Está en la página 1de 9

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/282904990

Relevance of non-equilibrium defect generation


processes to resistive switching in TiO2

Article in Journal of Applied Physics October 2015


DOI: 10.1063/1.4932225

CITATIONS READS

0 39

2 authors, including:

Samir Abdelouahed
The University of York
10 PUBLICATIONS 152 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Samir Abdelouahed
letting you access and read them immediately. Retrieved on: 31 October 2016
Relevance of non-equilibrium defect generation processes to resistive switching in
TiO2
Samir Abdelouahed and Keith P. McKenna

Citation: Journal of Applied Physics 118, 134103 (2015); doi: 10.1063/1.4932225


View online: http://dx.doi.org/10.1063/1.4932225
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/118/13?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in


Anomalous C-V response correlated to relaxation processes in TiO2 thin film based-metal-insulator-metal
capacitor: Effect of titanium and oxygen defects
J. Appl. Phys. 117, 154101 (2015); 10.1063/1.4917531

Highly uniform bipolar resistive switching characteristics in TiO2/BaTiO3/TiO2 multilayer


Appl. Phys. Lett. 103, 262903 (2013); 10.1063/1.4852695

Role of oxygen vacancies in TiO2-based resistive switches


J. Appl. Phys. 113, 033707 (2013); 10.1063/1.4779767

In-operando and non-destructive analysis of the resistive switching in the Ti/HfO2/TiN-based system by hard x-
ray photoelectron spectroscopy
Appl. Phys. Lett. 101, 143501 (2012); 10.1063/1.4756897

Understanding the intermediate initial state in TiO2/La2/3Sr1/3MnO3 stack-based bipolar resistive switching
devices
Appl. Phys. Lett. 99, 072113 (2011); 10.1063/1.3626597

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47
JOURNAL OF APPLIED PHYSICS 118, 134103 (2015)

Relevance of non-equilibrium defect generation processes to resistive


switching in TiO2
Samir Abdelouahed and Keith P. McKennaa)
Department of Physics, University of York, Heslington, York YO10 5DD, United Kingdom
(Received 24 July 2015; accepted 21 September 2015; published online 6 October 2015)
First principles calculations are employed to identify atomistic pathways for the generation of
vacancy-interstitial pair defects in TiO2. We find that the formation of both oxygen and titanium
defects induces a net dipole moment indicating that their formation can be assisted by an electric
field. We also show that the activation barrier to formation of an oxygen vacancy defect can be
reduced by trapping of holes which may be injected by the electrode. The calculated activation
energies suggest that generation of titanium defects is more favorable than generation oxygen
defects although activation energies in both cases are relatively high (>3.3 eV). These results pro-
vide much needed insight into an issue that has been widely debated but for which little definitive
experimental information is available. V C 2015 Author(s). All article content, except where

otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License.
[http://dx.doi.org/10.1063/1.4932225]

I. INTRODUCTION In this article, we provide insight into this fundamental


issue by investigating defect generation processes using first
The resistive switching effect in metal oxide films is
principles theoretical approaches. In particular, we consider
currently the focus of intense research owing to its potential
TiO2, which is a useful model system as it is both a candidate
as a low-power, high-speed, high-endurance, and non-
material for ReRAM and finds many other applications
volatile memory technology (ReRAM).15 First documented
where non-equilibrium defect generation under electrical
in the 1960s,69 the effect involves reversible switching
bias may be relevant (e.g., electrodes in rechargeable bat-
between high and low resistance states by application of teries, gas sensors, photocatalysts, and solar-cells3032). The
voltage pulses and has been observed in many oxide materi- equilibrium properties of a wide range of defects in TiO2
als, including TiO2, HfO2, SrTiO3, and Ta2O5.1015 Recent have been addressed theoretically using both semi-empirical
years have seen significant progress in understanding resis- and density functional theory based approaches.14,3350 Here,
tive switching; however, uncertainty still remains over the we determine the lowest energy pathway to creation of a va-
microscopic mechanisms responsible for the effect. There is cancy defect on both oxygen and titanium lattice sites, with
a consensus that for many oxide materials the low resistance the displaced ion occupying a neighboring interstitial site.
state involves an oxygen deficient conductive filament that The predicted activation barriers associated with generation
bridges the electrodes.4,1618 In most cases, as grown devices of the initial vacancy-interstitial defect pair are 7.3 eV (oxy-
are subjected to a forming step (in which a voltage stress is gen) and 3.3 eV (titanium). Furthermore, we show that the
applied) to establish this filament.11 Switching to the high re- activation barrier to formation of an oxygen vacancy defect
sistance state then involves partial reoxidation of the fila- can be reduced to 4.9 eV by trapping of holes which may be
ment.4,19 One key issue often debated is whether defect injected by the electrode. These relatively high activation
generation under electrical bias, i.e., the rupturing of bonds energies suggest that diffusion of pre-existing oxygen vacan-
leading to the creation of vacancy defects in the lattice, is cies (characterized by an activation energy of about 2.6 eV
important (particularly for forming). The generation of such (Refs. 5154)) is much more likely than defect generation in
defects, assisted by a non-equilibrium electric field and TiO2. However, both oxygen and titanium defect creation
charge injection, is often discussed in the literature and processes are associated with generation of a net dipole
included explicitly in some device simulations.11,2026 moment which may serve to reduce the activation barrier
However, resistive switching models are also developed con- under very high local electric fields (for example, as may be
sidering field driven diffusion of pre-existing defects only, present during the set operation). These results provide much
without defect generation.2729 Experimentally, definitive needed atomistic models of defect generation processes in a
evidence for one view over the other has proven extremely metal oxide material and provide key parameters needed to
challenging to obtain, and the uncertainty that remains assess the role they play in resistive switching.
presents an obstacle to addressing pressing technological The rest of this article in organized in the following way.
issues, such as controlling cycle-to-cycle and device-to-de- In Sec. II, we detail the first principles methods; we employ to
vice variability, increasing endurance, and improving predict the properties of defects in TiO2. In Sec. III, we pres-
reliability. ent our results, including the structure and formation energies
of intrinsic defects in TiO2, the structure and activation ener-
a)
Electronic mail: keith.mckenna@york.ac.uk gies associated with formation of vacancy-interstitial defect

0021-8979/2015/118(13)/134103/7 118, 134103-1 C Author(s) 2015


V

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47
134103-2 S. Abdelouahed and K. P. McKenna J. Appl. Phys. 118, 134103 (2015)

pairs, and the effect of charge trapping on defect formation. Some of the defects discussed in this article induce a net
Finally, in Sec. IV, we discuss the results and implications in dipole moment relative to the bulk crystal. We estimate this
detail before presenting our conclusions. dipole moment using the optimized coordinates from the
constrained minimization described above ropt i t and the
II. METHODS ionic charges obtained by Bader analysis of the correspond-
ing charge density64 Qi(t). The induced dipole moment is
The properties of defects in TiO2 are described using calculated as
density functional theory (DFT) and the projector augmented X opt
wave method as implemented in the VASP code.55,56 The pt ri tQi t  ropt
i 0Qi 0; (3)
Perdew, Burke, and Ernzerhof (PBE) approximation is i
employed for exchange and correlation,57 and electronic
where the sum is restricted over ions within a spherical
states are expanded in a plane wave basis with an energy cut- centered on the defect.
region of radius 6 A
off of 500 eV. To correct the self-interaction error associated
with localized electrons and holes, DFT U corrections III. RESULTS
implemented within the Dudarev scheme are applied on both
O 2p and Ti 3d states.58 We employ values of U that have A. Intrinsic defects
been shown to reproduce spectroscopic properties of TiO2 Before presenting results on defect formation, we sum-
defects and also ensure good linearity of the total energy marize the calculated structure and formation energies of the
with electron occupation as required by a self-interaction intrinsic vacancies and interstitial defects in TiO2, which can
free functional (UTi 4.20 eV and UO 5.25 eV).59 Using be considered as the building blocks of the vacancy-
this approach, the optimized lattice parameters of rutile TiO2 interstitial pair defects discussed in Sec. III B. We consider
(space group P42/mnm) are obtained as a b 4.670 A and
both Ti and O vacancies and intersitials representing the four

c 3.034 A and the single electron band gap is 2.4 eV. These key intrinsic defects in the TiO2 crystal. We perform a series
predicted lattice constants are within 2.5% of experiment of structural optimizations starting from different initial geo-
while the band gap is underestimated by about 0.6 eV.60,61 metries and spin densities in order to determine the most sta-
All defect calculations are performed using a 3  3  4 ble structure for each defect type and charge state. Defect
supercell containing 72 Ti atoms and 144 O atoms, and a formation energies are calculated using lO E(O2)/2 (i.e.,
2  2  2 Monkhorst-Pack grid is used for k-point sampling. half the energy of an oxygen molecule) corresponding to ox-
The stability of intrinsic defects are characterized by calcula- ygen rich conditions. Once lO is fixed, the chemical poten-
tion of the formation energy tial of Ti is also defined since lTiO2 lTi 2lO . We note
X that since the DFT U approach is parameterized, to
Ef fli g; EF Eqdef  Eideal Dni li qEF : (1) describe defects in TiO2 rather than the oxygen molecule for-
i
mation energies defined in this way may suffer from a sys-
In this equation, Eqdef is the total energy of the supercell con- tematic error. In principle, this could be corrected for by
taining a defect in relative charge state q and Eideal is the comparison to experiment; however, we do not attempt such
total energy of the ideal bulk supercell. Dni is the difference a correction here since the main focus of this study is on
between the number of atoms of species i in the defective defects which maintain the stoichiometry of TiO2 and so
and ideal supercells, li is the chemical potential of species i, their stability does not depend on the choice of the reference
and EF is the electron Fermi energy. Total energies are cor- chemical potential.
rected for potential alignment and image charge interactions The most stable structure of each defect is shown in Fig.
as described in Ref. 62. 1 with corresponding defect formation energies summarized
In order to estimate activation energies for defect forma- in Table I. The dependence of the formation energy on the
tion, we first perform a linear interpolation of atom coordi- electron Fermi energy for all intrinsic defect types is shown
nates between the ideal bulk supercell (rideal ) and the in Fig. 2. In the following, we discuss the properties of each
i
def
defective supercell (ri ), of the defects in detail with comparison to previous results
wherever available.
ri t trdef ideal
i 1  tri ; (2) The neutral oxygen vacancy (VO in Kroger-Vink nota-
tion) is predicted to have a formation energy of 4.0 eV with a
where t is a parameter ranging from 0 (ideal crystal) to 1 (de- triplet ground state (S 1) involving two electrons localized
fective). We then perform a constrained minimization where on neighboring Ti ions consistent with recent results
the diffusing ion forming the vacancy is held fixed but all obtained by electron paramagnetic resonance65 (Fig. 1(a)).
ions within a spherical shell of radius 6 A are fully opti- This result is also in agreement with numerous previous
mized. The minimized total energy of a set of configurations studies using a range of methods such as GGA U, HSE,
spanning range t 0  1 represents an upper bound to the and sX (with formation energies ranging from
minimum potential energy surface for defect formation. This 4.45.7 eV).4143,45,46 Also consistent with previous studies,
approach was adopted as attempts to employ the nudged we find that only the q 0 and q 2 charge states are sta-
elastic band method, which in general is a more rigorous ble across the entire range of Fermi energies, characteristic
approach,63 met issues with convergence. of a negative-U defect (Fig. 2). The charge transition level

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47
134103-3 S. Abdelouahed and K. P. McKenna J. Appl. Phys. 118, 134103 (2015)

FIG. 1. The predicted structure of intrinsic defects in rutile TiO2 (red and green spheres represent oxygen and titanium ions respectively). (a) Oxygen vacancy
(q 0). (b) Oxygen interstitial (q 0). (c) Titanium vacancy (q 0). (d) Titanium interstitial (q 4). For defects with unpaired electrons, spin density isosur-
faces are shown in blue.

(i.e., the Fermi energy for which the q 0 and q 2 have range of electron Fermi energies (Fig. 2). Again, the agree-
equal formation energy) is located 0.65 eV below the con- ment between these predictions and non-local screened
duction band minimum. This is consistent with recent results exchange functional results is extremely good.45
obtained using the non-local screened exchange functional.45 The neutral titanium vacancy (VTi) is predicted to have
The neutral oxygen interstitial (Oi) defect is predicted to a formation energy of 5.0 eV and adopts a S 2 spin configu-
adopt a dimer configuration by displacing one of the lattice ration with four holes localized on neighboring O ions (Fig.
oxygen ions. The defect can be considered as an O2 2 species 1(c)). We are not aware of any previous studies of the Ti va-
formed through the bonding of the neutral oxygen interstitial cancy in rutile TiO2, but the predictions are qualitatively
with a lattice oxygen ion which has a formal charge of 2 similar to results obtained for the same defect in anatase
(i.e., O0 O2 ! O2 2 ). This species which is not stable in TiO2.67 As the charge of the defect is decreased from q 0
the gas phase is stabilized in the TiO2 lattice by the electro- to q 4, each of the holes surrounding the Ti vacancy is
static environment and is characterized by an O-O bond removed one by one reducing the net spin from S 2 to
length of 1.49 A . This prediction is in good agreement with S 0. These charge transitions involve the sequential filling
previous studies;45 however, our exhaustive search for the of hole states predominantly localized on a single oxygen
most stable defect configuration has revealed two possible ion providing a good indication that the self-interaction error
neutral oxygen interstitial dimer configurations. One in in these calculations is small. The charge transition levels
which the axis of the O-O bond is aligned with respect to the are clustered around 0.5 eV above the valance band maxi-
local crystal symmetry (which we refer to as planar) and one mum (Fig. 2).
in which the dimer axis of the O-O bond is slightly rotated The titanium interstitial defect (Tii) is predicted to
(which we refer to as tilted). We find that these two configu- occupy an octahedral position (Fig. 1(d)) and can exist in ei-
rations have similar formation energies (within 70 meV) but ther a 3 or 4 charge state consistent with previous stud-
with the tilted dimer being the ground state with a formation ies.45 Further electrons added to the defect localize on
energy of 2.4 eV (Fig. 1(b)). We estimate that the barrier to
rotate the dimer between the planar to the tilted orientation is
around 100 meV, suggesting both configurations may be
readily accessible under typical conditions. The oxygen in-
terstitial is able to trap an additional one or two electrons
(leading to a corresponding increase in O-O bond length to
1.90 and 2.15 A ), consistent with previous studies.46,66 For
the negatively charged oxygen interstitial defects, only the
planar dimer structure is found to be stable. Similar to the
case for the oxygen vacancy, only the q 0 and q 2
charge states are thermodynamically favored over the full

TABLE I. The formation energies of intrinsic defects in rutile TiO2 calcu-


lated using Eq. (1). Values are given in eV for oxygen rich conditions
(lO E(O2)/2) and EF EVBM (where EVBM is the energy of the valence
band maximum in bulk TiO2).

Defect q Ef(E(O2)/2, EVBM)

VO 0/1/2 4.02/2.47/0.52
Oi (tilted) 0 2.36
FIG. 2. The calculated formation energy of intrinsic defects in rutile TiO2
Oi (planar) 0/1/2 2.43/4.26/5.55 under oxygen rich conditions (lO E(O2)/2). For each type of defect, only
VTi 0/1/2/3/4 5.03/5.57/5.83/6.37/6.92 the most stable charge state at any given electron Fermi energy is shown
Tii 0/1/2/3/4 6.83/4.76/2.99/1.26/0.18 (the charge is also indicated on the figure). The filled circles indicate the
charge transition levels.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47
134103-4 S. Abdelouahed and K. P. McKenna J. Appl. Phys. 118, 134103 (2015)

nearby Ti lattice sites and so are better described as weakly formation. To address this, we consider the stability of meta-
bound electron polarons rather than true defect charge states. stable configurations involving closely separated vacancy
The formation energies and charge transition levels predicted and interstitial defects and characterize the activation ener-
are again in good agreement with previous calculations45 gies associated with defect generation. First, we perform a
(Fig. 2). systematic search of possible vacancy-interstitial configura-
The results presented in this section demonstrate that the tions in order to determine the metastable configurations
DFT U method we employ is able to describe the proper- with the lowest formation energy. For oxygen Frenkel
ties of all four types of intrinsic defect in rutile TiO2 in very defects, we consider as initial geometries the optimized
reasonable agreement with previous calculations employing structures of the q 0, q  1, and q 2 oxygen intersti-
higher level methods such as non-local hybrid functionals. tial with an oxygen vacancy created in lattice sites up to the
Importantly, the DFT U method is much less computation- seventh nearest neighbor. For the titanium Frenkel defects,
ally expensive than hybrid functional approaches, allowing we consider the optimized structure of the q 4 and
us to perform calculations on relatively large supercells, q 3 titanium interstitial with Ti vacancies created in lat-
which is important for modeling the defect formation proc- tice sites up to the second nearest neighbor. Following struc-
esses discussed in Sec. III B. tural optimization, many of these initial geometries relax
to the bulk crystal structure, indicating the lack of any nearby
B. Frenkel defect formation potential energy minimum; however, a number of metastable
With the structure of intrinsic defects in rutile TiO2 configurations are obtained.
determined, we now turn to address the possibility of defect Following our search, we find that the most stable oxy-
formation in the perfect lattice. In particular, we consider gen Frenkel defect involves a q 2 oxygen interstitial and
processes where an atom is displaced from its lattice site into a q 2 oxygen vacancy located at the fourth nearest neigh-
a neighboring interstitial site (i.e., the creation of a Frenkel bor lattice site (Ef 5.97 eV). The most stable titanium
defect). Since vacancy and interstitial defects are stable in a Frenkel defect consists of a q 4 interstitial and a q 4
number of charge states in rutile TiO2, there are different vacancy at the neighboring lattice site (Ef 3.15 eV). The
possibilities for the Frenkel defect formation which can con- structures of both defects are shown in Fig. 3. The fact that
veniently be expressed in Kroger-Vink notation as follows: the most stable defects involve vacancies and interstitials
with the highest charge is consistent with the result obtained
OXO ! VO  O00i 6:1 eV; (4) for isolated defects above. However, the formation energies
are in both cases lower as a result of mutual electrostatic and
OXO ! VO  O0i 6:7 eV; (5) elastic interactions between the vacancy and interstitial. The
OXO ! VXO  OXi 6:4 eV; (6)
0000
TiXTi ! V
Ti  Tii 6:7 eV; (7)
000
TiXTi ! V
Ti  Tii 7:6 eV: (8)

We note that since formation of a Frenkel defect involves no


net change in the number of ions or electrons in the system,
the corresponding formation energy does not depend on the
electron or oxygen chemical potential. The numbers given in
the square brackets above are the Frenkel defect formation
energies for the reactions calculated by summing the forma-
tion energies of the individual intrinsic defects (Table I).
Calculated in this way, these formation energies correspond
to the formation of vacancy and interstitial pairs that are suf-
ficiently separated that there is negligible interaction
between them. For both oxygen and titanium, the most stable
Frenkel defect is comprised of the intrinsic defects with the
highest charge, jq 2j in the case of oxygen (Ef 6.1 eV)
and jq 4j in the case of titanium (Ef 6.7 eV). These
results are comparable to previous results obtained using the
PBE functional.68 Importantly, in the presence of a non-
equilibrium electric field, the stability of Frenkel defects
comprised of oppositely charged interstitial and vacancy
defects would be enhanced and could provide a driving force
for defect generation in films under electrical bias. FIG. 3. (a) Structure of the most stable oxygen Frenkel defect and (b) the
most stable titanium Frenkel defect in rutile TiO2. Red and green spheres
While the calculations above are helpful for identifying
represent oxygen and titanium ions, respectively. The ion which is displaced
the most likely Frenkel defects to form in TiO2, they do not out of its lattice site in highlighted in blue. The arrows indicate the displace-
allow us to say, anything about the mechanism of defect ment of ions associated with defect formation.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47
134103-5 S. Abdelouahed and K. P. McKenna J. Appl. Phys. 118, 134103 (2015)

binding of the interstitial to the vacancy is small for oxygen C. Role of charge trapping
but high for the titanium defect due to the smaller distance
versus 3.85 A ). We now extend the calculations above to consider
between the vacancy and interstitial (1.82 A
whether the trapping of electrons or holes which may be
With the most stable oxygen and titanium Frenkel
injected under a non-equilibrium bias can assist the forma-
defect configurations identified, we calculate the activation
tion of defects. This is an issue that has seen considerable
barriers associated with the initial stage of defect formation
using the constrained minimization procedure outlined in speculation, but there is little quantitative known. Modeling
Sec. II. Fig. 4 shows the potential energy surface for both the non-equilibrium injection and trapping of charge directly
defects from which we estimate activation energies of 7.3 is an extremely challenging problem requiring approaches
and 3.3 eV for oxygen and titanium, respectively. As noted that can bridge the timescales associated with electron trans-
above, the binding energy of the titanium interstitial to the fer and ionic motion.69 However, to provide some insight,
vacancy is higher than for oxygen and further separation of we consider a simpler model where we assess the formation
the vacancy and interstitial requires overcoming an addi- energies associated with creating Frenkel defects in the pres-
tional activation barrier of 3.4 eV. To assess the possible ence of an additional electron or hole. In particular, in
role of a non-equilibrium electric field in assisting the Kroger-Vink notation
defect formation process, we calculate the corresponding
induced dipole moment jpj (Fig. 4). The dipole moment OXO e ! Oi VO ; (9)
increases in a monotonous fashion to about 5.3 eA and 4.0
OXO h ! Oi VO ; (10)

eA for the oxygen and titanium defect, respectively. In the
presence of an electric field, these induced dipole moments TiXTi e ! Tii VTi ; (11)
will reduce both the defect formation energy and barrier to
TiXTi h ! Tii VTi : (12)
formation. For example, for both the O and Ti Frenkel
defects shown in Fig. 4, the dipole moment close to the On the left hand side of the reactions above, the electron ini-
transition state is about 3 eA , and for electric fields of 1, 5,
tially occupies a state at the conduction band minimum and
and 10 MV/cm, the expected reductions in the activation the hole occupies a state at the valence band maximum.
energy (E.p) are 0.03, 0.15, and 0.30 eV, respectively. We As in Sec. III B, for each of the reactions above, we
return to discuss this point in more detail in Sec. IV. have identified the most stable defect involving a vacancy
and interstitial at close separation. For titanium, we find little
reduction in defect formation energy by addition of electrons
or holes. However, for oxygen defects, addition of a hole
leads to a significant reduction in formation energy. In par-
ticular, we identify the following reaction as the most stable:

OXO h ! OXI VO ; (13)

with a formation energy of 4.70 eV. The defect is qualita-


tively different to the one formed in the absence of a hole as
it involves a neutral oxygen interstitial and a positively
charged vacancy (Fig. 5(a)). The much lower formation
energy (4.70 eV compared to 5.97 eV) can be understood in
terms of the energy gained by localization of the hole.
The potential energy surface associated with defect for-
mation is calculated in the same was as described previously
(Fig. 5(b)). The estimated activation energy is 4.9 eV, con-
siderably lower than the 7.3 eV obtained without the hole
present. This is explained by the fact that in the latter case the
displaced oxygen carries a 2 charge and therefore experien-
ces stronger electrostatic repulsion with surrounding lattice ox-
ygen ions during the transition, whereas in the former case, it is
neutral. Again, we also calculate the corresponding induced
dipole moment jpj, which reaches a maximum of 2.3 eA .

IV. DISCUSSION AND CONCLUSIONS


We start with a discussion of some of the factors which
may affect the accuracy of the calculations presented in this
article. The approximations used in the DFT implementation,
FIG. 4. (a) Potential energy surface associated with generation of an oxygen
Frenkel defect and (b) a titanium Frenkel defect in rutile TiO2. The corre- in particular, the use the DFT U approach, may leave
sponding induced dipole moment jpj is also shown. residual self-interaction error which will affect the formation

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47
134103-6 S. Abdelouahed and K. P. McKenna J. Appl. Phys. 118, 134103 (2015)

therefore, under an electric field, effective activation ener-


gies may be reduced. Considering the electric fields that may
realistically exist in devices (130 MVcm1), this field effect
can only offer a reduction of the order 1 eV at most. After
activation, the binding between the vacancy and interstitial is
relatively weak in the case of oxygen, and so the charged in-
terstitial would be free to diffuse away driven by the electric
field. For the titanium defect, the binding is stronger and a
subsequent barrier of 3.4 eV must be overcome before the Ti
ion is free to diffuse away. Typical activation energies asso-
ciated with oxygen vacancy diffusion in TiO2 are of the
order 2.6 eV.5154 Even considering the effect of hole trap-
ping which can reduce the activation energy to oxygen
vacancy generation, these results suggest that diffusion of
pre-existing oxygen vacancies is always much more facile
than defect generation in bulk TiO2. However, activation
energies near the electrode interface7072 or at grain bounda-
ries12,73,74 may be considerably lower. Of the defect genera-
tion processes considered, Ti Frenkel formation appears as
the most favorable. This is especially true in the presence of
a high local electric field, suggesting this process may play a
role in the set process where a high field may exist between
the filament apex and the electrode.
One of the reasons the results presented in this article
are useful is that there is little experimental information
available on defect generation process which may take place
under electrical bias in TiO2 and oxides more generally. On
the other hand, this makes direct validation of the predicted
FIG. 5. (a) Structure of the oxygen Frenkel defect formed in the presence of defect creation pathways and activation energies extremely
a hole. Red and green spheres represent oxygen and titanium ions, respec-
tively. The ion which is displaced out of its lattice site in highlighted in blue.
difficult. One of the few examples where an activation
The arrows indicate the displacement of ions associated with defect forma- energy for Frenkel defect generation has been extracted
tion. (b) Potential energy surface associated with defect generation. The cor- experimentally is in a study of forming in HfO2 films.11 In
responding induced dipole moment jpj is also shown. this work, the temperature dependence of the forming pro-
cess was investigated and fitted to a trap-assisted-tunneling
energy of defects and the nature of charge localization. The simulation including field enhanced Frenkel defect genera-
band gap is also underestimated using this approach. tion. An activation energy attributed to oxygen Frenkel
However, the properties of intrinsic defects calculated in Sec. defect formation of 4.4 eV was obtained. Although HfO2 is a
III A are in very good agreement with a range of alternative quite different material with a much wider gap, we note that
self-interaction corrected methods suggesting the approxima- this activation energy is broadly similar to the energies
tions employed are reasonable. Another potential source of obtained for TiO2 (4.9 eV and 7.3 eV).
error is the use of periodic boundary conditions which leads In summary, we have performed a first principles theoret-
to artificial interactions between the defect and its periodic ical investigation into non-equilibrium defect generation proc-
images. To limit these effects, we have used a relatively large esses in TiO2. We have identified atomistic pathways for the
supercell (216 atoms) and the high dielectric constant of rutile initial step in the generation of oxygen and titanium Frenkel
TiO2 means electrostatic interactions are screened effectively. defects and have characterized the corresponding activation
Finally, another possible source of error is the constrained energies. Our key conclusions are that the formation of both
minimization approach employed to characterize the potential oxygen and titanium Frenkel defects induces a net dipole
energy surface associated with defect formation. This moment, and therefore, their formation can be assisted by an
approach provides a reliable upper estimate of the activation electric field. In the case of oxygen, we also find that trapping
barrier but cannot exclude lower energy pathways involving of holes can further reduce the barrier to defect formation.
more complex ionic displacements (although we were unable However, the activation energies remain relatively high. The
to find any after a thorough search). lower barrier to formation of titanium defects and the larger
We now turn to discussion of some of the potential associated dipole moment suggests such processes may be rel-
implications of our results. Considering the possibility of evant in very high local electric fields (for example, as may be
electric field assisted defect generation in TiO2, we have present during a set operation). These results provide much
identified that the initial step in the formation of a charged needed atomistic models of defect generation processes in
vacancy and interstitial pair is associated with activation bar- metal oxide materials and provide key parameters needed to
riers of 7.3 eV (oxygen) and 3.3 eV (titanium). The dipole assess the role they play in resistive switching.28 More gener-
moment associated with these defects is about 4 to 5 eA ; ally, the results are relevant to a far wider range of TiO2

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47
134103-7 S. Abdelouahed and K. P. McKenna J. Appl. Phys. 118, 134103 (2015)

28
applications where non-equilibrium electric field driven defect D. Li, M. Li, F. Zahid, J. Wang, and H. Guo, J. Appl. Phys. 112, 073512
generation may contribute to material degradation, for exam- (2012).
29
Y. Guo and J. Robertson, Microelectron. Eng. 147, 339 (2015).
ple, in photocatalysts, dye-sensitized solar cells and electrodes 30
A. Fujishima and K. Honda, Nature 238, 37 (1972).
in rechargeable batteries.3032 31
B. ORegan and M. Gratzel, Nature 353, 737 (1991).
32
K. Schierbaum, U. Kirner, J. Geiger, and W. G opel, Sens. Actuators B 4,
ACKNOWLEDGMENTS 87 (1991).
33
N. Yu and J. W. Halley, Phys. Rev. B 51, 4768 (1995).
34
K.P.M. acknowledges support from EPSRC (EP/ J. Chen, L.-B. Lin, and F.-Q. Jing, J. Phys. Chem. Solids 62, 1257 (2001).
35
E. Cho, S. Han, H.-S. Ahn, K.-R. Lee, S. K. Kim, and C. S. Hwang, Phys.
K003151) and COST Action CM1104. This work made use
Rev. B 73, 193202 (2006).
of the facilities of Archer, the UKs national high- 36
N. A. Deskins and M. Dupuis, Phys. Rev. B 75, 195212 (2007).
37
performance computing service, via our membership in the M. M. Islam, T. Bredow, and A. Gerson, Phys. Rev. B 76, 045217 (2007).
38
UK HPC Materials Chemistry Consortium, which is funded E. Finazzi, C. Di Valentin, G. Pacchioni, and A. Selloni, J. Chem. Phys.
129, 154113 (2008).
by EPSRC (EP/L000202). All data created during this 39
B. J. Morgan and G. W. Watson, J. Phys. Chem. C 113, 7322 (2009).
research are available by request from the University of York 40
S.-G. Park, B. Magyari-K ope, and Y. Nishi, Phys. Rev. B 82, 115109
Research database http://dx.doi.org/10.15124/0bede454-b78e- (2010).
41
474c-8d93-38b9911700f5. B. J. Morgan and G. W. Watson, J. Phys. Chem. C 114, 2321 (2010).
42
A. Janotti, J. B. Varley, P. Rinke, N. Umezawa, G. Kresse, and C. G. Van
1 de Walle, Phys. Rev. B 81, 085212 (2010).
R. Waser and M. Aono, Nat. Mater. 6, 833 (2007). 43
2 G. Mattioli, P. Alippi, F. Filippone, R. Caminiti, and A. Amore Bonapasta,
A. Sawa, Mater. Today 11, 28 (2008).
3 J. Phys. Chem. C 114, 21694 (2010).
D.-H. Kwon, K. M. Kim, J. H. Jang, J. M. Jeon, M. H. Lee, G. H. Kim, 44
P. Deak, B. Aradi, and T. Frauenheim, Phys. Rev. B 86, 195206 (2012).
X.-S. Li, G.-S. Park, B. Lee, S. Han, M. Kim, and C. S. Hwang, Nat. 45
H.-Y. Lee, S. J. Clark, and J. Robertson, Phys. Rev. B 86, 075209 (2012).
Nanotechnol. 5, 148 (2010). 46
4 T. S. Bjrheim, A. Kuwabara, and T. Norby, J. Phys. Chem. C 117, 5919
H. Akinaga and H. Shima, Proc. IEEE 98, 2237 (2010).
5 (2013).
F. Pan, S. Gao, C. Chen, C. Song, and F. Zeng, Mat. Sci. Eng. R 83, 1 47
A. Janotti, C. Franchini, J. B. Varley, G. Kresse, and C. G. de Walle, Phys.
(2014).
6 Status Solidi-RRL 7, 199 (2013).
T. W. Hickmott, J. Appl. Phys. 33, 2669 (1962). 48
7 M. Setvin, C. Franchini, X. Hao, M. Schmid, A. Janotti, M. Kaltak, C. G.
J. Gibbons and W. Beadle, Solid State Electron. 7, 785 (1964).
8 Van de Walle, G. Kresse, and U. Diebold, Phys. Rev. Lett. 113, 086402
F. Argall, Solid State Electron. 11, 535 (1968).
9
G. Dearnaley, A. M. Stoneham, and D. V. Morgan, Rep. Prog. Phys. 33, (2014).
49
1129 (1970). A. Malashevich, M. Jain, and S. G. Louie, Phys. Rev. B 89, 075205
10
M. D. Pickett, D. B. Strukov, J. L. Borghetti, J. J. Yang, G. S. Snider, D. (2014).
50
R. Stewart, and R. S. Williams, J. Appl. Phys. 106, 074508 (2009). C. Lin, D. Shin, and A. A. Demkov, J. Appl. Phys. 117, 225703 (2015).
51
11
G. Bersuker, D. C. Gilmer, D. Veksler, P. Kirsch, L. Vandelli, A. M. Arita, M. Hosoya, M. Kobayashi, and M. Someno, J. Am. Ceram. Soc.
Padovani, L. Larcher, K. P. McKenna, A. Shluger, V. Iglesias, M. Porti, 62, 443 (1979).
52
and M. Nafra, J. Appl. Phys. 110, 124518 (2011). D. Derry, D. Lees, and J. Calvert, J. Phys. Chem. Solids 42, 57 (1981).
53
12
K. Szot, W. Speier, G. Bihlmayer, and R. Waser, Nat. Mater. 5, 312 (2006). D.-K. Lee and H.-I. Yoo, Solid State Ionics 177, 1 (2006).
54
13
D. Choi, D. Lee, H. Sim, M. Chang, and H. Hwang, Appl. Phys. Lett. 88, A. G. Hollister, P. Gorai, and E. G. Seebauer, Appl. Phys. Lett. 102,
082904 (2006). 231601 (2013).
55
14
K. Yang, Y. Dai, B. Huang, and Y. P. Feng, Phys. Rev. B 81, 033202 (2010). G. Kresse and J. Furthm uller, Phys. Rev. B. 54, 11169 (1996).
56
15
A. C. Torrezan, J. P. Strachan, G. Medeiros-Ribeiro, and R. S. Williams, G. Kresse and J. Furthm uller, Comput. Mater. Sci. 6, 15 (1996).
57
Nanotechnology 22, 485203 (2011). J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865
16
L. Liborio and N. Harrison, Phys. Rev. B 77, 104104 (2008). (1996).
58
17
J. P. Strachan, G. Medeiros-Ribeiro, J. J. Yang, M.-X. Zhang, F. Miao, I. S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, and A. P.
Goldfarb, M. Holt, V. Rose, and R. S. Williams, Appl. Phys. Lett. 98, Sutton, Phys. Rev. B 57, 1505 (1998).
59
242114 (2011). B. Morgan and G. Watson, Phys. Rev. B 80, 233102 (2009).
60
18
K. Kamiya, M. Y. Yang, S.-G. Park, B. Magyari-K ope, Y. Nishi, M. D. C. Cronemeyer, Phys. Rev. 87, 876 (1952).
61
Niwa, and K. Shiraishi, Appl. Phys. Lett. 100, 073502 (2012). Z.-W. Lu, D. Singh, and H. Krakauer, Phys. Rev. B 51, 6842 (1995).
62
19
D. Ielmini, F. Nardi, and C. Cagli, Nanotechnology 22, 254022 (2011). S. Lany and A. Zunger, Model. Simul. Mater. Sci. Eng. 17, 084002
20
B. J. Choi, D. S. Jeong, S. K. Kim, C. Rohde, S. Choi, J. H. Oh, H. J. Kim, (2009).
63
C. S. Hwang, K. Szot, R. Waser, B. Reichenberg, and S. Tiedke, J. Appl. G. Henkelman and H. J onsson, J. Chem. Phys. 113, 9978 (2000).
64
Phys. 98, 033715 (2005). G. Henkelman, A. Arnaldsson, and H. J onsson, Comput. Mater. Sci. 36,
21
J. Yang, F. Miao, M. Pickett, D. A. A. Ohlberg, D. R. Stewart, C. N. Lau, 354 (2006).
65
and R. S. Williams, Nanotechnology 20, 215201 (2009). A. T. Brant, E. M. Golden, N. C. Giles, S. Yang, M. A. R. Sarker, S.
22 Watauchi, M. Nagao, I. Tanaka, D. A. Tryk, A. Manivannan, and L. E.
L. Goux, P. Czarnecki, Y. Y. Chen, L. Pantisano, X. P. Wang, R.
Degraeve, B. Govoreanu, M. Jurczak, D. J. Wouters, and L. Altimime, Halliburton, Phys. Rev. B 89, 115206 (2014).
66
Appl. Phys. Lett. 97, 243509 (2010). H. Kamisaka and K. Yamashita, J. Phys. Chem. C 115, 8265 (2011).
23 67
P. Huang, X. Liu, B. Chen, H. Li, and Y. Wang, IEEE Trans. Electron S. Na-Phattalung, M. Smith, K. Kim, M.-H. Du, S.-H. Wei, S. Zhang, and
Devices 60, 4090 (2013). S. Limpijumnong, Phys. Rev. B 73, 125205 (2006).
24 68
K.-H. Xue, B. Traore, P. Blaise, L. R. C. Fonseca, E. Vianello, G. Molas, J. He, R. Behera, M. Finnis, X. Li, E. Dickey, S. Phillpot, and S. Sinnott,
B. De Salvo, G. Ghibaudo, B. Magyari-K ope, and Y. Nishi, IEEE Trans. Acta Mater. 55, 4325 (2007).
69
Electron Devices 61, 1394 (2014). L. G. C. Rego and V. S. Batista, J. Am. Chem. Soc. 125, 7989 (2003).
25 70
B. Traore, P. Blaise, E. Vianello, E. Jalaguier, G. Molas, J. F. Nodin, L. O. Sharia, K. Tse, J. Robertson, and A. A. Demkov, Phys. Rev. B 79,
Perniola, B. D. Salvo, and Y. Nishi, in International Reliability Physics 125305 (2009).
71
Symposium Proceedings (IEEE, 2014), p. 5E.2.1. D. Jeong, H. Schroeder, and R. Waser, Phys. Rev. B 79, 195317 (2009).
26 72
R. Ottking, S. Kupke, E. Nadimi, R. Leitsmann, F. Lazarevic, P. Planitz, S. R. Bradley, K. P. McKenna, and A. L. Shluger, Microelectron. Eng.
G. Roll, S. Slesazeck, M. Trentzsch, and T. Mikolajick, Phys. Status Solidi 109, 346 (2013).
73
212, 547 (2015). K. P. McKenna, A. Shluger, V. Iglesias, M. Porti, M. Nafra, M. Lanza,
27
S. Clima, Y. Y. Chen, R. Degraeve, M. Mees, K. Sankaran, B. Govoreanu, and G. Bersuker, Microelectron. Eng. 88, 1272 (2011).
74
M. Jurczak, S. De Gendt, and G. Pourtois, Appl. Phys. Lett. 100, 133102 S. K. Wallace and K. P. McKenna, Adv. Mater. Interfaces 1, 1400078
(2012). (2014).

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
54.152.109.166 On: Sat, 02 Jan 2016 01:45:47

También podría gustarte