You are on page 1of 6

Electrochemical Applications 1/02 page 1

1/2002 Advances in electrochemical applications of impedance spectroscopy

Issued and  by ZAHNER-elektrik GmbH & Co. KG in June 2002

First of all … work. You will find him as a co-author of various

… we have to say sorry for the long time you had to scientific presentations.
wait for this new issue of Electrochemical Applica-
tions. The delay was mainly caused by two facts: on The recent issue of Electrochemical Applications does focus
the one hand we developed a lot of new products like not only on EIS but on associated methods which are
the high current potentiostats of the EL- and the PP- able to complement EIS in specific applications. First
series, the NOISE probe with outstanding features of all, there is the HCI (High Current Interrupt)
(not only for noise measurements) and the new PC method, which allows you to find results in low im-
interface which is running under Windows95/98 as pedance objects such as power generating devices
well as under WindowsNT/2000 now. On the other even at high frequencies. The relaxation voltammetry
hand we had a lot of scientific presentations at well- is an alternative method for investigating high imped-
known meetings which yield enough stuff for the next ance systems such as coatings. The EIS chapter of
issues of Electrochemical Applications. this issue comes from Dr. Hoffmann (MTU-
Friedrichshafen GmbH) who is demonstrating the
Due to these facts we increased the Zahner team, application of AC impedance spectroscopy on molten
and I am proud to introduce to you Dr. Werner carbonate fuel cells.
Strunz, our new co-worker. He studied chemistry in
Heidelberg, Germany and took his doctor’s degree on We hope, the topics of this issue are a good excuse
“Electrical Conducting Coordination Polymers“. He is for the long waiting period.
supporting the Zahner team in several software pro- Your Zahner Team
jects and – above all – he is enhancing our scientific

Current Interrupt Technique -

Measuring low impedances at high
performance of the device and accounts some-
frequencies times for the main part of the overall losses. It
is very sensitive to degradation caused by corro-
F. Richter, Siemens AG, KWU, Erlangen, Germany, sion and thermal stress.
an o de
C.-A. Schiller, Zahner-elektrik, Kronach, Germany, c o n n e c to r
in d u c ta n c e an o de

N. Wagner, DLR, Stuttgart, Germany, c o n n e c to r

re s is ta n c e an o de
F a r a d a y ic a no de a n o d e s id e
proc es s * d o u b l e la y e r * po rou s
in s id e t h e c a p a c ita n c e d i s tr ib u te d
If a power generating device is examined, its dynamic p o ro u s
s y s te m
im p e d a n c e

electrical equivalence generally will appear as a net-

work which represents anode, cathode, membrane, m e m b ra n e +
e le c t ro ly te e q u iv a le n c e total
o h m ic
re s is ta n c e los s
electrolyte, and connectors.
b u lk total
in d u c ta n c e i n d u c tiv e
s h are
The specific losses of every partial impedance of the
c a th o d e
network contribute to the overall efficiency of the F a r a d a y ic
proc es s c a th o d e
c a th o d e s id e
po rou s
device. The porous layers of anode and cathode, in s id e t h e
p o ro u s
* d o u b l e la y e r
c a p a c ita n c e
* d i s tr ib u te d
im p e d a n c e

responsible for the charge transfer reaction, normally s y s te m

play a major role. Other contributions seem to be c a th o d e

c o n n e c to r
much less important. There is the resistance of the re s is ta n c e

electrolyte or the membrane as well as the resistance c a th o d e

c o n n e c to r
of contacts and connectors. A dynamic part is added in d u c ta n c e

by the inductance of the body and the connectors.

Fig. 1: Simplified electrical equivalent circuit of a
Nevertheless, the ohmic part (electrolyte, membrane, typical electrochemical power source device
connectors) plays an important role regarding the
Represented by TEST SOLUTIONS (770)410-9166
Electrochemical Applications 1/02 page 2

In applications with dynamic load changes the induc- It is rather contaminated by dynamically induced
tive parts, for example in a laptop computer battery or error voltages. These errors are caused by unavoid-
in electromotive applications, are limiting the maxi- able mutual induction from the magnetic field of the
mum pulse load available. current circuit.

The interference increases with increasing frequency

and with the strength of the magnetic field of the
- imaginary part

c a th o d e a rc
current. It depends on the geometry and grows with
the dimensions of the object. For the investigation of
a n o d e a rc power sources this means: The more you scale up,
the lower is the available upper frequency limit (fg).
Finally, there is a limit for the EIS at low ohmic ob-
jects. At a rough estimation you can calculate with:
L⋅ ω re a l p a rt
R fg ≈ 1 MHz * |Z|min / Ohm
a n o d e s id e c a th o d e s id e
p o ro u s to ta l to ta l
in d u c tiv e O h m ic
p o ro u s As a consequence, the window for getting an Ohmic
d is trib u te d d is trib u te d
im p e d a n c e s h a re lo s s im p e d a n c e resistance information by means of the EIS gets
smaller for bigger cells. For certain systems, the
L R window will be closed.

What can be done to complement the EIS under

. these conditions? The question is answered by the
Fig. 2: The appearance of ohmic share and stray induc- well known current interrupt technique, which does
tance in a fuel cell spectrum not need the knowledge of two signals simultane-
ously. The principle is depicted in figure 4 and ex-
EIS allows to separate all contributions and to deter- plained in the following:
mine the ohmic part in the high frequency region of a
spectrum, where the impedance curve intersects the elec tro nic s w itc h po ten tia l
real axis. The inductance is shown at successive m e as ure m en t
higher frequencies in the diagram. But there is one
great restriction: Measuring impedance always
means to measure two signals, current and voltage.

* E
Z = ∼


loa d o bjec t

s tea dy s ta te
c u rre nt I

ze ro c urr en t

Z = I m p e d a n c e o f t h e O b je c t o f In t e r e s t
W i th th e E x te n s i o n s in S p a c e x , y , z .
C u rre n t line ar
∼ s tep ∆ E
E n o is e ∝ ω , x, y, z, I p o ten tial
u n de r loa d O hm ic s h are =
I n te r f e r in g ∆ E
Current Flow

P o te n ti a l
E n o is e 1 tim e o f I
in te rru p tio n T im e
P otential

M agn. M e a s u re d Fig. 4: Principle of current interrupt technique to deter-

Z P o te n t ia l
C o u p l.
mine the ohmic share

I n te r f e r in g
P o te n ti a l
E n o is e 2
A steady-state current is interrupted by a switch. The
step response of the potential is sampled and ana-
lysed assuming that the current drops instantane-
Fig. 3: Basic impedance measurement circuit – ously from its stationary value to zero.
principle (top) and detail (bottom)
In practice, the settling time depends on the electro-
A closer look at the circuit in figure 3 shows that the magnetic energy stored in the parasitic capacity and
potential information does not only contain the inter- inductivity of the cell arrangement on the one hand
esting part from the site of the connecting terminals. and the damping process on the other hand.

Represented by TEST SOLUTIONS (770)410-9166

Electrochemical Applications 1/02 page 3

Provided that the set-up is built appropriate, the inter- spectrum and all parasitic effects can be analysed by
ruption results in a breakdown of the current to at means of EIS methods.
least small values within a short time. In this case,
total total
the potential will be disturbed much less by mutual porous damped
induction compared with an EIS measurement. distributed

In theory, the ohmic contribution to the overall im- L R

pedance can be easily seen from the height of the

fast rectangular step of the potential. For the evalua-
tion a linear step model is commonly used.

E /m V T im e / µ s C sw
R dmp R sw

-5 0 C sw,R : Capacitance
sw and resistance of the electronic switch.
R dmp: Damping of the inductive share due to spatial distribution.
-1 0 0

Fig. 6: Approximate equivalent circuit for a complete high

-1 5 0
current interrupt measurement set-up of an electrochemi-
cal power source device.
-2 0 0

0 200 400 600 800

In figure 6, a simplified equivalent circuit for a com-
E /m V T im e / µ s plete HCI measurement set-up of a power cell is
shown. It contains the impedance of the active cell
part (circle), the integral inductance (L) and resis-
tance (R) and the parasitic effects of the switch cir-
-5 0
cuit. The resonance circuit is mainly built of the series
inductance, the double layer capacity and the capac-
-1 0 0
ity of the electronic switch. It is responsible for the
overshot and “ringing” in the pulse response signal.
-1 5 0

T im e
-2 0 0 E (t)
10 15 20 I0 fro m
c alib ra tio n
I(t) E0 pro c e du re
Fig. 5: Typical current interrupt potential step response. 0

Long (top) and short (bottom) term response of a single

cell PEM fuel cell at 80 A. H a n n in g W in d o w
f(t) = E (t) ⋅ H (t)
But this evaluation suffers from the fact that the h ig h prec is ion
c oa x ia l re s is to r
analysis of the time domain data is interfered by the Z o om -F F T
“ringing” in the signal as a result of the parasitic G ( j ω ) = F (f(t))
resonance. In addition, the early phase of the re-
sponse is characterized by a non-linear behavior due A m plitud e R efe re nc e
to the imperfect characteristics of the electronic S pe c tru m S pe c trum
switch. Furthermore, the response of both double-
layers may follow soon after the interruption when the
freq ue nc y freq ue nc y
concentrations of the involved species turn from their
steady state values to new ones without load. All
S c a la r Im p ed an c e
these effects bend and distort the expected ideal S p e ctru m
shape of the potential step. Therefore, the automatic
im pedance

analysis of pulse measurements by means of a sim-

ple fit to a linear step model often leads to inaccurate
results. freq ue nc y

Our aim was to improve the method in order to get

results of comparable reliability to the EIS. The basic Fig. 7: Principle scheme of the Zahner high current inter-
idea is not to evaluate the distorted step function in rupt data processing.
the time domain. Instead, after a transformation of
the data into the frequency domain, the resulting Figure 7 illustrates the essential steps for the trans-
formation of the time domain data into the frequency
Represented by TEST SOLUTIONS (770)410-9166
Electrochemical Applications 1/02 page 4

domain. The potential response signal E is sampled pared to the automatic minimum detection in the
by a transient recorder. scalar impedance function.
The numeric algorithms use discrete Fourier trans- Figure 10 is a sketch of the practical set-up of our
form methods to achieve an effective analysis. In high current interrupt measurement arrangement of a
order to minimize the errors caused by their applica- fuel cell. The electrochemical cell (A) is supplied by
tion on single events, a weighing function has to be means of an electronic load (B) or another type of
applied. At least, a Zoom FFT calculates the ampli- high power potentiostat to force the steady state load
tude spectrum in the frequency domain. A similar conditions. Additionally, the potentiostat acts as fast
procedure using a reference resistor was done for electronic switch for the current interruption.
calibration. The quotient of both spectra finally leads |impedance| / Ω
to the modulus of the impedance of the unknown 10m

|impedance| / Ω
10m 1m

f0 f1
1m 1K 10K 100K 1M 8M
frequency / Hz

Zmin ⇓
2 π dϕ
π η∫
100u ln | Z (η) | ≈ const. + ϕ (ω )d ln ω − ⋅ (η)
6 d ln ω
1K 10K 100K 1M 8M 0

frequency / Hz ⇓
Fig. 8: Automatic evaluation of the ohmic share from the phase ο
scalar impedance function. 90

This scalar impedance spectrum can be used to

evaluate the ohmic share in a simple, automated 45
way: The user selects a reliable frequency range for
analysis, which excludes the parasitic resonance at
the high frequency end. The impedance minimum
within this range represents the ohmic share with an
acceptable accuracy of about 1 to 3%. -45

If you want to evaluate the response spectrum with

the standard methods of the EIS, beside the imped- -90
ance, the phase data will be necessary: For this cal- 1K 10K 100K 1M 8M
culation a relation between impedance and phase for frequency / Hz
all two pole impedance objects of minimum phase
can be used. The ZHIT relation allows to calculate
the modulus of the impedance course from the Fig. 9: Calculation of the phase angle from the impedance
course of the phase angle. It will also be able to per- modulus by means of the inverse application of the ZHIT
form the inverse application, if one uses the ZHIT in transform. The ZHIT equation shows that the impedance
modulus at the frequency η can be calculated from the
an iterative numeric way. This is the way, our analy-
integral of the phase angle course ϕ(ω) within limited
sis software obtains a complete spectrum. frequency boundaries (η 0 to η). A correction term propor-
tional to dϕ/dlnω enhances the accuracy.
The complex spectrum can now be analyzed in the
usual way, for instance, by means of simulation and
fitting of the equivalent circuit. According to our ex- The potentiostat is controlled by an electrochemical
perience, the ohmic share can be determined in this workstation (D) including a high resolution transient
way, with typically the double or triple accuracy com- recorder. The recorder input is connected to the po-
tential sense lines of the cell along a so called pulse
W. Ehm, R. Kaus, C.-A. Schiller, W. Strunz: ZHIT - a simple probe (C). The main task of the pulse probe is the
relation between impedance modulus and phase angle... Electro- galvanic isolation of the potential sense circuit from
chemical Society Proceedings 2000-24, 1
Represented by TEST SOLUTIONS (770)410-9166
Electrochemical Applications 1/02 page 5

the instrument in order to minimize electromagnetic We also found that the HCI worked fine with power
interference. On the other hand, it is responsible for generating devices. As an example, the results of an
the protection of the instrument input by means of an EIS and a HCI experiment at a high temperature fuel
energy consuming clipping circuit. cell are depicted in figure 12. The cell has been
driven with air and humidified hydrogen and gener-
After a short ‘sampling shot’ during the interrupt, the ated more than four Amperes at a potential exceeding
instrument switches on the current again, in order to 0.73 Volts. In both experiments the best case
re-establish the steady state and to avoid potential
damage of the cell. The pulse response is analyzed
then by the software of the workstation as described wiring was used to reduce the mutual induction con-
before. tribution.

The original HCI potential step response is plotted on

top of figure 12. At the bottom, the transformed im-
pedance (circles) as well as a comparative imped-
ance measurement (rhombi) are drawn. As one can
see, the methods complement each other for the
different frequency ranges. In this special case, the
frequency limit for the EIS experiment to get accurate
information on the ohmic share is not exceeded.
Therefore, both methods deliver the correct results.

impedance Ω
10m Ω


10 1K 100K
frequency / Hz
Current load lines
Potential sense lines

Fig. 10: Practical set-up of a high current interrupt fuel cell phase ο
measurement arrangement
10m Ω

A lot of test experiments under controlled conditions
have been done. The major experience is that, com-
B -45

pared with the standard EIS, the HCI is much less

sensitive to mutual induction artefacts. This is illus- 10 1K 100K
trated by the example depicted in figure 11. Here, we frequency / Hz
made test measurements at resistors using two dif-
ferent, intentionally non-optimized connection geome- Fig. 11: Results of test measurements at reference resis-
tries. As you can see, the strong in-phase mutual tors incl. mutual induction (MI) components.
induction of the above example leads to an inductive
component which seems unrealistically high in the A: In phase MI leads to a high inductive response in the
case of the standard EIS. The HCI method, however, case of EIS (full dots) whereas the HCI spectrum (squares)
matches the theory.
leads to the almost exact value. B: Out-of-phase MI causes paradox capacitive (-) EIS
phase courses (squares) while the HCI phase course
With the other example, which shows a strong out-of- shows the correct phase sign for inductance (+).
phase mutual induction, one obtains almost the same
impedance functions, which has been omitted here.
Yet, the phase diagram shows paradox behaviour for
the standard EIS curve: The course indicates capaci-
tive characteristics! The phase curve of the HCI ex-
periment shows the correct sign due to its origin from
the ZHIT transform.

Represented by TEST SOLUTIONS (770)410-9166

Electrochemical Applications 1/02 page 6

E /m V T im e / µ s on a big PEM fuel cell at a current of 80 A were per-

formed. The shift of the increase of the impedance for
the transformed HCI-data (circles) to higher frequen-
cies indicates the smaller sensitivity of the HCI
0 measurement against the mutual induction.

-1 0 0 In our opinion, the missing correspondence between

the EIS- and the HCI-data at the low frequency end of
-2 0 0 1.1 KHz results from a non-linear component in the
long term pulse response signal. HCI analysis has to
rely on the rule of linearity. The transient changes of
0 200 400 600 800
about -10 mV within the 0.85-milliseconds-analysis-
interval may be enough to violate this rule.
im p e d a n c e Ω

30m 1. EIS capabilities are basically limited by mutual
induction at the high-frequency low-impedance edge.
This falsifies significantly the results for ohmic and
inductive share.
2. The HCI capability is limited by the magnetic en-
100m 10 1K 100K ergy stored in the load circuit.
fr e q u e n c y / H z 3. The HCI analysis can be automatically analyzed
reliably by transforming the time data into the fre-
Fig. 12: High current interrupt measurement of a single quency domain.
solid oxide fuel cell at 866°C. 4. According to our experience, HCI can extend the
EIS: 100 mHz – 100 kHz (rhombi) available frequency range about a factor of three to
HIC: 1 kHz – 800 kHz (circles) ten in a carefully optimized experimental set-up.
5. HCI data interpretation should not be extended to
The last example demonstrates that the dominating the low frequency response. The unavoidable viola-
error from the unavoidable mutual induction falsifies tion of the EIS linearity rule after a certain interruption
the result of standard EIS measurements at very low time may lead to misinterpretations.
impedance objects. In the experiments depicted in 6. Thus, an arrangement which performs both, the
figure 13, a HCI measurement (top) as well as a standard EIS and the HCI measurement within one
comparative EIS measurement (bottom, triangles) set-up, is the best choice for the challenges of elec-
trochemical power source device testing.
E /m V T im e / µ s

-5 0

-1 0 0

-1 5 0

-2 0 0

0 2 0 0 4 0 0 6 0 0 8 0 0

im p e d a n c e Ω
For N. American support contact:

1 m 5665 Hwy 9 N, Ste. 103-181
Alpharetta GA 30004
3 0 0 u
T: 770-410-9166
1 0 0 m 1 0 1 K 1 0 0 K
fre q u e n c y / H z
F: 443-947-0745
C: 678-231-5783
Fig. 13: High current interrupt measurement of a single
PEM fuel cell at 85° C and 80 A. The resulting spectrum
(C) is compared with a standard EIS (D).

Represented by TEST SOLUTIONS (770)410-9166