The release process for the fabrication of freestanding oxide microstructures relies on appropriate, controllable, and repeatable wet etching procedures. SrTiO3 (STO) is among the most employed substrates for oxide thin films growth and can be decomposed in HF:water solution. Such a process is strongly anisotropic and is affected by local defects and substrate cut-planes. We analyze the etching behavior of SrTiO3 substrates having (100), (110), and (111) cut-planes during immersion in a 5% HF:water solution. The etching process over the three substrates is compared in terms of pitting, anisotropy, macroscopic etch rate, and underetching effects around HF-resistant (La,Sr)MnO3 thin film micropatterns. The release of targeted structures, such as the reported (La,Sr)MnO3 freestanding microbridges, depends on the substrate crystallographic symmetry and on the in-plane orientation of the structures themselves along the planar directions. By comparing the etching evolution at two different length scales, we distinguish two regimes for the propagation of the etching front: an intrinsic one, owing to a specific lattice direction, and a macroscopic one, resulting from the mixing of different etching fronts. We report the morphologies of the etched SrTiO3 surfaces and the geometries of the underetched regions as well as of the microbridge clamping zones. The reported analysis will enable the design of complex MEMS devices by allowing to model the evolution of the etching process required for the release of arbitrary structures made of oxide thin films deposited on top of STO.

SrTiO3 (STO) is a standard substrate for the deposition of many transition metal oxides (TMO) thin films1 and for the realization of oxide-based devices due to its convenient lattice parameter, variety of surface preparation procedures, wide bandgap, and large dielectric constant2 and, recently, it has been employed as the preferential substrate for developing new oxide MEMS devices3–9 without requiring integration procedures with Si technology.10–15 Commercially available common SrTiO3 substrate crystal cuts are (100), (110), and (111). STO(100) is the most studied and used one; it is weakly polar and has the highest surface symmetry. In contrast, STO(110) and STO(111) crystal cuts are polar and have a reduced in-plane symmetry that makes them prone to surface reconstructions and adsorbates that influence device fabrication steps.16–18 The proposed etching model of single-crystal SrTiO3 is the so-called “dislocation hierarchical tree structure,”19 which takes into account both rate dispersion and induced surface morphology, and many critical details of this behavior have been tested so far.20–22 A complementary concept to this crystallographic approach is the “reactive surface area,” associated with methods suitable to study the variability of etch rates and mechanisms on a bigger scale.23–25 Notwithstanding the significative overall results briefly reviewed, SrTiO3 specific results regarding crystal cut, anisotropy, and etch rate variability in the case of MEMS fabrication are still lacking. In such a case, many parameters are fixed due to device specification constraints. Hence, the specific effects of crystal cut and device orientation are very relevant. In order to develop a MEMS technology based on crystalline oxide thin films deposited on top of SrTiO3 substrates, it is, thus, crucial to understand the evolution of the substrate etching process along different lattice directions, in particular, with respect to the planned device geometry and the feature size.

In this work, we analyze faceting, etch rates, pitting, and underetching effects of chemically etched (5% HF aqueous solution) STO substrates having different cut-planes, namely, (100), (110), and (111). We employ 5 × 5 × 0.5 mm3 single crystal substrates from CrysTec GmbH with a miscut angle below 0.1°. We compare the etching evolution on two different device geometries designed to measure both the STO etching rates at the macroscale and the etching dynamics at the microscale. For the former, we use a hard mask made from a 50 nm thick (La0.7, Sr0.3)MnO3 (LSMO) film with a 200 × 200 μm2 square hole array. For the latter, we investigate the under-etching and the release process of 100 nm thick and 5 μm wide LSMO microbridges.

LSMO films are grown by pulsed laser deposition (PLD) from a sintered powder target using an excimer laser (248 nm wavelength) having a beam energy density of 0.6 J/cm2 and 2 Hz repetition rate. The deposition temperature is 800 °C, and the oxygen background pressure is 10−4 mbar. After the growth, the films are in situ annealed for 20 min at 600 °C and 200 mbar of oxygen pressure. Both the hard mask and microbridge samples are fabricated by standard optical lithography as described in previous works from our group.3,7 The hard mask etched samples are obtained by put soaking together (100), (110), and (111) STO samples in HF (5% in aqueous solution) kept at 30 °C. This allows us to perform some statistical analysis and test for homogeneity over the sample surface. At given times (1, 2, 5, 10, 20, 40, and 60 min), they are all removed from the bath, cleaned in de-ionized water, dried under nitrogen flow, and inspected at the optical microscope. LSMO microbridges oriented along three in-plane orientations (0°, 45°, and 90°) were realized on three substrates having different cut-planes and fabricated following the same procedure discussed above with the only difference that after 60 min of etching, they were dried using a critical point drying system. HF decomposes the STO, leaving water-insoluble salts on the surfaces that are identified as strontium fluoride (see the supplementary material, Sec. I). The removal of such salts is achieved mechanically during the etching process by magnetic stirring at the relatively slow rotational speed of 200 rpm, so as to avoid the break of the microstructures by drag forces.

We start by discussing the etching process at the macroscale by monitoring two characteristics of the STO substrate regions that are not covered by the LSMO hard mask: (1) the formation, alignment, and evolution of pyramidal etch pits and (2) the overall substrate etching in the out-of-plane direction.

The pits form and progressively grow on the STO(100) and STO(111) substrates, resulting in surfaces covered with well-formed pyramidal pits of quite uniform lateral dimensions after 10  and 40 min for STO(100) and STO(111), respectively [see Fig. 1(a)]. STO(100) pits have square symmetry with a centered apex as shown in Fig. 1(a) (α, β, γ) and (b)(η), while STO(111) pits have triangular section with the apex mostly shifted from the symmetry center, as visible from the shadowed bounding plane in Fig. 1(b) (θ, ι, κ, λ) and in some case with a remarkable symmetry, as in Fig. 1(b) (μ). For both STO(100) and STO(111) substrates, the pits bounding planes are parallel to (hh0) substrate planes, and all the pits are identically oriented with respect to the substrate. As the etching proceeds, already existing pits grow in size, but nucleation of new pits is not observed over the pristine STO surface. During pit growth, other localized events can be found that are able to change the pit bounding planes' direction, as in Fig. 1(b) (λ), or even the pit morphology (i.e., a pyramidal pit within a flat pit), as in Fig. 1(b) (ν, η, π, o). By following the evolution of individual pits, as shown in Fig. 1(a) (α, β, γ) and (δ, ε, ζ), it is possible to estimate their in-plane growth rate [RPIT of Fig. 1(c)], reported in Table I, as long as they do not coalescence. The indicated growth rates are in agreement with the literature.19,26 Finally, once the surface is cluttered with overlapping pits, the etching proceeds distributed all over the exposed substrate surface. In contrast, the STO(110) substrate remains pretty flat even at the end of the etching process with a rms roughness of about 4 nm, as measured by AFM, and no detectable pit formation (see the supplementary material, Sec. II). Such striking difference shows, in accordance with previous literature reports, that the [110] direction is a fast etching direction, since no pit eventually nucleated on this face would be able to survive on its own.26 

FIG. 1.

Optical microscopy images of etched square holes of 200 μm lateral size. (a) Pictures taken at different etching times. Insets are magnified pit details. The visible dark-gray marks on the STO(110) surface are not pits but residues of strontium fluoride coming from the STO chemical etching process (see the supplementary material, Sec. I). (b) Details of the etched surface, where the STO(100) pictures were acquired after 5 min of etching while the STO(111) ones after 60 min. In-plane angles are 90° (pink), 110° (green), 140° (yellow), and 120° (red). (c) Simplified illustration of the etching evolution in the out-of-plane direction for a STO(100) substrate.

FIG. 1.

Optical microscopy images of etched square holes of 200 μm lateral size. (a) Pictures taken at different etching times. Insets are magnified pit details. The visible dark-gray marks on the STO(110) surface are not pits but residues of strontium fluoride coming from the STO chemical etching process (see the supplementary material, Sec. I). (b) Details of the etched surface, where the STO(100) pictures were acquired after 5 min of etching while the STO(111) ones after 60 min. In-plane angles are 90° (pink), 110° (green), 140° (yellow), and 120° (red). (c) Simplified illustration of the etching evolution in the out-of-plane direction for a STO(100) substrate.

Close modal
TABLE I.

Comparison etching rate for a SrTiO3 single crystal in HF bath along the [001], [011], and [111] directions. In our work, the etch front rate is calculated by considering the final depth after 60 min. The RMS evaluates the surface roughness and is calculated over the entire 200 × 200 μm2 square mask (see also the supplementary material, Secs. II and III). For an in-depth comparison, a particular care with respect to how the rate is defined in each of the referenced works is required.

DirectionMethodRate eq.RMSEtchantTemperatureReference
[001] RMACRO 13.5 μm/h 1 μ5% (2.8M) HF 30 °C This work 
[011] RMACRO 9 μm/h 4 nm 5% (2.8M) HF 30 °C This work 
[111] RMACRO 0.6 μm/h 0.75 μ5% (2.8M) HF 30 °C This work 
[001] Surface 1.2 μm/h N/A 2.5M HF 30 °C Ref. 26  
[001] Pits depth 11.1 μm/h N/A 2.5M HF 30 °C Ref. 26  
[001] Pits depth 21.6 μm/h N/A N/A N/A Calculated from Ref. 19  
[001] Terraces 0.01 μm/h N/A Buffered HF “Room temperature” Calculated from Ref. 27  
[011] Surface 6.0 μm/h N/A 2.5M HF 30 °C Ref. 26  
[001] RPIT 5 μm/min N/A 5% (2.8M) HF 30 °C This work 
[111] RPIT 1.3 μm/min N/A 5% (2.8M) HF 30 °C This work 
DirectionMethodRate eq.RMSEtchantTemperatureReference
[001] RMACRO 13.5 μm/h 1 μ5% (2.8M) HF 30 °C This work 
[011] RMACRO 9 μm/h 4 nm 5% (2.8M) HF 30 °C This work 
[111] RMACRO 0.6 μm/h 0.75 μ5% (2.8M) HF 30 °C This work 
[001] Surface 1.2 μm/h N/A 2.5M HF 30 °C Ref. 26  
[001] Pits depth 11.1 μm/h N/A 2.5M HF 30 °C Ref. 26  
[001] Pits depth 21.6 μm/h N/A N/A N/A Calculated from Ref. 19  
[001] Terraces 0.01 μm/h N/A Buffered HF “Room temperature” Calculated from Ref. 27  
[011] Surface 6.0 μm/h N/A 2.5M HF 30 °C Ref. 26  
[001] RPIT 5 μm/min N/A 5% (2.8M) HF 30 °C This work 
[111] RPIT 1.3 μm/min N/A 5% (2.8M) HF 30 °C This work 

As schematically illustrated in Fig. 1(c) [showing the STO(100) case], the overall out-of-plane etching rate on a large scale (several tens of micrometers) is a combination of different processes, also including the evolution of the pits. In fact, the etch front along a direction (intrinsic rate RINT) given by a crystallographic face may proceed by the removal of crystalline layers in the same crystallographic direction or by removing layers in different and faster etching directions, enhancing the overall etching rate through the formation of pits (in plane pit growth rate RPIT). Moreover, during the pitting process, a flat surface along a slow etching direction may form (i.e., truncated pyramidal pits) decreasing the etching speed unless new pits are formed. Considering these mechanisms, the nucleation and the growth of different etching planes may depend on local conditions (i.e., etchant concentration gradients, local fluid velocity, salt precipitation, and contamination, etc.). However, taking into account that the observed pit characteristic length is well below the size of the exposed regions in our hard mask, we can evaluate the average large-area etching (RMACRO) rate for STO(100), STO(110), and STO(111) substrates by inspecting our samples after 60 min of etching. This was performed by an optical profilometer (see the supplementary material, Sec. III), and the results of this analysis are reported in Table I, together with a comparison with etching rates reported in previous works.

A second important aspect, directly related to the release process of suspended microstructure, is the evolution of the underetch profile, i.e., the STO region etched below the mask at the edges of patterned structures. We now analyze the evolution of the underetching below the LSMO microbridge and the geometry of the underetch profile around the LSMO microbridge clamping regions for different substrate types and microbridge directions.

In Figs. 2 and 3, suspended regions appear as light-gray colored and progressively evolve around all the borders of the LSMO patterns. The optical inspection (Fig. 2) monitors the 2D projection of the underetch profiles only, while their tridimensional structure is better evidenced in the SEM pictures reported in Fig. 3. In the following, in-plane directions are defined as “x,” “y,” and “xy,” corresponding to the substrate edges and diagonal, respectively, while “z” is the out-of-plane direction, as indicated in Fig. 2(a). In the STO(100), the extension of the underetched region is similar to both the equivalent directions x and y [Fig. 2(b) (α, β)], but smaller for bridges oriented along the xy direction [Fig. 2(b) (γ)]. STO(110), instead, presents a less symmetric pattern with a straight underetch profile in the y direction, characterized by vertical walls [Figs. 2(b) (δ) and 3(c)]. This is in contrast with the rough underetch profile in x and xy directions [Figs. 2(b) (ε) and 3(d)]. For the STO(111) case, the underetch profile is characterized by a larger extension below the LSMO pattern and a marked asymmetry between the +x and −x directions, as evidenced in Fig. 2(b) (ζ). The etching in the z direction is limited and characterized by a faceted surface, as already discussed and visible in Fig. 3(f). The faceted STO underetch profile has different orientations with respect to the bounding planes observed on pits forming in open regions of the substrates (Fig. 3); this fact is a clear indication that geometric boundary conditions significantly contribute in determining the final clamping profile of the etched structures.

FIG. 2.

(a) Scheme of the experiment, indicating the three STO samples and crystallographic directions: STO substrate surface (blue), sample directions (red), crystal unit cell (red), and crystal basis directions (orange). (b) Optical microscopy at 40× magnification after 40 min etching time of LSMO microbridges of different orientations (rows) on STO substrates with different crystal cuts (columns). (c) Magnified detail of the clamping zone of microbridges (profile sketch in red). Light gray regions are freestanding.

FIG. 2.

(a) Scheme of the experiment, indicating the three STO samples and crystallographic directions: STO substrate surface (blue), sample directions (red), crystal unit cell (red), and crystal basis directions (orange). (b) Optical microscopy at 40× magnification after 40 min etching time of LSMO microbridges of different orientations (rows) on STO substrates with different crystal cuts (columns). (c) Magnified detail of the clamping zone of microbridges (profile sketch in red). Light gray regions are freestanding.

Close modal
FIG. 3.

SEM images of LSMO microbridges between square pads. In-plane directions are reported in red color.

FIG. 3.

SEM images of LSMO microbridges between square pads. In-plane directions are reported in red color.

Close modal

The time required to obtain a complete release from the substrate depends on the device geometry and size. As a general reference, we can consider the reported structures, i.e., 5 μm-wide LSMO microbridges, etching time to be about 30 min for all the cases except the two combinations: STO(100) xy-aligned and STO(110) y-aligned. In such cases, etching times exceeding 1 h could be required depending on local defects. Considering these etching times, in general, the underetch profile is not the equilibrium one, as predicted by the Wulff–Jaccodine construction.28 Instead, the profile is determined by the system anisotropy and varies, among other factors, with the in-plane orientation of the pattern, crystal cut-plane, and etching time.

Regarding the etching at the clamping zones, the process proceeds even after the microbridge is suspended. The precise geometry of the clamping regions and the symmetry between the two edges of the bridges are discussed in Fig. 2(c). They depend on the microbridge orientation and the substrate crystal cut as follows:

  • STO(100): the clamping region is symmetric for all the three microbridge orientations x, y, and xy. In the case of the x and y-oriented microbridges, the clamping zone has a polygonal shape, while in those oriented in the xy direction, it has a net flat shape that progressively evolves toward a complete release of the microbridge.

  • STO(110): the geometry of the clamping region, although symmetric in both cases, differs significantly between the x and y directions. For x-aligned microbridges, it is flat and aligned with the borders of the pads, while for y-aligned microbridges, it protrudes from the pads, outlining a polygon. This is not the case for STO(110) xy-oriented microbridges, where asymmetry arises.

  • STO(111): a varied mix of shapes and symmetries at the clamping regions can be observed. Microbridges aligned in the x direction have symmetric clamping regions but one with a flat profile and the other with a protruding profile. Microbridges aligned along the y and xy directions have asymmetric clamping regions with the former showing larger under-etch.

In the following, we show that the complex faceting observed in the clamping regions results from the interplay among substrate cut-planes, device geometry, and anisotropic etching. In addition, at small scales, the etching rate measured along different cut planes (nearby pattern features) is different from what extracted from the macroscopic rates RMACRO measured reported in Table I, where the averaging effects over different crystallographic directions play a major role. This also indicates that the relevant length scale determining the geometry of the edges in micrometric devices is below the threshold required to observe the large-area rates.

The symmetric underetching behavior for STO(100) and the similarity between the x and y directions reflect the crystallographic equivalence between [001] and [010] directions. The lower underetch along xy corresponding to the [011] direction can be explained considering that this is a fast etching direction, and the xy oriented microbridges progressively exhibit walls parallel to the [1hh] directions that have a slower etching rate than the pure [011] and [001] ones. The trapezoidal-shaped (in-plane) clamping zones are the result of the [111] face etching that is slower than the [110] face one, hence prevailing the concave corner.28 The 90° angle between the underetch profiles [Fig. 2(c) (η)] observed in the clamped regions for the microbridges aligned along the xy direction can be explained considering the formation of underetch faces oriented along equivalent [111] directions. The profile symmetry is guaranteed by the crystallographic equivalence between the involved directions on both sides of the same clamping zone.

The STO(110) underetch in-plane asymmetry arises from the fact that the x and y directions are aligned along the slow [100] and the fast [110] etching directions, respectively. The flat clamping zone of x-aligned microbridges is the result of reaching a steady state condition with the exposure of the (100) faces, which have a slower etching rate. The faceting of the profile observed around the clamping regions along the y direction results from the faceting vicinal to the [110] direction, which determines deep trenches into the substrate along the y-directed border regions. The clamping regions of y-aligned microbridges have a completely different shape, because in this case the underetch vertical walls are the steady state slow-etching faces ([100] equivalent directions). In this case, other residual crystallographic planes are exposed such as the [111]. The symmetry of the clamping regions in both x and y aligned microbridges is due to the fact that, contrary to xy-aligned microbridges, sagittal planes containing microbridges axis are symmetry planes of the crystal lattice.

STO(111) has in-plane crystallographic trigonal symmetry that does not match the symmetry of the microbridge patterns, leading to a variety of clamping zone shapes. Due to the lower etch rate in the z direction and the relevant in-plane underetch, it is difficult to measure the extension of the underetch profile. However, by applying the same analysis used for both STO(100) and STO(110), we may argue that microbridges aligned along the x direction have one flat clamping zone probably limited by the slow etching [100] face, while the other clamping region, with a protruding profile, is limited by the equivalent [010] and [001] directions. The y-aligned microbridges' clamping regions have asymmetric profiles and are probably limited by the slow etching rate along the [010] and [001] directions in one side and by the fast etching rate along the [110] and [101] on the other side of the microbridge. In the case of xy-aligned microbridges, analogous analysis leads to infer that the clamping regions are limited by the slow etching rate along the [010] and [001] directions, each one at different angle with the microbridge sagittal plane, resulting in an asymmetrical profile.

Finally, in order to demonstrate the design opportunity enabled by our study, we show in the supplementary material Sec. IV the realization of a LSMO micromechanical bridge resonator that can be measured optically by focusing a laser through a double-polished SrTiO3(110) substrate. This was made possible thanks to the low RMS of the bottom surface, which is preserved after the chemical etching and the transparency of SrTiO3 to visible light. Such configurations will allow to couple TMO-based resonators in proximity to other systems while maintaining an optical readout scheme.

In conclusion, we showed that the choice of the STO crystal cut-plane has significant effects in terms of etching rate, pitting of large areas, microbridge release time, underetching, and shape of the clamping regions. With the shrinking of the device size, the presence of pits and their size become increasingly important and, for increasing complexity of MEMS geometries, achieving desired morphology of the released structures will require trade-off solutions based on choosing the proper crystal cut. Lattice defects also play a crucial role in the propagation of the etching front, and their statistical spatial distribution determines a critical scale separating large and small-area behaviors of the etching process. Moreover, the in-plane orientation of the microbridges has considerable implications on the final geometry of the clamping regions, likely affecting its mechanical behavior. Below, we provide few general considerations summarizing the behavior of STO substrates having different cut-planes with respect to the investigated chemical etching process. As a general reference, the lattice symmetry of the substrate is the main characteristic determining how freestanding regions are released, and this should be taken into account for oxide MEMS design.

  • STO(100) shows 90° in-plane symmetry and the fastest macroscopic etching direction along the out-of-plane direction and good in-plane etching rates.

  • STO(110) shows smooth etched substrate surfaces and good out-of-plane etching rates. It also shows the possibility to obtain sharp underetch walls, taking into account for the device geometry or symmetry.

  • STO(111) has the fastest in-plane underetching rate and the lower out-of-plane etching rate

See the supplementary material for the following: a SEM image of the deposits of salts scattered over the SrTiO3 surface and energy dispersive x-ray (EDX) spectra, roughness analysis of etched STO(110), optical profilometry images over large-area etched regions, and mechanical measurement of a LSMO MEMS through the STO(110) substrate.

We thank Flavio Gatti, Francesco Buatier de Mongeot, and Lorenzo Ferrari Barusso for providing access to the optical profilometer. This work was carried out under the OXiNEMS project (www.oxinems.eu). This project has received funding from the European Union's Horizon 2020 research and innovation programme under Grant Agreement No. 828784.

The data that support the findings of this study are openly available in Zenodo at http://dx.doi.org/10.5281/zenodo.4738478.

1.
D. G.
Schlom
,
L.-Q.
Chen
,
X.
Pan
,
A.
Schmehl
, and
M. A.
Zurbuchen
, “
A thin film approach to engineering functionality into oxides
,”
J. Am. Ceram. Soc.
91
,
2429
2454
(
2008
).
2.
A.
Biswas
,
C.-H.
Yang
,
R.
Ramesh
, and
Y. H.
Jeong
, “
Atomically flat single terminated oxide substrate surfaces
,”
Prog. Surf. Sci.
92
,
117
141
(
2017
).
3.
L.
Pellegrino
,
M.
Biasotti
,
E.
Bellingeri
,
C.
Bernini
,
A. S.
Siri
, and
D.
Marré
, “
All-oxide crystalline microelectromechanical systems: Bending the functionalities of transition-metal oxide thin films
,”
Adv. Mater.
21
,
2377
2381
(
2009
).
4.
M.
Biasotti
,
L.
Pellegrino
,
E.
Bellingeri
,
N.
Manca
,
A. S.
Siri
, and
D.
Marrè
, “
Strain response of La0.7Sr0.3CoO3 epitaxial thin films probed by SrTiO3 crystalline microcantilevers
,”
Appl. Phys. Lett.
97
,
223503
(
2010
).
5.
C.
Deneke
,
E.
Wild
,
K.
Boldyreva
,
S.
Baunack
,
P.
Cendula
,
I.
Mönch
,
M.
Simon
,
A.
Malachias
,
K.
Dörr
, and
O. G.
Schmidt
, “
Rolled-up tubes and cantilevers by releasing SrRuO3-Pr0.7Ca0.3MnO3 nanomembranes
,”
Nanoscale Res. Lett.
6
,
621
(
2011
).
6.
N.
Manca
,
L.
Pellegrino
,
T.
Kanki
,
S.
Yamasaki
,
H.
Tanaka
,
A. S.
Siri
, and
D.
Marré
, “
Programmable mechanical resonances in MEMS by localized joule heating of phase change materials
,”
Adv. Mater.
25
,
6430
6435
(
2013
).
7.
V.
Ceriale
,
L.
Pellegrino
,
N.
Manca
, and
D.
Marré
, “
Electro-thermal bistability in (La0.7Sr0.3)MnO3 suspended microbridges: Thermal characterization and transient analysis
,”
J. Appl. Phys.
115
,
054511
(
2014
).
8.
N.
Manca
,
T.
Kanki
,
H.
Tanaka
,
D.
Marré
, and
L.
Pellegrino
, “
Influence of thermal boundary conditions on the current-driven resistive transition in VO2 microbridges
,”
Appl. Phys. Lett.
107
,
143509
(
2015
).
9.
N.
Manca
,
G.
Mattoni
,
M.
Pelassa
,
W. J.
Venstra
,
H. S. J.
van der Zant
, and
A. D.
Caviglia
, “
Large tunability of strain in WO3 single-crystal microresonators controlled by exposure to H2 gas
,”
ACS Appl. Mater. Interfaces
11
,
44438
44443
(
2019
).
10.
S. H.
Baek
,
J.
Park
,
D. M.
Kim
,
V. A.
Aksyuk
,
R. R.
Das
,
S. D.
Bu
,
D. A.
Felker
,
J.
Lettieri
,
V.
Vaithyanathan
,
S. S. N.
Bharadwaja
,
N.
Bassiri-Gharb
,
Y. B.
Chen
,
H. P.
Sun
,
C. M.
Folkman
,
H. W.
Jang
,
D. J.
Kreft
,
S. K.
Streiffer
,
R.
Ramesh
,
X. Q.
Pan
,
S.
Trolier-McKinstry
,
D. G.
Schlom
,
M. S.
Rzchowski
,
R. H.
Blick
, and
C. B.
Eom
, “
Giant piezoelectricity on Si for hyperactive MEMS
,”
Science
334
,
958
961
(
2011
).
11.
D. L.
Bourdais
,
G.
Agnus
,
T.
Maroutian
,
V.
Pillard
,
P.
Aubert
,
R.
Bachelet
,
G.
Saint-Girons
,
B.
Vilquin
,
E.
Lefeuvre
, and
P.
Lecoeur
, “
Epitaxial manganite freestanding bridges for low power pressure sensors
,”
J. Appl. Phys.
118
,
124509
(
2015
).
12.
U. K.
Bhaskar
,
N.
Banerjee
,
A.
Abdollahi
,
Z.
Wang
,
D. G.
Schlom
,
G.
Rijnders
, and
G.
Catalan
, “
A flexoelectric microelectromechanical system on silicon
,”
Nat. Nanotechnol.
11
,
263
266
(
2016
).
13.
S.
Liu
,
B.
Guillet
,
C.
Adamo
,
V. M.
Nascimento
,
S.
Lebargy
,
G.
Brasse
,
F.
Lemarié
,
J. E.
Fallah
,
D. G.
Schlom
, and
L.
Méchin
, “
Free-standing La0.7Sr0.3MnO3 suspended micro-bridges on buffered silicon substrates showing undegraded low frequency noise properties
,”
J. Micromech. Microeng.
29
,
065008
(
2019
).
14.
Z. H.
Lim
,
M.
Chrysler
,
A.
Kumar
,
J. P.
Mauthe
,
D. P.
Kumah
,
C.
Richardson
,
J. M.
LeBeau
, and
J. H.
Ngai
, “
Suspended single-crystalline oxide structures on silicon through wet-etch techniques: Effects of oxygen vacancies and dislocations on etch rates
,”
J. Vac. Sci. Technol. A
38
,
013406
(
2020
).
15.
V. M.
Nascimento
,
L.
Méchin
,
S.
Liu
,
A.
Aryan
,
C.
Adamo
,
D. G.
Schlom
, and
B.
Guillet
, “
Electro-thermal and optical characterization of an uncooled suspended bolometer based on an epitaxial La0.7Sr0.3MnO3 film grown on CaTiO3/Si
,”
J. Phys. D
54
,
055301
(
2021
).
16.
P. W.
Tasker
, “
The stability of ionic crystal surfaces
,”
J. Phys. C
12
,
4977
4984
(
1979
).
17.
F.
Sánchez
,
C.
Ocal
, and
J.
Fontcuberta
, “
Tailored surfaces of perovskite oxide substrates for conducted growth of thin films
,”
Chem. Soc. Rev.
43
,
2272
2285
(
2014
).
18.
J.
Chang
,
Y.-S.
Park
, and
S.-K.
Kim
, “
Atomically flat single-terminated SrTiO3 (111) surface
,”
Appl. Phys. Lett.
92
,
152910
(
2008
).
19.
K.
Szot
,
C.
Rodenbücher
,
G.
Bihlmayer
,
W.
Speier
,
R.
Ishikawa
,
N.
Shibata
, and
Y.
Ikuhara
, “
Influence of dislocations in transition metal oxides on selected physical and chemical properties
,”
Crystals
8
,
241
(
2018
).
20.
R.
Wang
,
Y.
Zhu
, and
S. M.
Shapiro
, “
Structural defects and the origin of the second length scale in SrTiO3
,”
Phys. Rev. Lett.
80
,
2370
2373
(
1998
).
21.
L.
Jin
,
X.
Guo
, and
C.
Jia
, “
TEM study of 110-type 35.26° dislocations specially induced by polishing of SrTiO3 single crystals
,”
Ultramicroscopy
134
,
77
85
(
2013
).
22.
R.
Waser
,
R.
Dittmann
,
G.
Staikov
, and
K.
Szot
, “
Redox-based resistive switching memories-nanoionic mechanisms, prospects, and challenges
,”
Adv. Mater.
21
,
2632
2663
(
2009
).
23.
E.
Trindade Pedrosa
,
I.
Kurganskaya
,
C.
Fischer
, and
A.
Luttge
, “
A statistical approach for analysis of dissolution rates including surface morphology
,”
Minerals
9
,
458
(
2019
).
24.
A.
Luttge
,
R. S.
Arvidson
,
C.
Fischer
, and
I.
Kurganskaya
, “
Kinetic concepts for quantitative prediction of fluid-solid interactions
,”
Chem. Geol.
504
,
216
235
(
2019
).
25.
W. H.
Casey
,
M. J.
Carr
, and
R. A.
Graham
, “
Crystal defects and the dissolution kinetics of rutile
,”
Geochim. Cosmochim. Acta
52
,
1545
1556
(
1988
).
26.
G. C.
Spalding
,
W. L.
Murphy
,
T. M.
Davidsmeier
, and
J. E.
Elenewski
, “
Faceting of single-crystal SrTiO3 during wet chemical etching
,”
MRS Proc.
587
,
O6.4
(
1999
).
27.
M.
Lippmaa
,
K.
Takahashi
,
A.
Ohtomo
,
S.
Ohashi
,
T.
Ohnishi
,
N.
Nakagawa
,
T.
Sato
,
M.
Iwatsuki
,
H.
Koinuma
, and
M.
Kawasaki
, “
Atom technology for Josephson tunnel junctions: SrTiO3 substrate surface
,”
Mater. Sci. Eng.: B
56
,
111
116
(
1998
).
28.
R. J.
Jaccodine
, “
Use of modified free energy theorems to predict equilibrium growing and etching shapes
,”
J. Appl. Phys.
33
,
2643
2647
(
1962
).

Supplementary Material