Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

38th European Photovoltaic Solar Energy Conference and Exhibition

IMPROVED MEASUREMENTS OF THE CONTACT RESISTIVITY OF ECA-BASED JOINTS

María Ignacia Devoto*, Tudor Timofte, Andreas Halm, Daniel Tune


International Solar Energy Research Center (ISC) Konstanz
Rudolf-Diesel-Str. 15, D-78467 Konstanz, Germany
Corresponding author*: +49 (0) 7531 36183341; ignacia.devoto@isc-konstanz.de

ABSTRACT: Electrically conductive adhesives (ECAs) can replace soldering for the interconnection of cells within
photovoltaic modules such as those based on IBC-based technology or the shingling interconnection approach. They
can also be used for applications in which soldering is no longer appropriate or lacks long-term reliability, such as
temperature-sensitive SHJ-based technology or vehicle-integrated photovoltaics. The aim of this work is to show
further improvements in the development of a method to characterize the contact resistivity of joints based on these
adhesives. The developed method, which is based in the Transmission Line Method, is shortly explained while the
treatment of the raw data and the obtained results are clarified in detail. Samples were manufactured using nine different
commercially available adhesives that contained different polymer matrices (acryl and epoxy) and conductive filler
(silver and copper with silver coating). It is concluded that the introduction of the adhesive significantly increases the
complexity of the system under study and increases the number of calculations required to remove non-important
resistive components so that the contact resistivity of the joint can be extrapolated. It is also shown that although
measurements per sample are statistically significant, a larger number of probes should be manufactured so as to
extrapolate a non-misleading value. This means that many measurements per sample with several samples per ECA
composition are required.

Keywords: see the list of keywords

1 INTRODUCTION neither simple nor straightforward.


In this work, previous manual studies are expanded by
Rather than the traditional soldering approach, several performing the measurements in an automated way. Only
commercially available alternative module concepts use two contacting events are required to obtain a very large
electrically conductive adhesives (ECAs) as a means of number of measurements from each sample, thereby
cell interconnection. Typically, concepts such as minimizing any measurement differences due to contact
shingling, conductive backsheets, and heterojunction cells damage. Many measurements are carried out after each
use adhesive because soldering is not suitable for them. contacting event and these are used to reduce the standard
The shingling approach requires direct joints between deviation of the outcomes, increasing their statistical
busbars while conductive adhesives need to perform the significance. The sample design is chosen according to
interconnection process during lamination. Heterojunction what previous work has shown is best. A revised
cells are sensitive to high temperatures and their methodology is also used to treat the raw data.
passivation can be damaged by soldering. ECAs possess
several advantages over soldering; they are lead-free,
mechanically flexible, and can better tolerate thermal 2 SAMPLE DESIGN AND ADHESIVES
cycling and vibrating environments. In addition, they can
be precisely dispensed through screen/stencil printing or 2.1 Sample design
pressure-time/jet dispensing. Among the main In our previous work, several designs were tested in
disadvantages of ECAs compared to soldering are the order to find the best sample design, in which the contact
higher per-module cost of the material and the initial behavior is found to be ohmic. The sample design used in
investment for new tools/machinery. this work is shown in Figure 1.
For ECAs to gain wider acceptance in photovoltaic
(PV) manufacturing they need to be further studied.
Physical, electrical, and mechanical characterization is
necessary to compare different adhesive compositions, test
electrical and mechanical performance of ECA-based
joints, and to have a means of quantifying the impact of
accelerated aging tests at adhesive and joint level. To
quantify the electrical performance of conductive joints,
the contact resistivity is a useful figure of merit.
Geipel et al. used the Stacked Greek Cross (or Cross
Bridge) structure with a four-point probe measurement to
extract the contact resistivity, concluding that due to
current crowding, contact resistivity is overestimated.[1]
The transmission line method (TLM) has proved to be a
useful method for electrical characterization of ECA-
based joints and similar applications. [2]–[5] Nevertheless,
TLM has been revised several times in the last decades and Figure 1. Sample design. The contact under study
several improvements have been made.[6]– [11] However, is highlighted with a green dotted line.
although its use is widespread in the microelectronics
community and it is also used in the PV field, its usage is

627
38th European Photovoltaic Solar Energy Conference and Exhibition

To study different contacting area sizes, six different R@2PHF = 2𝑅𝑐′ + 2𝑅𝐻𝐹 + 2𝑅𝑐 + 𝑅𝑏 (2)
samples were designed with different finger widths (0.2,
0.4, 0.8, 1.0, 1.2, and 1.5 mm). R@4PWF = 𝑅𝑏 (3)

2.2 ECA composition under study R@4PHF = 𝑅𝑏 (4)


Nine commercially available ECAs were tested. Two
different polymeric materials were chosen: acryl and where 𝑅𝑐′ is the contact resistance between the pin and the
epoxy, and two different metals were chosen as fillers: finger (in Ω); 𝑅𝑊𝐹 and 𝑅𝐻𝐹 correspond to the resistance
pure silver and copper core with silver coating. A measured when contacting on the whole and half finger,
summary of the tested ECAs is shown in Table I. respectively (in Ω); 𝑅𝑐 is the resistance of the contact (in
Ω); and 𝑅𝑏 is the bulk resistance of the ECA line (in Ω).
Table I. ECA composition of adhesives under test. Equations 1 to 4 assume the following:
Polymer Filler Best performance
Name
material material application  The resistance of the pins is negligible.
ECA 1 Acryl Ag Silicon-to-Silicon  For R@4PWF and R@4PHF, the voltage drop
ECA 2 Acryl Ag Silicon-to-Silicon
along the sensed fingers is negligible (this was
Silicon-to-Silicon
ECA 3 Acryl Ag tested with Keithley 2602A and is in accordance
Silicon-to-(Sn)Ribbon
ECA 4 Epoxy Ag Silicon-to-Ribbon with manufacture references).
ECA 5 Epoxy Ag Silicon-to-Ribbon
Silicon-to- To extrapolate the contact resistivity for a particular
ECA 6 Acryl Cu(Ag)
(Sn/SnPb)Ribbon sample using TLM, only 2𝑅𝑐 + 𝑅𝑏 should be plotted. To
Silicon-to-Silicon remove the 𝑅𝑊𝐹 and 𝑅𝐻𝐹 components, another TLM is
ECA 7 Epoxy Cu(Ag) Silicon-to- performed using R@2PWF and R@2PHF, as shown in
(Ag/Ni)Ribbon Figure 3. The distance on the x-axis corresponds to twice
ECA 8 Epoxy Cu(Ag) Conductive backsheet
the distance between the pin positions for whole and half
ECA 9 Epoxy Cu(Ag) Conductive backsheet
finger measurements. The extrapolated y-axis intercept of
this first plot corresponds to the hypothetical case of finger
3 METHODOLOGY
length = 0 and thus yields 2𝑅𝑐′ + 2𝑅𝑐 + 𝑅𝑏 .
3.1 Equipment
A source–measure unit (SMU, Keithley 2602A) was
used to inject current between two pins (high and low
source) and to sense the voltage between two pins (high
and low sense). A custom contacting fixture with 37 pins
was designed to simultaneously contact 2 or 4 fingers for
each measurement. To assign a role to each pin, a
multiplexer was used between the Keithley 2602A unit
and the fixture.

3.2 Types of measurements


Four different types of measurements were performed as
illustrated in Figure 2: Figure 3. TLM to remove finger resistance component.

 Two pins on whole finger: R@2PWF To remove the 𝑅𝑐′ term, an additional sample was used
 Two pins on half finger: R@2PHF to perform another TLM along a longer line of the same
 Four pins on whole finger: R@4PWF metallization paste as used in the fingers and with the same
 Four pins on half finger: R@4PHF , width as the fingers of the sample. As shown in Figure 4,
for each sample several measurements were taken at a
distance of 4–20 mm. The minimum distance can be
measured 10 times (e.g. between fingers 1 and 2 in both
directions, 2 and 3 in both directions, etc.) while the
maximum distance can only be measured twice (between
fingers 1 and 6 in both directions). For each distance the
median is calculated.

Figure 2. Types of measurements carried out.

Considering the sample circuits shown Figure 2, the Figure 4. Measurements required to perform a final TLM
following equations are expected to be true: and extrapolate 𝑅𝑐 and 𝑅𝑏 .

R@2PWF = 2𝑅𝑐′ + 2𝑅𝑊𝐹 + 2𝑅𝑐 + 𝑅𝑏 (1)

628
38th European Photovoltaic Solar Energy Conference and Exhibition

The distance used to plot 𝑅𝑥𝑚𝑚 (the resistance between Effective area Transfer length
pins with a separation of x mm) corresponds to the distance 𝛽= = ⋅ 100 [%]
Geometrical area Finger width
between the inner edges of the fingers (𝑑𝑥𝑚𝑚 ). This
distance depends on the width of the finger 𝑊𝐹 in mm and As shown in Figure 6, when the contact is overflowed
decreases when the finger is wider. That is, with a high current density (𝛽 > 100 %), a condition
𝑑𝑥𝑚𝑚 = 𝑥 − 𝑊𝐹 , where 𝑥 can be from 4 mm up to commonly known as current crowding, TLM tends to
20 mm. The resulting TLM plot is shown in Figure 5. underestimate the bulk resistance of the adhesive (𝛼 <
100 %). Hence, the line resistance of the ECA is also
underestimated. On the other hand, when the current
density is small and it occupies a small fraction of the
geometrical area (𝛽 < 100 %), TLM overestimates the
bulk resistance of the ECA (𝛼 > 100 %). However, when
the current occupies somewhat more than the whole
geometric contact area (𝛽~100 %, red oval), TLM tends
to extrapolate the bulk resistance as expected (𝛼~100 %).
Hence, the extrapolated line resistance of the ECA via
TLM under these conditions is comparable to the values
measured using four pins.

Figure 5. TLM plot using the measurements shown in


Error! Reference source not found. to extrapolate 𝑅𝑐
and 𝑅𝑏 .

3.3 Validation of measurements

Because the method is under development, resistance


values are not known in advance and literature values can
vary significantly for different sample geometries. For this
reason, the SMU was configured to sense using auto
ranging. To know that a sensed voltage value is valid, the
range in which the SMU had performed the measurement Figure 6. ECA line resistance ratio (𝛼) depending on the
is registered. If the range used was equal to the voltage percentage of the used contact area (𝛽). Data is shown for
upper limit, the value was excluded. If a whole (half) a current range of 4.00 × 10-3 to 1.00 × 10-1 A and a TLM
finger measurement was excluded, then its respective half squared Pearson correlation coefficient larger than 0.9
(whole) finger measurement was also excluded. (𝑅2 > 0.9).
Before performing any calculations, the following
conditions were required to be true:
It is then concluded that in order to calculate an
accurate ECA line resistance, the injected current should
 R@2PWF > R@2PHF (condition 1)
be selected so that the current flows through the whole
 R@4PWF = R@4PHF (condition 2)
area, avoiding current crowding due to currents that are too
large as well as underestimation due to current injections
Condition 1 was used exactly as stated here. However,
that are too small.
condition 2 was relaxed and ± 5 % deviation was
accepted.
4.2 Effective contact resistivity versus current injection
The dependence of the effective contact resistivity,
4 RESULTS
calculated with TLM, on the current injection is shown in
Figure 7. So that the area of the contact does not influence
4.1 ECA line resistance versus effective and geometrical
the analysis, the curves are separated by their area. The
contact area
contact resistivity tends to stabilize at the upper current
As shown in the methodology, the ECA line resistance
injection levels between 4.00 × 10-3 A and 1.00 × 10-1 A,
(in Ω mm-1) can be calculated from the slope of the TLM
independent of the area of the contact. For the cases in
plot (𝑅𝐿[𝑇𝐿𝑀] ) as shown in Figure 5 or directly calculated which the current injection levels are small (see red ovals
using either R@4PWF or R@4PHF or the median of both for ECA 3, 4, and 6 in Figure 7), TLM tends to
(𝑅𝐿[4𝑃𝑋𝐹] ). The ratio of these two values is referred to as underestimate the contact resistivity.
the ECA line resistance ratio (𝛼). Figure 6 shows the
variation of 𝛼 with the percentage of the geometrical 4.3 Contact resistivity of all tested ECA
contact area used (𝛽), which depends on the transfer Table II shows the effective contact resistivity and its
length. The two ratios are calculated as follow: statistical values. The median and the median absolute
deviation are in mΩ-cm². All values within the table are in
𝑅𝐿[𝑇𝐿𝑀] the current range of 4.00 × 10-3 A to 1.00 × 10-1 A and R² >
𝛼= ⋅ 100 [%] 0.9. However, if results are filtered to only consider
𝑅𝐿[4𝑃𝑋𝐹]
and, contact resistivities calculated when the usage of the
geometrical contact area is 100 % (𝛼~100 %), then there
are only a few ECAs with valid results. For this reason,

629
38th European Photovoltaic Solar Energy Conference and Exhibition

Table II shows values that are calculated for any contact data sets it is clear that although there are certain
area usage. That means, that some contact resistivities are combinations of geometry, current injection, and
calculated on current crowding conditions and other are measurement methodology that can produce apparently
not. sensible results, such conditions do not reliably produce
As it can be seen in the table, the median deviation is sensible results for all samples and there remains
still in the same order of magnitude when compared to the significant deviation in the contact resistivity calculated
median, and we believe that this is related to sample when different contact areas are used.
manufacture and also to the number of measured samples. Now that a measure system has been developed that is
fast and automated, future work will focus on further
improving and expanding sample manufacture to obtain
greater confidence in the statistical significance of
calculated contact resistivities independent of sample
geometry or current injection.

6 ACKNOWLEDGEMENTS

This work is supported by the German Federal


Ministry for Economic Affairs and Energy (BMWi) as part
of the HoSSa and Zquadrat projects with funding reference
numbers 03EE1014D and 03EE1005A.

7 REFERENCES

Figure 7. Dependence of the effective contact resistivity [1] T. Geipel, M. Meinert, A. Kraft, and U. Eitner,
on the current injection. Data is shown for effective “Optimization of Electrically Conductive Adhesive
contact areas that are smaller than the geometrical contact Bonds in Photovoltaic Modules,” IEEE J.
areas and a TLM squared Pearson correlation coefficient Photovoltaics, vol. 8, no. 4, pp. 1074–1081, Sep.
larger than 0.9 (𝑅2 > 0.9). 2018.
[2] M. I. Devoto, T. Timofte, A. Halm, and D. Tune,
Table II. Statistics (median and median absolute deviation) “Measuring the contact resistivity of ECA-based
of effective contact resistivity (in mΩ-cm²) for all ECAs joints,” in 37th European Photovoltaic Solar Energy
under test. Conference and Exhibition, 2020, pp. 1001–1008.
Median [3] M. I. Devoto, T. Timofte, A. Halm, and D. Tune,
ECA N Median Absolute “Contact resistivity of ECA bonded joints,” AIP
Deviation Conf. Proc., vol. 2367, 2021.
[4] G. Beaucarne et al., “Study of compatibility of
ECA 1 10 2.9587 2.4512 silicone-based electrically conductive adhesives and
ECA 2 14 7.9956 1.6845 conductive backsheets for MWT modules,” Energy
Procedia, vol. 55, pp. 444–450, 2014.
ECA 3 15 5.7412 1.3821 [5] M. Estruga, L. Theunissen, A. Ardizzone, B.
ECA 4 18 1.4631 1.0091 Willems, and A. Henckens, “Electrically
Conductive Adhesives for Cell Interconnection in
ECA 5 20 0.4439 0.3483
Shingled Module Technology: The Impact of
ECA 6 25 3.2895 1.7915 Material Properties on Mini-Module Performance,”
ECA 7 15 2.9285 0.4581 in 35th European Photovoltaic Solar Energy
Conference and Exhibition, pp. 67–70.
ECA 8 20 10.7865 3.6106 [6] W. Schockley, A. Goetzberger, R. Scarlett, R.
ECA 9 25 6.0917 3.4497 Gereth, V. Williams, and N. Zetterquist, “Research
and investigation of inverse epitaxial UHF power
5 CONCLUSIONS transistors,” 1964.
[7] H. Murrmann and D. Widmann, “Current Crowding
As also shown previously, TLM is a method that can on Metal Contacts to Planar Devices,” IEEE Trans.
be used to extrapolate the contact resistivity of ECA-based Electron Devices, vol. 16, no. 12, pp. 1022–1024,
joints. However, the complex sample geometry and 1969.
microscale variations introduce new levels of complexity [8] H. H. Berger, “Contact resistance and contact
into the system, and this added complexity requires more resistivity,” J. Electrochem. Soc., vol. 119, no. 4,
steps to properly treat the raw data to obtain reliable and 1972.
consistent values with statistical confidence. [9] H. H. Berger, “Models for contacts to planar
In this work, the contact resistivity of several ECA- devices,” Solid State Electron., vol. 15, no. 2, pp.
based joints was extrapolated via TLM. The samples were 145–158, 1972.
manufactured by pressure-time dispensing and the [10] G. K. Reeves and H. B. Harrison, “Obtaining the
measurements were carried out automatically with the use specific contact resistance from transmission line
of specially designed samples and measurement model measurements,” IEEE Electron Device Lett.,
equipment. vol. 3, no. No. 5, pp. 111–113, 1982.
Many different measurements were performed for [11] Y. Pan, “Versatile Circular Test Structure for Ohmic
each sample, but it remains necessary to increase the Contact Characterisation,” no. March, 2015.
number of samples under study. By obtaining very large

630

You might also like