Page 256 - Read Online

Page 4 of 15 Keeney et al. Microstructures 2023;3:2023041 https://dx.doi.org/10.20517/microstructures.2023.41

In recent years, AFM tip-based nano-machining has been recognized as an emerging technology for
machining nanostructures, nanometer-scale 3D imaging of material properties and photomask repair [34-39] .
In this contribution, we employ a commercially available probe to perform atomic force microscopy
(AFM)-based nano-machining to remove surface artifacts and expose the underlying ferroelectric domain
configuration of 7.9 nm (1.5 u.c.) and 5.6 nm (1 u.c.) B6TFMO films. We perform PFM investigations as a
function of milled depth (from surface to substrate) to investigate domain structures and persistence of
ferroelectricity through the ultrathin B6TFMO films.

MATERIALS AND METHODS
Ultrathin film synthesis
B6TFMO thin films were fabricated using a horizontal flow AIXTRON AIX 200/4FE AVD direct liquid
injection chemical vapor deposition (DLI-CVD) system using processes similar to those previously
-3
described by Keeney et al. [27,28] , Bismuth, iron, and titanium precursor solutions [0.1 mol dm solutions of
Bi(thd) , Fe(thd) , and Ti(O-iPr) (thd) (where thd = 2,2,6,6-tetramethyl 3,5-heptanedionate;
2
3
3
2
O-iPr = isopropoxide) in toluene] were purchased from Epivalence Ltd. (The Wilton Centre, Redcar,
Cleveland, TS10 4RF, United Kingdom) and were maintained under N (g) in stainless steel bubbler
2
containers. 0.025 mol dm manganese precursor was prepared by dissolving Mn(thd) (99%, STREM
-3
3
chemicals) in anhydrous toluene (99.8%, Sigma-Aldrich) under N (g) and transferring to a stainless steel
2
bubbler. The pressure of the growth chamber was 9 mbar, and the temperature of the susceptor [rotated
with 60 sccm N (g)] heated substrate holder was maintained at 710 °C via an infra-red lamp assembly. The
2
liquid precursors were injected using inert N (g) carrier gas into a vaporizer kept at 220 °C, using four
2
individual injectors, one for each precursor source. The micro-liter injected precursor volumes were
regulated by tuning the numbers of pulses [between 75 and 100 injections for each precursor, depending on
desired film thickness (between 5 nm and 8 nm)] and the opening time of the injectors (between 1.3 and
11.3 ms) during a continuous injection mode (frequency of 1 Hz). The net volumetric precursor ratios were
approximately 2.9:1.5:0.6 for Ti(O-iPr) (thd) , Mn(thd) , and Fe(thd) , respectively. Between 5 and 8% excess
2
3
2
3
of Bi(thd) was utilized to mitigate bismuth migration during deposition. The total N (g)/O (g) flow in the
2
3
2
growth chamber was 3,000 sccm, of which O (g) flow was 1,000 sccm. The substrates used in this study were
2
single-side polished (001)-orientated NdGaO (NGO) with average dimensions 10 mm × 10 mm × 0.5 mm
3
(Crystal GmBH, Ostendstraße 25, D-12459 Berlin, Germany). The substrates were cleaned with isopropyl
alcohol followed by distilled water and dried using N (g).
2
Physical characterization
The crystallographic structure of the films was determined using a Panalytical MRD X-Pert Pro X-ray
diffraction (XRD) system with a filtered Cu K radiation source (λ = 0.1541876 nm). The setup included a
vertical magnetic strip mount, a 2 mm mask, and a divergence slit of 0.5°. Scans were conducted over a 2θ
(2 theta) range of 5° to 50°, employing a precise step size of 0.006°. Each step was captured within 0.5 s,
resulting in a total of 7,500 data points collected across the scan. To ensure accurate intensity comparisons,
a uniform baseline correction was applied to both spectra using the second derivative method. Full Width at
Half Maximum (FWHM) values for the peaks were established by integrating the peaks using a Gaussian
function in the Origin software. The X-ray reflectivity (XRR) method was used to determine the film
thickness of the B6TFMO film deposited using 75 injections. The same XRD system was used to detect the
reflectometry on the ultrathin film and substrate. 2Theta-Omega scans were performed over an angular 2θ
range of 0.2° to 7° with an angular Δθ step of 0.005°. A parabolic mirror was used as the incident optics,
while a parallel plate collimator with a 0.09° slit was used as the diffracted optics. The XRR data was then
analyzed through automatic fitting procedures implemented in the Panalytical XRR software package,
where the number of layers, the density of each layer, their thickness, and the interface or surface roughness
can be adjusted within a reasonable user-defined range. During the simulation, three layers were assumed: