Implementation of Multisine EIS in a toolbox for electrochemical studies

Student thesis: Doctoral Thesis

Abstract

In this work a toolbox is developed that can serve as a guideline to perform
adequate electrochemical impedance spectroscopy (EIS) measurements. This
goal is obtained in dierent stages. At rst data quality is maximised by using
an optimised setup and an appropriate signal: odd random phase multisine.
This signal allows to derive levels of noise, noise + non-linearities as well as an
indication for the presence of non-stationarities from the data. This information
gives a rst estimate of the quality of the recorded data. In parallel the toolbox
relies on the input of complementary experimental information to obtain a better
insight in the physical parameters involved in the system. Based on the recorded
impedance data and the information from additional experiments a model for the
system under test can be proposed. This forms a mathematical representation
of the relation between the various physical quantities believed to be involved in
the system. These physical quantities form (or are related to) the parameters of
the proposed model.
The next step consists of deriving numerical values for these parameters by
matching the measured data as good as possible with the proposed model. This
procedure is called the tting of the data. In this work a tting procedure is
developed that can use information about noise levels or levels of noise + nonlinearities
as a weighting factor. As such, the tting procedure takes account
of the intrinsic value of each data point when deriving the parameter values.
Additionally the uncertainty on the obtained parameters can be estimated.
Traditionally this tting is performed using a Levenberg-Marquardt algorithm.
To maximise the chance to obtain a satisfactory t, a simulated annealing (SA)
step was introduced in the tting procedure to provide starting values for the
Levenberg-Marquardt algorithm. Following the philosophy of the toolbox, the
derived parameter values are subsequently matched with the additional physical
information available from the complementary experiments. If this match is
satisfying, the model can be used to describe the investigated system with con-
dence. If not, the model will have to be adapted. This might invoke the need
to gather additional experimental information. If this delivers no satisfactory
information, even a redesign of the EIS experiment must be considered.
In the scope of this toolbox two of the possible experimental methods that can
generate additional information are developed as well.
At rst the use of target factor analysis (TFA) is put forward. This is a mathematical
operation performed on a set of Auger electron spectroscopy (AES)
experiments. This operation compares a set of standard spectra of preferably
known components with the measured data. As such it is possible to derive the
distribution of certain compounds on the surface with a high lateral as well as
in-depth resolution.
The second source of additional information developed in the scope of this work
is local EIS (or LEIS). The use of LEIS can yield information about the nature
of reactions taking place in dierent positions of a heterogeneously reacting
surface. This way a number of features detected in the global EIS spectra can
be identied. A new setup was developed in the scope of this work to enable
performing LEIS measurements at higher lateral resolution than is standard.
This enhanced electrochemical resolution is coupled with optimal imaging of
the surface as the probe to perform LEIS measurements can be used as well
to perform atomic force microscopy (AFM) on this surface. This allows the
coupling of surface structure and electrochemical information.
To validate the approach proposed in this toolbox, in the second part of this
work it was implemented on the study of a copper surface evolving during the
rst 24 hours in a chloride environment.
This system has been studied since long. A part of these studies was performed
using EIS experiments. A literature study showed however that the interpretations
given to the parameters in the model (an equivalent circuit) diers according
to the consulted work. The toolbox was implemented to try and solve these
contradictions.
At rst multisine EIS experiments were performed and compared to the proposed
model structure. This did not allow to draw denitive conclusions about the
system, thus additional steps were taken.
Scanning electron microscopy (SEM) learned that in the rst 24 hours local
corrosion processes as well as the formation of a global corrosion product layer
take place. Local corrosion activity cedes however, resulting in a copper surface
almost entirely covered by a general corrosion product layer.
The subsequent AES experiments showed that the corrosion product layer grows
thicker as a function of time. TFA performed on the recorded AES spectra
learned that the corrosion product layer was mainly an oxide layer (Cu2O and
CuO). The layer was found to have a heterogeneous composition. Lateral and
in depth dierences could be identied. Raman spectroscopy revealed that in
the solution, close to the oxide layer, a Clinoatacamite layer is present.
After gathering this additional information, a tting operation was performed
to match the recorded EIS data with this experimental information. The evolution
of the parameters derived from the t were compared to this information.
It was concluded that none of the proposed interpretations of the model was
satisfactory. This lead us to proposing an alternative model, taking into account
the experimental data obtained before.
The data were matched with the model by a tting operation and the resulting
parameter values and their evolution in time evaluated. From this operation it
proved that the new model seems to match the system more adequately. Still a
number of points persist where the model cannot yet fully describe the system.
These involve the reactions taking place at the surface and a number of observed
distributions of time constants.
To gather information about this latter eect, in a last step the newly developed
LEIS was implemented on the studied system. Although the technique is still
suering from a number of limitations, its potential was shown since a dierence
in behaviour between grains and grain boundaries could be evidenced.
A more detailed investigation of the nature of the distributed elements in the
proposed model is the rst priority in future work. If properly used, the approach
presented in this toolbox presents itself as a powerful method to obtain a reliable
model for many systems involving EIS investigations by showing new insights in
system behaviour and a high potential to avoid pitfalls.
Date of Award30 Jun 2009
Original languageEnglish
SupervisorAnnick Hubin (Promotor), Jean Vereecken (Co-promotor), Rik Pintelon (Jury), Johan Deconinck (Jury), Isabelle Vandendael (Jury), Nadine Pebere (Jury), Isabelle Frateur (Jury) & Bernard Boukamp (Jury)

Keywords

  • Impedance spectroscopy
  • surface science
  • toolbox
  • EIS
  • multisine
  • electrochemistry
  • LEIS

Cite this

'