Professional Documents
Culture Documents
The efficiency of fuel cells and metal–air batteries is significantly limited by the activation of oxygen reduction and
evolution reactions. Despite the well-recognized role of oxygen reaction kinetics on the viability of energy technologies,
the governing mechanisms remain elusive and until now have been addressable only by macroscopic studies. This lack of
nanoscale understanding precludes optimization of material architecture. Here, we report direct measurements of oxygen
reduction/evolution reactions and oxygen vacancy diffusion on oxygen-ion conductive solid surfaces with sub-10 nm
resolution. In electrochemical strain microscopy, the biased scanning probe microscopy tip acts as a moving,
electrocatalytically active probe exploring local electrochemical activity. The probe concentrates an electric field in a
nanometre-scale volume of material, and bias-induced, picometre-level surface displacements provide information on local
electrochemical processes. Systematic mapping of oxygen activity on bare and platinum-functionalized yttria-stabilized
zirconia surfaces is demonstrated. This approach allows direct visualization of the oxygen reduction/evolution reaction
activation process at the triple-phase boundary, and can be extended to a broad spectrum of oxygen-conductive and
electrocatalytic materials.
T
he broad implementation of electric and hybrid vehicle tech- studies, which have already demonstrated their potential in energy
nology, development of distributed energy sources, and storage10–12 and catalytic systems13.
optimization of mobile electronic devices requires a significant Here, we demonstrate that ORR/OER phenomena on oxygen-
increase in the energy and power densities of batteries and fuel cells, conductive surfaces can be probed and mapped on the scale of
ideally approaching that of hydrocarbon fuels1,2. The use of atmos- several nanometres, well below the limit of microcontact measure-
pheric oxygen as an active component of the cathode in fuel cells3 ments14,15, allowing for direct identification of local electrochemical
and lithium–air batteries4,5 directly addresses this challenge by reactivity and providing insight into local kinetic parameters. This
making use of the oxidizer present in the atmosphere. A compara- scanning probe microscopy (SPM)-based approach is based on
tive analysis of the individual contributions to the overall perform- the detection of electrochemical strain associated with ionic and
ance of fuel cells and related systems suggests that one of the most vacancy movement during bias-induced ORR/OER in a process
significant limiting factors is the activation of oxygen reduction called electrochemical strain microscopy (ESM). The SPM tip con-
and evolution reactions (ORR and OER)6. Similarly, large charge– centrates an electric field in a nanoscale volume of material, result-
discharge hysteresis (and therefore low efficiencies and lifetimes) ing in the injection/annihilation of oxygen vacancies, and
in lithium–air systems is largely related to ORR/OER activation7,8. subsequent vacancy transport and migration induced by the bias,
Classical solutions for improving ORR kinetics include the use of as illustrated in Fig. 1. The vacancy movement results in localized
electrocatalytic materials, including (most prominently) platinum strain under the tip, detected through dynamic surface displace-
and its alloys3, and some transition-metal oxides2,3. The need for ment. This strain-detection approach enables volumes to be
an expensive electrocatalyst is perhaps the most important factor probed that are 1 × 106 to 1 × 108 times smaller than those accessi-
limiting the large-scale introduction of fuel-cell systems, and is ble using current-based electrochemical methods16–18.
therefore driving a huge R&D effort in the United States and world-
wide3. Despite the well-recognized role of ORR/OER kinetics on the Dynamic ESM of ORR/OER
viability of energy technologies, the underpinning mechanisms ESM probing of ORR/OER reactivity is illustrated in Fig. 1a. The
remain elusive and until now have been amenable only to macro- SPM tip is brought into contact with the surface, and application
scopic studies. The fundamental reason for this dearth of infor- of an electric bias shifts the electrochemical potential of mobile
mation is the complex geometry of active regions such as the oxygen vacancies at the tip–surface junction compared to the
three-phase junction between the gas phase, catalyst/cathode and bulk. At sufficiently high probe bias, the potential drop at the junc-
electrolyte9, which hinders both structure and property studies tion activates the ORR/OER process, resulting in the generation or
and the development of macroscopic model systems. On the meso- annihilation of vacancies depending on the sign of the bias. The
scopic level, this complexity largely precludes phenomenological vacancies diffuse and migrate through the material under the com-
optimization of the material architecture, because the target micro- bined effect of the electric field and concentration gradient. The
structural parameters defining cathode functionality are unknown. associated changes in molar volume19 result in electrochemical
On the atomic level, this lack of structural information limits the strains, and associated dynamic surface deformation is detected by
potential impact of first-principle density functional theory a scanning probe microscope (SPM) at the 2–5 pm level. Note
1
The Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA, 2 Heidelberg Graduate School of
Mathematical and Computational Methods for the Sciences, Institut für Angewandte Mathematik, Ruprecht-Karls-Universität Heidelberg, Im Neuenheimer
Feld 293, 69120 Heidelberg, Germany, 3 Institute of Semiconductor Physics, National Academy of Science of Ukraine, 41, pr. Nauki, 03028 Kiev, Ukraine.
* e-mail: ka7@ornl.gov; sergei2@ornl.gov
–20
b
262 40
Resonance (kHz)
response (a.u.)
264
Cantilever
266 20
268
270 0
c
10
ESM response (a.u.)
R1(V) (a.u.)
6 6
0
5 4
–2
4 2
3 –4 0
–30 –20 –10 0 10 20 30 –30 –20 –10 0 10 20 30 20 10 0 –10 –20 –30
Applied voltage (V) Applied voltage (V) Applied voltage (V)
Figure 2 | Dynamic ESM—separating local kinetics from thermodynamics. a, Tip bias waveform in a D-ESM measurement. Each voltage pulse is followed
by train of band excitation waveforms to detect the relaxing electrochemical strain. b, 2D spectrogram of D-ESM response as a function of frequency and
time. Only the data in the bias-off states are shown. c, Time dependence of the ESM amplitude response averaged over frequency space. The relaxation after
each bias pulse is clearly seen. The colour code in c corresponds to the delay time after the pulse. d,e, The response after each pulse is fit to a relaxation
model and the bias dependence of R0(V) and R1(V) are shown in d and e, respectively. f, ESM loops as a function of time delay illustrating phenomenological
relaxation dynamics. Colour scale as in c.
in Fig. 1. For low d.c. tip biases, the potential drop in the tip– are then tD ¼ L 2/D(T ) ≈ 0.13 s, comparable with those observed
surface junction is small, and the local surface overpotential is experimentally. The nucleation biases on the hysteresis loops are
insufficient to activate ORR/OER. The bias-induced vacancy then identified with the critical biases required to activate
motion is limited to redistribution of the oxygen vacancies in ORR/OER. Note that the potential drop is divided between the
the volume of material below the probe, and is responsible for gen- tip–surface junction and the material, so the actual overpotential
erating the ESM response signal. However, the hysteresis loop is (or local polarization) is significantly smaller than the tip bias.
closed because d.c. bias pulses do not change the total number Finally, the hysteresis loop shape is controlled by relaxation kin-
of vacancies. For higher bias windows, the ORR/OER at the tip– etics (see Fig. 2 and the evolution of curves in Fig. 3a). The
surface junction is activated (as indicated by the inflection static hysteresis loop R0(V ) then describes the thermodynamics
points in the ESM loops), resulting in large changes in oxygen of the process, whereas R1(V ) and the corresponding relaxation
vacancy concentration and thus large changes in ESM amplitude. times describe the kinetics of vacancy redistribution. The response
After the bias pulse has been removed, a subsequent slow diffusion saturates once the diffusion length becomes larger that the
in the material takes place. The kinetics of this process is controlled detection volume, which in ESM is controlled by the tip–surface
by the diffusion times of the vacancies, and is therefore expected to contact area26.
be relaxational in nature. To estimate the corresponding relaxation
time, we use D(T ) ¼ D0exp(–Ea/kT), where D0 ¼ 2 × 1025 m2 s21 Numerical modelling of the tip–surface junction
(ref. 24) and Ea ¼ 0.62 eV for nanocrystalline YSZ25 (note that for The critical potentials measured in ESM are significantly larger
bulk material, Ea ¼ 1.09 eV). For diffusion lengths of L ≈ 5–10 nm than those expected in classical electrochemical experiments.
(comparable to the tip–surface contact radius) the diffusion times This behaviour can be readily rationalized by considering that the
a –2 b
5V 8 YSZ Pt tip
15 V YSZ Au tip
–4
30 V Glass Pt tip
–10
–8
–12
–12
–60 –40 –20 0 20 40 60 –40 –20 0 20 40
Applied voltage (V) Applied voltage (V)
Figure 3 | Effects of bias magnitude and tip coating on local ESM. a, Electrochemical strain (ESM) loops obtained on an YSZ (100) surface at different peak
biasing voltages. b, Comparison of ESM loops obtained on YSZ using tips with a platinum and gold coating. A reference loop on a glass surface is also
shown. Error bars indicate standard deviation over 100 locations.
·ΦÒrxn/Vth
a e 2.0 0.9
Tip 1.5
Vtip Vtip 0.8
Reaction
1.0
ΔVs log10 (KΦ/K0) 0.7
0.5
Potential
ΔVb 0 0.6
–0.5 0.5
–1.0 0.4
–1.5
0.3
0 –2.0
0 Coordinate –2 –1 0 1 2
log10 (Kc/K0)
b μ* c Φ d cion/[cion]
500 1.0 500 1.0 500
0.0
400 0.8 400 0.8 498
–0.1
300 0.6 300 0.6 496
z/RD
z/RD
z/RD
–0.2
200 0.4 200 0.4 494
–0.3
100 0.2 100 0.2 492
–0.4
0 0.0 0 0.0 490
0 100 200 300 400 500 0 100 200 300 400 500 0 50 100 150
r/RD r/RD r/RD
Figure 4 | Numerical modelling of electrochemical potentials at the tip–surface junction. a, Potential distribution under the tip, composed of the junction
potential (electrochemically active) and the bulk potential drop. b–d, Normalized electrochemical potential of vacancies (b), electrostatic potential (c) and
vacancy concentration (d) at the tip–surface junction in ESM of YSZ surface at room temperature. e, Normalized electrostatic potential at the TPB region
versus surface reaction rates. For low reactivity, the potential drop in the bulk dominates and the effective driving force for the ORR/OER process is well
below tip bias.
highly localized nature of the SPM tip leads to a potential drop both Spatially resolved mapping of electrochemical activity
at the tip–surface junction (inducing the electrochemical process) The ESM approach enables spatially resolved mapping of electro-
and in the bulk, as schematically illustrated in Fig. 4a. The potential chemical activity across a solid surface. In this, the dynamic (Fig. 2)
and vacancy concentration can be analysed using the numerical or standard (Fig. 3) hysteresis loops are acquired over a dense 2D
model developed by Ciucci et al.27,28, extended here to a purely spatial grid, yielding a 3D (or 4D for D-ESM) array of hysteresis
ionic conductor for material parameters extrapolated from refs 29, data. Descriptive parameters such as area inside the loop (reactivity),
30 and 31 to room temperature (see Supplementary Information), remanent response or nucleation biases, can be plotted as 2D maps.
as illustrated in Fig. 4b–d. The parametric plot of the effective poten- The reactivity map obtained on a 1 mm2 area of a (100)-oriented
tial versus reaction rates in Fig. 4e illustrates that for low reactivity single-crystal YSZ surface with a pixel size of 20 nm is shown in
(inevitable at room temperature) the potential drop at the junction Fig. 5. Clearly discernible variations of response across the crystal
is reduced compared to the tip potential, resulting in ‘stretching’ of surface are seen, indicating the presence of 100–200-nm-scale inho-
the voltage axis compared to macroscopic experiments. The positive mogeneties in the surface reactivity. The hysteresis loops from selected
nucleation bias (PNB) and negative nucleation bias (NNB) in ESM areas are shown in Fig. 5c, illustrating the difference in spectroscopic
thus provide local analogues of macroscopic cathodic and anodic responses between locations. The contact frequency variation and dis-
polarization voltages. persion are shown in Fig. 5d and e, respectively. The frequency map
d e 0.6 g 7 (V)
f
266 4 (V)
4.5
σ (f ) (kHz)
0.4 1
f (kHz)
265
−2 2.0
264 0.2
−5 −0.5
−8 −3.0
263 0
Figure 5 | Local ESM mapping on an YSZ surface with nanometre-scale resolution. a,b, Topography of the (100) YSZ surface (a) and map of ESM
hysteresis loop area (b, reactivity map). Pixel spacing is 20 nm. c, ESM hysteresis loops extracted from the characteristic spots in b. d,e, Contact resonance
frequency (d) and contact resonance frequency dispersion (e) maps. f,g, Negative (f) and positive (g) ESM nucleation voltage maps.
a 60 (nm) b
15 Pt
60 nm YSZ
TPB 1
10 TPB 2
ESM response (a.u.)
40
0
20
–5
0 –10
–20 –10 0 10 20
Applied voltage (V)
c
Pt
3.5
3.0
ESM activity (a.u.)
TPB 2.5
2.0
1.5
YSZ 1.0
0.5
Figure 6 | Mapping electrochemical activity near a triple-phase boundary. a, Topography of platinum nanoparticles deposited on an YSZ (100) surface.
b, ESM loops collected on YSZ, a platinum particle, and two different positions along the triple-phase boundary (TPB) using a platinum tip. c, Overlay of
electrochemical activity on the topography of the nanoparticles reveals enhancement of activity along the triple-phase boundaries.
illustrates the combined effect of surface topography and elastic prop- origins of this inhomogeneity are non-uniform dopant distributions
erties. Because the topography is essentially uniform, the contrast indi- or possible contamination of the surface, for example, by silicon
cates the presence of intrinsic inhomogeneities in the material. Maps of (epitaxial films of SDC, for instance, do not exhibit these variations)32.
the negative and positive nucleation voltages in Fig. 5f,g illustrate that
only the positive nucleation bias varies across the mapped surface, with Reactivity mapping near a triple-phase boundary in Pt-YSZ
the negative nucleation bias being approximately constant. This To illustrate reactivity mapping in more complex structures, we
suggests the thermodynamic potential for the activation of OER, but applied ESM mapping to a platinum-nanoparticle-coated YSZ
not ORR, varies across the studied YSZ surface region. The likely surface (Fig. 6). We aimed to probe the triple-phase boundary at
which the catalyst electrode is in contact with the reactants and the spacings and image sizes were also used. All measurements were performed with the
electrolyte, forming a critical unit in a catalysed reaction. The ESM biased tip in direct contact with the YSZ and Pt-YSZ surface in ambient air and
without any additional protective coating. The implementation and data analysis in
response was mapped in a 300 nm region containing the nanopar- D-ESM are detailed in Supplementary Section II.
ticles. The corresponding ESM hysteresis loops are shown in Fig. 6b.
The reactivity map is overlaid on the topography in Fig. 6c, revealing
Received 18 May 2011; accepted 7 July 2011;
increased reactivity close to the triple-phase boundary in compari-
son to the free YSZ surface. Note the relatively weak ESM response published online 14 August 2011
of the particle itself, the sharp increase of reactivity at the triple-
phase junction, and the low reactivity in the YSZ surface. References
This behaviour can be readily explained following the logic illus- 1. Basic Research Needs for Electrical Energy Storage, DOE BES Workshop 2007.
http://science.energy.gov/~/media/bes/pdf/reports/files/ees_rpt.pdf.
trated in Fig. 1b–d. For the tip positioned on the nanoparticle, the 2. Armand, M. & Tarascon, J. M. Key challenges in future Li-battery research.
particle is biased as a whole, and the ORR/OER process is activated. Nature 451, 652–657 (2008).
However, because the particle size is much larger than the tip size, 3. Bagotsky, V. S. Fuel Cells: Problems and Solutions (Wiley, 2009).
the field within the YSZ (and hence migration motion) is weaker. 4. Abraham, K. M. & Jiang, Z. A polymer electrolyte-based rechargeable lithium/
For the tip positioned on the YSZ surface, the applied bias is insuf- oxygen battery. J. Electrochem. Soc. 143, 1–5 (1996).
5. Wang, D. Y., Xiao, J., Xu, W., Zhang, J. G. High capacity pouch-type Li–air
ficient to activate the OER/ORR process, so the hysteresis loop is batteries. J. Electrochem. Soc. 157, A760–A764 (2010).
only partially open. Finally, for the tip positioned on the junction 6. Adler, S. B. Factors governing oxygen reduction in solid oxide fuel cell cathodes.
between the nanoparticle and substrate, the electrocatalytic effect Chem. Rev. 104, 4791–4843 (2004).
of the particle is activated and the electric field also penetrates the 7. Zhang, D., Fu, Z. H., Wei, Z., Huang, T. & Yu, A. S. Polarization of oxygen
electrode in rechargeable lithium oxygen batteries. J. Electrochem. Soc. 157,
material, resulting in an enhanced response. In this manner, an
A362–A365 (2010).
increased ORR activity is present around the boundary of a single 8. Hummelshøj, J. S. et al. Elementary oxygen electrode reactions in the aprotic
catalytic nanoparticle. Li–air battery. J. Chem. Phys. 132, 071101 (2010).
9. Wilson, J. R. et al. Three-dimensional reconstruction of a solid-oxide fuel-cell
Conclusions anode. Nature Mater. 5, 541–544 (2006).
To summarize, using electrochemical strain detection we have 10. Reed, J. & Ceder, G. Role of electronic structure in the susceptibility of
metastable transition-metal oxide structures to transformation. Chem. Rev. 104,
demonstrated direct measurements of ORR/OER on the nanometre 4513–4534 (2004).
scale, in volumes six to eight orders of magnitude smaller than is 11. Singh, G. K., Ceder, G. & Bazant, M. Z. Intercalation dynamics in rechargeable
possible by conventional electrochemical methods. The mapping battery materials: general theory and phase-transformation waves in LiFePO4.
of the YSZ surface demonstrated the presence of 100–200-nm- Electrochim. Acta 53, 7599–7613 (2008).
scale inhomogeneities, presumably associated with surface contami- 12. Goodenough, J. B. Electronic and ionic transport properties and other physical
aspects of perovskites. Rep. Prog. Phys. 67, 1915–1993 (2004).
nation. The enhanced ORR activity at the triple-phase junction of 13. Nørskov, J. K., Bligaard, T., Rossmeisl, J. & Christensen, C. H. Towards the
Pt-YSZ has been demonstrated, for the first time providing infor- computational design of solid catalysts. Nature Chem. 1, 37–46 (2009).
mation on the local characteristics of the ORR/OER process. 14. Opitz, A. K. & Fleig, J. Investigation of O2 reduction on Pt/YSZ by means of thin
Nanoscale probing of ORR and OER kinetics and oxygen film microelectrodes: the geometry dependence of the electrode impedance.
Solid State Ionics 181, 684–693 (2010).
vacancy diffusion will provide a transformative change in the
15. La O’, G. J., Yildiz, B., McEuen, S. & Shao-Horn, Y. Probing oxygen reduction
ability to explore and control the mechanisms underpinning the reaction kinetics of Sr-doped LaMnO3 supported on Y2O3-stabilized ZrO2.
efficiency of air-based fuel-cell systems and metal–air batteries. In J. Electrochem. Soc. 154, B427–B438 (2007).
particular, the capability to directly link local structure and electro- 16. Morozovska, A. N., Eliseev, E. A. & Kalinin, S. V. Electromechanical probing of
chemical activity will allow for systematic high-throughput studies ionic currents in energy storage materials. Appl. Phys. Lett. 96, 222906 (2010).
17. Balke, N. et al. Nanoscale mapping of ion diffusion in a lithium-ion battery
of electrocatalysis and will establish the bridge between atomistic cathode. Nature Nanotech. 5, 749–754 (2010).
theory and macroscopic electrochemical measurements. Together, 18. Balke, N. et al. Real space mapping of Li-ion transport in amorphous Si anodes
this will allow for knowledge-driven design and optimization of with nanometer resolution. Nano Lett. 10, 3420–3425 (2010).
energy conversion and storage systems. The breakthrough in effi- 19. Bishop, S. R., Duncan, K. L. & Wachsman, E. D. Defect equilibria and chemical
ciency of ORR and associated cost reductions can bring back to expansion in non-stoichiometric undoped and gadolinium-doped cerium oxide.
Electrochim. Acta 54, 1436–1443 (2009).
the forefront of research the development and commercialization 20. Jesse, S., Kalinin, S. V., Proksch, R., Baddorf, A. P. & Rodriguez, B. J. The band
of multiple fuel-cell technologies. In the situation where renewable excitation method in scanning probe microscopy for rapid mapping of energy
energy technologies are rapidly gaining broad adoption (for dissipation on the nanoscale. Nanotechnology 18, 435503 (2007).
example, 62% of new generation capacity in the EU is renewable33), 21. Weppner, W. & Huggins, R. A. Electrochemical methods for determining kinetic
the role of large-scale implementation of lithium–air and fuel-cell- properties of solids. Ann. Rev. Mater. Sci. 8, 269–311 (1978).
22. Guo, S. et al. Spatially resolved probing of Preisach density in polycrystalline
based vehicles and local and grid-storage components is hard ferroelectric thin films. J. Appl. Phys. 108, 084103 (2010).
to overestimate. 23. Jesse, S., Lee, H. N. & Kalinin, S. V. Quantitative mapping of switching behavior
in piezoresponse force microscopy. Rev. Sci. Instrum. 77, 073702 (2006).
Methods 24. Knöner, G., Reimann, K., Röwer, R., Södervall, U. & Schaefer, H.-E. Enhanced
Materials. Platinum nanoparticles, 50 nm in size, were deposited on a YSZ (100) oxygen diffusivity in interfaces of nanocrystalline ZrO2–Y2O3. Proc. Natl Acad.
crystal using direct sputter deposition and laser annealing. A thin film (3 nm) of Sci. USA 100, 3870–3873 (2003).
platinum was deposited on (100)-oriented YSZ single crystals using d.c. magnetron 25. Kosacki, I., Rouleau, C. M., Becher, P. F., Bentley, J. & Lowndes, D. H. Nanoscale
sputtering. Subsequent laser annealing (single pulse: KrF, 248 nm wavelength, effects on the ionic conductivity in highly textured YSZ thin films. Solid State
650 mJ cm22 fluence, 25 ns pulse width) resulted in the melting and dewetting of the Ionics 176, 1319–1326 (2005).
platinum film into regularly spaced (pitch ¼ 50–100 nm), regularly sized (50 nm) 26. Morozovska, A. N., Eliseev, E. A., Balke, N. & Kalinin, S. V. Local probing of
nanoparticles as shown in Supplementary Fig. S1. ionic diffusion by electrochemical strain microscopy: spatial resolution and
signal formation mechanisms. J. Appl. Phys. 108, 053712–21 (2010).
ESM. Atomic force microscopy (AFM) and ESM measurements were performed 27. Ciucci, F., Chueh, W. C., Goodwin, D. G. & Haile, S. M. Surface reaction and
with a commercial system (Asylum Research Cypher) additionally equipped with a transport in mixed conductors with electrochemically-active surfaces: a 2-D
LabView/MatLab based band excitation controller implemented on NI-5122/5412 numerical study of ceria. Phys. Chem. Chem. Phys. 13, 2121–2135 (2011).
fast AWG and DAQ cards. ESM imaging and spectroscopy were performed with a 28. Ciucci, F. & Goodwin, D. G. Non linear modeling of mixed ionic electronic
200–400 kHz 2Vpp band excitation signal applied to a metal-coated tip. The conductors. ECS Trans. 7, 2075–2082 (2007).
spectroscopic measurements were performed with an 1 s/pixel waveform with 29. Lai, W. & Haile, S. M. Impedance spectroscopy as a tool for chemical and
2 ms for each d.c. voltage step. Mapping of the electromechanical response was electrochemical analysis of mixed conductors: a case study of ceria. J. Am.
carried out typically on a 50 × 50-point grid with a spacing of 20 nm, although other Ceram. Soc. 88, 2979–2997 (2005).