The α-MoO3 has a layered structure consisting of sheets of MoO6 bipyramids with normal along the [010] direction. The sheets are bonded by weak vdW forces while the bonds within each layer are covalent and ionic as shown in the side view schematic in Fig. 1(a). α-MoO3 can therefore be exfoliated to form sheets with [010] normal.
Figure 1(b) shows the X-ray diffraction (XRD) patterns of 100-nm-thick films of sputtered MoO3 target on STO grown at a substrate temperature of 600°C, working pressure of 20 mTorr, and argon-to-oxygen ratio of 10:1. The orthorhombic α-MoO3 peaks are located at 2θ = 12.7°, 25.5°, 38.7°, and 67.0°, which correspond to (020), (040), (060), and (0 10 0) of the Pbnm space group (JCPDS #05-0508, with a = 3.962 Å, b = 13.858 Å, c = 3.697 Å in bulk and α = β = γ = 90°). The strong (0k0) reflection indicates that the planes are parallel to the surface of the (001)-oriented STO substrates.
The peak positions of α-MoO3 thin films yield an (010) interplanar spacing of 13.884 Å, larger than the bulk value implying tensile strain along the out-of-plane direction. The lattice constant of α-MoO3 along the [001] direction is significantly smaller than that of STO (a = b = c = 3.905 Å) but along the [100] direction, the lattice constant of α-MoO3 is larger. Figure 1(a) shows top view and side view schematics of the structural relationship between α-MoO3 and STO. We believe that the strain of the α-MoO3 layer is partially or completely relaxed along the [001] direction, whereas a compressive strain is generated along the [100] direction, increasing the out-of-plane lattice constant when the films grow coherently. From the 4-fold symmetry of the STO, two variants of the α-MoO3 can form with [100] of α-MoO3 parallel to [100] and [010] of STO respectively.
A flat surface morphology with protruding faceted edges is observed for the 100-nm-thick α-MoO3 film in the top-view SEM image in Fig. 1(c). The [001] and [010] directions for the STO are indicated. The large-area (010) facets of α-MoO3 are parallel to the surface of the STO, consistent with the strong (0k0) peaks in XRD. In contrast, α-MoO3 films grown under different sputter conditions consisted of crystals with a range of orientations and had low (0k0) peak intensities (Figures S1(a)–(d)).
The XRD relative peak intensities of a 30-nm-thick α-MoO3 film are the same as those of the 100-nm-thick film, with dominant (0k0) peaks; however, the peak position yields an out-of-plane lattice parameter of 13.923 Å indicating a higher out-of-plane tensile strain. Faceted rectangular crystallites on the flat surface are observed, although they are smaller than those in the 100-nm-thick film (Fig. 1(f)). EBSD mapping identifies the crystalline orientations of grains, as shown in the right images of Fig. 1(c) and (e). The two epitaxial relationships of α-MoO3 (010) || STO (001), α-MoO3 [100] || STO [100] and α-MoO3 (010) || STO (001), α-MoO3 [001] || STO [100] are visible in both 30- and 100-nm-thick films, which consist of the two orientational variants rotated in-plane by 90°. The size of the mosaic domains in EBSD mapping is comparable to that of the dark and bright grains in the SEM image for the 30-nm-thick α-MoO3 film, Fig. 1(e), suggesting that the SEM contrast originates from the orientation of the domains.
The feasibility of using the as-grown α-MoO3 thin films on STO substrates as a sacrificial layer for transferring epitaxial oxide thin films was investigated by etching in deionized water. No discernible change was observed when the films were in room-temperature water, whereas α-MoO3 thin films were etched in water heated to 45°C on a hot plate, as verified by the top-view SEM images. Pits and islands with edges parallel to the (100) and (001) planes were observed in films dipped for 30 min and 2 h (Figures S2(a) and (b)), indicating anisotropic etching. Small-sized grains remained for a longer time (6 h), and a smooth surface was observed after etching for 24 h, as shown in Figures S2(c) and (d).
When the hydroxyl group in water interacts with the Mo cations in the α-MoO3 lattice at the surface of the films, a water-soluble \({MoO}_{4}^{2-}\) anion is formed. [25] Figures S3(a) and (b) demonstrate the spread of the removed areas. The pits are first formed because the etching along the vertical direction is dominant due to the rapid etching rate along the [010] direction; the removed area then widens along the lateral direction.
The peaks from the α-MoO3 phase disappeared when the as-grown film was immersed for 24 h while the STO peaks remained, Fig. 2(a) and (b). From the energy dispersive spectroscopy (EDS) data in Figure S4(a), Mo was detected in the α-MoO3 thin film etched for 30 min but was absent after 24 h etching. The chemical etching behavior and mechanism of α-MoO3 single crystals in NaOH aqueous and KOH buffer solutions have been previously reported, [26–28] but our work is the first report of the etching of an epitaxial α-MoO3 film in heated water. In contrast, amorphous or polycrystalline α-MoO3 phases did not etch in water. The selective removal of α-MoO3 in warm water is promising for its use as a sacrificial layer.
Transfer of an Au film was demonstrated as indicated in the schematic illustration in Fig. 2(d). Polyethylene terephthalate (PET) tape was attached to an 800-nm-thick Au-coated α-MoO3 layer on the STO substrate. After 24 h immersion in 45˚C water the PET/Au floats to the surface whereas the STO substrate sinks, Figure S4(b). Figure 2(e) shows the Au thin film on α-MoO3 before etching and Fig. 2(f) is the transferred Au on PET after etching. The STO substrate after etching appears indistinguishable from the bare substrate as shown in Fig. 2(g).
We then redeposited α-MoO3 on the 24 h etched substrate. The regrown film was visually similar to that of the first grown film (Fig. 2(h)), the XRD pattern indicated (0k0) reflections (Fig. 2(c)) and the regrown film had a flat surface morphology with rectangular crystallites in the top-view SEM image (Figure S4(c)), i.e., the regrown film is indistinguishable from the first as-grown thin film. This demonstrates the potential for recycling the single-crystalline STO substrates.
An alternative transfer process was demonstrated by peeling the Au off the α-MoO3 with tape, Figure S4(d), similar to the transfer using scotch tape of films grown on graphene. The top-view SEM image in Figure S4(e) exhibited traces of α-MoO3 remaining on the substrate, and the EDS spectra of the transferred layer showed both Au and Mo (Figure S4(f)), indicating that the thin-film transfer was achieved by rupturing the weak vdW bonding between the α-MoO3 nanosheets instead of separation between the Au and α-MoO3 interfaces. These two examples therefore demonstrate the versatility of α-MoO3 as a sacrificial layer, which can either be dissolved in warm water or split apart by mechanical forces.
We now apply α-MoO3 to the exfoliation of an oxide, spinel-structured CFO. First, for comparison, a CFO thin film was grown directly on a (001)-oriented STO substrate. The CFO is under compressive strain along the in-plane directions because the lattice parameter of CFO (abulk = 8.392 Å, JCPDS # 22-1086) is greater than the doubled lattice parameter of STO (Fig. 3(a)). Figure S5(a) shows the θ–2θ XRD scan of a 100-nm-thick CFO thin film, which reveals the (004) peak of CFO without additional reflections, consistent with epitaxial or textured growth. The out-of-plane lattice parameter of the sputtered CFO was 8.469 Å, which is larger than that of bulk CFO, due to Poisson expansion and possibly the presence of oxygen vacancies. Top-view SEM images of a CFO thin film, Fig. 3(a), show 30–50 nm diameter faceted crystals within a smooth matrix, which are assumed to form to relieve strain.
Room temperature magnetic hysteresis loops show an in-plane easy axis. The in-plane loop shows a coercivity of 3.3 kOe and a magnetic moment of 230 emu cm− 3 at 10 kOe, the maximum field of the magnetometer, though the loop is not saturated at this field. The out-of-plane direction is a hard axis with an anisotropy field much greater than 10 kOe, based on the low magnetic moment compared to the easy axis loop. The in-plane easy axis is attributed to the combination of shape anisotropy and magnetoelastic anisotropy. Bulk CFO has a magnetization of 400 emu cm− 3 [29] implying that a thin film of CFO has a shape anisotropy field of 5.0 kOe. The magnetoelastic anisotropy also favors an in-plane easy axis for a compressively strained film owing to the negative magnetostriction constant λ100,CFO = (-250 to -590) × 10− 6 [30–32].
Comparing a CFO film on a 100-nm-thick α-MoO3 layer (Fig. 3(b)), we observe only the (004) CFO reflection in the XRD pattern, Figure S5(b); however, the surface morphology showed a higher density of crystallites, Fig. 3(b). The (004) peak position is at a lower angle than that of CFO/STO (out-of-plane CFO lattice parameter is 8.412 Å), suggesting greater strain relaxation of CFO/100-nm-thick α-MoO3/STO vs. CFO/STO. The strain relaxation in CFO/100-nm-thick α-MoO3/STO results in a decrease in magnetic anisotropy but the easy axis remains in-plane. The out-of-plane loop is still not saturated at 10 kOe but reaches a higher magnetic moment than that of CFO/STO.
The strain state of the CFO thin film depends on the thickness of the α-MoO3 layer. In Fig. 3(c,d), CFO/30-nm-thick α-MoO3/STO had an out-of-plane lattice parameter of 8.426 Å, intermediate between 8.469 Å for CFO/STO and 8.412 Å for CFO/30-nm-thick α-MoO3/STO. The hysteresis loop follows the trend, indicating an anisotropy field between that of the other two samples attributed to the intermediate strain state of CFO/30-nm-thick α-MoO3/STO. Furthermore, the CFO/30-nm-thick α-MoO3/STO exhibited fewer crystallites than the CFO/100-nm-thick α-MoO3/STO, consistent with these surface features being associated with strain relaxation.
EBSD maps of the CFO/30-nm-thick α-MoO3/STO measured with respect to all three-coordinate axis verified the single crystallinity of CFO with a cube-on-cube epitaxial orientation over a large area (Fig. 3(e)). Some points in the map were not indexed (shown in black color) most likely due to local variations in morphology. A (001) pole figure (Figure S5(c)) confirms the epitaxial growth.
Figure 4(a) shows the transfer process of CFO thin films onto flexible PET substrates. An Au layer was deposited over the CFO/α-MoO3/STO then an adhesive-coated PET substrate was applied to the surface of the Au layer. The PET was gently lifted to peel off the CFO layer from the STO substrate by mechanically exfoliating the α-MoO3 nanosheets to break the vdW force. The PET/Au/CFO/α-MoO3/STO was separated into two parts: the STO substrate part, which was dark in color compared to the bare STO or α-MoO3 coated STO, and the PET part, which exhibited a maroon color, as shown in Fig. 4(b). The out-of-plane lattice parameters of the CFO layer did not change after exfoliation, as shown by a comparison of the magnified XRD pattern in Fig. 4(c). The CFO films were detached from the substrates regardless of the thickness of the α-MoO3; however, exfoliation was unsuccessful for CFO grown on polycrystalline or amorphous α-MoO3.
Figure S6(a,d) show that α-MoO3 remained on both the PET/Au/CFO and the STO after exfoliation. A SEM image of the separated α-MoO3/CFO/Au/PET shows steps parallel to the [100] and [001] directions of α-MoO3 (Figure S6(b)). When the tape is lifted, the nanosheets separate along the (010) vdW planes, but the sheets also break along the in-plane [001] and [100] directions leaving steps in the surface consisting of multiple nanosheets. The step structure is also visible at the surface of the detached STO substrate (Figure S6(c)).
TEM was used to image a cross-section of exfoliated α-MoO3/CFO/Au/PET which included regions with different heights of residual α-MoO3, Figure S7(a-c), after coating with carbon. Figure 4(d) indicates that the film consisted of a 100 nm-thick CFO, a thin interlayer, and 50 nm-thick blocks of α-MoO3 left when the α-MoO3 was split apart. Figure S7(d) shows steps in the side of the residual α-MoO3 block where the α-MoO3 separated.
A high-resolution transmission electron microscopy (HRTEM) image along the [100] zone axis of STO, α-MoO3 and CFO is shown in Fig. 4(e). Considering the two possible orientational variants of α-MoO3, we expect to see regions of α-MoO3 with [100] and [001] zone axes parallel to STO [100], but the area of sample we examined contained only the [100] variant. The region of α-MoO3 shown by the gray square displayed two orthogonal planes with spacing of 6.94 Å and 3.70 Å, which correspond to (020) and (001) of α-MoO3, respectively. The (020), (001), and (021) spots observed in the Fast Fourier Transform (FFT) pattern in Fig. 4(e) are consistent with the orthorhombic α-MoO3 phase, in which the (021) planes formed an angle of 62° with the (020) plane. The bottom middle HRTEM image shows perpendicular (02\(\stackrel{-}{2}\)) and (0\(\stackrel{-}{2}\stackrel{-}{2}\)) planes with spacings of 2.97 Å. The bottom right panel shows the FFT corresponding to the CFO [100] direction.
The interface layer between CFO and α-MoO3 extends across the entire sample and showed bright contrast with a thickness of 5 nm, and a different crystal structure from the α-MoO3, Fig. 4(e) lower left. The EDS line scan in Figure S7(e) indicates Mo, Co and Fe are present at the interface. The layer most likely results from intermixing of the CFO into the MoO3. It has been reported that Co and Fe are soluble in α-MoO3, and that bombardment (which occurs during sputtering) can generate oxygen vacancies and modify the structure of the oxide. [33, 34]
We now describe the magnetic behavior of a CFO film transferred on a flexible PET substrate using a 30-nm-thick α-MoO3 layer. The magnetic hysteresis loops of the as-grown CFO/α-MoO3/STO and the transferred CFO/Au/PET are similar, showing an in-plane easy axis (Figs. 3(c) and 5(a)), though the transferred film loops are less anisotropic, consistent with strain relaxation. The magnetic anisotropy of the transferred film was modulated by bending the film through a 1 cm radius of curvature. In Fig. 5(b), the in-plane (tangential to the bend) and out-of-plane loops of the film become more similar as a result of the curved geometry, because each loop measurement samples regions of the film tilted at an angle to the applied field. The effect of bending can be seen from a comparison of the in-plane (axial) loop of the bent film vs. the in-plane loop of the flat film, Fig. 5(c). The bent film shows lower in-plane coercivity, remanence and loop area compared to the flat film. This is consistent with the bending-induced introduction of a tensile in-plane (tangential) strain, which reduces the in-plane anisotropy energy. The magnetic anisotropy modulation of the CFO/Au/PET by bending is larger than that reported for exfoliated CFO grown on a sacrificial layer of MgO or SAO by ALD or PLD. [18, 35]
Our results show that mechanical exfoliation of a α-MoO3 sacrificial layer can be used to effectively transfer CFO thin films onto flexible substrates. The CFO thin film retains the majority of its magnetic anisotropy after transfer, but the magnetic anisotropy can be modulated by bending. The transfer process of epitaxial oxide thin films onto flexible substrates using vdW epitaxy described here is relevant to achieving high-performance flexible electronic devices. Furthermore, the solubility of epitaxial α-MoO3 films in warm water provides a guideline for the growth of highly crystallized oxide thin films using sputtering, paving the way for large-area flexible electronics with low cost, good performance and durability.