|
|
Conclusions
Introduction
EIS is becoming a well-accepted and widely used technique for evaluating the
corrosion resistance of coatings applied to metals. It offers a number of advantages when
compared to the older exposure tests:
- EIS can detect failures
before they become visible -- decreasing test time,
- EIS gives a numerical result that is
independent of the tester’s subjective judgment,
- EIS can give information about the
failure mechanism.
EIS tests are inherently
non-destructive, so repeated tests can follow the degradation of a coating with exposure
time. Analysis of the changes in the EIS spectrum over time often yields better
information than a single measurement after a set exposure time. The time series can
improve the accuracy of the test, increase the relevance of the information obtained and
decrease the exposure time necessary to produce a detectable failure.
Even with the availability of high
performance, lower cost EIS instrumentation, an EIS instrument generally cannot be
dedicated to a single painted sample. Instead, multiple samples are exposed to a corrosive
environment and periodically connected to the EIS instrument and tested. Unfortunately,
each test takes some time to set up and a significant time to run, making this type of
test program technician time intensive.
An Electrochemical Multiplexer, such as the
Gamry Instruments’ ECM8, can help lower the labor requirements and instrumentation
costs of an EIS test program. The ECM8 allows 8 samples to be connected to a single EIS
instrument. An EIS spectrum can be measured on each sample in a sequential (not
simultaneous) manner.
The software controlling the ECM8 will
record an EIS spectrum on each sample connected to it. An extension of the software allows
repetition of this set of measurements at set time intervals. In the former case,
technician time is lowered because cell connections are all made at once and software
setup is only done once for all eight samples. In the latter case, periodic EIS
measurements are recorded over a long time period with no operator intervention at all.
The ECM8 Electrochemical Multiplexer was
originally developed for use with DC corrosion measurements of bare metals. It can be used
in an EIS system testing coated metal specimens with some limitations (caused by the
higher sensitivity required for EIS measurements). This Note describes these limitations
and offers suggestions for optimal use of the ECM8 with the Gamry Instruments
EIS300 EIS Software operating with an FAS2
Femtostat.
EIS
Background
Gamry Instruments’ Web site contains both a
primer on EIS measurements and a discussion
of EIS on coatings. The information in these
application notes will not be repeated here.
Experimental Techniques
This application note includes some EIS spectra recorded on a dummy cell
connected to an ECM8. These spectra were recorded using a Gamry EIS system
equipped with an FAS2 Femtostat. Both the FAS2 and ECM8 were unmodified instruments
randomly pulled from Gamry’s stock.
The dummy cell was enclosed in a
Faraday shield, which was in turn connected to both earth ground and to the floating
ground terminal on the PCI4 cell cable.
The FAS2 was connected to the ECM8
with a standard ECM8 to potentiostat cable (P/N 985-13) which was in turn connected to
an FAS2 to ECM8 Cable Adapter (P/N 985-00062). The adapter connected to the three
FAS2 cell connections and also connected the FAS2 Floating Ground terminal to the ECM8's
Cell Connector ground.
The ECM8 was connected to the dummy
cells using standard PCI4 cell cables. Only the Working, Reference and Counter electrodes
leads were connected to the cell. Unused leads were connected to the Faraday cage.
The dummy cell consisted of a dipped
mica capacitor (with ± 10% tolerance) in parallel with a carbon film, high Meg resistor
(± 10% tolerance). In some experiments, the spectrum of an open lead dummy cell was
measured. In this case, the reference and counter electrode leads were shorted together
and the working electrode lead was left dangling in the air.
In all cases, the dummy cell’s
spectrum was recorded using the EIS300’s standard Potentiostatic EIS
experiment. Prior to recording the spectrum, an ECM8 channel was selected by ABORTing
the MUXEISP script when it had switched the ECM8 to the channel of interest. In some
cases, modified scripts were used as described in the Results and Discussion section.
All spectra were recorded at a DC
potential of zero volts versus the reference electrode potential.
Results and Discussion
FAS2 Baseline Data –
No ECM8
An FAS2 without an ECM8 is
specified to measure 15 pF in parallel with 1011 ohms with errors of less than 5% in magnitude
and 10° of phase. A stock FAS2 was used to record the
spectrum of a parallel RC dummy cell built using these values. The resulting Bode plot is
seen in Figure 1. The FAS2 performance exceeds its specifications.
Figure 1
Baseline Spectrum of Dummy Cell
The open lead Bode plot recorded using the same
FAS2 is
seen in Figure 2. In this figure, the line is the result of a visual fit to a parallel RC
model. The equivalent capacitance of this open lead curve is less than 1 pF and the
limiting resistance is greater than 8 TW. The loss of phase information at low frequency/high impedance is
typical of FAS2 open lead spectra.
Figure 2
Base Line – Open Lead Curve
FAS2
Connected to ECM8 – Stock Script and ECM8 Jumpers
The same
FAS2 was then connected to an ECM8.
The next test attempted to record the spectrum of the dummy cell using this setup. The
resulting Bode plot is seen in Figure 3.
Comparing Figure 1 and Figure 3, you
can see that the spectrum recorded though the ECM8 was very poor indeed. The magnitude
should approach 1011 W at 100 mHz. In Figure 5, the magnitude at 0.1
Hz is only about 109 W.
Figure 3
Stock ECM Based System – Dummy Cell
The Lissajous figures shown while the spectrum in Figure 3 was being recorded
provided the clue used to diagnose the problem. A typical "bad" Lissajous figure
is seen in Figure 4. Large spikes in the current occur as the voltage crosses zero. The
charge in one of these spikes is about 10 pC.
The local potentiostat in the ECM8 is
left in an "open loop" condition when it is not in use. Its output is at
positive or negative saturation (about +12.5 volts or -12.5 volts). This output is not
directly connected to the cell, but can couple to the cell through the capacitance of open
relays and adjacent PC board traces.
As the potentiostat’s input
voltage crosses the local potentiostat’s set point voltage (which is zero volts by
default), the local potentiostat’s output abruptly swings from one saturation point
to the other. The swing is thus about 25 volts. This voltage change will couple 10 pC of
charge through a 0.4 pF capacitor. 0.4 pF is a reasonable value for stray coupling
capacitance in the ECM8.
Figure 4
Bad Lissajous Figure Recorded with Stock System
FAS2 Connected to an ECM8 – Jumpered to Disable the Local Pstats
The local potentiostats in the ECM8 can be disconnected from the cell by moving
internal jumpers. Moving the jumpers disconnects the potentiostats’ inputs from the
cells’ reference electrode inputs.
When the dummy cell experiment was
repeated using an ECM8 with a disabled local potentiostat, a much better spectrum was
recorded (most of the time). Figure 5 is an example. Note that the fit reports a high
capacitor value and a low resistor value.
However, in some cases, distorted
Lissajous figures and very noisy spectra were seen even though the local potentiostats
were disabled. I believe that the disconnected potentiostat input was drifting and picking
up noise, resulting in local potentiostat output swings and coupling of noise into the
cell.
Figure 5
EIS Spectrum of Dummy Cell with Local Potentiostats Disabled
Open Lead Curve with the Local Potentiostat Voltage at +5
Volts.
A modified MUXEISP script was
generated to help eliminate the drift and pickup problems mentioned above. This script
changed the set point voltage of all local potentiostats to +5 volts (in place of the
default setting of zero volts). As before, ABORT was selected to halt this script on a
desired ECM8 channel.
With the local potentiostats still
disabled, use of this script this script eliminated the occasional glitch seen in the
previous test. Spectra of similar to that in Figure 5 were recorded on the dummy cell.
Making this set point change and
leaving the local potentiostats enabled also led to good quality spectra, but resulted in
higher DC background currents.
Figure 6 shows an open lead spectrum
recorded with the local potentiostats both biased at +5 volts and disabled. Compare
this spectrum with Figure 2. Connecting the ECM8 increases the open lead capacitance to
more than 5 pF. Without the ECM8, the open lead capacitance was less than 0.5
pF. Looking
back at Figure 5, the reported capacitance is about 5 pF too high - most likely the result
of the open lead capacitance adding to the cell’s capacitance.
Figure 6
Open Lead Spectrum – No Local Pstat and 5V Potentiostat Setting
The low frequency region of the open lead curve in Figure 6 does not fit the
parallel RC model very well. The following discussion is an attempt to explain this
phenomena.
The impedance at frequencies below 1
Hz was in excess of 1011 W. The AC current through 1011
W with a
10 mV excitation voltage is 100 fA rms.
The DC background current on this
ECM8 channel was much larger than this AC signal. It was generally between 40 and
80 pA. This background current drifts substantially with time and temperature. The
open lead curve was recorded overnight. The ambient temperature in the building fell
substantially (by at least 4 ° C) while the spectrum was recorded. The DC background
current also fell, because it is highly temperature dependent. Distorted Lissajous figures
are observed when low frequency data was recorded on top of a shifting background. I
believe that this effect caused the poor low frequency performance in the open lead
experiment.
Conclusions
Script
and Jumper Setting Changes
The Gamry
EIS300 with the default ECM8 jumper settings and standard
MUXEISP script has significant problems measuring the EIS spectra of high impedance
samples.
Both disabling the ECM8’s local
potentiostats and modifying the MUXEISP script to move the local potentiostat set point
away from zero substantially improve the system’s performance. We recommend that both
changes be made in all systems. Note that the local potentiostat is not required or used
in most EIS coatings tests.
The ECM8 Operator’s Manual
contains instructions telling you how to disable the ECM8’s local
potentiostats. The
set point of the ECM8’s local potentiostat’s is set to 5 volts by adding the
following line to the MUXEISP script just prior to the loop in which data is taken:
Mux.SetDac(NIL,5.0)
This change should already be
incorporated in all revisions of the EIS300 later than revision 3.03.
Typical Accuracy Limits
The ECM8 characteristics dominate the performance of the
FAS2/ECM8 based EIS
system. The FAS2 performance is degraded by at least one order of magnitude when it
has to contend with the added leakage currents and capacitance on the ECM8 and long
interconnecting cables.
Conservatively extrapolating from the
open lead spectrum, the typical accuracy contour map will be no
worse than that in Figure 7. Note that EIS through an ECM8 is not recommended for
coatings with a capacitance less than 60 pF or a low frequency resistance of greater than
5 x 109 ohms. System performance is similar to
that of older, single channel systems built around Gamry's PC3 Potentiostat or EG&G's
Model 273. Tests done for this Application Note were all performed with the
FAS2 Femtostat.
Similar effects
were obtained with a PCI4 Potentiostat. With either potentiostat, the ECM8 characteristics
dominate the accuracy limits, since the potentiostat’s solo performance is at least
two times better than that in an ECM8 system.
Note - Gamry Instruments cannot
guarantee this "typical" performance. ECM8 channels exhibit significant
variation in leakage currents both within a specific ECM8 and when comparing ECM8s.
However, a second FAS2 /ECM8 measuring an 82 pF capacitor (in the 10% and 10° region of
Figure 7, gave performance closer to 4% and 3° up to a DC limit of 20 G ohms.
Figure 7
Typical Accuracy Contour Map – EIS System with ECM8
|
