Development of a microfabrication process is essential to embed fascinating physical properties of functional materials into mesoscopic devices. Different from well-investigated materials with established microfabrication process, newly-discovered materials often meet difficulty when scaling down into a mesoscopic size, because process damages cause serious deterioration of their functionalities. Here, we demonstrate a versatile lift-off method using a carefully designed sacrificial bilayer, composed of an easily soluble layer and a thermally stable rigid layer. In this method, the target films can be grown in optimum conditions, such as high temperature and high oxygen partial pressure, on the stable pre-patterned substrate with the inorganic sacrificial bilayer. After film deposition, measurable patterned devices can be obtained just by a short-time lift-off in a mild chemical solution. We carried out micron-scale patterning and electrical measurements by applying this technique to one of perovskite oxides, SrRuO3, and Fe-based chalcogenide superconductors, FeSe, both of which are incompatible with conventional photolithography and dry-etching processes. The demonstrated narrowest line width of 5 μm is successfully patterned with maintaining the almost identical properties of the pristine films, exemplifying that process damage is minimized. The demonstrated versatile patterning process expands the range of application of emerging functional materials in thin film devices.

Continuous advances in microfabrication techniques have played an important role in the discovery of various physical phenomena and significant functionalities in the micro devices based on semiconductors,1,2 superconductors,3,4 metal spintronic devices,5 and recent layered materials.6 In particular, silicon and other semiconductor electronics have been established on robust device fabrication processes. Recently the combination of exfoliation and electron-beam lithography techniques has been widely applied to layered materials, graphene and chalcogenides, for observation of exotic electron transport phenomena.7 Such a widely applicable device processing method leads an explosive growth of the research field. As a minimum prerequisite for development of versatile process, it is desirable to suppress damages to the quality of target materials during microfabrication. However, in functional oxides and chalcogenides, development of applicable device processes often meets difficulty because it has to be adapted to the different chemical trend of each target material. For example, strongly correlated oxides and metal chalcogenides drastically deteriorate their physical properties by the formation of oxygen or selenium vacancies. Therefore, development of a versatile device process has a great potential to open a new arena for realization of a superior device performance based on emerging quantum materials such as strongly-correlated electron systems, superconductors, and topological insulators. In this study, we propose a versatile method based on a lift-off process using easily soluble compounds, which is compatible to the matured microfabrication technology.

The developed process is based on lift-off of a sacrificial bilayer, stacking of an easily soluble layer and a shape-defining rigid layer.8–15 The bottom layer works as a lift-off layer, which should be soluble in a mild chemical solution. The covering rigid layer works as a protection layer for the bottom layer and prevent thermal deformation of the designed device structure during the growth of the target materials. Inorganic compounds are appropriate candidates that satisfy the above-mentioned two requirements of solubility and stability. There is a significant feature of this process that we can employ the optimum growth condition when fabricating a target material, such as high temperature and different gas conditions, because a stable bilayer under these conditions is selectable. To achieve the patterned device structure after the growth, the sacrificial bilayer region is removed by lift-off with gentle chemical agents or water. As representative examples, we examined two types of sacrificial bilayers LaAlO3/BaOx and AlOx/Al for device patterning of a perovskite oxide SrRuO3 and a Fe-based superconductor FeSe.

Figure 1(a) illustrates the schematics of the procedure for device patterning using the sacrificial bilayer. Firstly, we prepared a substrate patterned with organic photoresist by a conventional photolithography technique. Then, a sacrificial bilayer is deposited on the substrate in a vacuum chamber at room temperature. Removing the photoresist with organic solvent results in a patterned sacrificial bilayer on the substrate. After installing the pre-patterned substrate to the deposition chamber, the target material is fabricated in the optimum growth condition. By selecting appropriate inorganic bilayers, growth temperature as high as hundreds of Celsius degrees does not cause deformation of the patterned template, maintaining the desired device structure. Finally, lift-off is carried out to remove the sacrificial bilayer within short time scale about a few minutes by gentle chemical agents, obtaining the desired device structure of the target material. This process only requires standard photolithography and thin film growth techniques. To minimize damage in the patterned film, selection of a proper lift-off layer is important. Figure 1(b) summarizes the melting point (TMP) and the chemical property of the typical etchant for some candidates of the lift-off layer. Higher TMP is preferable in terms of thermal stability to maintain the designed structure at the growth temperature. The chemical stability of the target material and the lift-off layer should be also considered to satisfy that the lift-off layer can be removed by a mild etchant without damaging the target material.

FIG. 1.

(a) The scheme of patterning process using a sacrificial bilayer from an initial substrate (left) to lift-off after thin film growth (right). (b) Candidate materials for the bottom soluble layer summarized in terms of melting point16 and the acid-base properties of typical etchant:12,17 dilute KOH (aq) for Al, H2O2 (aq) in acidic condition for Mo, H2O for CaO, SrO and BaO, dilute HCl (aq) for ZnO, (NH4)2[Ce(NO3)6] (aq) with acid for Cr, and HCl (aq) for Ti. BaO and Al, which are focused in this work, are plotted by red symbols. (c) Optical microscope images of the LaAlO3/BaOx sacrificial bilayer (dark blue region) and the bare SrTiO3 substrate surface open for channel deposition (whitish region), (d) the 10 μm SrRuO3 channel (dark gray region), and (e) the sample wired to a measurement holder.

FIG. 1.

(a) The scheme of patterning process using a sacrificial bilayer from an initial substrate (left) to lift-off after thin film growth (right). (b) Candidate materials for the bottom soluble layer summarized in terms of melting point16 and the acid-base properties of typical etchant:12,17 dilute KOH (aq) for Al, H2O2 (aq) in acidic condition for Mo, H2O for CaO, SrO and BaO, dilute HCl (aq) for ZnO, (NH4)2[Ce(NO3)6] (aq) with acid for Cr, and HCl (aq) for Ti. BaO and Al, which are focused in this work, are plotted by red symbols. (c) Optical microscope images of the LaAlO3/BaOx sacrificial bilayer (dark blue region) and the bare SrTiO3 substrate surface open for channel deposition (whitish region), (d) the 10 μm SrRuO3 channel (dark gray region), and (e) the sample wired to a measurement holder.

Close modal

Among the candidate materials shown in Fig. 1(b),16,17 we employed two examples of the lift-off layers, BaO; high TMP around 2000 oC and the etchant pH of about 7, and Al; low TMP around 660 oC and the etchant pH of about 12. BaO is a simple ionic oxide that is soluble in water through the following chemical reaction: BaO (s) + H2O → Ba(OH)2 (aq). We prepared BaOx film by pulsed laser deposition (PLD) at room temperature using a BaO2 pellet as a source material.18 Aluminum is also a good candidate, being removable by a basic solution with pH ∼ 12. The merit of the aluminum is the easiness of the film deposition using standard thermal or electron beam evaporation. As the top rigid layer, we selected amorphous LaAlO3 (TMP ∼ 2110 oC) for BaOx and AlOx (TMP ∼ 2054 oC) for Al, because of their robust thermal stability owing to high TMP. Both LaAlO3 and AlOx were deposited by PLD at room temperature.

The optical micrographs at each step of the LaAlO3 (20 nm) /BaOx (190 nm) bilayer process for patterning SrRuO3 film are displayed in Fig. 1(c)–1(e). Firstly, the LaAlO3/BaOx bilayer was patterned on a SrTiO3 substrate using conventional photolithography and resist lift-off, which is clearly seen as the deep blue region in Fig. 1(c). The whitish region corresponds to the bare SrTiO3 surface where SrRuO3 channels are to be formed. After the SrRuO3 deposition by PLD at 850 °C under the oxygen partial pressure of 100 mTorr, the sample was put in water under ultrasound for ∼ 10 sec to remove SrRuO3/LaAlO3/BaOx sacrificial regions. As a result, the 10-μm-wide SrRuO3 channel was obtained in the dark region of Fig. 1(d), representing a well-defined channel width. Figure 1(e) is a photograph of the wire-bonded sample on a typical sample holder for electrical transport measurements in PPMS. In total after the SrRuO3 growth, it takes typically as short as 30 minutes to make a sample ready for electrical measurements. Such a quick process is effective when the target material suffers from time degradation.

One of the target materials in this study, SrRuO3 has a perovskite structure as schematically shown in Fig. 2(a).19 In thin film growth of perovskite oxides, growth temperature as high as several hundred degree Celsius is usually required to enhance the migration of adatoms on a substrate and achieve an atomically flat surface and high crystalline quality. Here we use SrTiO3 substrates, which are widely employed for perovskite oxide thin films growth owing to the commercially available atomically flat surface.20 To process high-temperature grown SrRuO3 thin films, we examine two kinds of sacrificial bilayers: AlOx/Al and LaAlO3/BaOx. The optimum growth temperature (Tg) for SrRuO3 film in our apparatus is about 850 oC, which is much higher than the melting point of Al. In fact, AlOx/Al bilayer could not completely be removed by lift-off process (not shown), probably due to partial melting of Al. Considering the thermal stability, we grew SrRuO3 layers at 700 oC for AlOx/Al and 850 oC for LaAlO3/BaOx sacrificial layers. Figures 2(b) and 2(c) show the structural characterization results for the patterned SrRuO3 channels using the AlOx (20 nm) /Al (50 nm) and LaAlO3/BaOx sacrificial layers, respectively. Well-defined SrRuO3 channels are obtained in both samples as seen in the right inset of the wide scale atomic force microscope (AFM) images. Moreover, the flat SrRuO3 surface with step-and-terrace structure (the narrow scale AFM images in the left insets) indicates that the growth temperature of 700 and 850 oC is high enough to allow adatoms to migrate sufficiently. The height of the channel obtained from the AFM (tAFM = 21 nm) image in Fig. 2(c) is in good agreement with the SrRuO3 thickness determined from the thickness fringes in X-ray diffraction (XRD) pattern (tXRD = 20 nm) shown in Fig. 2(d). Figure 2(e) displays the element mappings of Al (right upper panel) and Ru (right lower panel) for the AlOx/Al bilayer before SrRuO3 deposition. The AlOx/Al bilayer patterned by photolithography is clearly visible in Al mapping as the dark blue region in top right panel of Fig. 2(e). After lift-off of the sacrificial region of SrRuO3/AlOx/Al layer by a basic solution of KOH with pH ∼ 12, Al signal is completely disappeared (right upper panel of Fig. 2(f)). Instead, the SrRuO3 channel structure appears in Ru mapping as red region (right lower panel of Fig. 2(f)). Judging from these structural and elemental investigations, AlOx/Al and LaAlO3/BaOx work well as soluble sacrificial bilayers for patterning SrRuO3.

FIG. 2.

(a) The crystal structure of SrRuO3 depicted by VESTA.19 The cross-section profiles (blue lines) of the SrRuO3 channels patterned by the (b) AlOx/Al and (c) LaAlO3/BaOx sacrificial bilayers. The left and right insets show narrow and wide AFM images of the SrRuO3 channel, respectively. (d) The XRD pattern of the SrRuO3 channels patterned by a LaAlO3/BaOx bilayer. (e) Left: the schematic of the AlOx/Al bilayer on a substrate. Right top and bottom panels show element mappings of Al (blue) and Ru (red) in the sample. (f) Left: the schematic of SrRuO3 channel after lift-off process. Right top and bottom panels show element mappings of Al (blue) and Ru (red) in the sample. The elemental mappings were measured by scanning electron microscope energy dispersive X-ray spectroscopy (SEM-EDX).

FIG. 2.

(a) The crystal structure of SrRuO3 depicted by VESTA.19 The cross-section profiles (blue lines) of the SrRuO3 channels patterned by the (b) AlOx/Al and (c) LaAlO3/BaOx sacrificial bilayers. The left and right insets show narrow and wide AFM images of the SrRuO3 channel, respectively. (d) The XRD pattern of the SrRuO3 channels patterned by a LaAlO3/BaOx bilayer. (e) Left: the schematic of the AlOx/Al bilayer on a substrate. Right top and bottom panels show element mappings of Al (blue) and Ru (red) in the sample. (f) Left: the schematic of SrRuO3 channel after lift-off process. Right top and bottom panels show element mappings of Al (blue) and Ru (red) in the sample. The elemental mappings were measured by scanning electron microscope energy dispersive X-ray spectroscopy (SEM-EDX).

Close modal

Figure 3(a) shows the temperature dependences of resistance (R-T curves) in SrRuO3 channels patterned by different processes. The designed widths of the channels w are varied in the range of w = 5-100 μm. For SrRuO3 channels patterned by AlOx/Al (red lines in the left panel of Fig. 3(a)) and LaAlO3/BaOx (blue lines in the central panel of Fig. 3(a)) sacrificial bilayers, all the R-T curves show metallic behaviors with a clear kink corresponding to the ferromagnetic transition. Here, ferromagnetic transition temperature (TFM) is defined as the peak temperature in the derivative of R-T curves, which is shown as black arrows in Fig. 3(a). The TFM of SrRuO3 channels fabricated with LaAlO3/BaOx (blue lines in center of Fig. 3(a)) represents higher temperature at about 130 K than that with AlOx/Al at 110 K. Both TFM are in good agreement with that of pristine SrRuO3 films grown at 700 oC or 850 oC (gray lines), respectively, implying a process damage, for example formation of oxygen vacancy or non-stoichiometry of cations, is minimized. Here, although difference between the obtained TFM and the bulk value of TFM (about 160 K)21 could partially originate from epitaxial strain effect in films from the substrate,22TFM at about 130 K is comparable to that reported in previous thin film studies.23,24 In addition, thanks to the higher growth temperature of 850 oC by employing a LaAlO3/BaOx bilayer, higher TFM is obtained. The proper selection of lift-off layer is a key to apply optimum growth temperature i.e. high temperature with high oxygen partial pressure. The room-temperature channel conductance (G300K), plotted in the insets of Fig. 3(a), scales well with channel width for the both samples. This linear scaling in narrow channels clearly evidences that the device width is well controlled as designed in pre-patterned sacrificial bilayer. In contrast, the shapes of R-T curves for the SrRuO3 channels patterned by standard Ar ion milling with photolithography (green lines in the right panel of Fig. 3(a)) show non-systematic temperature and width dependence in particular at high temperature region above 200 K. Moreover, G300K does not follow the width variation as shown in the inset. Such non-systematic behaviors indicate the existence of parasitic conduction at the ion-milled surface region of the SrTiO3 substrate.25 

FIG. 3.

(a) The R-T curves of the SrRuO3 channels patterned by AlOx/Al (red in left), LaAlO3/BaOx (blue in center) sacrificial bilayers, and conventional Ar ion milling (green in right). The length of the SrRuO3 channels l is fixed at l = 500 μm. The nominal widths of the SrRuO3 channels are varied from w = 5 to 100 μm, which is noted in the unit of μm. The thickness of SrRuO3 layers are 39, 20, and 18 nm for the channels patterned by AlOx/Al, LaAlO3/BaOx, and Ar ion milling, respectively. The higher resistance in the blue lines than in the red lines originates from the difference in the SrRuO3 thickness. In the left and middle plots, temperature dependences of the sheet resistance for the pristine SrRuO3 films are plotted in gray, after multiplying by the factor of 2.5. The thickness of the pristine SrRuO3 films are 35 (left) and 21 nm (center). The black arrows indicate the ferromagnetic transition temperature. The insets are channel width w dependences of the channel conductance at 300 K. (b) Comparison of cross-talk resistances between neighboring SrRuO3 channels patterned by the AlOx/Al (red) and LaAlO3/BaOx (blue) sacrificial bilayers, and the conventional Ar ion milling (green).

FIG. 3.

(a) The R-T curves of the SrRuO3 channels patterned by AlOx/Al (red in left), LaAlO3/BaOx (blue in center) sacrificial bilayers, and conventional Ar ion milling (green in right). The length of the SrRuO3 channels l is fixed at l = 500 μm. The nominal widths of the SrRuO3 channels are varied from w = 5 to 100 μm, which is noted in the unit of μm. The thickness of SrRuO3 layers are 39, 20, and 18 nm for the channels patterned by AlOx/Al, LaAlO3/BaOx, and Ar ion milling, respectively. The higher resistance in the blue lines than in the red lines originates from the difference in the SrRuO3 thickness. In the left and middle plots, temperature dependences of the sheet resistance for the pristine SrRuO3 films are plotted in gray, after multiplying by the factor of 2.5. The thickness of the pristine SrRuO3 films are 35 (left) and 21 nm (center). The black arrows indicate the ferromagnetic transition temperature. The insets are channel width w dependences of the channel conductance at 300 K. (b) Comparison of cross-talk resistances between neighboring SrRuO3 channels patterned by the AlOx/Al (red) and LaAlO3/BaOx (blue) sacrificial bilayers, and the conventional Ar ion milling (green).

Close modal

To probe the parasitic conduction directly, cross-talk resistance between neighboring channels were plotted in Fig. 3(b). In the sample patterned by Ar ion milling (green), the cross-talk resistance is in the order of kΩ, which is as low as the channel resistance of SrRuO3. Such low resistance in the ion-milled SrTiO3 surface could originate from oxygen vacancies formed by Ar ion bombardment.25 In contrast, the resistances in the samples patterned by AlOx/Al (red) and LaAlO3/BaOx (blue) bilayers were as high as 100 GΩ, which is the upper limit of our measurement set up, indicating negligible parasitic conduction in the substrates. This surface damage problem is limiting the available device structures of thin films grown on SrTiO3 substrates so far, because additional post-annealing step is necessary to remove it.26 From this comparison, the proposed lift-off based process demonstrates its versatility for such perovskite oxide thin films on SrTiO3 substrates.

As another example from a different family of materials, we demonstrate microfabrication of FeSe, which is a two-dimensional layered material where each layer is weakly bonded by van der Waals interaction (Fig. 4(a)). The bulk FeSe shows superconducting transition with the critical temperature Tc = 8 - 9 K.27,28 In this study, the FeSe thin films were grown by PLD at the growth temperature of 300 °C followed by in-situ annealing at 450 °C.29 The FeSe thin films (t = 14 nm) suffer from time degradation in air as shown in Fig. 4(b), changing their temperature dependent sheet resistance (Rs-T curves) from metallic to insulating in half a day. To keep metallic conduction and superconductivity in FeSe devices, microfabrication should be completed in a short time. Besides, such degradation is inevitably accelerated by heating in air. To evaluate the degradation by heating, the Rs-T curves for a FeSe film (t = 35 nm) were measured after baking at typical temperature 110, 130, and 150 oC as plotted in Fig. 4(c). The as-grown pristine FeSe thin film shows metallic properties with the superconducting transition at Tcmid ∼ 7 K, where Tcmid is the temperature where resistance drops to the half of the normal state resistance. After baking at 110 °C for 2 min, the Rs-T curve shows small upturn at around 27 K with broad superconducting transition at Tcmid ∼ 2 K. Increasing the baking temperature to 130 °C and 150 °C, FeSe films become insulating without showing any signature of superconductivity. Although the origin for the degradation is unclear at present, it can be at least concluded that the conventional lithography process with organic resist baking above 100 °C is not applicable. To solve this problem, the proposed lift-off process using the sacrificial bilayers is quite beneficial because no heating process is necessary after the FeSe growth as shown in Fig. 1(a). As a practical example, the Rs-T curve (red curve in Fig. 4(c)) of the 100 μm-channel fabricated with LaAlO3/BaOx bilayer (photograph of the device is shown in the inset of Fig. 4(c)) is comparable to that of the as-grown film, showing the metallic properties with the clear superconducting transition Tcmid ∼ 9 K. This result indicates that the lift-off process enables us to pattern FeSe films into μm-scale devices with minimum deterioration, owing to absence of any heating step and rapid installation into measurement apparatus. We believe that the proposed microfabrication process is of great use when fabricating mesoscopic devices based on chalcogenide thin films such as Fe-based superconductor and topological insulators.

FIG. 4.

(a) The crystal structure of FeSe. (b) The time degradation of Rs-T curves of the FeSe (14 nm) thin film grown on a SrTiO3 substrate. Each Rs-T curve is recorded after keeping the identical FeSe sample in atmosphere for the interval (hours) shown as the notations. (c) Comparison of the Rs-T curves of FeSe (35 nm) thin films: just after growth (as-grown, dark blue), w = 100 μm, l = 250 μm channel patterned by the LaAlO3/BaOx process (patterned, red), and the non-processed film after annealing on a hot plate at Tanneal = 110 °C, 130 °C, and 150 °C for 2 min. The inset is the optical micrograph of the patterned FeSe channel.

FIG. 4.

(a) The crystal structure of FeSe. (b) The time degradation of Rs-T curves of the FeSe (14 nm) thin film grown on a SrTiO3 substrate. Each Rs-T curve is recorded after keeping the identical FeSe sample in atmosphere for the interval (hours) shown as the notations. (c) Comparison of the Rs-T curves of FeSe (35 nm) thin films: just after growth (as-grown, dark blue), w = 100 μm, l = 250 μm channel patterned by the LaAlO3/BaOx process (patterned, red), and the non-processed film after annealing on a hot plate at Tanneal = 110 °C, 130 °C, and 150 °C for 2 min. The inset is the optical micrograph of the patterned FeSe channel.

Close modal

Finally, we address the limitation of the developed versatile patterning process in terms of target material thickness and achievable lateral resolution. As for the film thickness, we have succeeded in patterning ∼ 35 nm of FeSe using LaAlO3 (20nm)/BaOx (190 nm) sacrificial bilayers in this study. We believe that the process is applicable to thicker film fabrication when the thickness of target films is thinner than the thickness of the bottom lift-off layer. In addition, the lateral resolution is limited by sub-micron scale winding of the patterned bilayer. We demonstrated well-defined 5 μm Hall-bar devices by using conventional photolithography as shown in Fig. 2(b). Next step for this versatile device fabrication process is a challenge to fabrication of sub-micron scale devices with electron-beam lithography. Further optimization is necessary to suppress the winding of the edge of bilayer but the process will potentially apply to sub-micron scale fabrication.

In summary, we have demonstrated that inorganic sacrificial bilayers are beneficial to perform device patterning in minimal time, without Ar ion etching and heating after target material growth. By using a template of a soluble sacrificial bilayer, target materials can be patterned into device structures after growth in the optimal condition, e.g. at high temperatures. Simple and compatible with the conventional lithography technique, the versatile device processing method will be widely applicable to emerging quantum materials such as oxides and chalcogenides including correlated electron systems, Fe-based superconductors, and topological insulators toward development of new functional devices.

The authors thank J. Mannhart and K. Fujiwara for valuable discussions, and T. Seki and K. Takanashi for help in the cleanroom facilities. This work is a cooperative program (Proposal No. 16G0404) of the CRDAM-IMR, Tohoku University. This work is partly supported by a Grant-in-Aid for Research Activity Start-up (15H06029) and a Grant-in-Aid for Specially Promoted Research (No. 25000003) from the Japan Society for the Promotion of Science (JSPS).

1.
B. J.
van Wees
,
H.
van Houten
,
C. W. J.
Beenakker
,
J. G.
Williamson
,
L. P.
Kouwenhoven
,
D.
van der Marel
, and
C. T.
Foxon
,
Phys. Rev. Lett.
60
,
848
(
1988
).
2.
D.
Hisamoto
,
T.
Kaga
,
Y.
Kawamoto
, and
E.
Takeda
,
IEEE Electron Device Lett.
11
,
36
(
1990
).
3.
J. R.
Kirtley
,
C. C.
Tsuei
,
A.
Ariando
,
C. J. M.
Verwijs
,
S.
Harkema
, and
H.
Hilgenkamp
,
Nat. Phys.
2
,
190
(
2006
).
4.
L. P.
Rokhinson
,
X.
Liu
, and
J. K.
Furdyna
,
Nat. Phys.
8
,
795
(
2012
).
5.
S. O.
Valenzuela
and
M.
Tinkham
,
Nature
442
,
176
(
2006
).
6.
C.
Stampfer
,
E.
Schurtenberger
,
F.
Molitor
,
J.
Güttinger
,
T.
Ihn
, and
K.
Ensslin
,
Nano. Lett.
8
,
2378
(
2008
).
7.
K. S.
Novoselov
,
D.
Jiang
,
F.
Schedin
,
T. J.
Booth
,
V. V.
Khotkevich
,
S. V.
Morozov
, and
A. K.
Geim
,
Proc. Natl. Acad. Sci. USA
102
,
10451
(
2005
).
8.
D.
Lu
,
D. J.
Baek
,
S. S.
Hong
,
L. F.
Kourkoutis
,
Y.
Hikita
, and
H. Y.
Hwang
,
Nat. Mater.
15
,
1255
(
2016
).
9.
N.
Banerjee
,
E. P.
Houwman
,
G.
Koster
, and
G.
Rijnders
,
Appl. Phys. Lett.
102
,
142909
(
2013
).
10.
S. Y.
Chou
,
P. R.
Krauss
, and
P. J.
Renstrom
,
Science
272
,
85
(
1996
).
11.
N.
Suzuki
,
H.
Tanaka
,
S.
Yamanaka
,
M.
Kanai
,
B. K.
Lee
,
H. Y.
Lee
, and
T.
Kawai
,
Small
4
,
1661
(
2008
).
12.
M.
Nijland
,
A.
George
,
S.
Thomas
,
E. P.
Houwman
,
J.
Xia
,
D. H. A.
Blank
,
G.
Rijnders
,
G.
Koster
, and
J. E.
ten Elshof
,
Adv. Funct. Mater.
24
,
6853
(
2014
).
13.
K.
Goto
,
H.
Tanaka
, and
T.
Kawai
,
Nano. Lett.
9
,
1962
(
2009
).
14.
R.
Nechache
,
C. V.
Cojocaru
,
C.
Harnagea
,
C.
Nauenheim
,
M.
Nicklaus
,
A.
Ruediger
,
F.
Rosei
, and
A.
Pignolet
,
Adv. Mater.
23
,
1724
(
2011
).
15.
I.
Vrejoiu
,
A.
Morelli
,
D.
Biggemann
, and
E.
Pippel
,
Nano Rev.
2
,
7364
(
2011
).
16.
W. M.
Haynes
,
D. R.
Lide
, and
T. J.
Bruno
,
CRC Handbook of Chemistry and Physics 97th Edition
(
CRC Press
,
2016
).
17.
P.
Walker
and
W. H.
Tarn
,
CRC Handbook of Metal Etchants
(
CRC Press
,
1990
).
18.
T.
Ohnishi
,
M.
Yoshimoto
,
G. H.
Lee
,
T.
Maeda
, and
H.
Koinuma
,
J. Vac. Sci. Technol. A
15
,
2469
(
1997
).
19.
K.
Momma
and
F.
Izumi
,
J. Appl. Crystallogr.
44
,
1272
(
2011
).
20.
M.
Kawasaki
,
K.
Takahashi
,
T.
Maeda
,
R.
Tsuchiya
,
M.
Shinohara
,
O.
Ishiyama
,
T.
Yonezawa
,
M.
Yoshimoto
, and
H.
Koinuma
,
Science
266
,
1540
(
1994
).
21.
G.
Cao
,
S.
McCall
,
M.
Shepard
,
J. E.
Crow
, and
R. P.
Guertin
,
Phys. Rev. B
56
,
321
(
1997
).
22.
Q.
Gan
,
R. A.
Rao
,
C. B.
Eom
,
J. L.
Garrett
, and
M.
Lee
,
Appl. Phys. Lett.
72
,
978
(
1998
).
23.
C. B.
Eom
,
R. J.
Cava
,
R. M.
Fleming
,
J. M.
Phillips
,
R. B.
van Dover
,
J. H.
Marshall
,
J. W. P.
Hsu
,
J. J.
Krajewski
, and
W. F.
Peck
, Jr.
,
Science
258
,
1766
(
1992
).
24.
J.
Xia
,
W.
Siemons
,
G.
Koster
,
M. R.
Beasley
, and
A.
Kapitulnik
,
Phys. Rev. B
79
,
140407
(
2009
).
25.
D. W.
Reagor
and
V. Y.
Butko
,
Nat. Mater.
4
,
593
(
2005
).
26.
D. A.
Tenne
,
I. E.
Gonenli
,
A.
Soukiassian
,
D. G.
Schlom
,
S. M.
Nakhmanson
,
K. M.
Rabe
, and
X. X.
Xi
,
Phys. Rev. B
76
,
024303
(
2007
).
27.
F.-C.
Hsu
,
J.-Y.
Luo
,
K.-W.
Yeh
,
T.-K.
Chen
,
T.-W.
Huang
,
P. M.
Wu
,
Y.-C.
Lee
,
Y.-L.
Huang
,
Y.-Y.
Chu
,
D.-C.
Yan
, and
M.-K.
Wu
,
Proc. Natl. Acad. Sci. USA
105
,
14262
(
2008
).
28.
Y.
Mizuguchi
and
Y.
Takano
,
J. Phys. Soc. Jpn.
79
,
102001
(
2010
).
29.
J.
Shiogai
,
Y.
Ito
,
T.
Mitsuhashi
,
T.
Nojima
, and
A.
Tsukazaki
,
Nat. Phys.
12
,
42
(
2016
).