Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

LETTERS

PUBLISHED ONLINE: 10 AUGUST 2015 | DOI: 10.1038/NMAT4365

Negative-pressure-induced enhancement in a
freestanding ferroelectric
Jin Wang1,2*, Ben Wylie-van Eerd3, Tomas Sluka1, Cosmin Sandu1, Marco Cantoni4, Xian-Kui Wei1,5,
Alexander Kvasov1, Leo John McGilly1, Pascale Gemeiner6, Brahim Dkhil6, Alexander Tagantsev1,
Joe Trodahl3 and Nava Setter1*
Ferroelectrics are widespread in technology1 , being used in corners (Fig. 1a), whereas in the PX phase, oxygen octahedra are
electronics and communications2 , medical diagnostics and connected by both their corners and their edges (Fig. 1b). Therefore,
industrial automation. However, extension of their operational although their cation to oxygen ratio is identical, the conversion
temperature range and useful properties is desired3–5 . Recent from PX to perovskite requires the presence of extra oxygen, which is
developments have exploited ultrathin epitaxial films on absorbed at the beginning of the conversion and released again on
lattice-mismatched substrates, imposing tensile or compres- its completion13 . These two features, the difference in density and
sive biaxial strain, to enhance ferroelectric properties6,7 . the oxygen required for the phase conversion, provide the condition
Much larger hydrostatic compression can be achieved by for negative-pressure creation. When a body of PX phase is heated in
diamond anvil cells8,9 , but hydrostatic tensile stress is regarded air, the perovskite phase first forms at the surface where the oxygen
as unachievable. Theory and ab initio treatments10 predict is abundant; the released oxygen is gradually supplied to the inner
enhanced properties for perovskite ferroelectrics under hydro- parts of the body and the conversion proceeds inwards. However,
static tensile stress. Here we report negative-pressure-driven whereas the outer part accommodates the shrinkage, the inner part,
enhancement of the tetragonality, Curie temperature and when transforming, is bound by the rigid outer part, resulting in a
spontaneous polarization in freestanding PbTiO3 nanowires, strong tensile stress in the core.
driven by stress that develops during transformation of PbTiO3 nanowires were initially prepared in the PX phase
the material from a lower-density crystal structure to the (see Methods and Supplementary Information). The perovskite
perovskite phase. This study suggests a simple route to obtain nanowires, with thicknesses ranging from 20 to 500 nm (Fig. 1c)
negative pressure in other materials, potentially extending were obtained by annealing the PX nanowires in air at ∼540 ◦ C.
their exploitable properties beyond their present levels. Nearly spherical pores with a typical diameter of around 10 nm
Pressure influences the properties of ferroelectrics as a simple did not break their monocrystallinity (Supplementary Fig. 1a).
consequence of elasto-electric coupling. Biaxial stress has been Annealed nanowires showed a tetragonal perovskite structure,
achieved in films on lattice-mismatched substrates, resulting in confirmed by X-ray diffraction (XRD) and Raman spectroscopy
strongly enhanced properties, whereas negative pressure is consid- (Supplementary Fig. 1b,c). The spontaneous polarization direction,
ered practically unrealizable. However, when a material undergoes the c axis of the tetragonal perovskite cell, did not adopt any
a phase transformation from a low-density structure to a denser specific orientation with respect to the nanowire geometry. Bending
one, the conditions of the transformation can dictate stretching of occurred in some of the perovskite nanowires and 90◦ domains were
the dense structure, resulting in a built-in negative pressure and its frequently found across the bent regions (Supplementary Fig. 1d).
corresponding property tuning. Here we implement this concept for The c/a ratio of individual nanowires was investigated by
the prototypical ferroelectric PbTiO3 (PT). We produced freestand- high-resolution transmission electron microscopy (HR-TEM)
ing PbTiO3 particles under the appropriate conditions, studied the for nanowires thinner than 120 nm and selected area electron
structural fingerprints of the negative pressure, and found that these diffraction (SAED) for thicker wires. Remarkably, in a small range
particles had enhanced tetragonality, spontaneous polarization and of diameters near 110 nm, there was an enormously enhanced
Curie temperature. The experimental work was complemented by tetragonality (c/a − 1) up to ∼0.13 (Fig. 1d), almost doubling the
theory, ab initio modelling and numerical simulations. tetragonality from bulk PbTiO3 (ref. 14). Both thinner and thicker
PbTiO3 exists in two different crystalline structures, the common wires exhibited c/a ratios similar to the bulk PbTiO3 value.
ferroelectric perovskite and a body-centred tetragonal structure, Polarization switching under an applied field confirmed
named the ‘PX’ phase11,12 . The unit-cell volume of the PX phase is ferroelectricity of the wires. Supplementary Fig. 2 demonstrates
approximately nine times that of perovskite, whereas the number of polarization switching with a scanning probe in a wire of diameter
atoms in the unit cell is eight times that in perovskite. The remainder ∼105 nm. Off-centring of Ti with respect to the four nearest
of the volume is taken up by a longitudinal central pore along the Pb columns in a wire of 116 nm diameter was about twice the
c-axis. Thus, the density of perovskite is 13% higher than that of bulk value as shown by the high-resolution high-angle annular
PX. In perovskites, oxygen octahedra are interconnected by their dark-field (HAADF) measurement, and a spontaneous polarization

1 CeramicsLaboratory, Swiss Federal Institute of Technology (EPFL), 1015 Lausanne, Switzerland. 2 Graduate School at Shenzhen, Tsinghua University,
518055 Shenzhen, China. 3 MacDiarmid Institute for Advanced Materials and Nanotechnology, Victoria University of Wellington, PO Box 600,
Wellington 6140, New Zealand. 4 Interdisciplinary Centre for Electron Microscopy, Swiss Federal Institute of Technology (EPFL), 1015 Lausanne,
Switzerland. 5 Peter Grünberg Institute and Ernst Ruska Center for Microscopy and Spectroscopy with Electrons, Research Center Jülich, 52425 Jülich,
Germany. 6 Laboratoire Structures, Propriétés et Modélisation des Solides, CNRS-UMR8580, Ecole Centrale Paris, Grande Voie des Vignes,
92290 Châtenay-Malabry, France. *e-mail: wang.jin@sz.tsinghua.edu.cn; nava.setter@epfl.ch

NATURE MATERIALS | VOL 14 | OCTOBER 2015 | www.nature.com/naturematerials 985


LETTERS NATURE MATERIALS DOI: 10.1038/NMAT4365

a b

b b

a a
c c

c d
1.14

1.12

1.10

c/a
1.08

1.06
c/a in bulk PT
1.04
200 nm 0 25 50 75 100 125 150 175 200
Thickness (nm)

Figure 1 | Crystalline structure and microstructure of the PbTiO3 nanowires. a, The crystalline structure of the perovskite phase viewed along the c axis.
b, The crystalline structure of the PX phase viewed along the c axis. c, Scanning electron microscopy image of perovskite PbTiO3 nanowires. d, c/a versus
the thickness of the nanowires. The error bars reflect uncertainty in determining c/a values and their spatial variation on given wires.

of ∼160 µC cm−2 was obtained according to the relative cationic bringing TC enhancement into agreement with that measured on the
displacements (Supplementary Information and Supplementary 250 nm single wire. The enhancement is stable with time: samples
Fig. 3). The ferroelectric–paraelectric phase transition was showed the same increased tetragonality after having been stored
investigated in individual nanowires by temperature-dependent for more than two years.
Raman spectroscopy. Figure 2a shows spectra for three wires with It is untenable that the enhancements are related to a purely size
thicknesses of 650, 250 and 125 nm, at temperatures that span effect, which can appear only on a much shorter length scale19–22 .
the bulk Curie temperature of 490 ◦ C (refs 15–17). There are no On the other hand, the expanded unit-cell volume relative to bulk
Raman active modes in the paraelectric cubic phase, and thus the PT shown in Fig. 3a seems to be a fingerprint of tensile stress; indeed,
thickest wire loses all Raman bands above the bulk TC . However, negative pressure provides a natural scenario in which the entire
the signal from the ferroelectric tetragonal phase persists to higher data set is understood. Within the Landau–Ginzburg–Devonshire
temperatures in thinner wires, and is still clearly seen in the 125 nm (LGD) phenomenological description, the effect of the hydrostatic
nanowire at 600 ◦ C in the top panel of Fig. 2b. stress on tetragonality and the shift of Curie temperature 1TC (see
The E(1TO) mode below 100 cm−1 provides a convenient Methods and Supplementary Information) are plotted in Fig. 3c.
signature of the impending ferroelectric–paraelectric transition. In The measured 1TC of 40 ◦ C in the 250 nm wire and the multiple-
bulk PbTiO3 it is a narrow line near 90 cm−1 at ambient temperature, wire samples corresponds to a hydrostatic tensile stress ∼0.4 GPa,
and it softens and broadens as TC is approached, disappearing after which agrees also with the tetragonality (c/a − 1 = 0.07) determined
reaching ∼55 cm−1 at TC (refs 15,16). Figure 2b shows its frequency by multiple-wire XRD. Similar agreement is found near 2 GPa for
dependence on temperature, compared with single-crystal PbTiO3 the enhanced tetragonality (c/a − 1 = 0.08, Fig. 1d) and Curie
(ref. 15). Data for the 650 nm nanowire agree with bulk PbTiO3 , temperature (1TC > 100 ◦ C, Fig. 2b) in 125 nm wires.
whereas the 250 nm nanowire exhibits a frequency shifted by Temperature-dependent lattice parameters measured for the
∼40 ◦ C relative to bulk PbTiO3 , with TC ∼ 520–530 ◦ C. The 125 nm multiple-wire sample also agree well with the predictions of the
nanowire with c/a ≈ 1.08 (Fig. 1d) has an enormously enhanced TC , LGD expansion (solid curves in Fig. 3a,b). Note that the only
with the signal seen well above 55 cm−1 in spectra at the 600 ◦ C limit input parameter is the hydrostatic stress σ , which is ∼0.4 GPa
above which Pb starts to be lost. as extracted from the measured shift of TC of this sample. It
An enhanced TC is corroborated by XRD on a multiple-wire is clear that the experimental data are in full agreement with a
sample compared in Fig. 3a,b with standard bulk data14,18 . The negative-pressure scenario as determined by the semi-empirical
lattice parameter c and c/a are larger than in the bulk over the LGD description in these moderately stressed wires. The different
whole measured temperature range, and the Curie temperature deformation characteristics under hydrostatic tension in the c and a
is enhanced by ∼40 ◦ C. As shown in Supplementary Fig. 4, the axes due to an anisotropic electrostriction effect are exactly observed
volume-weighted average diameter of the wires is 220–270 nm, in the experimental data.

986 NATURE MATERIALS | VOL 14 | OCTOBER 2015 | www.nature.com/naturematerials


NATURE MATERIALS DOI: 10.1038/NMAT4365 LETTERS
a b 90
600 °C
3 80

2 70 125 nm

1 60

0 90
500 °C
Raman intensity (a.u.)

3 80

Frequency (cm−1)
2 125 nm 70
250 nm
250 nm
1 60
650 nm

0 90
475 °C

3 80

2 70
650 nm

1 60

50
0 100 200 300 400 100 200 300 400 500 600
Frequency (cm−1) Temperature (°C)

Figure 2 | Curie temperature of single wires probed by Raman spectroscopy. a, Spectra on single wires of diameters 125, 250 and 650 nm, at
temperatures spanning the Curie temperature of bulk PT. The lowest frequency E(1TO) feature was followed to investigate the diameter-dependent TC .
The occasional peaks around 40–50 cm−1 are substrate lines. b, The frequency of the lowest E(1TO) line plotted versus temperature for the three wires,
with the temperature dependence of the pure E(1TO) mode in bulk PT (ref. 15; red dashed lines) for comparison. The uncertainty bars marked are based on
several spectra on the same wire at fixed T as well as the uncertainties associated with fitting lines on a background that rises towards low frequency.

The maximal measured tetragonality, (c/a = 1.130 in the 115 nm of a few GPa. These cavitation voids are formed by a strongly
wire versus 1.065 at zero pressure), however, cannot be reconciled tensile isotropic component of the stress tensor during the PX to
with the results of the LGD expansion; it extends into the region perovskite transformation23–25 . An analogous situation is found in
above 2 GPa which is certain to violate the LGD linear stress– the solidification of glasses, where cavitation occurs, although at
strain curve in Fig. 3c. Thus, to understand these very highly much larger scales26 . In that case the phase front proceeds from
stressed wires we appeal to the ab initio results10 . The blue nonlinear the surface inwards, and the outermost regions form a rigid shell
curve in Fig. 3c is the result of that calculation, shifted by 0.36 to which then prevents the core from collapse. When the phase front
align it with the experimental zero-pressure tetragonality14 , as local- propagates further inwards, large negative pressure is built up,
density approximation (LDA) calculations are known to give an resulting in cavitation—that is, the appearance of voids, or even
underestimated tetragonality in PbTiO3 (ref. 10). The tensile stress fractures26 . A similar transformation from the surface inwards is also
corresponding to c/a = 1.13 is ≈4.5 GPa, at the foot of the ab initio a feature of the present system, because the transformation from
nonlinearity. As ab initio calculations of the dielectric permittivity the PX phase to the perovskite phase cannot proceed without the
of PT show no significant variation of the dielectric permittivity as assistance of catalytic oxygen13 .
a function of negative pressure (see Methods and Supplementary Traces of a different phase transformation history in small and
Fig. 5), the lattice softness and the ferroelectricity of PT are expected large diameters can be seen in the characteristic shapes of wires in
to be maintained beyond the nonlinear regime of the negative- the cross-sectional TEM (Supplementary Fig. 6). The square profile
pressure effect. in the PX phase12 transforms to circular in the thinnest wires. At the
The data above clearly indicate that the wires are under strong other extreme, for larger wires, we find either a large central void
hydrostatic tensile stress; we now demonstrate that such a negative- or cracks (Supplementary Fig. 7). These larger wires retain much of
pressure build-up is signalled also by both the nanopore formation their square cross-section, and many show inward bowing of the flat
and the cross-sectional deformation of the wires. Turning first to surface, which can no longer maintain the inner tensile stress. Such
the distribution of approximately spherical voids shown in Fig. 4a–c, convex bowing, the first step of failure, is already seen in wires within
there is a clear signature that the cores of the wires need to the band of maximal stress, as shown in Fig. 4d.
accommodate a volume deficit. In particular, the image of a 115 nm To obtain a quantitative view, the negative pressure build-up is
nanowire, with a c/a about 1.13, shows a remarkable density of simulated by a plane-strain elasto-plastic model (see Supplementary
∼10 nm voids separated by about two to three times their diameter. Information for details) which considers 13% volume shrinkage at
The voids account for less than 10% of the net 13% deficit, and the a defined phase transition. The simulation (Fig. 4e) confirms that
remaining few per cent is provided by tensile stress in the range the negative pressure can result solely from the volume shrinkage

NATURE MATERIALS | VOL 14 | OCTOBER 2015 | www.nature.com/naturematerials 987


LETTERS NATURE MATERIALS DOI: 10.1038/NMAT4365

a 4.20 b 1.07

1.06
4.15

1.05
4.10
Lattice parameter (Å)

1.04
c axis of nanowires
4.05
a axis of nanowires
1.03

c/a
c axis of bulk (work of ref. 18) Nanowires
4.00 Bulk (work of ref. 18)
a axis of bulk (work of ref. 18) 1.02
Simulated
Simulated c axis
3.95
Simulated a axis 1.01

3.90 1.00

3.85 0.99
0 150 300 450 600 0 150 300 450 600
Temperature (°C) Temperature (°C)

c 1.35 700

1.30 600

1.25 500

1.20 400

ΔTC (°C)
c/a

1.15 300

1.10 200

1.05 100

1.00 0
0 1 2 3 4 5 6 7 8
Hydrostatic tensile stress (GPa)

Figure 3 | Temperature-dependent lattice parameters and tetragonality of the multiple-wire sample and predicted effect of hydrostatic tensile stress by
LGD phenomenology and first-principles calculations. a,b, Temperature dependence of lattice constants (a) and c/a (b) of the multiple-wire sample
compared with the bulk (the powder data from ref. 18 is used as the reference for bulk PbTiO3 , as the temperature dependence of lattice parameters
measured on PbTiO3 single-crystal13 is in good agreement with that measured on PbTiO3 bulk powders18 ). The simulated lattice constants and c/a of PT
under 0.4 GPa hydrostatic tensile stress are also included (see the text for further discussions). c, TC (black curve) and c/a at room temperature (red curve)
under hydrostatic tensile stress predicted by the LGD theory. c/a of PbTiO3 under hydrostatic tensile stress simulated by first-principles calculations (blue
curve, reconstructed according to the data in ref. 10 and shifted up to fit the measured c/a value of bulk PT at atmospheric pressure). The dashed parts
above 2 GPa in the LGD results are considered less reliable owing to the limited length of the expansion series used.

progressing from the surface inwards (Supplementary Fig. 8) and modelling and semi-empirical LGD theory for tensile stress of a
that the key difference between individual nanowires is the ratio few GPa. The negative-pressure scenario is further signalled by the
of their diameter to the phase-boundary width. In thicker wires, appearance of ∼10 nm cavitation voids seen in TEM images of the
the progressing shrinkage develops larger local stresses and plastic wires. Negative pressure is developed in the wire interiors as the
deformation, and the phase-boundary region occupies a smaller transformation front migrates inwards. It is technically important
volume fraction than in thinner wires. As a result, the simulated that effective hydrostatic tensile stress is created in freestanding
built-in negative pressure grows with the wire diameter (Supple- nanowires. The technique is not limited to lead titanate, but can
mentary Fig. 9), inducing also enhancement of the average c/a ratio develop in any transformation from a low- to high-density phase
(Supplementary Fig. 10). The negative pressure in real wires cannot that proceeds inwards from the surface. The PX to perovskite
grow without limit, leading to central pore formation and wire transformation is possible also in other widely used ferroelectrics;
splitting at specific stress thresholds, and the corresponding sharp a similar approach might also be applied to a large number of
drop of c/a back to the normal bulk value (Fig. 1d). other materials with controlled transformations: for example, from
In summary, we have reported the preparation of PbTiO3 pyrochlore phases (such as Na2 Nb2 O6 ) to the denser perovskite
nanowires formed by oxygen-mediated transformation from the phase. Amorphous phases of many useful oxides are typically some
PX phase, which has a 13% larger specific volume than the 10–30% less dense than their corresponding crystalline phases27 ,
final ferroelectric perovskite phase. The wires with diameters of therefore extension to materials useful in diverse functionalities
∼100–125 nm show strongly enhanced tetragonality, spontaneous might be reached. Negative pressure seems to be a feasible way to
polarization and Curie temperature, in accordance with ab initio modify materials and to extend their exploitable properties.

988 NATURE MATERIALS | VOL 14 | OCTOBER 2015 | www.nature.com/naturematerials


NATURE MATERIALS DOI: 10.1038/NMAT4365 LETTERS
a b c
D = 20 nm D = 35 nm D = 115 nm

20 nm 20 nm 20 nm

d e

Pressure (GPa)
0

−2

20 nm 20 nm −4

Figure 4 | Pore profile at different wire thicknesses and simulation of the hydrostatic pressure during the PX–perovskite phase transition. a–c, TEM
images showing the pore distribution in the 20 nm (a), 35 nm (b) and 115 nm (c) nanowires. d, Cross-section of a 116-nm-thick nanowire. e, Simulated
distribution of hydrostatic pressure in a 100-nm-thick nanowire after transformation from PX to the perovskite phase.

Methods 14. Mabud, S. A. & Glazer, A. M. Lattice parameters and birefringence in PbTiO3
Methods and any associated references are available in the online single crystal. J. Appl. Phys. 62, 49–53 (1979).
15. Burns, G. & Scott, B. A. Lattice modes in ferroelectric perovskites: PbTiO3 .
version of the paper.
Phys. Rev. B 7, 3088–3101 (1973).
16. Fontana, M. D., Idrissi, H., Kugel, G. E. & Wojcik, K. Raman spectrum in
Received 5 May 2015; accepted 26 June 2015; PbTiO3 re-examined: Dynamics of the soft phonon and the central peak.
published online 10 August 2015 J. Phys.: Condens. Matter 3, 8695–8705 (1991).
17. Foster, C. M., Li, Z., Grimsditch, M., Chan, S.-K. & Lam, D. J. Anharmonicity
of the lowest-frequency A1 (TO) phonon in PbTiO3 . Phys. Rev. B 48,
References 10160–10167 (1993).
1. Haertling, G. H. Ferroelectric ceramics: History and technology. J. Am. Ceram. 18. Haun, M. J., Furman, E., Jang, S. J., McKinstry, H. A. & Cross, L. E.
Soc. 82, 797–818 (1999). Thermodynamic theory of PbTiO3 . J. Appl. Phys. 62, 3331–3338 (1987).
2. Garcia, V. et al. Giant tunnel electroresistance for non-destructive readout of 19. Naumov, I. I. & Fu, H. X. Spontaneous polarization in one-dimensional
ferroelectric states. Nature 460, 81–84 (2009). Pb(ZrTi)O3 nanowires. Phys. Rev. Lett. 95, 247602 (2005).
3. Grinberg, I. Perovskite oxides for visible-light absorbing ferroelectric and 20. Geneste, G., Bousquet, E., Junquera, J. & Ghosez, P. Finite-size effects in
photovoltaic materials. Nature 503, 509–512 (2013). BaTiO3 nanowires. Appl. Phys. Lett. 88, 112906 (2006).
4. Anton, S. R. & Sodano, H. A. A review of power harvesting using piezoelectric 21. Spanier, J. E. et al. Ferroelectric phase transition in individual single-crystalline
materials (2003–2006). Smart Mater. Struct. 16, R1–R21 (2007). BaTiO3 nanowires. Nano Lett. 6, 735–739 (2006).
5. Yan, H. X. et al. A lead-free high-Curie-point ferroelectric ceramic, 22. Junquera, J. & Ghosez, P. Critical thickness for ferroelectricity in perovskite
CaBi2 Nb2 O9 . Adv. Mater. 17, 1261–1265 (2005). ultrathin films. Nature 422, 506–509 (2003).
6. Choi, K. J. et al. Enhancement of ferroelectricity in strained BaTiO3 thin films. 23. Huang, Y., Hutchinson, J. W. & Tvergaard, V. Cavitation instabilities in elastic
Science 306, 1005–1009 (2004). plastic solids. J. Mech. Phys. Solids 39, 223–241 (1991).
7. Schlom, D. G. et al. Strain tuning of ferroelectric thin films. Annu. Rev. Mater. 24. Tszeng, T. C. Quasistatic and dynamic growth of sub-microscale spherical
Res. 37, 589–626 (2007). voids. Mech. Mater. 41, 584–598 (2009).
8. Ramrez, R., Lapena, M. F. & Gonzalo, J. A. Pressure dependence of free-energy 25. Cohen, T. & Durban, D. Hypervelocity cavity expansion in porous elastoplastic
expansion coefficients in PbTiO3 and BaTiO3 and tricritical-point behavior. solids. Trans. ASME 80, 011017 (2013).
Phys. Rev. B 42, 2604–2606 (1990). 26. Abyzov, A. S., Schmelzer, J. W. P. & Fokin, V. M. Theory of pore formation in
9. Sanjurjo, J. A., Lopez-Cruz, E. & Burns, G. High-pressure Raman study of glass under tensile stress: Generalized Gibbs approach. J. Non-Cryst. Solids 357,
zone-center phonons in PbTiO3 . Phys. Rev. B 28, 7260–7268 (1983). 3474–3479 (2011).
10. Tinte, S., Rabe, K. M. & Vanderbilt, D. Anomalous enhancement of 27. Nakamura, A. et al. Self-elongated growth of nanopores in annealed
tetragonality in PbTiO3 induced by negative pressure. Phys. Rev. B 68, amorphous Ta2 O5 films. Scr. Mater. 66, 182–185 (2012).
144105 (2003).
11. Cheng, H. M. et al. Hydrothermal synthesis of acicular lead titanate fine
powders. J. Am. Ceram. Soc. 75, 1123–1128 (1992).
Acknowledgements
The research leading to these results has received funding from the European Research
12. Wang, J. et al. Structure determination and compositional modification of
Council under the European Union’s Seventh Framework Programme
body-centered tetragonal PX-phase lead titanate. Chem. Mater. 23, (FP7/2007-2013)/ERC grant agreement no. [268058]. Research within the MacDiarmid
2529–2535 (2011). Institute is supported under the New Zealand Centres of Research Excellence fund.
13. Wang, J. et al. Mechanism of hydrothermal growth of ferroelectric PZT J. Wang acknowledges the National Natural Science Foundation of China (Grant
nanowires. J. Cryst. Growth 347, 1–6 (2012). No. 51302143) for financial support. Additional financial support was received from the

NATURE MATERIALS | VOL 14 | OCTOBER 2015 | www.nature.com/naturematerials 989


LETTERS NATURE MATERIALS DOI: 10.1038/NMAT4365

Swiss National Science Foundation (Grant No. 200020_153177). We thank inputs from T.S. and A.T. N.S. was responsible for overall direction and coordination. All
K. Vaideeswaran for sample preparation using focused ion beam and A. Duncan for authors contributed to the manuscript and the interpretation of the data.
nano-beam electron diffraction experiments using the JEOL 2200FS electron microscope.

Additional information
Author contributions Supplementary information is available in the online version of the paper. Reprints and
J.W. prepared the samples and merged the theory/modelling with experimental results. permissions information is available online at www.nature.com/reprints.
B.W.-v.E. and J.T. performed Raman spectroscopy on single wires. T.S. and A.T. Correspondence and requests for materials should be addressed to J.W. or N.S.
conducted the modelling. C.S., X.-K.W. and M.C. characterized the sample by TEM. A.K.
performed ab initio calculations advised by A.T. L.J.M. conducted piezoresponse
measurements on individual wires. P.G. and B.D. carried out XRD measurements. J.W. Competing financial interests
and J.T. developed the concept of the paper. J.W., J.T. and N.S. wrote the manuscript with The authors declare no competing financial interests.

990 NATURE MATERIALS | VOL 14 | OCTOBER 2015 | www.nature.com/naturematerials


NATURE MATERIALS DOI: 10.1038/NMAT4365 LETTERS
Methods account. As a result, they are not expected to give a good description of the effect
Wires used here were prepared initially as PX (ref. 11), with the same stoichiometry of the hydrostatic enhancement in the proximity of the critical region (above 2 GPa,
as perovskite PT but a crystalline structure that has a reduced density (by 13%). dashed lines in Fig. 3c) where, according to the ab initio results, an anomalous
When annealed in air at 540 ◦ C for 30 min they transformed into conventional variation of structural parameters occurs. The TEM cross-sectional specimens
perovskite PT. Raman spectroscopy was performed with a Horiba Jobin-Yvon were prepared using a focused ion beam system. The high-resolution STEM image
T64000 spectrometer, excited with the 514.5 nm line from an Ar+ ion laser. was obtained on a probe-corrected FEI Titan-80-300 microscope with an operating
Measurements were performed on single wires of 125, 250 and 650 nm diameter, voltage of 300 kV. Ab initio calculations were performed using the Quantum
previously selected by scanning electron microscopy. The samples were supported ESPRESSO (QE) ab initio package to study the dielectric properties of perovskite PT
on sapphire substrates placed on the hearth of a Linkam variable-temperature under negative pressure. For the structural full-relaxation calculations, the PWscf
microscope stage capable of being heated to 600 ◦ C. The power code, which is a set of programs for electronic structure calculations within density
used was reduced to avoid heating the wire, and carefully checked to establish that functional theory (DFT), was exploited. The PWscf calculations are made within
increasing it by factors of two did not affect the results. The temperature-dependent the generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof
XRD measurements on the multiple-wire sample were performed using (PBE) exchange–correlation functional using the ultrasoft pseudopotential from
a high-resolution two-axis X-ray diffractometer in the Bragg–Brentano geometry PSLibrary 0.3.1. An automatically generated uniform Monkhorst–Pack 10×10×10
with Cu Kβ radiation (λ = 1.3922 Å). The {200}, {220} and {222} family peaks were grid of k-points was used, and the kinetic energy cutoff for wavefunctions is 80 Ry.
recorded in the temperature range between 25 ◦ C and 640 ◦ C in temperature steps The atomic positions are allowed to relax until the forces on atoms are smaller than
of 10 ◦ C. After correction of the zero shift, the lattice parameters in both tetragonal 10−5 Ry/bohr (0.3 meV A−1 ). The process of negative-pressure formation during
and cubic phases were extracted. The local piezoelectric and ferroelectric the structural phase transition from PX to the perovskite phase was simulated
activity of the PT wires were examined by a piezoresponse force microscope with a two-dimensional numerical model of an elasto-plastic material whose
(Cypher, Asylum Research) with a Ti/Ir-coated tip (3 N m−1 in force constant). The spontaneous strain is coupled (through a phase parameter) with the concentration
spontaneous polarization (PS ) in the d = 116 nm nanowire was estimated based of diffusing catalytic oxygen. The simulation domain has a square shape of varying
on a high-resolution HAADF image following the linear relationship between size. The boundary conditions are set as stress-free with the influx of oxygen.
PS and the relative displacements of the B-site cations in displacive ferroelectrics. In The phase change is defined at a specified oxygen concentration which induces the
LDG calculations, the coefficients of the free-energy expansion of PbTiO3 are taken change of spontaneous strain. The model was implemented in the finite-element
from ref. 18. Note that the LGD predictions are based on the classic expansion method software COMSOL 4.3a by modification of its standard packages. More
series in ref. 18, where only lowest-order coupling between σ and P was taken into detailed descriptions on the methods can be found in Supplementary Information.

NATURE MATERIALS | www.nature.com/naturematerials

You might also like