Professional Documents
Culture Documents
PII: S0263-2241(16)30373-6
DOI: http://dx.doi.org/10.1016/j.measurement.2016.07.014
Reference: MEASUR 4201
Please cite this article as: S. Grassini, S. Corbellini, M. Parvis, E. Angelini, F. Zucchi, A simple Arduino-based EIS
system for in situ corrosion monitoring of metallic works of art, Measurement (2016), doi: http://dx.doi.org/10.1016/
j.measurement.2016.07.014
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A simple Arduino-based EIS system for in situ
corrosion monitoring of metallic works of art
Abstract
Metallic artifacts of archaeological and historical interest, exposed to outdoor
environmental conditions, can be affected to a great extent by degradation, due
to the presence of aggressive compounds in the surrounding environment.The
development of suitable preventive conservation strategies for these important
signs of our culture, requires their corrosion behavior to be investigated in order
to understand the electrochemical phenomena that occur on the metallic surface.
This investigation can easily be performed in a laboratory by means of different
chemico-physical techniques, which provide information on the composition,
microstructure and morphology of the corrosion products, as well as by means of
electrochemical measurements, which allow the corrosion resistance of the metal
to be estimated. Unfortunately, these measurements are often invasive and
require microsampling. Electrochemical Impedance Spectroscopy (EIS), thanks
to the its high degree of sensitivity, and to the small perturbation applied to the
corroding system, can be considered a useful non-destructive testing method to
obtain valuable information on the conservation state of metallic works of art.
This paper describes an innovative, portable, low-cost and user-friendly solu-
tion for EIS in situ measurements, based on the widely diffused Arduino board,
and on specifically designed measuring probes. The proposed solution can-
not provide the detailed knowledge of the corrosion mechanisms that occur on
metallic surface, but it can be an extremely useful tool for conservators and art
historians for the assessment and long-time monitoring of the stability of the
artifacts.
Keywords: EIS, Metallic artefacts, Corrosion measurement, Cultural
Heritage, Portable instrument
This paper deals with the conservation state of metallic artefacts of archae-
ological and historical interest, exposed outdoors, which are affected by severe
degradation, due to exposure to aggressive environmental conditions, in par-
ticular in most modern cities. Different corrosion product layers and patinas
can grown on an artifact surface, depending on alloy composition and environ-
mental conditions, and these can affect the corrosion behaviour of the metallic
substrate. For example, patinas formed on Cu-based alloys can have a par-
tially protective effect, while the atmospheric corrosion phenomena that occur
on iron artifacts can lead to the formation of active corrosion product phases,
whose contact with metal can accelerate degradation.
The development of suitable preventive conservation strategies for this im-
portant evidence of our culture requires their corrosion behaviour to be inves-
tigated in order to understand the electrochemical phenomena that occur on
the metallic surface [1]. In this context, the possibility of performing in situ
measurements, which allow an artefact to be characterized in its environment,
and long-lasting monitoring campaigns to be set up to assess its stability, as a
function of the exposure time, is of utmost importance [2], [3].
Moreover, it should be taken into account that large artefacts, such as bronze
statues, iron structures and reinforcements in historical buildings and archaeo-
logical sites cannot be moved easily to a laboratory; therefore, the assessment
of their conservation state requires a completely different approach. Microsam-
pling carried out to obtain small specimens for the analysis by means of differ-
ent analytical techniques can only be performed in particular conditions, and
for these reasons it is important to find solutions that are suitable for in-situ
measurements and which respect the integrity and aesthetic appearance of the
artefact[4].
This paper describes a possible solution that relies on Electrochemical Im-
pedance Spectroscopy (EIS) measurements, and which is based on a simple
portable instrument, and on some polymeric measuring probes, specifically de-
signed and developed for non-invasive in-situ measurements.
EIS is generally used to investigate the protective properties of organic coat-
ings [5, 6, 7], but can also be employed to study the corrosion mechanisms
that occur on metallic surfaces, and to estimate the corrosion rates [8], [9]. EIS
measurements are capable of discriminating between different kinds of corrosion
behaviour when coupled with a suitable data analysis. They can be proposed
as a simple solution for the monitoring of the conservation state of an artifact
to help restorers and conservators develop tailored conservation strategies.
2
ICW Counter
Reference
electrode electrode
VAC
Electrolytic
RS
solution
VWR
Corrosion Cp1 R1
VDC layer
Cp2 R2
Metallic artifact (working electrode)
Figure 1: EIS measuring principle: an electrolytic solution is put in contact with the sample
and the impedance is measured between the artefact, which is usually referred to as “working
electrode”, and a “counter electrode” usually made of a noble metal, such as Pt. A third
electrode referred to as reference electrode is employed to measure the so called open corrosion
potential, however the impedance spectrum can be obtained also in the absence of the reference
electrode.
3
is limited to a few millivolts, usually no more than 10 mV, to avoid stimulat-
ing any non linearity of the system under investigation and accelerating the
corrosion process.
The EIS system measures the ICW current that flows through the metallic
surface, i.e. between the counter and working electrodes, and computes the
surface impedance as:
VWR
Z= (1)
ICW
where VWR is the alternating voltage between the working and reference elec-
trodes. All the values in eqn. 1 are shown in boldface as they are complex
numbers. If the impedance between the counter and reference electrodes is
negligible with respect to the impedance of the corrosion layer, VWR can be
measured between the counter and working electrodes, thus avoiding the neces-
sity of using of the reference electrode.
The development of a two-electrode EIS measuring approach simplifies the
procedure used to perform in situ measurements and, consequently, should en-
courage the employment of the EIS technique in the cultural heritage field,
where simple to use and fast response methodologies are required.
When EIS measurements have to be performed on historical and archaeo-
logical works of art to asses their conservation state, non aggressive electrolytic
solutions should be used, in order to minimize their effect on the degradation
process. In principle, the measurements can be performed in mineral water, but
its very low conductivity has to be taken into account, in particular if the metal-
lic artefact is coated with a very thin corrosion product layer. In this case, low
aggressive electrolytic solutions, such as a sodium sulphate solution (N a2 SO4 ),
can be employed.
However, it should be considered that it is not possible to measure the solu-
tion resistance with a two-electrode cell, thus, since this may affect the measure-
ment, such a resistance has to be estimated on the basis of the electrode/cell
geometry, as well as the electrolyte chemical composition and conductivity.
In principle, the measurement can be performed on any surface area, but
the impedance values are generally scaled to an exposed area of 1 cm2 in order
to enable easy comparisons. The impedance is measured for frequencies in the
range of 0.01 Hz to 100 kHz. The impedance measured values obviously depend
on the nature of the surface layer and can vary from a few tens of ohms per
square centimetre, for bare metal, up to several gigaohms for a metallic substrate
coated with a thick, compact and non-conductive layer.
The EIS spectra are then analysed to determine simple electrical compo-
nents, such as resistors, capacitors, and inductors, of the equivalent circuit
models that represent the coated metal/solution interface. These circuits re-
produce the electrical properties of the system under investigation, and the
different elements are assigned to different physical elements, but there is not
a unique correspondence between the EIS data and the equivalent circuit, and
different circuits can be employed to model the same impedance spectrum.
4
Several approaches can be followed for this identification (see for example [10,
11]) but, due to the low electronic conductivity and thickness of the corrosion
product layers that are usually present on the surface of metallic archaeological
and historical artefacts, the impedance spectra tend to have a mainly capacitive-
like behaviour, at least at mid frequencies. However, the actual behaviour over
a large frequency range, e.g. spanning 7 decades, requires equivalent circuit
models with several electrical components, in order to take into account the
electrode kinetics, diffusion phenomena and the heterogeneous nature of the
corrosion product layer.
Electrochemical processes often exhibit complex behaviour and ideal capac-
itors are not able to model all the EIS spectra, as for example, in the case of a
non-uniform surface layer, surface roughness or an inhomogeneous current dis-
tribution; thus, some specialized electrical elements such as the Warburg and
Constant Phase Element (CPE) have been introduced to overcome this short-
coming and to develop equivalent circuit models that provide useful and mean-
ingful results.
The CPE has the following impedance:
1
Z= (2)
Q(jω)α
where j is the imaginary unit, Q is the CPE value and α is an exponent in the
range of 0 to 1 which characterizes the CPE. The CPE can model a resistor
(α = 0), a capacitor (α = 1), a pure diffusion phenomenon (α = 0.5) and/or
mixed conditions when α assumes intermediate values.
As an example, fig. 1 shows a typical circuit, based on the circuits pro-
posed by Randles in 1947 [12]. The circuit contains a resistor that models the
ohmic resistance of the electrolytic solution, and two cells, composed of a re-
sistor and a CPE, which model the phenomena that occur inside the corrosion
product layer and at the interface between the corrosion product layer and the
solution, respectively. This circuit can be employed, for example, to model the
electrochemical behaviour of an iron artefact exposed to outdoor atmospheric
corrosion. It relies on the flexibility of the CPE to model the presence of a
heterogeneous surface, and the diffusion processes through the porous material
in the corrosion product layer.
This circuit has 7 parameters (i.e. 4 parameters for the 2 CPEs, plus 3
parameters for the resistances) which have to be adjusted to fit the experimental
data [13].
In this case, the data fitting is not easy, and the values obtained for the
circuit components are often questionable since, as already mentioned before,
the impedance values can span 6-8 decades and a fitting procedure that is simply
devoted to minimizing the sum of squares would therefore mainly weight on the
high impedance values, i.e. on the low frequency values:
X 2
min |Ze − Zm | (3)
5
where Ze are the measured experimental points and Zm are the impedance
values obtained by the model.
Several approaches can be followed to minimize this effect such as the ones
described in [11]. Another simpler possibility is the use of a different metric,
such as:
X |Ze − Zm |2
min 2 (4)
|Ze |
which weights the experimental data regardless of their actual values, and is
used in the examples considered in this paper.
It is important to observe that the presence of noise on the measured data,
which mainly arises from the extremely small current amplitudes, makes the
fitting problematic: for example, when the impedance is of the order of 109 Ω,
with a voltage amplitude of 10 mV, the current is of the order of 10 pA and the
noise contribution can therefore become quite important.
All these problems are usually tackled by designing costly instruments, equip-
ped with complex input front-ends [14], [15] based on variable gain amplifiers
and variable filters. Such a solution is effective, but limits the possibility of
arranging multiple measuring systems that can used in parallel to reduce the
measuring time. In fact, since the lower frequency used for the measurement is
usually 0.01 Hz, which corresponds to a period of 100 s, a measurement session
generally requires no less than 20 min.
6
Logarithmic processing
Digitized signals
scaled to the same
amplitude
Log Amplifier
AntiLog &
DC shifter
DC rem.
Signals with
In (µA)
amplitude ADC
ratio 1:100
Out (V)
compressed signal is digitally converted and the obtained values are eventually
passed through a digital anti-log function to recreate the original signal. It is
easy to see how in this case both signals can be recreated making the minimal
quantization noise.
The overall process obviously heavily relies on the correct choice of the DC
shift [16] and the proposed solution therefore offers real advantages when sine
signals are used; nevertheless, in the proposed application, it permits quite
interesting results to be obtained.
The block diagram of the proposed instrument is shown in fig. 3. The system
is composed of two boards: one board is an off-the-shelf Arduino Due microcon-
troller, while the other board is arranged as a shield that has to be connected to
Arduino Due. Arduino Due is based on a 32 bit microcontroller, which contains
a 12 bit ADC and two 12 bit DACs. A complete EIS solution which employs the
internal DACs and ADC of the Arduino board can be arranged [16], with good
results. However, the system flexibility can be increased if the ADC embedded
inside the Arduino microcontroller is used to sample the signals, and a separate
DDS, allocated on the shield, is employed to generate the stimulus.
The DDS is paced by the Arduino clock to ensure a synchronous sampling
with the generated signal and to enable an equivalent time acquisition that also
permits several samples to be acquired at the maximum measuring frequency,
thus overpassing the limit connected to the maximum sampling frequency of
the ADC, which is of 1 MHz. Since the complete system works in synchronous
mode, the frequency of each signal is known in advance and the acquired samples
can therefore be processed by using a simple conventional 3 parameter sine fit.
The bias that is required to keep the input signal positive and to allow
the logarithmic converter to work correctly, is obtained by using a 16bit DAC,
7
Shield Arduino
Buffer Switch
DDS DA DA Cell
LogAmp
Arduino Due
512kB
Flash Digital I/O ADC
Clock
96kB
SRAM USB
PC
installed on the shield. The DAC output is fed into a diode, so that it is easy
to obtain the very large range of bias currents that are required to cope with
the large range of input currents. The relationship between the DAC output
voltage and the bias current is of course not know and it depends on the actual
temperature. The bias is therefore continuously updated, by means of a feedback
loop, which keeps the input signal in the correct logarithmic amplifier interval.
This bias can be easily removed from the sampled values by means of the sine-fit
algorithm [16].
A digital switch has to be used to disconnect the signal generator output
from the artifact, thus permitting the open corrosion potential (E OCP ) to be
monitored: when a measurement is started, the instrument measures the E OCP ,
sets the the value of the DC component to such value, closes the switch, and
then starts the measurements.
Fig. 4 shows the proposed system, along with a small box, which is used to
enclose the Arduino Due board and the shield that contains the analogue cir-
cuits. Table 1 shows the most important specifications of the proposed system.
The impedance magnitude and phase uncertainty have been measured by using
standard components thus in the absence of varying corrosion potentials, and by
adjusting the parameters of the logarithmic amplifier which is the most critical
component so that worse values can be expected on less favourable conditions.
8
Arduino Shield
9
Some commercially available probes that exploit a fixing solution are based
on a magnetic ring. However, these devices can only be used on ferromagnetic
materials. In addition, in order to have an acceptable magnetic force, these
probes have to have a diameter of the order of 70 mm, so that they only can be
used on almost completely flat surfaces.
Fig. 5 shows the measuring cell specifically designed to be fixed onto complex
shaped metallic structures by means of a removable double-side bonding tape.
The cell is made of acrylonitrile butadiene styrene (ABS) and has an external
diameter of about 30 mm, and a thickness of about 20 mm. The cell has a
measuring chamber with a diameter of about 8 mm, and, therefore an exposed
surface area of about 0.5 cm2 . The cell can be filled with the electrolytic solution
by injecting it into the inlet pipe; the air in the cell flows out from the outlet
pipe shown in fig. 5, where a section of the electrode is depicted.
The cell is equipped with a 2 − 3 mm thick adhesive closed-cell polyurethane
disk which makes it possible for the cell to be used on non-perfectly flat and/or
rough surfaces. This pad is fixed to the artefact by means of a 0.3 mm thick
bi-adhesive disk. The selection of the adhesive disk is of utmost importance
since the glue must be strong enough to prevent the electrolytic solution from
pouring out, but not too strong to prevent the cell from being removed after
the measurement.
Outlet tube
Inlet tube
Measuring
Pt counter chamber
10
Reditus ad origines
5
10
|Z| (Ω)
50
θ (deg)
0
−2 −1 0 1 2 3 4 5
10 10 10 10 10 10 10 10
Frequency (Hz)
Figure 6: EIS spectra collected on the Reditus ad origines weathering steel sculpture, exposed
to outdoor environmental conditions in Ferrara, Italy.
11
area of 1 cm2 .
The impedance modulus, |Z| value, indicates the extent of the corrosion
phenomenon at the metal/corrosion product layer interface. The higher the
impedance value, the higher the electrochemical stability of the rust layer and,
consequently, the higher the protective effectiveness against further corrosion.
The EIS spectra recorded on the red-brown area (black and red lines in the
Bode plot) highlight the higher stability of the corrosion product layer, a result
that is confirmed by the higher impedance value, while the orange-red layer
(pink line in the Bode plot) shows different electrochemical behaviour, and a
lower impedance value. Moreover, it should be noted that the black line has a
clearly different phase trend from that of the red and pink plot traces. This can
be explained by considering the different porosity of the surface layers, which
can easily be observed by looking at the small insets, which show the artefact
surface where the plots have been obtained.
Fig. 7 shows the results of the data fitting, performed using the two-cell
model described in fig. 1, which accounts for the impedance of the two in-
terfaces: metal/corrosion product layer and corrosion product layer/solution,
respectively.
The three pictures show the impedance modulus of the three points, using
the same colours as fig. 6 (i.e. black, red, and magenta). In addition, the
impedance values of the two cells for each plot are shown in green and blue,
and the values of the corresponding parameters are reported on the right. It
is easy to see how the CPE cell, whose contribution is reported in blue, is
responsible for the impedance at low frequency, while the other CPE cells are
mainly responsible for the high frequency behaviour. As expected the CPE that
accounts for the behaviour at low frequencies has an exponent which makes it
almost appearing like a capacitor thus accounting for a double layer capacitance.
Such a capacitance is lower for the sloped surfaces, as expected, due to the
greater thickness, and the polarization resistance varies accordingly. A real time
constant cannot be determined for these CPE cells since the reactive component
is not a capacitor, nevertheless values of the order of 35 − 55 s are obtained in
all cases.
The second CPE cell, which accounts for the corrosion layer morphology,
only appears to be important in the case of the black line: the parallel resistance,
which is usually interpreted as a charge transfer resistance, has a non-negligible
value in this case. Therefore, the Impedance Spectrum highlights that, only
in this case, the corrosion layer reaches a consolidated state, which makes it
protective with respect to the corrosion process. In the other two cases, either
due to the vertical orientation, which does not permit the formation of a thicker
corrosion layer, or due to a still under-developed layer, the corrosion process
can still proceed.
5. Conclusions
The monitoring of the corrosion processes of artefacts exposed to the en-
vironment for very long times is a problem of utmost importance, both as far
12
5
10 R1=3.5MΩ
Cp1= 16µF α= 0.9
|Z| (Ω)
2
10 R2=12kΩ
Cp2= 110µF α= 0.4
−1
10
5
10 R1=2.0MΩ
Cp1= 16µF α= 0.84
|Z| (Ω)
2
10 R2=0.5kΩ
Cp2= 500µF α= 0.2
−1
10
5
10 R1=0.5MΩ
Cp1= 65µF α= 0.84
|Z| (Ω)
2
10 R2=0.1kΩ
Cp2= 150µF α= 0.7
−1
10
−2 −1 0 1 2 3 4 5
10 10 10 10 10 10 10 10
Frequency (Hz)
Figure 7: Result of EIS data fitting using the the two-cell equivalent model of fig. 1.
13
as the stability of the artefacts and their aesthetic appearance are concerned.
Unfortunately, most of the analytical techniques generally used in laboratories
require an artefact sampling and therefore cannot be used routinely.
From this point of view, EIS could be an interesting approach, provided that
a comprehensive solution is available for in-situ use, directly on the artefacts.
This paper proposes the use of a cheap and simple Arduino board, coupled
with a shield to set-up a simple portable EIS instrument, and an adhesive elec-
trode, which permits measurements to be obtained in the field in an almost
non-invasive way. The cost of the instrument is of the order of e 150 while
the electrodes have a negligible cost. The system does not require a dedicated
power supply and just needs to be connected to a normal PC equipped with a
USB port to work.
14
[10] G. Qiao, Y. Hong, J. Ou - Corrosion monitoring of the RC structures
in time domain: Part I. Response analysis of the electrochemical transfer
function based on complex function approximation - Measurement, Vol. 67,
pp. 78-83, 2015
[11] G. Qiao, Y. Hong, J. Ou, X. Guan - Corrosion monitoring of the RC struc-
tures in time domain: Part II. Recognition algorithm based on fractional
derivative theory - Measurement, Vol 67, pp. 84-91, 2015
[12] J.E. B. Randles Kinetics of rapid electrode reactions Discussions of the
Faraday Society, vol. 1, no 11, 1947
15