water-infiltrated CAN film to change the valence of the metal oxides by applying a gate voltage because H⁺/OH⁻ ions are strong reducing/oxidizing agents for metal oxides. As the valence of the transition metal ion changes, the metal oxide changes from an insulator to a metal, a nonmagnetic to a magnetic material, or from transparent to a black color. Although the present device requires a relatively long storage time (a few seconds) since it utilizes mobile ion diffusion in the functional metal oxide, it has great merits. For example, it has nonvolatile operations, which mean no standby power is required after storing information. The present multi-information storage devices would be useful for IoT technologies.

The authors have developed two multi-information memory devices, which use a magnetic [67] or optical [68] signal along with an electronic signal to double the storage capacity in these “multiplex writing/reading” devices. In addition to the binary 0/1 method of storing information in a state-of-the-art memory device, the present devices can also store A/B for the information. More details for each type of memory device are provided below.

4.1 Utilizing Antiferromagnetic Insulator/Ferromagnetic Metal Conversion in SrCoO_2.5+δ [67]

To realize “multiplex writing/reading” devices, material selection is the most information factor. Katase and Ohta et al. choose strontium cobaltite, SrCoO_2.5+δ , for this purpose because SrCoO_2.5 is an antiferromagnetic insulator and SrCoO₃ is a ferromagnetic metal [69, 70]. The valence state of the cobalt ion in SrCoO_2.5+δ can be controlled from 3 + (SrCoO_2.5) to 4 + (SrCoO₃) by changing the excess oxygen content (δ) from 0 to 0.5. Since the crystal structures of SrCoO_2.5 (brownmillerite) and SrCoO₃ (perovskite) are similar, the authors expected that the topotactic redox reaction between SrCoO_2.5 and SrCoO₃ can be controlled electrochemically. By utilizing this phenomenon for three-terminal thin-film transistors with water containing a mesoporous glass gate insulator, we developed a multi-information memory device. This device can be utilized not only to change the electrical resistivity (0/1) but also to change of magnetic property (A/B), as schematically shown in Fig. 10.17.

Fig. 10.17

Reprinted with permission from [67]. © 2016 John Wiley and Sons

Principle of a multi-memory device using antiferromagnetic insulator/ferromagnetic metal conversion in SrCoO_2.5+δ . This device would store both A/B and 0/1 information.

The three-terminal TFT device, which is composed of an epitaxial SrCoO_2.5 film (30 nm, active channel material), an amorphous Na-Ta-O film with a mesoporous structure (300 nm, gate insulator), and an amorphous WO₃ film (20 nm, proton absorber), was prepared by PLD on (001) SrTiO₃ single-crystal substrate. It should be noted that we recently developed an amorphous Na-Ta-O film with a mesoporous structure, which can be used as an alkaline solution. When a negative gate voltage (−3 V) is applied between the gate and source electrodes, OH⁻ ions, which are contained in the mesoporous glass, penetrate into SrCoO_2.5+δ . Finally, SrCoO₃ is formed in 3 s. On the contrary, a positive gate voltage (+3 V) application to the gate–source electrodes reduces SrCoO₃ into SrCoO_2.5 in 3 s [67].

Figure 10.18a shows a schematic of the device structure, which is similar to conventional three-terminal thin-film transistors. The channel (source–drain) length and width are 800 μm and 400 μm, respectively. The electrodes E1–E4 are used to measure the sheet resistance (R _s). Figure 10.18b shows the changes in R _s of the device. Before the device operation (state A), R _s increases with decreasing temperature, indicating an insulating behavior. When a negative gate voltage (−3 V) is applied for 3 s (state B), R _s decreases by three orders of magnitude and shows a metallic temperature dependence. After that, the device returns to the original state when a positive gate voltage (+3 V) is applied for 3 s. The device is reversibly operable (Fig. 10.18b, inset).

Fig. 10.18

Reprinted with permission from [67]. © 2016 John Wiley and Sons

a Schematic device structure similar to conventional three-terminal thin-film transistors. b Temperature dependence of the sheet resistance of the device. a Virgin state, b after applying a negative Vg of −3 V, and c subsequent application of +3 V (dotted line). The inset shows the cyclability at RT in air. c m−T curves of the SrCoOx layer at states A−C in (b) measured under H = 20 Oe applied parallel to the in-plane direction. The inset shows a magnetic hysteresis loop at 10 K at states A and B.

Figure 10.18c shows the changes in the magnetic state of the device at states A, B, and C. At states A and C, the magnetic moment is zero, indicating that SrCoO_2.5+δ (δ = 0) is an antiferromagnetic state. At state B, the device shows a ferromagnetic behavior with a Curie temperature of 275 K, indicating that SrCoO_2.5+δ (δ ~ 0.5) is a ferromagnetic state. These results clearly demonstrate that both electrical resistivity and magnetism changes can be used in the present device.

4.2 Utilizing a Colorless Transparent Insulator/Dark Blue Metal Conversion in H _x WO₃ [68]

Katase and Ohta et al. [68] have also developed a new information display/storage device using a three-terminal thin-film transistor structure on an electrochromic material, which has been attracted attention as an “electric curtain”. The device shows a color change (colorless transparent/dark blue) together with an electrical conductivity change (insulator/metal) by applying a gate voltage. Since the device can be fabricated at room temperature, low cost fabrication is possible. Thus, larger area devices are easily fabricated. For example, the present device is applicable as an information display/storage on a window glass.

Protonation/deprotonation of tungsten trioxide (WO₃), known as an electrochromic material, is converted reversibly from a colorless transparent insulator to a dark blue metal [71]. By utilizing this phenomenon in a three-terminal thin-film transistor with a water-containing mesoporous glass gate insulator, a multi-information memory device has been realized. This device can be utilized using not only a change in the electrical resistivity (0/1) but also a change in optical transmittance (A/B), as schematically shown in Fig. 10.19.

Fig. 10.19

Reprinted from [68]. © 2016 NPG

Principle of a multi-memory device using a colorless transparent insulator (amorphous WO₃)/dark blue metal (amorphous HWO₃) conversion in H _x WO₃. This device would store both A/B and 0/1 information. This device should be suitable as a smart window or a smart mirror, which can display or store information. Furthermore, the present device can be used as an “electronic curtain” as the whole surface of window glass can be switched reversibly from colorless and transparent to dark blue.

The three-terminal TFT device composed of an amorphous WO₃ film (100 nm, active channel material), a mesoporous CAN film (300 nm, gate insulator), a polycrystalline NiO film (50 nm, oxygen absorber), and amorphous ITO films (20 nm, gate, source, and drain electrodes) was prepared by PLD at room temperature on a glass substrate. When a positive gate voltage (a few volts) is applied between the gate and the source electrodes, H⁺ and OH⁻ ions, which are contained in the mesoporous glass, diffuse to the WO₃ and NiO sides, respectively, forming HWO₃ and NiOOH. Since the resultant dark blue colored HWO₃ shows a metallic electrical conductivity, the channel (drain–source) becomes electrically conductive. On the contrary, a negative gate voltage (a few volts) application to the gate–source electrodes results in HWO₃ and NiOOH returning to WO₃ and NiO, respectively. This conversion can be reversibly operated and the degree of change can be controlled by the applied gate voltage.

Figure 10.20a schematically depicts the device structure composed of a-WO₃ (80 nm), CAN (300 nm), and NiO (20 nm)/ITO (20 nm) layers. Transparent ITO thin films are used for all the electrodes. All the films are deposited by PLD at room temperature. The device was fabricated on a transparent glass substrate (1 cm × 1 cm) as shown in Fig. 10.20b. The channel (source–drain) length and width are 800 μm and 400 μm, respectively. Note that the device is colorless and transparent. Figure 10.20c shows the changes in R _s. When a positive gate voltage is applied to the device for 10 s, R _s decreases by several orders of magnitude. The color becomes dark blue (Fig. 10.20d). The device reverts to the original state (transparent, insulator) when a negative gate voltage is applied for 10 s. The device is reversibly operable. Although the present device requires a relatively long storage time (a few seconds) since it utilizes mobile ion diffusion in the functional metal oxide, it has great merits. For example, it employs nonvolatile operation, which means standby power is not required after storing information. The present multi-information storage devices would be useful for IoT technologies.

Fig. 10.20

Reprinted from [68]. © 2016 NPG

a Schematic device structure. b Transparent device on a glass substrate. c Repeatable switching of the sheet resistance. d Optical transmission spectra. Before the operation, the device is an insulator (R _s ~ 10⁸ Ω sq⁻¹) and fully transparent in the visible light region. When a positive gate voltage is applied to the device for 10 s, R _s decreases several orders of magnitude and the device becomes dark blue due to electrochemical protonation of WO₃.