This project, performed in collaboration with Ho Nyung Lee and Hans Christen (CMSD) addresses the surface stability of epitaxial strontium ruthenate (SRO) thin films. These materials are considered as perspective oxide electrodes, e.g. for ferroelectric and high-k materials. The close chemical and structural similarity between electrode and functional material in this case minimizes interface electrochemical reactions, charge injection in oxide and other detrimental processes, thus improving retention performance, fatigue resistance, number of operational cycles, etc. However, device integration utilizing SRO electrodes necessitates high surface stability of the material. Indeed, the interface properties in the SRO based heterostructure will be strongly dependent on the interface properties and atomically smooth interface without interface states, etc. can not be achieved unless growth was performed on the atomically flat stable SRO surface. Thus, stability of the latter in vacuum and under the conditions of growth of second component is vital. Another reason for our interest to this material is that it is one of the end members for the familily of layered strontium ruthenates SrnRun-1O3n-2, which exibit a range of interesting bulk electronic properties, ranging from p-wave superconductivity in Sr 214 to metamagnetic behavior in Sr 327 and ferromagnetism in 113. These materials exhibit extremely interesting surface behavior, including surface specific Mott transition, etc. However, while the surfaces of layered members of strontium ruthenate family (n = 2,3,4) can be prepared in-situ by cleaving, perovskite 113 is not amenable to this procedure and thus atomically clean surface must be prepared using alternative ways.
Surface science
studies of SrRuO3 as well as its integration in device fabrication
process are hindered by the chemical stability of oxide surfaces. Indeed,
the well known problem in the surface science of oxide materials is that,
unlike metals or semiconductors, oxide surfaces can generally not be prepared
by sputtering and annealing. Application of these methods results in an
oxygen loss, cation segregation and other processes detrimental to the
properties and chemical integrity of material. Similarly, for subsequent
growth of oxide heterostructures reduction of SRO or oxidation and loss
of volatile RuO4 will irreversibly destroy the material. To
get the first insight into chemical behavior of SRO thin films, we have
analyzed the thermodynamic behavior in the SrO-Ru-O2 system.
Depending of the partial pressure of oxygen, decomposition pathway
for strontium ruthenate can include oxidation and loss of RuO4
in an oxygen rich atmosphere (average oxidation state of Ru increases),
disproportionation with formation of metallic Ru and RuO4 (average
oxidation state of Ru is constant) and reduction with formation of
metallic Ru (average oxidation state of Ru decreases). Using published
thermodynamic data for SrRuO3, corresponding stability diagram
(Ellingham diagram) can be constructed as shown below. Note that while
disproportionation of SRO (independent of oxygen pressure) doesn't occur
below ~ 2000 ºK, volatilization of RuO4
is possible under high oxygen pressures at elevated temperatures. Also,
shown in comparison (dotted line) is partial pressure of oxygen expected
in a well-baked vacuum chamber under ion- or sublimation pump. Note that
this line is very close to the stability line of SRO. This means that stability
of the material with respect to reduction is strongly dependent of specific
conditions in vacuum chamber. In other words, it will be stable under turbo
pump or in the unbaked chamber, but baking or using oil pump will lower
the partial pressure of oxygen below the stability limit of the material,
which will thus become thermodynamically unstable. Another important conclusion
of thermodynamic analysis is that 113 and SrO can not coexist and decomposition
of SrRuO3 is likely to proceed through the stage of strontium
rich ruthenates (decreasing n in SrnRun-1O3n-2)
In this project, SrRuO3
epitaxial thin films were grown using Pulsed Laser Deposition (PLD) by
H.N. Lee and H.M. Christen. Shown on the left is ambient AFM image of the
SrTiO3 substrate prior to the deposition. The steps on the image
are single unit cell high (0.4 nm) separated by the ~500 nm terraces, indicative
of extremely high quality of substrate. The topography of the deposited
10 nm sample is shown on the right. Note that apart from slight roughening
of the terrace edges, the two are virtually identical. This (as well as
in-plane XRD) confirms extremely high quality of the SrRuO3
film.
After the films were grown, they were transferred to the surface analysis chamber. Interestingly enough, even though the samples were exposed to air for several hours, a clear Low Energy Electron Diffraction pattern can be seen. LEED is sensitive to the crystallographic structure of the several top atomic layers of the surface (in a sense, it is an analog of XRD for bulk crystallographic structure). The fact that LEED pattern can be observed after air exposure indicates that the surface is very stable, thus resembling behavior of noble metals such as Au. Also very interesting is the fact that even moderate heating of the surface to ~200 ºC led to the disappearance of the LEED pattern. A comparison with the thermodynamic data suggests that this temperature is at least 400-600º lower than expected decomposition temperature for the material.
To establish
the chemical nature of the chemical species on the SRO surface, the material
was analysed using reflection High Resolution Electron Energy Loss Spectroscopy
(HREELS). This technique effectively probes the vibrational excitations
of the several top atomic layers, thus providing a surface analog for infra-red
spectroscopy of bulk materials. The HREELS spectrum shows the presence
of the C-H vibrational modes. At the same time, there is no detectable
-OH group vibrations and vibration modes at ~30-70 mV attributable to bulk
phonons of SRO can be clearly seen. Thus, HREELS data suggests that the
surface is
- stable with respect to hydroxilation
(SrO terminated, RuO2 termination being thermodynamically unstable)
- partically covered by hydrocarbons
(C-H bonds)
- the hydrocarbon coverage is less
than 1 mL (we can still see phonons)
In order to
get some further insight in the thermal behavior of these films, the decomposition
of SRO was studied using thermal desorption spectroscopy (TDS). In this
technique, the sample is heated at controlled rate and the chemical identity
and quantity of the desorbing species is determined using mass-spectrometer.
This technqiue is basically a surface analog of the thermogravimetric analysis
for bulk materials. The TDS spectrum for SrRuO3 thin film is
shown below. Note that at 400 ºC there
is a peak corresponding to emission of SrO and metallic Ru. The second
peak corresponding to Ru emission is at ~600 ºC.
In agreement with thermodynamic arguments, there is no peaks corresponding
to metallic Sr or ruthenium oxides. Interestingly, the amount of desorbed
material is very small and can be estimated as a fraction of the monolayer.
At the same time, annealing above 1200 ºC
results in massive emission of SrO, Ru and O2, indicative of
the onset of bulk decomposition of material. This data suggests that decomposition
of SRO proceeds in two steps, one corresponding to the top most surface
layer(s), the other corresponding to bulk material.
The evolution
of surface morphology during the decomposition was studied by high vacuum
STM. The samples were annealed from 100 ºC
to 800 ºC in 100 ºC
steps and STM images were acquired at RT after each annealing. At room
temperature, the STM image clearly indicates the presence of atomic steps
and multiple white particles. On increasing the temperature to 100 ºC
and 200 ºC, the area density o fwhite
particles decreases. At 300 ºC and 400
ºC, the particles disapper and the surface
developes a number of irregular 1 unit cell deep pits. Finally, at higher
temperatures a number of spherical particles form and their size increases
with annealing temperature. Interestingly, the unit cell steps can be observed
even at the late stages of decomposition process.
The changes
in surface composition during annealing was studies by X-ray Photoelectron
Spectroscopy, which provides information on the identity and oxidation
states of surface.
The evolution
of O1s, Sr3d, Ru3p and Ru 3d peaks is shown below. The biggest problem
in the XPS studies of strontium ruthenates is that Ru3d, Sr3p3/2 and C1s
peaks coincide (both in XPS and Auger), thus precluding quantitative and
even qualitative detection of carbon on these surfaces (hence the need
for HREELS). The XPS analysis suggests that th ebinding energy of Ru decresases
during annealing, indicative of reduction process. At the same time, BE
for strontium increases, as can be expected for transition from SRO 113
to SrO or Sr-rich ruthenates. The quantitative analysis of O, Sr, and Ru
peaks ratio suggest that:
- the final stoichiometry of the
surface is the mixture of Sr2RuO4 and metallic Ru.
- carbon burn-out occurs at ~400C.
To summarize these observations, LEED
and HREELS data indicate that SrRuO3 surfaces are stable with
respect to hydroxylation when exposed to air; however, the surface is susceptible
to contamination by carbon-containing species. The characteristic contamination
levels are small and both LEED patterns and energy loss peaks attributable
to SrRuO3 phonons can be observed. As expected for oxides, traditional
cleaning procedures such as vacuum annealing cannot remove the contamination
without compromising the stability of the film, as evidence by disappearance
of the LEED pattern even after moderate (~100 -200 ºC) annealing.
The reduction of the SrRuO3 surface starts at temperatures as low as 300
- 400 ºC and is confined predominantly to a one unit cell thick surface
layer. Decomposition is accompanied by the desorption of SrO and metallic
Ru and formation of characteristic one-layer deep pits in STM images. At
higher temperatures (~500 – 700 ºC), decomposition proceeds with formation
of nanoparticles, presumably of Sr-rich ruthenates, SrO and metallic Ru.
The onset of bulk decomposition occurs at much higher temperatures around
~1000 ºC in agreement with results of thermodynamic analysis.
These observations suggest that the
surface stability of SrRuO3 is markedly different from the bulk
thermodynamic stability. This can be attributed to the presence of carbon
contaminates that reduce the thermal stability of the surface layer. This
contamination problem is now shown be resolved by a low temperature anneal
of films in a moderate oxygen pressure (e.g. similar to deposition conditions)
prior to future processing or characterization.
Finally, getting back to our original
mission of studying the physics of perovskite surfaces, currently we have
(almost) finished the NanoTransport system, which is specifically designed
for in-situ PLD growth and surface characterization of perovskite thin
films, thus opening a pathway to study these materials.
Research was performed at the Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the U.S. Department of Energy under Contract DE-AC05-00OR22725. Support as a Eugene P. Wigner Fellow is acknowledged (SVK). Research was partially sponsored as part of a BES NSET initiative on Nanoscale Cooperative Phenomena (HNL and HMC).