Available via license: CC BY 3.0
Content may be subject to copyright.
Journal of Physics: Conference Series
PAPER • OPEN ACCESS
High-resolution Scanning Transmission EBIC Analysis of Misfit
Dislocations at Perovskite pn-Heterojunctions
To cite this article: T Meyer et al 2019 J. Phys.: Conf. Ser. 1190 012009
View the article online for updates and enhancements.
This content was downloaded from IP address 178.171.60.62 on 24/05/2019 at 02:44
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd
19th International Conference on Extended Defects in Semiconductors (EDS2018)
IOP Conf. Series: Journal of Physics: Conf. Series 1190 (2019) 012009
IOP Publishing
doi:10.1088/1742-6596/1190/1/012009
1
High-resolution Scanning Transmission EBIC Analysis of
Misfit Dislocations at Perovskite pn-Heterojunctions
T Meyer1, B Kressdorf2, J Lindner2, P Peretzki1, V Roddatis2, C Jooss2
and M Seibt1
1 4th Institute of Physics – Solids and Nanostructures, University of Goettingen,
Friedrich-Hund-Platz 1, 37077 Goettingen, Germany
2 Institute of Materials Physics, University of Goettingen, Friedrich-Hund-Platz 1,
37077 Goettingen, Germany
tmeyer@uni-goettingen.de
Abstract. Fundamental losses of photovoltaic energy conversion are transmission of sub band
gap photons and thermalisation which are the underlying physics of the Shockley-Queisser
limit defining maximum conversion efficiency of single-junction solar cells. Strongly
correlated materials such as perovskites are promising candidates to exceed this limit by
exploiting (i) long wavelength light absorption and (ii) the existence of long-living intraband
excitations indicating that harvesting hot excess carriers might be feasible in such systems. In
this work, we study pn-heterojunctions produced from Pr1-xCaxMnO3 on SrTi1-yNbyO3 by
means of microscopic techniques. Such systems exhibit relevant quantities such as space
charge layer width, screening lengths and excess carrier diffusion lengths in the 1-10 nm range
which makes the use of standard methods such as electron beam induced current a challenging
task. We report scanning transmission electron beam induced current experiments of misfit
dislocations at the heterojunction. The dislocation-induced reduction of the charge collection is
studied with nanometer spatial resolution. Effects of surface recombination and the
heterojunction electric field are discussed.
1. Introduction
Transition metal oxide perovskites, in particular manganites, became popular in the 1950s due to their
extraordinary magnetic properties [1]. Ever since then the interest in this class of strongly correlated
materials has grown continuously and various extraordinary phenomena such as high-temperature
superconductivity [2], colossal magnetoresistance [3], and photovoltaic activity [4] have been
discovered. One reason for the diversity of research projects related to perovskites is the high
adjustability of different degrees of freedom: The ideal perovskite structure ABC3 is cubic with A
occupying the primitive, B the body-centred, and C, symbolising oxygen for all compounds
considered in this work, the face-centred sites. In addition, tetragonal and even orthorhombic
perovskites exist resulting from small lattice deformations. The latter are typically related to rotations
and distortions of the octahedron spanned by species C and depend highly on the particular choice of
constituents as well as doping of the respective lattice sites [5]. Thus, bonding lengths and angles of
neighbouring atoms can be adjusted by the samples’ stoichiometry resulting in different overlaps of
orbitals and hence a tunable interplay of exchange interactions [6-8].
Using organic halide compounds, the efficiency of perovskite based solar cells has been increased
beyond 20% during the last decade [9] offering a competitive solution to silicon solar cells. Still,
19th International Conference on Extended Defects in Semiconductors (EDS2018)
IOP Conf. Series: Journal of Physics: Conf. Series 1190 (2019) 012009
IOP Publishing
doi:10.1088/1742-6596/1190/1/012009
2
achieving long-term stability and explaining the underlying physical transport processes remain open
challenges. Despite much lower photovoltaic energy conversion efficiencies, transition metal oxide
perovskites can be considered as a model system to accomplish the second task. In fact, it has been
shown that Pr1-xCaxMnO3 exhibit strong infrared absorption [10]. While a thorough review on
manganites in general can be found in [11], this article focuses on the particular compound
Pr0.66Ca0.34MnO3, subsequently referred to as PCMO, exhibiting a positive Seebeck coefficient and
thus being considered as a p-type material [12]. Used as an absorbing material, PCMO shows a
tremendously better photovoltaic performance in a charge-ordered state where the lattice distortions
mentioned above become cooperative [13,14]. Most noticeably, the corresponding polaronic
excitations are intraband transitions. Although they could potentially relax via phonon scattering, they
seem to be stabilised by the cooperative configuration. Hence, the basic assumption of hot carrier
thermalisation being faster than typical charge collection times resulting in the Shockley-Queisser
limit [15] might be overcome in strongly correlated materials. Consequently, perovskites are a
promising candidate to exceed this limitation. In this work, the considered PCMO films are grown on
SrTi0.998Nb0.002O3 substrates, subsequently referred to as STNO, an n-type material showing polaronic
transport behaviour [16]. Besides the advantageous characteristics for photovoltaic applications
mentioned above, due to typically high doping levels and the low mobility of polaronic charge
carriers, important physical properties of the resulting PCMO-STNO heterojunction such as the space
charge region width as well as the screening and excess carrier diffusion lengths are in the 1-10 nm
range [17] demanding for high-resolution technique to investigate the fundamental electric transport
phenomena.
Electron beam induced current (EBIC) is a widespread technique to map the electronic response of
rectifying devices stimulated by a local excitation with energetic electrons and can be easily
implemented in scanning electron microscopes (SEM) [18]. Despite the well-focussed incident beam,
resolutions in the nanometer range cannot be achieved, which is due to multiple scattering of electrons
in the material resulting in a pear-shaped spread of the energy deposited by the beam [19]. The
generation volume increases with increasing acceleration voltage, also referred to as the Bethe limit
[20]. Complementary approaches to decrease the latter have been made to decrease the samples’
dimension along the beam propagation direction resulting in transmission of the impinging electrons
before the beam spreads severely [21,22]. This idea is not only applicable to EBIC but also to
cathodoluminescence (CL) measurements [23] and in contrast to non-transmitting methods the
generation volume decreases with increasing acceleration voltage, also referred to as the Goldstein
limit [24].
In this work, we present how scanning transmission EBIC (STEBIC) can be applied to PCMO-
STNO heterojunctions in order to visualise enhanced recombination at misfit dislocations on the
nanometer scale.
2. Experiment
A 100 nm thick Pr0.66Ca0.34MnO3 film has been deposited on a commercial SrTi0.998Nb0.002O3
substrate by reactive Xe-ion beam sputtering with an oxygen partial pressure of 1.4∙10-4 mbar at 700°
C. Subsequently, the heterojunction has been annealed for 20 hours at 900° C in air and equipped with
a 200 nm thick Pt front contact layer via room temperature Xe-ion beam sputtering.
TEM lamellae have been prepared on DENSsolutions four contacts biasing chips suitable for the
Lightning D7+ holder inside an FEI Nova 600 NanoLab Dual Beam FIB and investigated with a
Philips CM200 UT (BF-TEM) and an FEI Titan80-300 (DF-STEM and EBIC) at 200 kV and 80
kV/300 kV, respectively. Figure 1 shows a DF-STEM image of a final TEM lamella after Ga-ion
milling following the recipe summarised in table 1 and reactive plasma cleaning for 2 min. Electron
energy loss spectroscopy confirms a quite uniform lamella thickness of roughly 40 nm.
In order to measure the EBIC signal, the DENSsolutions holder was connected to a Stanford
Research Systems SR570 current preamplifier, which converts the input current into an output voltage.
19th International Conference on Extended Defects in Semiconductors (EDS2018)
IOP Conf. Series: Journal of Physics: Conf. Series 1190 (2019) 012009
IOP Publishing
doi:10.1088/1742-6596/1190/1/012009
3
The latter was read out using a Keithley 2000 digital multimeter controlled by a homemade National
Instruments LabVIEW script. Synchronously, the electron beam position inside the TEM was adjusted
by a homemade Gatan DigitalMicrograph script communicating with the LabVIEW script via a shared
folder.
Figure 1. DF-STEM image of
the FIB-prepared lamella: The
enlargement of the small
rectangular box marked
illustrates the different layers
of the sample. Vertical cuts
have been made utilising low-
dose (30 pA) Ga-ion milling
with an acceleration voltage
of 30 kV in order to prevent
short circuits across the pn-
junction.
Acceleration
Voltage (kV)
Ga-ion beam
current (pA)
Tilt offset
(°)
Residual
thickness (nm)
30
300
1.2
800
30
100
1.2
300
5
29
2
<50
Table 1. Milling recipe for the TEM
lamellae preparation using Ga-ions. The
residual thickness values are rough
estimations obtained by the measurement
tool of the FIB software and should be
considered as a guide.
Figure 2. (a) BF-TEM image of the pn-junction revealing frequently occurring misfit dislocations
with (b) corresponding diffraction pattern. Only the systematic row of (010)STNO reflections is excited,
i.e. the interface is inclined by 20° with respect to the incident electron beam leading to the vertical
elongation of the dark contrasts.
(a)
(b)
19th International Conference on Extended Defects in Semiconductors (EDS2018)
IOP Conf. Series: Journal of Physics: Conf. Series 1190 (2019) 012009
IOP Publishing
doi:10.1088/1742-6596/1190/1/012009
4
3. Results and Discussion
Given the high spatial resolution of STEBIC, local variations of the junction’s short-circuit current can
be detected. Typical causes for such variations are perturbations of the crystal symmetry resulting in
enhanced recombination of excess charge carriers [25]. Due to the lattice misfit between film and
substrate of approximately f = -1.5% [26], defect formation to relax the corresponding stress are likely
to occur. In fact, in PCMO/STNO thin film couples various relaxation possibilities like oxygen
vacancy formation [27], twinning [28], and plastic relaxation due to dislocations [12] have been
reported. Figure 2 shows that misfit dislocations occur frequently under the growth conditions
described in the previous section. The mean distance between the observed dislocations of 33 nm is
only 27% larger than the characteristic dislocation distance d = a/f = 26 nm for total stress relaxation
via misfit dislocation formation (with a being the mean in-plane lattice constant of PCMO [29]).
Hence, a substantial fraction of the epitaxial stress is relaxed via this channel.
High-resolution DF-STEM images reveal that, despite having the same Burgers vector [010]STNO,
the dislocations have different characters: Figure 3(b) shows an interfacial dislocation (subsequently
referred to as ‘dislocation I’) with an added half plane in the film. Since the unit cell of PCMO is
larger than in STNO due to alternatingly tilted oxygen octahedra [30], the Burgers vector [010]STNO
corresponds to a partial dislocation in the PCMO which is probably accompanied by a stacking fault in
the tilt system. The related strain could induce chemical disorder, e.g. vacancies, leading to the dark
vertical contrast emerging at the interface. On the contrary, the core of the buried dislocation in figure
4(b) (subsequently referred to as ‘dislocation II’) lies approximately five unit cells below the junction
and the alternating tilting pattern of the PCMO film above seems to be undisturbed. The corresponding
EBIC maps in figure 3(a) and figure 4(a) show that the induced recombination behaviour is
significantly different.
Dislocation I leads to a small kink in the EBIC signal marked by the black circle but no medium
range effect. A straight forward explanation would be to assume that dislocation I does not introduce
deep states into the gap of PCMO. It should be kept in mind, however, that the screening length in
PCMO [31] as well as the excess carrier diffusion length [14] are in the nanometer range indicating
that the effect of deep states on the EBIC signal should be small. In contrast, dislocation II reduces the
EBIC signal by roughly 30% in the vicinity of the core as shown by the averaged horizontal line
Figure 3. (a) EBIC map of an interfacial edge dislocation and (b) the corresponding DF-STEM
micrograph. The dislocation is accompanied by a stacking fault in the PCMO layer due to its larger
unit cell and has little effect on the EBIC signal.
(a)
(b)
19th International Conference on Extended Defects in Semiconductors (EDS2018)
IOP Conf. Series: Journal of Physics: Conf. Series 1190 (2019) 012009
IOP Publishing
doi:10.1088/1742-6596/1190/1/012009
5
Figure 4. (a) EBIC map of a buried edge dislocation and (b) the corresponding DF-STEM
micrograph. The dislocation reduces the EBIC signal significantly. The black arrows in (a) indicate
spatial directions considered in figure 5.
Figure 5. EBIC profile taken along (a) the horizontally aligned arrows in figure 4(a) (averaged over
five lines) and (b) along the vertically aligned arrows in figure 4(a) (averaged over five lines); roman
numbers refer to arrows in figure 4(a). In addition, the cyan curve in (b) shows an EBIC profile taken
across the pn-heterojunction of a different lamella with 80 kV instead of 300 kV in a region without
any dislocations. Note that the EBIC magnitude for 80 kV is higher due to a different beam current
and inelastic scattering cross section.
profile (i) in figure 5(a). The apparently stronger recombination behaviour of dislocation II can be
partially explained by the slightly higher excess carrier diffusion length of STNO [14]. However, the
intersection of the profiles in figure 5(a) shows that the dislocation induced recombination below the
core is stronger than at its actual position, which is not the case for dislocation I. A possible
explanation for this behaviour is the interplay between the dislocation itself and the electric field of the
space charge region of the pn-heterojunction. The latter is supposed to be maximal at the chemical
interface and to decay monotonously resulting in a potentially smaller charge carrier separation and
hence an effectively larger influence of the dislocation induced recombination. The black and yellow
vertical profiles in figure 5(b) confirm indeed that the main part of the reduced EBIC signal is located
inside the space charge region expected to extend at least 10 nm into the STNO [14].
(a)
(b)
(a)
(b)
19th International Conference on Extended Defects in Semiconductors (EDS2018)
IOP Conf. Series: Journal of Physics: Conf. Series 1190 (2019) 012009
IOP Publishing
doi:10.1088/1742-6596/1190/1/012009
6
Finally, we would like to emphasise the reproducibility of the experiment and an experimental
evidence that beam spread effects of the electron probe resulting from (multiple) scattering can be
ruled out as the resolution-limiting effect in these experiments: The cyan curve in figure 5(b) shows
the EBIC profile across the pn-heterojunction taken from another lamella with an acceleration voltage
of 80 kV instead of 300 kV. Since any beam broadening effects are expected to strongly depend on the
acceleration voltage, the strong matching of the shape of the two measured profiles provides evidence
that beam spread is negligible as well as that different lamellae lead to identical results within
experimental error.
4. Summary and Conclusion
We have reported an experimental setup achieving a resolution of at least a few nanometers by means
of scanning transmission EBIC (STEBIC) investigations offering the possibility to map the
recombination and transport behaviour of transition metal oxide perovskites with relevant physical
quantities in the 1-10 nm range. The spatial extension of STEBIC signals is limited by sample
properties such as the excess carrier diffusion lengths of the materials rather than by the size of the
generation volume resulting from electron scattering. This is evidenced by showing that the shape of
the resulting STEBIC profiles is independent of the chosen acceleration voltage.
Applying this technique to misfit dislocations, we have shown that the position of the dislocation
core can have a tremendous impact on the resulting excess carrier recombination. In fact, in contrast to
interfacial dislocations, buried dislocations shifted inside the STNO show an enhanced EBIC decrease
and, in addition, a medium range recombination effect. The obvious effect of different bulk
recombination properties of the two materials on the dislocation recombination behaviour has been
discussed qualitatively. Nevertheless, the underlying mechanisms are not yet fully understood.
A sufficiently precise model has to describe the concerted action of (i) the presence of the pn-
junction’s space charge region, (ii) potential surface recombination at the two lamella surfaces, (iii)
additional electric field below the lamella surfaces as a result of Fermi level pinning, and (iv) the
formation of 'dead layers' due to sample preparation. The latter have been discussed to some extent in
[19] and related to experimental results obtained on the PCMO/STNO system and a silicon pn-
junction. Surface recombination in the absence of any electric field can be described in terms of a
Green’s function approach which has been evaluated numerically in [32] by placing a large number of
mirror charges in order to satisfy the boundary conditions at the surfaces resulting in a quantitative
description of EBIC signals obtained outside the charge collecting junction's space charge region.
Other models considering effective excess carrier diffusion lengths or a surface recombination factor
in the EBIC profile have been introduced in [33] and [34]. Furthermore, sophisticated and
quantitatively proven models of excess carrier recombination at dislocations [35] assume excitations in
the neutral bulk and cannot be readily adopted to the situation in our experiments. Hence, all existing
models are based on the assumption of exciting the sample outside the electric field of the space
charge region. A possibility to fill this gap is the employment of finite element simulations, which we
are currently working on.
Acknowledgements
We thank Arne Ahrens for valuable discussions. Funding by the Deutsche Forschungsgemeinschaft
(DFG) via the CRC 1073, project B02 and Z02, is gratefully acknowledged.
References
[1] van Santen J H and Jonker G H 1950 Physica 16 337
[2] Bednorz J G and Müller K A 1986 Z. Physik B - Condensed Matter 64 189
[3] von Helmolt R, Wecker J, Holzapfel B, Schultz L and Samwer K 1993 Phys. Rev. Lett. 71 2331
[4] Sawa A, Fujii T, Kawasaki M and Tokura Y 2005 Appl. Phys. Lett. 86 112508
[5] Röder H, Zang J and Bishop A R 1996 Phys. Rev. Lett. 76 1356
[6] Zener C 1951 Phys. Rev. 82 403
19th International Conference on Extended Defects in Semiconductors (EDS2018)
IOP Conf. Series: Journal of Physics: Conf. Series 1190 (2019) 012009
IOP Publishing
doi:10.1088/1742-6596/1190/1/012009
7
[7] Anderson P W and Hasegawa H 1955 Phys. Rev. 100 675
[8] Gennes P G 1960 Phys. Rev. 118 141
[9] Abdi-Jalebi M et al. 2018 Nature 555 497
[10] Mildner S, Hoffmann J, Blöchl P, Techert S and Jooss C 2015 Phys. Rev. B 92 035145
[11] Dagotto E 2013 Nanoscale Phase Separation and Colossal Magnetoresistance (Berlin: Springer
Science & Business Media)
[12] Yamada S, Arima T H, Ikeda H and Takita K 2000 Journal of the Physical Society of Japan 69
1278-1281
[13] Raiser D, Mildner S, Ifland B, Sotoudeh M, Blöchl P, Techert S and Jooss C 2017 Adv. Energy
Mater. 7 1602174
[14] Ifland B, Hoffmann J, Kressdorf B, Roddatis V, Seibt M and Jooss C 2017 New J. Phys. 19
063046
[15] Shockley W and Queisser H J 1961 Journal of Applied Physics 32 510
[16] Bi C Z, Ma J Y, Yan J, Fang X, Zhao B R, Yao D Z and Qiu X G 2006 Journal of Physics:
Condensed Matter 18 2553
[17] Ifland B, Peretzki P, Kressdorf B, Saring P, Kelling A, Seibt M and Jooss C 2015 Beilstein
Journal of Nanotechnology 6 1467
[18] Leamy H J 1982 Journal of Applied Physics 53 R51
[19] Donolato C 1981 phys. stat. sol. (a) 65 649
[20] Bethe H 1930 Annalen der Physik 4 325
[21] Cabanel C, Brouri D and Laval J 2006 The European Physical Journal 34 107
[22] Peretzki P, Ifland B, Jooss C and Seibt M 2017 Phys. Status Solidi RRL 11 1600358
[23] Chin A K, Temkin H and Roedel R J 1979 Appl. Phys. Lett. 34 476
[24] Goldstein J I 1977 Scanning Electron Microscopy 1977 315
[25] Falkenberg M A, Schuhmann H, Seibt M and Radisch V 2010 Review of Scientific Instruments
81 063705
[26] Ifland B, Hoffmann J, Kramer T, Scherff M, Mildner S and Jooss C 2015 J. Nanomaterials 3 1
[27] Chandrasena R U et al. 2017 Nano Lett. 17 794
[28] Hoffmann J, Moschkau P, Mildner S, Norpoth J, Jooss C, Wu L and Zhu Y 2014 Mater. Res.
Express 1 046403
[29] Jirak Z, Krupicka S, Simsa Z, Glouha M and Vratislav S 1985 Journal of Magnetism and
Magnetic Materials 53 153
[30] Glazer A M 1972 Acta Crystallographica Section B 28 3384
[31] Saucke G, Norpoth J, Jooss C, Su D and Zhu Y 2012 Phys. Rev. B 85 165315
[32] Tan C C, Ong V K and Radhakrishnan K 2013 Prog. Photovolt: Res. Appl. 21 986
[33] Donolato C 1983 Appl. Phys. Lett. 43 120
[34] Ong V K S, Phang J C H and Chan D S H 1994 Solid-State Electronics 37 1
[35] Kveder V V, Kittler M and Schröter W 2001 Phys. Rev. B. 63 115208
