Advanced Materials Research Vol. 1101 (2015) pp 120-123
© (2015) Trans Tech Publications, Switzerland
doi:10.4028/www.scientific.net/AMR.1101.120
Submitted: 06.01.2015
Accepted: 06.01.2015
Morphological Transformations of Top Electrodes on YMnO3 caused by
Filamentary Resistive Switching in the Oxide Matrix
Agnieszka Bogusz1, 2, a, Daniel Blaschke1, 2, Danilo Bürger2,
Oliver G. Schmidt2,3 and Heidemarie Schmidt2
1
Institute of Ion-Beam Physics and Materials Research, Helmholtz-Zentrum Dresden-Rossendorf,
01328 Dresden, Germany
2
Material systems for Nanoelectronics, Chemnitz University of Technology, 09126 Chemnitz,
Germany
3
Institute for Integrative Nanosciences, IFW Dresden, 01069 Dresden, Germany
a
a.bogusz@hzdr.de
Keywords: resistive switching, Joule heating, morphological changes of electrode, multiferroic
YMnO3.
Abstract. Unipolar resistive switching in YMnO3 with large-scale bottom and small-scale top
electrodes is analyzed in detail by tracking the morphological transformations of the top electrodes
induced by applied writing voltages. Micro-scale digital images are taken after each subsequent
quasi-static current-voltage sweep. Current mapping after electrical investigations indicates a shift in
the conductivity at the localized areas of the morphologically transformed top electrodes. Those
changes are assigned to the heat induced structural and compositional changes within YMnO3 which
lead to the formation and rupture of conductive filaments observed as unipolar resistive switching.
Presented results underline the importance of Joule heating in the fostering of resistive switching and
its adverse impact on the device endurance.
Introduction
Resistive switching (RS) phenomena have been intensively studied in the field of novel solutions for
electronics. It refers to the nonvolatile and reversible change of resistance between at least two
distinguishable states induced by the current or voltage [1]. Several mechanisms have been proposed
in order to explain RS observed in a wide range of insulating and semiconducting materials. One of
the many classifications divides the RS into the filamentary and interface-change related phenomena
[1,2]. Formation and rupture of conductive, whether single filament or multi filaments within the
oxide matrix might induce both unipolar and bipolar (i.e. bias polarity independent and dependent,
respectively) modes of RS. Several experimental methods have been used for the investigation of the
filament(s) nature, e.g. transmission electron microscopy [3], current sensing atomic force
microscopy [4,5] or spectroscopic techniques [6]. Theoretical works focus on the mechanisms of
filament formation and rupture [6,7]. It is generally accepted that the filamentary RS in most cases is
related to the local redox processes induced by large electrical potential gradient and Joule heating.
Electroforming process, i.e. the first electrical stimulus activating the RS is very often assisted by
creation of oxygen vacancies and removal of the oxygen from RS oxide. The latter process might
result in evolution of oxygen gas bubbles which are causing physical deformation of the
electrode/oxide junctions [5]. Recently, the triple phase boundary (TPB) located at the contact point
of oxide, electrode and ambient atmosphere, has been indicated as the only place of the localized
reduction and release of oxygen during both, electroforming and filamentary RS [6].
This work investigates the unipolar RS of multiferroic YMnO3 thin films, a promising candidate
for a new generation of multifunctional materials [8]. Morphological changes of the top electrode
were tracked after subsequent biasing of the sample. Results allow to link the high electric field and
Joule heating with observed RS and point towards its redox nature.
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of TTP,
www.ttp.net. (ID: 134.109.33.226-23/02/15,07:40:45)
Advanced Materials Research Vol. 1101
121
Experimental Details
Sample Preparation. 300 nm thick YMnO3 (YMO) film was grown in a pulsed laser deposition
(PLD) process on 100 nm Pt/ (0001) Al2O3 substrate, where Pt serves as a bottom electrode (BE).
During PLD process, 6000 pulses of a KrF excimer laser at a wavelength of 248 nm were employed
on a ceramic YMO-target. Laser density energy, its frequency, substrate temperature (TS) and oxygen
partial pressure (pO2) were fixed to 2.5 J/cm2, 3 Hz, 800°C and 0.018 mbar, respectively. Circular
shaped top electrodes (TE) of a diameter ranging between 240 µm and 760 µm were prepared by
electron beam evaporation through a shadow mask at room temperature. The electrodes consist of 50
nm Pt coated with 50 nm Ti. The scheme of the investigated sample is presented in Fig. 1. a.
Fig. 1. Scheme of the sample under investigation (a),
SEM image of as-grown YMO film (b) and CS-AFM
images presenting topography (c) with corresponding
current map (d) of sample under bias of 1 V.
Fig. 2. I-V characteristics of TE/YMO/BE
junction in pristine state and in the following HRS 1,
LRS 1 and HRS 2 states (a) after reset 1, set 1, and
reset 2, respectively (b). ROFF/RON ratios are up to 103.
Solid and dotted lines correspond to the applied
positive and negative bias, respectively.
Sample Characterization. The as-grown YMO thin film was characterized by X-ray
diffractometry (XRD) using Cu Kα radiation, scanning electron microscopy (SEM) and current
sensing atomic force microscopy (CS-AFM). Resistive switching properties of the sample were
investigated in a metal-insulator-metal (MIM) configuration, i.e. TE/YMO/BE, in a two-point probe
geometry using a Keithley 2400 source meter. Throughout all electrical measurements BE and TE
were grounded and biased, respectively. Morphological changes of the TE induced by applied bias
were captured by a digital camera attached to the probe station. Properties of the TE after series of
resistive switching events were investigated by SEM, energy-dispersive X-ray spectroscopy (EDX)
and CS-AFM.
Results and Discussion
According to the obtained XRD data and SEM images (Fig. 1. b), the as-grown YMO film is of
nanocrystalline, hexagonal phase (International Centre for Diffraction Data, JCPDF 25-1079).
Low-field electrical investigations indicate that the pristine TE/YMO/BE junctions are mostly in low
resistance state (LRS), as presented in Fig. 2. a, where for comparison also a low-field current-voltage
(I-V) characteristics of the following high resistance state (HRS) 1, LRS 1 and HRS 2 are shown.
Observed high conductivity of the as-prepared TE/YMO/BE structures might be explained by
segregation of point defects, i.e. oxygen vacancies, manganese cations of other than nominal valence
(i.e. +3) and impurities along grain boundaries and/or charged domain walls that would lead to the
shunting between TE and BE. Due to the TS and low pO2 during PLD of YMO, presence of oxygen
vacancies and fluctuations in the valence of Mn cations in as-prepared film are expected. Results of
CS-AFM of the YMO film presented in Fig. 1. c, d reveal high conductivity regions at the grain
boundaries and at the small crystallites. In addition, electrical characterization of analogous MIM
structure with the YMO film grown at higher pO2 (not shown in this work) indicates the HRS of
pristine junctions. This observation supports our argumentation for the origin of the initial LRS of
sample under investigation. Further electrical measurements of the TE/YMO/BE junctions with
increased applied voltage reveal a nonvolatile, unipolar resistive switching with the ROFF/RON ratios
up to 103. Because of the LRS of as-prepared structures, electroforming process is not required. The
first switching event is therefore a reset process, i.e. transition from LRS to HRS. Reset process is
122
Material Science and Engineering Technology III
triggered at VRESET below 5 V with compliance current (CC) fixed to 0.5 A. The reverse process, i.e.
set, occurs at VSET higher than 10 V. As presented in Fig. 2. b both set and reset processes are abrupt
(within the time resolution of measurement set up, i.e. 0.1 s) and independent of the bias polarity. It is
generally accepted that the unipolar RS can be attributed to the formation and rupture of conductive
filaments within the oxide matrix. The latter process is ascribed to the release of the Joule heating
which causes filament to break at its weakest point. In addition, temperature and area dependent
measurements of the ROFF and RON of the analogues Au/YMO/BE [9] and Pt/YMO/BE [10] samples
point towards filamentary origin of the observed RS phenomena where conductivity of LRS and HRS
is of metallic and semiconducting nature, respectively. In order to gain a better understanding of the
observed RS and of the reasons of limited endurance, possibly caused by the damage of TE,
morphological changes of the TE during electrical measurements have been tracked. Fig. 3 presents
the top view of as-prepared TE (Fig. 3. a) and its most significant changes (indicated by arrows on Fig
3. b-f) induced by applied bias and local increase of the temperature. It can be seen that already the
first switching process, i.e. reset 1, causes a microstructural change located at the edge of TE (Fig. 3.
b). The next, visible at the micro scale, modification of the TE appears in the form of big blown-off
region along with few smaller local alterations after applying a voltage sweep with amplitude of 20 V
(Fig. 3. c). It is worth to note that at this value of voltage, the current level reaches approximately 3
mA. The first set process is accompanied by further modification of the TE (Fig. 3. d) which can be
described as the growth of primary changed area observed after reset 1. Subsequent switching of the
junction lead to continued, morphological transformations and degradation of the TE. It can be seen
that each main, step-wise structural change appears as continuation of the preceding transition
process. After the 15th switching event (reset 8) irreversible breakdown of the investigated junction
occurs resulting in the resistance value of approx. 10 GΩ and in the capacitor-like low-field I-V
characteristics. This is explained by physical contact of the probe needle with Al2O3 used as a
substrate in the PLD of YMO film, marked by the circle in Fig. 3. f and confirmed by the EDX
mapping of the TE (Fig. 4. e, f). Interestingly, continuous monitoring of the TE at the contact point
with the probe needle prior reset 8 did not reveal any macro-scale changes. In addition, when the
contact between TE and a probe needle was established again (by changing its position on TE),
junction exhibited further RS. The final state of the TE, i.e. after many RS events and re-adjusting of
the probe needle/TE contact is presented in Fig. 4. The SEM images (Fig. 4. a, b) show a significant
damage of TE in several, independent areas. The CS-AFM investigations covering part of the blown
off region (Fig. 4. c, d) reveal the changes in both morphology and local conductivity. It can be seen
that the height and conductivity of edge of such a region are enhanced when compared to the
unchanged TE and uncovered oxide. Moreover, droplets of even higher conductivity can be seen. This
points towards local increase of temperature what causes TE to melt and form the aforementioned
droplets. In addition, it is well accepted that the damage of the electrode is caused by evolution of the
oxygen gas bubbles formed in the discharge of O2- at the anode (positively biased electrode) [7].
Based on the obtained results of electrical characterization and corresponding changes of the TE it
might be concluded that both reset and set processes involve the local redox reactions which result in
evolution of the O2 gas, preferably at the TPB. It is expected that the nanoscale reduction concerns the
B-site cations, however it remains unclear whether it involves possibly present Mn4+ or Mn3+, or both.
Set process requires both high electric field and Joule heating while the reset process is related only to
the local increase of temperature. The observed breakdown of the junction leading to the limited
endurance performance is ascribed to the local increase of the temperature and delamination of TE
and possibly BE. This limitation might be overcome by decreasing the cell size, i.e. film thickness and
electrode size [7]. In the future, we will focus on investigating smaller size RS cells and on possible
local delamination of the Pt bottom electrode (Fig. 1 a).
Advanced Materials Research Vol. 1101
Fig. 3. Digital images presenting as-prepared TE
on YMnO3 (a) and morphological changes (b-f)
induced by subsequent quasi-static current-voltage
sweeps. Arrows point at new alterations of TE due
to the process defined in the bottom of each image.
Circle in (f) indicates a contact point of probe
needle/TE. Scale bars correspond to 100 µm.
123
Fig. 4. The SEM images (a, b) of the TE after the
electrical characterization. Area marked in (a) by
square is magnified in (b). CS-AFM reveals
topographical (c) and electrical (d) differences of
the altered areas of TE whereas EDX mapping (e,
f) confirms that the breakdown of the sample
occurred by physical contact of probe needle with
Al2O3 substrate.
Summary
Unipolar resistive switching properties of YMnO3 thin film were investigated. Morphological
changes of the top electrode due to the applied bias were tracked subsequently. The obtained results
indicate a local redox reaction accompanied by release of oxygen gas for both set and reset processes.
The importance of a coupled effect of Joule heating and high electric fields in the formation of
conductive filaments was shown. The presented results concern the micro scale investigations but yet
contribute to a better understanding of the observed phenomena.
Acknowledgements
The authors acknowledge Mrs. Ilona Skorupa (Helmholtz-Zentrum Dresden-Rossendorf) for the
sample preparation and the financial support from the Initiative and Networking Fund of the
Helmholtz Association (VH-VI-422) and from Deutsche Forschungsgemeinschaft (DFG
SCHM1663/4-1 and BU 2956/1).
References
[1] R. Waser, R. Dittman, G. Staikov and K. Szot: Adv. Mater. Vol. 21 (2009), p. 2632
[2] A. Sawa: Mater. Today Vol. 11 (2008), p. 28
[3] D.H. Kwon, K.M. Kim, J.H. Jang, J.M. Jeon, M.H. Lee, G.H. Kim, H.S. Li, G.S. Park, B. Lee, S.
Han, M. Kim and C.S. Hwang: Nature Nanotechnology Vol. 5 (2010), p. 148
[4] R. Münstermann, J.J. Yang, J.P. Strachan, G. Medeiros-Ribeiro, R. Dittmann and R. Waser:
Phys. Status Solidi RPL. Vol. 4 (2010), p. 16
[5] J.J. Yang, F. Miao, M.D. Pickett, D.A.A. Ohlberg, D.R. Stewart, C.N. Lau and R.S. Williams:
Nanotechnology Vol. 20 (2009), p. 215201
[6] C. Lenser, M. Patt, S. Menzel, A. Köhl, C. Wiemann, C.M. Schneider, R. Waser and R.
Dittmann: Adv. Funct. Mater. Vol. 24 (2014), p. 4466
[7] P.R. Mickel, A.J. Lohn, C.D. James and M. Marinella: Adv. Mater. Vol. 26 (2014), p. 4486
[8] N.A. Spaldin and M. Fiebig: Science Vol. 309 (2005), p. 391
[9] A. Bogusz, T. You, D. Blaschke, A. Scholz, Y. Shuai, W. Luo, N. Du, D. Bürger, I. Skorupa,
O.G. Schmidt and H. Schmidt: IEEE Xplore, 2013 International Semiconductor Conference
Dresden-Grenoble, (2013)
[10] Z.B. Yan, S.Z. Li, K.F. Wang and J.-M. Liu: Appl. Phys. Lett. Vol. 96 (2010), p. 012103