Content uploaded by Rodolphe Koehly
Author content
All content in this area was uploaded by Rodolphe Koehly on Sep 15, 2015
Content may be subject to copyright.
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
Carbon Black Loaded Paper: an intelligent substrate for Electronic Sensors
Design
Rodolphe Koehly
Music Technology, McGill University
550 Sherbrooke W, Eastern tower, Suite 500, Montreal, Quebec, H3A1E3, Canada
rodolphe.koehly@mail.mcgill.ca
Denis Curtil
LGP2, EFPG, INPG Grenoble
461 rue de la Papeterie - BP 65 38402 St Martin d'Hères Cedex, France
denis.curtil@efpg.inpg.fr
Theodorus G.M. van de Ven
Pulp and Paper Research Centre, McGill University
3420 University Street, Montreal, Quebec, H3A2A7, Canada
theo.vandeven@mcgill.ca
Marcelo M. Wanderley
Music Technology, McGill University
550 Sherbrooke W, Eastern tower, Suite 500, Montreal, Quebec, H3A1E3, Canada
marcelo.wanderley@mcgill.ca
Abstract
Pressure and position sensors are usual components of many electronic devices. They equip cars and planes
for security and presence sensing, industrial automated systems and many other Human-Interface Devices for
medicine and kinesiology robotics, video games, etc. We present alternative ways to develop our selves these
sensors using a low cost material: conductive paper loaded with carbon black pigments.
We show that such a paper can be easily used to develop position, pressure and flexion sensors and that it is a
good basis to develop more refined sensors such as accelerometers or tilt sensors for shock sensing with smart
packaging. We last compare a few trials to produce oneself this paper using conventional fibers such as TMP,
chemical or recycled pulp using classic laboratory handsheet formers and mixing those with industrial carbon
black loaded papers.
Keywords
Intelligent substrates; Conductive pigments; Piezoresistivity; Human Computer Interface.
1. Introduction
The most massively used flat pressure and position sensors are the Force Sensing Resistors (FSR) from
Interlink Electronics Company, Camarillo, USA. FSRs are printed components and are composed of
mixed polymers, organic pigments and metals. Unfortunately, due to their production process, they
cannot be easily recycled. Moreover one can only have such sensors produced for minimal quantities
corresponding to large-scale productions, and so, Interlink or most equivalent companies only develop
sensors solutions for the massive industries introduced earlier.
Researchers and designers of Human Interface Devices who need only a few sensors have to find the
appropriate one among those that were developed by major brands. They have to adapt their design to
available sensors but they would prefer some customizable piezoresistive material that would enable them
to design the size and shape of the sensor itself. This material would be inherently sensitive, available in
rolls or sheet and so completely open to any shape or dimension for the user.
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
Figure1: Various pressure sensors to be used or mounted onto a glove for musical application.
One can also notice an example of homemade paper FSR on the left *.
We have been looking for such materials [Koehly, 2005] and we have found that conductive paper would
be an ideal candidate here. Loading paper with electrically resistive pigments such as carbon black
provides us with an elastic and flexible volume resistive material, which simply needs to be placed
between metallic electrodes (A1 and A2) to be used as a pressure (Figure 2), position or flexion sensor
depending on the chosen design. Loading paper with conductive pigments does not prevent its re-use, as
most papers are loaded with white pigments for optical properties, and we can recycle them similarly to
produce cardboard or newspaper.
Figure 2: The resistance of carbon black loaded paper between two electrodes (A1 and A2) is
function of the dimension of the electrodes (constant C) and is inversely proportional to the force
applied onto the paper: R=f(C/F)
In a research project between EFPG in France and the department of Music Technology at McGill
University in Canada, we have been studying conductive paper to be used as position and strain sensors.
Our goal is to provide digital musical instrument designers with straightforward means to circumvent the
limited offering of commercial sensors, which are restricted to a few technologies with predefined
electrical characteristics, shapes and sizes [Jensenius, 2005]. Our goal is also to define the basis of
possible applications of our sensors for advances in printing and media technology. For instance, we can
imagine developing smart packaging for fragile packages that would integrate force sensors and tilts in
case of shock. We also think on the development of printed conductors as far as connections are
concerned.
2. Minimal considerations for building paper sensors
Carbon black loaded paper is a conventional paper that can be made using various types of fibers. The
difference with other papers is mainly that it is loaded with carbon black pigments instead of the usual
whitening pigments such as clay or CaCO3. If one wants to develop a paper for a position sensor, the best
paper will be highly refined and calendered to decrease as much as possible the surface porosity and
A1
A2
A1
A2
*
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
roughness, as the aim will be to get a linear variation of electrical resistance along the surface. In reality,
and as shown in figure 3, even a rough paper (industrial sample, thickness e=0.275mm, basis weight
W=160g/m²) can provide sufficient linearity to develop a position sensor with a few mm resolution.
Figure 3: Even a rough paper provides sufficient linearity to develop a position sensor with a
resolution under the pressure exerted by a human finger (2mm resolution equals around 50
Ohms).
Position sensors [Koehly, 2006] are made using three electrodes along a linear conductive sample. This
sample can be any flexible support, inked or coated with an electrically conductive component or it can
be a volume conductive material such as a band of conductive paper. One electrode is then fixed on each
end of the band (figure 4). The resistance between the two electrodes A1 and A2 is then fixed and
depends on the volume resistivity of respectively the ink coat or the whole conductive paper. The third
electrode A3 is placed all along the conductive length, and provides a variable output proportional to the
volume of resistive material that the current has to cross.
Figure 4: Schematic of a one-axis position sensor: pressing with the finger along the flexible
resistive material creates a contact with the anode A3 that enables to obtain a value of the position
of the sensor along the x-axis.
To develop efficient force or flexion sensors [Koehly, 2006], it is better to provide high roughness and
porosity, as it is the volume resistivity and the surface contact variation that will mainly affect the
resistance of the paper under strain. Pressure sensors are made using one or a stack of sheets of
conductive paper between 2 electrodes. This sandwich would then act as a variable resistance, decreasing
with compression. We show that this variation is partially due to the compression of the material along its
thickness, and most mainly to contact surface variation between each sheet.
A2
z
Resistive surface Rs [Ohms/m] and
flexible material (x cm width)
A3
A1
Potential contact zone
x
Spacer
Metal
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
Rsurface
Rvolume
Rsurface
Metal electrode or
conductive paper
Figure 5: a conductive paper in between electrodes or other sheets of paper is equivalent to
variable resistances in series, the variations of resistance being mostly due to surface contact
variations (see later figure 8: the variation of resistance due to contact surface is 100 times more
important than the one in volume).
A few sensors were developed using this configuration for which two conductive papers were
investigated: One (paper G) was loaded with carbon and was thick (0.275mm, W=160g/m², 12,6% ash
content for an ash test at 450°C) and the other (paper F) with graphite and was thinner (0.160 mm,
100g/m², 8.4% ash content). We varied the number of sheets stacked in order to see how we could extend
the range of those pressure sensors; each stack was inserted between 2 copper plates used as electrodes.
The paper samples were squares of 2.5cm² and the copper electrodes were 2cm² extended by a 1cm width
plate to solder a wire. The whole was covered with some plastic film to prevent from moisture variations
and copper oxidation. Figure 1 shows one of these sensors (*). 2 laboratory sheets were made from some
repulped paper G with an English Sheet Former to increase its bulk.
Table 1 shows their range for each of those sensors. For instance, a sensor called 4G3F is composed of a
stack of 4 sheets of G and 3 sheets of F. The table enables also to compare the range of the paper sensors
to the range of 2 industrial sensors. For each sensors we give some values of a resistance to use to make a
tension divider (usual electronic trick to adapt a sensor to the input tension). This enables to better
compare the range efficiency between sensors.
Table 1: Comparison of the range for various homemade sensors to two industrial sensors for
loads of 100g and 5kg. The samples were placed onto a weighing machine (compressed zone:
2cm2) while the resistance was measured. It is possible to reproduce the same sensor using the
same number of sheets stacked in a sandwich under 2 copper electrodes.
Paper
3G
3G
3G
7G
4G
+3F
4G
+3F
4G
+3F
Round
FSR
Square
FSR
1R
2R
R at 100g
(Ω)
550
450
500
1100
7, E 06
8, E 06
6, E
06
50, E 06
55000
2000
4000
R at 5000g
(Ω)
18
17
22
41
28000
24000
30000
3400
200
19
42
ΔR (Ω)
532
433
478
1059
6972k
7976k
5970k
49996k
54800
1981
3958
ΔR/Rmean
1,87
1,85
1,83
1,86
1,98
1,99
1,98
1,99
1,98
1,96
1,96
We can first notice that we obtain a good reproducibility of the sensors looking at the samples 3G or
4G3F that were duplicated three times and all provide around the same range. We can also notice that
industrial sensors can have quite a much wider range than the homemade ones. Last, we can see that
increasing the number of sheets stacked enable to improve the range, but here again, there is a limit
number of sheet to stack for optimal results. It is also interesting to notice that stacking the 2 sorts of
papers (F and G) together provided the best results.
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
The industrial paper G was repulped and some new sheets were realised in order to provide a thicker
fibrous mat and so to increase the range of resistance variation. the new paper was around 2.1mm thick
for w=1200g/m². The bulk then increased from 1.67 cm3/g to 1.75 cm3/g. Experiments show (paper 1R
and 2R) that we increased the possible range by this way (ΔR/Rmean went from 1.85 to 1.96), but we can
not know if this is due to the difference in pigments concentration (11.2 % ash content for an new ash test
at 450°C), the increase of the bulk or simply the increase of the basis weight or the thickness. We have
then produced a series of paper based on paper G and three types of pulps: TMP and chemical pulp to see
how this could vary the results in resistance variation.
3. Results & Discussion
We tried to improve paper G changing the pigment concentration and adding new virgin fibres to increase
the elasticity of the fibrous mat. The major parameter that would influence the paper conductivity is the
Carbon Black concentration. Preliminary ash tests (ISO 1762:1974) at 400°C on some samples showed
the pigment concentration of the paper is between 15 to 20%. However, according to the papermakers,
there is also some CaCO3 pigments in this paper, and an ash test does not able to differentiate carbon
black pigments from the other ones. We then chose to produce new papers, increasing their thickness and
their bulk by using bigger and less refined fibres. We used unbleached chemical pulp at 11°SR, and
mechanical pulp at 66°SR to check that high refining would really be bad for paper elasticity.
Starting from some repulped paper G, we made 10 different pulps adding each time from 10 to 50% of
each new pulp to the industrial one. From those pulps, we made each time two 500g/m² hand sheets (ISO
5269:1998) to check the influence of forming, pressing and drying. We expected to have higher bulk and
lower pigment concentration for the “a series” than for the “b series”. Table 2 shows some of the results.
Table 2: influence of the sheet formation and addition of new fibres into the conductive paper G.
Paper
%
Paper
G
%other
pulp
Basis
weight
(g/cm²)
Thicknes
s (µm)
Bulk
(cm3/g)
%
Pigments
%
Porosity
Gram-
mage
(g/cm²)
%
Moisture
Resist.
10cm
(Ω)
0a
100
630.8
1686
2.67
15.20%
74.0%
598.3
5.1%
575
0b
100
509.3
1019
2.00
12.10%
66.5%
476.4
6.5%
1500
1a
90
Chemical
660.8
1688
2.55
13.70%
73.1%
626.2
5.2%
1b
90
Chemical
459.7
1039
2.26
10.30%
70.8%
429.9
6.5%
4700
2a
80
Chemical
658.2
1878
2.85
12.10%
76.2%
622.8
5.4%
2b
80
Chemical
492.5
1162
2.36
9.50%
72.2%
460.5
6.5%
6000
3a
70
Chemical
645.9
1872
2.90
10.30%
76.9%
611.4
5.3%
1600
3b
70
Chemical
478.4
1193
2.49
7.60%
74.0%
447.8
6.4%
17000
6a
90
TMP
609.3
1615
2.65
13.30%
74.3%
574.7
5.7%
6b
90
TMP
451.0
1061
2.35
10.60%
71.9%
421.3
6.6%
3700
7a
80
TMP
543.0
1331
2.45
11.80%
72.4%
514.4
5.3%
7b
80
TMP
441.7
1033
2.34
9.60%
72.1%
410.2
7.1%
8200
8a
70
TMP
567.1
1325
2.34
10.10%
71.5%
534.5
5.7%
8b
70
TMP
434.9
1012
2.33
7.50%
72.3%
405.9
6.7%
18500
Pigments concentration
Carbon black particles that are present in the paper enable it to be conductive. These pigments have the
advantage to be very fine and very inexpensive. We can assume that the more a paper has pigments, the
more conductive it will be. However, this is not the only parameter that will influence the conductivity.
Indeed, we can also imagine that an increase in thickness, bulk or porosity will decrease the conductance.
Figure 6 plots the resistance of the various samples. We can see that there is a clear but not direct link
between pigment concentration and resistance value, and an exponential decrease of the resistance value
with an increase of the pigment concentration. The variations are certainly due to the variations of the
bulk and porosity between the samples (see table 2).
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
Figure 6: Variation of the volume resistance of various samples with the pigments concentration.
The resistance was measured over samples of 10 cm length and 2cm width. The thickness varying
between 1 and 2 mm, the values of resistance were normalised to the volume of the samples.
Extracted from table 2 and other samples tested at lower pigments concentration.
Repeatability
A fundamental parameter for a sensor is its repeatability: the device should provide the same result in
terms of resistance variation for the same applied compression. Conductive papers can provide such a
characteristic because of their high compressibility and memory-shape property. If calibrated, i.e. if the
paper is submitted to an excess of compression before its use, one can expect to obtain excellent results
with a paper force sensor for various forces as long as one does not exceed the compression load.
We mentioned earlier that the resistance variation would mainly be due to surface contact variations and
much less to the volume compression of the paper. We then developed an apparatus (figure 7) enabling to
control the pressure on a defined zone of the paper (4 or 10cm²) and measure the variation of
conductivity. The sample was placed on a high sensitivity pressure sensor that would send the pressure
value applied on the sample to the computer. A pneumatic system was controlled by the computer to
apply a force of 0 to 60 Newton through a pneumatic jack on which was screwed a flat contact piece
(5*2cm²) that would distribute the pneumatic force along a defined zone over the paper sample (4 or 10
cm² depending on the contact piece orientation). The contact piece in metal would vary the conductivity
of the paper sample and provide information on the variation of resistance due to contact surface
variations. This apparatus does not work efficiently for less than 4 Newton, because of some friction
forces in the pneumatic jack. This system enables to plot the variation of Force vs Conductance on a scale
of 4 to 60N.
Multimeter
Computer
Verrin
Measurement
Force Sensor
Computed Pneumatic Control System
Paper
Sample
Mechanic
Clip
Figure 7: Sensor testing apparatus developed at the EFPG by the LGP2
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
A first trial was made on paper G to determine the influence of the resistance variation in volume and the
one in surface in surface. For the test in volume, the contact piece was covered with an insulator. Figure 8
shows the results: while the variation of resistance in volume is only a few ohms, the variation in surface
is around 300 ohms.
Canson 1f 10cm?
600
700
800
900
1000
1100
1200
0 1 2 3 4 5
Masse (kg)
Résistance (Ohms)
Rsurface Rvolume
Figure 8: Variation of resistance for sample G tested with the apparatus of figure 7. When the
contact piece is insulated, we can only notice a few Ohms resistance variations, whereas when it is
not, the resistance variation drops from more than 1100 ohms to 800 ohms for a load gradually
applied from 0 to 5 kg and return (compressed zone 4 cm2).
We can also notice the hysteresis that is generated when compressing and releasing the force applied
which is inherent to any material in compression that is not purely elastic. Other industrial sensors would
suffer the same drawback. Also, the variations at the beginning of the test are due to the apparatus: when
the friction force of the jack is passed a sudden load is applied onto the paper and the system has to
readjust to the setting. However, it is interesting to see the results given over a few trials to check the
repeatability of the measurements for the same load setting. The test of figure 9 shows that, after a first
loading, that is different from the following (i.e. calibration test), the other trials are very similar thus
providing a good repeatability for the measurement of varying force from the paper G.
Figure 9: repeatability test: after a calibration load the paper will provide the same resistance
variation for the same applied force, and this as much when loading than when unloading.
Canson 1f 4cm_ 1000ms
1020
1040
1060
1080
1100
1120
1140
1160
1180
012345
Masse (kg)
Résistance (Ohms)
essai1 ess ai2 essai3
essai4 ess ai5
Paper G- 4cm2 contact piece
ΔRv= 3 Ω
ΔRs > 400Ω
First
trial
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
Influence of the fibres
The influence of the fibres is not easy to determine. Indeed, we did not make sheets with various pulps
and having the same pigment concentration, except for samples 3b and 8b:
Figure 10: influence of the fibres on the linear response of the resistance variation. 30% of
chemical pulp and TMP were added respectively for papers 3b and 8b. We can see that long and
less refined fibres enables better linear results and elasticity.
Sheets 3b being made with chemical pulp and sheet 8b with TMP, we can notice the importance of using
long and less refined fibres to obtain linear results and fibre elasticity. Indeed, the results for sheet 3b is
much better than for 8b in terms of linearity for 4cm² loading surface.
4. Perspectives & conclusion
We could notice that some industrial papers such as paper G already provided good results in terms of
repeatability for various applications. Another important point to use sensors for "expert" control gesture
or dynamic measurement is its time-response, which means the time the material takes to come back to its
original position and conduction response after being loaded. We measured this with an oscilloscope and
noticed that it takes 75ms for the sensor to come back to a steady conductivity value when the pressure is
released. This means that we can not measure with precision two data under this time interval. Then, if we
build for instance an electronic percussion, the sensor will enable to detect around 13 hit per second. A
professional percussionist might go faster than the sensor when he drums, but for toys and no professional
quality, this time-response would be sufficient for such an application. As far as tilts sensing for
packaging applications is concerned, it is also interesting to have a low time response as the sensors
should detect quick shocks due to bad manipulation during transportation.
New papers will be developed as well as another method to measure the content of pigments such as
using a thermogravimetric analyer. Paper G will be kept as a basis of material and lower concentrations of
long fibres such as chemical pulp will be added to improve its bulk, trying to find a compromise between
an optimised range, good repeatability and low hysteresis for the optimised sample.
Another machine will also be built to enable better measurements, replacing the pneumatic jack by an
electrically driven worm, thus eliminating the problem of friction and bad measurements under 4
Newtons.
References
3b 4cm_ 5kg 500ms
15900
16000
16100
16200
16300
16400
16500
0 1 2 3 4 5
Mas se ( kg)
Résistance (Ohms)
essai1 essai2 ess ai3 essai4 essai5
8b 4cm_ 5kg 500ms
18050
18100
18150
18200
18250
18300
18350
18400
18450
18500
18550
0 1 2 3 4 5
Mas se (kg )
Résistan ce (O hms)
essai1 essai2 es sai3
IARIGAI 2007
Carbon Black Loaded Paper: a intelligent substrate for Electronic Sensors Design
[Jensenius, 2005]
Jensenius, A. R., R. Koehly and Wanderley. M. M. 2005. “Building Low-Cost Music Controllers.” In R. Kronland-
Martinet, T. Voinier, and S. Ystad (Eds.): CMMR 2005 - Proc. of Computer Music Modeling and Retrieval 2005
Conference, LNCS 3902. Berlin Heidelberg: Springer-Verlag, pp. 123-129, 2006.
[Koehly, 2005]
Koehly, R. 2005. “Study of Various Technologies for Hom e-Made Sensors.” Masters Thesis. Master AST 2004-
2005, Grenoble, France. Available by request to the author.
[Koehly, 2006]
Koehly, R. & Al. 2006. “Paper FSRs and Latex/Fabric Traction Sensors: Methods for the Development of Home-
Made Touch Sensors.” Proceedings of the International Conference on New Interfaces for Musical Expression
(NIME’06), IRCAM, Paris, France, pp. 230-233.
