Detection of Ice Formation With the Polymeric Mixed Ionic‐Electronic Conductor PEDOT: PSS for Aeronautics

Ice formation detection is important in telecommunications and aeronautics, e.g., ice on the wings of an aircraft affects its aerodynamic performance and leads to fatal accidents. While many types of sensors exist, resistive sensors for ice detection have been poorly explored. They are however attractive because of their simplicity and the possibility to install an array of sensors on large areas to map the ice formation on wings. Hygroscopic ionic conductors have been demonstrated for resistive ice sensing but their high resistance prevents the readout of sensor arrays. In this work, mixed ionic‐electronic polymer conductors (MIEC) are considered for the first time for ice detection. The polymer blend poly(3,4‐ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) is solution deposited on a pair of electrodes. The sensor displays an abrupt rise in electrical resistance during the transition phase between water liquid to solid. It is proposed that the morphology and electronic transport in PEDOT are affected by the freezing event because the absorbed water in the PSS‐rich phase undergoes dilatation upon forming ice crystals. For the aeronautics application, successful tests of integration of sensing layer in pre‐preg layers of aeronautical grade and freezing detection are carried out to validate the ice detection principle.


Introduction
3] More especially in the airline industry with 10 00 00 flights daily, one of the most common reasons for accidents is so-called "loss of control in-flight" situations, where the flight crew for some reason lose control over the plan. [4]One reason for this could be the formation of ice on the aircraft body or in the engine parts. [5]To avoid such problems, most aircraft are equipped with de-icing protection systems consuming a lot of energy. [6]These deicing systems are connected to an ice formation detection system to function only when necessary, but there exists no mapping of the ice formation to give the correct information about the whole aeroplane surface.The consequence of this is that the whole aircraft body is heated, even though ice may be formed only at a small part of it. It is thus desirable to develop cost-effective sensor systems for ice formation in large areas.
The detection of ice formation cannot simply be achieved by measuring the temperature and the humidity since supercooling phenomena can happen to make water freeze at temperatures below 0 ˚C. [9]The most used physical methods for ice detection today are based on magnetostriction, [10] pyroelectric effects, [11] piezoelectric [12] technology or capacitor-antenna. [13]Those electrical methods require proper complex data analysis to interpret the data and identify ice formation.There is a lack of simple electrical methods to detect ice and map ice formation on a large surface.Optical systems such as infrared cameras [14] or photodiodes [15] work well too but are advanced, expensive and can in addition only measure a limited area.The ideal sensor system would be an array of sensors printed on large area and based on a resistive ice detection mechanism.
In the above-mentioned methods, the property of the material used in the sensor device is not changed; but the ice formation leads to a different physico-chemical response for the sensor device.Very few studies have identified a change in electrical resistance upon ice formation in a material.The ionic resistance of water in a soil decreases drastically upon freezing [16] and could then be used as sensing strategy for a simple electrical method.Indeed, ions in water undergo a drastic change in mobility at the freezing point from liquid to ice and only small ions like Li-ions keep a decent conductivity. [17]The huge drop in ionic conductivity of a hydrogel upon freezing is a problem for all electrochemical devices (battery, etc…) and special designs of the hydrogels can be made to prevent freezing down to −20 ο C. [18] Hence, the strategy to use ionic conductivity as a sensing property for freezing is obviously accompanied with the drawback of the supercooling phenomena in the electrolyte itself.Another drawback of this approach is the low ionic conductivity in a frozen water electrolyte (10 −3 −10 −7 S cm −1 ) which limits the size of the resistive sensor arrays [19] and makes impossible to use Joule heating for deicing.
Because of this last drawback, it is interesting to consider electronic conductors with higher electrical conductivities than ionic conductors.Nanocomposites based on carbon nanotubes in a matrix of polydimethylsiloxane (PDMS) enable electronic transport and they could display an increase in resistance decreasing the temperature in the range of the freezing temperature. [20]owever, it was not possible to observe a clear threshold increase in the resistance at the freezing point, thus leading to low accuracy for the detection of ice formation.Note that compared to the ionic conductors, this electronic composite conductor could transport enough current to produce a Joule effect for deicing.Note that in this example, we believe this is because the matrix of PDMS is rather hydrophobic and does not absorb water that the composite is not expected to change much its morphology upon freezing.
In this work, we investigate polymeric mixed ionic-electronic conductors (MIECs); which are polymers that conduct both ions and electronic charge carriers (electrons and/or holes) for their ability to detect ice formation through the measurement of electrical resistance (Figure 1a, Supporting Information).The ionic conduction parts are typically ensured by the ability of the material to absorb water and thus we expect this promotes a change in morphology in the electronic conducting path upon freezing. The morphology of PEDOT:PSS can be controlled to tune the electrical conductivity from 0.1 S cm −1 to 3000 S cm −1 . [23] We demonstrate that PEDOT:PSS can be formulated to be very sensitive to the detection of ice formation by displaying surprisingly sharp threshold in its electrical resistance at the freezing point.Both X-ray diffraction and scanning probe microscopies are used to investigate how freezing affects the morphology of the PEDOT:PSS.Finally, we built a demonstrator of a PEDOT:PSS ice detection sensor on the polymeric material used for coating airplane wings.

Film Fabrication
The polymer solution (Clevios PH1000 from Heraeus Holding GmbH, 1:2.5 PEDOT to PSS ratio) was first filtered through a polyvinylidene fluoride (PVDF) filter (0.45 mm).To get water-stable polymer films, GOPS ((3-Glycidyloxypropyl)trimethoxysilane) was added to reach a concentration of at least 0.1%.Samples of higher GOPS concentration (1%), higher PSS (polystyrene sulfonate) content (1:6 PEDOT to PSS ratio), and with DMSO (dimethylsulfoxide) (5%) were prepared by adding the GOPS/PSS/DMSO solution to the 1:2.5 PEDOT: PSS solution.Solutions were left stirring for at least 1 hour and finally sonicated for 5-10 min to become uniform before the spin coating.To provide electrical connections with the sample, glass substrates with four 15 × 1 mm electrodes (6 nm Cr and 60 nm Au), with a 1 mm distance were used.Spin coating was performed using a speed of 1500 rpm for 30s.All spin-coated films were baked at 120 °C for 10 min.
To approach a realistic prototype, the authors used a substrate made of the polymer composite used in aeronautics called "Prepreg," which was the collective term for a reinforcing fabric.Fabrics were preimpregnated with a resin system, typically epoxy.Prepreg 7781 E-glass from the company FibreGlast was used.The substrate used for our application results in two layers of prepreg 7781 E-glass (thickness of 0.22 mm) pressed together to eliminate any air bubble and cured in an oven at 150 °C for 3 h.On top of this "Prepeg" substrate, four graphene electrodes were glued that had been fabricated as ribbons of graphene GS50 Blue curve shows resistance, orange curve is temperature and grey solid curve is the optical reading (grey value in the camera analysis).b) Upon heating, the absorption of heat by the melting of ice on the glass surface from the ceramic appears as temperature rise delay by a time dt.c) Upon cooling, the temperature drops but at the freezing point heat is produced on the glass surface, heat passes through the glass and reaches the thermometer with a delayed dt.

Film Characterization
Considering the film in ambient temperature and relative humidity around 45%, the electrical conductivity () of film was measured using a Keithley 4200-SCS connected to a four-point probe configuration and calculated with relation: [26] where R, l, w, and t are respectively the resistance, length, with and thickness of the film.The thickness of the spin-coated film was of the order of 100 nm as measured using an optical profilometer (Sensofar PLux neox).Morphological characterizations were performed by using a Dimension 3100 atomic force microscope (AFM) operated with a Nanoscope IV system and working in tapping mode.Topography was recorded with Silicon tips having a force constant of 40 N m −1 .To cool down the sample under vacuum, the Janis cryogenic probe station had been used.The sample was left overnight in the chamber under pumping at 10 −5 .Grazing Incidence Wide Angle X-ray Scattering (GIWAXS) measurements were performed on the Dutch Beamline (DUB-BLE CRG), station BM26B, at the European Synchrotron Radiation Facility (ESRF), Grenoble, France.The energy of the X-rays was 12.42 keV, and the sample-to-detector distance was set as 12.07 cm. The diffracted intensity was recorded by a Frelon CCD camera and was normalized by the incident photon flux and the acquisition time.Flat field, polarization, solid angle, and efficiency corrections were subsequently applied to the 2D GIWAXS images.The scattering vector q was defined with respect to the center of the incident beam and has a magnitude of q = (4/)sin(), where 2 is the scattering angle and  is the wavelength of X-ray beam (0.998 Å).Herein we present the wedge-corrected images where q r (q r 2 = q x 2 +q y 2 ) and q z are the in-plane and near out-of-the-plane scattering vectors, respectively.The scattering vectors are defined as follows: where  f is the exit angle in the vertical direction and 2 f is the in-plane scattering angle, in agreement with standard GIWAXS notation.Exposure time was set at 30 sec.2D images have been collected during cooling and heating from +10°C to −10°C to +10°C, with a step of 1 °C.The sample of film had been placed in a homemade chamber with Kapton windows and constant N 2 flow, to prevent water condensation on the sample from the atmosphere.

Electrical Probing
A sketch of the experimental set up to record the resistance versus temperature is depicted in Figure 1a.The temperature was controlled by a Peltier cooler.The hot side of the Peltier cooler is cooled down by an external water-cooling system to discard the heat.The sample was placed on the cold side of the Peltier element and connected to four probes for electrical monitoring using a source meter.The temperature on top of the sample was measured using a thermocouple that is contacted with a thermally conductive grease with the substrate.The sample was video recorded using a simple webcam to distinguish changes in optical reflection (grey value image analysis) upon freezing.The whole setup was placed in a closed chamber in which the humidity could be controlled, even though most experiments were conducted in room humidity set in the closed chamber.
The PEDOT:PSS films on a glass substrate were set to +10 °C for a few minutes (R = 22 000 Ω) followed by a temperature lowering of ≈−10 °C per minute down to −20 °C.An illustration of ice formation on the film and set-up is found in Figure S1a, Supporting Information.Above the freezing point, the electrical resistance varied linearly with the temperature in correlation with their expected and well-documented thermistor behavior (Figure 2).The electronic charge carriers are transported by hopping between the PEDOT chains that form macroscopic percolation paths (Figure 1b), as expected for a doped disordered semiconductor material. [29]Close to freezing point, water condensates at the surface of the sample but the samples keep displaying a thermistor behavior, even though their sensitiveness (dR/dT) changed slightly.At the freezing point, a sharp spike is detected in the optical reflection (at 520 s on the gray curve in Figure 2) resulting from a change of the light reflection from a liquid water mirror to a solid ice mirror, later the optical properties is affected by the growth of ice crystals on the PEDOT:PSS film (Figure S1a, Supporting Information).Interestingly, exactly at the same time as the ice film is formed on PEDOT:PSS, a sudden rise in the resistance of the film was observed (at 520 s from R freezing = 25000 Ω to 30000 Ω within a second).Repeating the same experiment using an Au thermistor on glass shows no such behavior upon ice formation (Figure S2, Supporting Information), meaning this is not an experimental artifact but a true response of the polymer PEDOT:PSS.Interestingly, the temperature reading of the thermocouple on the Peltier ceramic displays a peak that is delayed by a time dt = ≈10s after the freezing.This corresponds to the heat generated by the freezing of water on the surface of PEDOT:PSS.Indeed, since freezing is exothermic, [30] that heat ΔQ(w) will take some time to be transported through the glass substrate and read the thermocouple (Figure 2c).The temperature on the glass substrate is also measured afterward with the thermocouple on the glass, but the delay of the temperature signal "dt" is then less obvious.The temperature difference between the the top of the glass substrate and the top of the ceramic is about 7 ± 0.5 °C, so the freezing point of the condensed water on the PEDOT:PSS film is about −5 ± 0.5 c C.
Note that after the freezing point, the Peltier device is still set to go even cooler but the rate of change of the temperature tends to saturate to −16 °C.This slowing down of the cooling is the result of the constant heat produced under freezing when water condensed from the atmosphere and the ice crystal continues to grow on the surface.The optical reading displays (gray line) a quasi-linear increase from 2 a.u. to 6 a.u.during the cooling step due to the continuous formation of ice on the surface.Note that the resistance saturates even if the temperature slightly decreases and we will call this R max .
At a time of 630 s, we reverse the current in the Peltier device to heat up the glass substrate.The temperature rises; and in the same time, the resistance decreases too.At 665 s, the optical signal is similar to the original situation for a liquid water mirror which indicates that all the ice crystals on PEDOT:PSS are now melted.Simultaneously, the resistance stabilizes to the value of the pristine PEDOT:PSS before freezing.Interestingly the temperature signal then sees a sudden slowdown of its rise during a time equals to dt = ≈10 s.This is attributed to the endothermic behavior of the melting that is using some of the heat ΔQ(w) from the Peltier device to melt the ice, thus the thermometer indicates a slower rise of increase corresponding to the heat transport through the glass substrate (Figure 2b).

Humid Versus Wet Situation
To further investigate how the freezing is affecting the polymer, the freezing experiment on the glass substrate was repeated in different conditions.PEDOT:PSS being a highly hygroscopic material, contains around 20 wt% water in ambient conditions (40%RH). [31]Freezing in ambient conditions as described previously, attracts of course a lot of condensed water and thus put the polymer in a wet condition.When repeating the same experiment in a dry environment (few RH%), linear thermistor behavior is observed, as there is no surrounding water and thereby no ice formation.Note, that this does not reveal whether the water inside the polymer freezes in ambient conditions, as the water level in the polymer is also lower in a dry environment.To get a better understanding, the very hydrophilic polymer film was covered with a hydrophobic oil (in ambient conditions) (Figure 3a).The oil is not getting into the polymer films because of their opposite affinity for water.It means that the addition of an oil drop is a way to encapsulate the PEDOT:PSS film and keep the water level inside the polymer the same as in normal ambient conditions (40%RH) while being able to tune the temperature of the film.The oil is preventing any contact of the PEDOT:PSS film with liquid condensed water.As a result, we could not record a freezing response in the resistance measurement (Figure 3b).In ambient conditions (around 40%RH), there is thus no freezable water available inside the polymer, which correlates well to the study of Siu et al. [32] that showed a relative humidity of at least 85% is needed to have freezable water in a PSS-containing material.The resistance behavior in an oil-covered sample in ambient is thus equivalent to what is observed in a low humid environment (see Figures S5b,c, Supporting Information).It is note-worthy to say that the morphology of the PEDOT:PSS film (even without GOPS) is unchanged when cooled down at −10 °C under vacuum (i.e., in a humidity-free atmosphere) (see Figure S3, Supporting Information).From these observations, we can conclude that it is important that the polymer is in contact with the condensed water to see the freezing sensing effect because the water content inside the film is not high enough.
The presence of absorbed water in hygroscopic polyelectrolyte like PSS is well known to act as a plasticizer for the polymer chains. [33]Water sorbed on the polymer can be of three different kinds: free water, freezable bound water, and nonfreezing water. [34]It is however not clear if the presence of the adsorbed water in PEDOT:PSS could induce a change in morphology.For that purpose, X-ray scattering patterns are analyzed at wide angles for a PEDOT:PSS film upon cooling in absence of contact with liquid condensed water at the freezing point of water.Figure 4 presents the wedge-corrected GIWAXS image recorded for 1:2.5 PEDOT:PSS + 0.1%GOPS film at 0 °C and the 1D scattering patterns obtained during cooling from 10 °C to 0 °C and back.No ice peak is observed indicating that no free water is present, but still bond water is expected to be present in the film without forming ice at that temperature.From 0 °C to −10 °C (Figure S4, Supporting Information), the intensity of the pattern does not change and no diffraction peak for ice is visible either.This indicates that the bond water is not freezable, likely because the water screens the electrostatic interaction with the polymer ions and counterions rather than forming a hydrogen bond to create an ice crystal.In other polycation-polyanion systems, bond water molecules are associated with translational motion at high temperature, still acting as plasticizer for the polymer chain dynamics.However, at low temperature, water motions become dominantly local vibrations and rotations which means that the polymer chains become mostly immobile. [35]This could be the reason why the scattering intensity changes above 0 °C but not below 0 °C.
Let's look more carefully at the peak intensity change between 10 °C to 0 °C.The broad peaks at 1.3 and 1.8 Å −1 are assigned to the amorphous halo of PSS and to the - stacking of PE-DOT, respectively.An important decrease of intensity with de- creasing temperature is observed for these peaks, which is recovered almost reproducibly during heating.The scattering peaks at 0.3 and 0.53 Å −1 are assigned to the 100 reflections of the layered PEDOT/PSS alternation; and, in particular, to two different types of lamellar stacking of PEDOT and PSS. This later peak remains unaffected by temperature (Figure 4c), suggesting that these crystallites are very well-packed, and water cannot penetrate within. [37]On the contrary, the 0.3 Å −1 peak intensity decreases and increases following temperature (Figure 4c).Recalling that these crystallites are less compact, we can assume that water can penetrate and wet the hydrophilic PSS-rich domains.

Cyclability of the Response
From the above experiment, it appears clearly that condensed water needs to be in contact with the PEDOT:PSS film upon freezing to induce a resistance jump.The reproducibility of the signal is a requirement for a sensor application.Detection of the freezing moments over time in a future sensor can easily be done by reading the current resistance variation per time unit (dR/dt) during successive cooling/heating cycles.In addition to the experiments carried out on the glass substrate (Figure 5a), prepreg substrate are also considered (Figure 5b).The prepreg is a composite material [38] based on epoxy and resin and used among others for coating aircraft wings.On the prepeg substrate, the PEDOT:PSS film is coated on four electrodes of graphene of 4 × 0.5 cm and space of 2 mm between them (Figure S1b, Supporting Information).The result obtained after several cycles of freezing and cooling of climatic chamber, present a similar curve of response than the one obtained with the film on glass substrate.Considering one cycle, analysis of the instant of ice formation reveals approximately the same percentage of freezing response.This shows that the concept of ice detection with PEDOT:PSS can be transferred on realistic material solutions for aerospace.
Figure 6 displays the topography, recorded by AFM, of thin films of PEDOT:PSS and PEDOT:PSS with GOPS on glass substrates.Both pristine films (Figure 6a,e) present a relatively flat and uniform morphology with comparable roughness.GOPS does not appear to phase separate and is likely homogenously distributed among the PEDOT:PSS nano-grains as previously observed. [26]Previous AFM and electron microscopy studies identified that a PEDOT:PSS film is formed of nanograins (30-50 nm in diameter) composed of PEDOT-rich core surrounded by a PSS-rich shell. [39,40]The dimension of the grains observed by microscopy techniques in the film corresponds to the smallest particle size measured a nanosuspension of PEDOT:PSS in water measured by dynamic light scattering. [41]The water nanosuspension is stable by the repulsive effect of the negatively charged PSS in excess surrounding the PEDOT-rich cores in the floating nanoparticles.Thus upon drying, the film is expected to be composed of plenty of PEDOT:PSS-rich particles connected in a PSS-rich matrix.The GOPS is soluble in water and likely penetrate on and in the nanoparticles of PEDOT:PSS, leading to an efficient cross-linking as demonstrated by photoelectron spectroscopy. [25]pon cooling/heating cycles, the morphology of pure PE-DOT:PSS films is drastically changing already after 10 cycles of temperature oscillation between 25 °C and −10 °C and relative humidity between 36% and 80%.A remarkable increase of the roughness is observed upon cycling, as shown in Figure 7, which is rising from 1.2 nm for the uncycled sample to 2.6 nm for the cycled samples.We attributed this change of roughness to a degradation of the mechanical integrity of the film upon freezing.The role of the GOPS as cross-linking agent is thus to make the film robust to the heating/cooling cycles.GOPS is known to make the PEDOT:PSS resistant to liquid water [26] which is of course a first effect present in this situation since the water condensation makes the film wet before freezing.But on top of the stability in liquid water, we believe that the GOPS crosslinker enable to withstand also the mechanical stress induced by the freezing.This mechanical stress is due to the volumetric dilatation of liquid water into ice crystals within the polymer film.The integrity of the polymer film is a key to obtaining reliable measurement and a stable sensor.Note that without GOPS, after vaporization of the melted water from the surface, the rigid PEDOT film cannot recover its initial morphology, and is left with higher roughness.

Discussion
Finally, we propose a mechanism for the sensing detection of ice for the mixed ionic-electronic conductor.The condensed water penetrates in the hygroscopic PSS-rich domains of PEDOT:PSS forming small liquid nanodomains.The electrical conduction is ensured by electronic charge carriers that undergo a hopping transport along percolation paths formed of PEDOT nanocrystals close to each other.At freezing, the water nanodroplets form ice crystals on the surface and inside the PEDOT:PSS films.Due to the dilatation of the ice crystals, some volumes of the percolation paths are accompanied by a separation of the PEDOT nanocrystals (Figure 8).Freezing water has the effect of breaking the percolation path for the electronic charge carriers and enhance the resistance.

Conclusion
This work presents an innovative approach to detect early ice formation on the surface based on a conductive polymer called PE-DOT:PSS as a sensing element.The ice detection principle exploits the charge carrier dynamics of PEDOT:PSS which leads to a sudden increase in its electrical resistance upon ice formation.The hygroscopicity of polymeric mixed ionic-electronic conductor enables water absorption in specific domains, while other domains stay hydrophobic and thus ensure a mechanical integrity and the absence of dissolution of the films in presence of water.The intimate proximity between the water-rich domains (PSSrich) and the electrically conducting domains (PEDOT-rich) enable an interplay between the ice formation, accompanied with a dilatation, and the modification of electronic percolation paths in the material.After preparation, characterization, and integration of the sensitive element on different substrates (glass and prepreg), several cycles of freezing and melting experiments were performed in a climate chamber.Results confirmed sensitivity and accuracy of sensing element to detect time of ice formation.More needs to be done to clearly determine the potential for application and its long-term stability, but the data are enough to claim the sensing effect and its reversibility.Although we have not investigated systematically many polymer systems, this first study suggests that polymeric MIECs could be a potential new class of material for resistive ice detection sensors.
This approach based on organic MIEC offers a novel low-cost solution of ice detection devices in different sectors including telecommunications, energy, marine, and air safety.Indeed, as PEDOT:PSS is solution processible, the sensors could be fully printed on large area to create resistive sensor arrays.Moreover the electronic conductivity of PEDOT:PSS could enable Joule heating as potential strategy to remove the ice at the sensor points and enable multiple reading of the ice detection.The potential for mapping large areas is exciting since PEDOT:PSS-based resistive ice detection sensors could identify the area that needs to be warmed up on the structure of the airplane; thus leading to energy savings.

Figure 1 .
Figure 1.a) Sketch of the experimental set-up for ice detection with a Peltier cooler that is water-cooled supporting a device for electrical resistance measurement and a thermocouple.The PEDOT:PSS film is coated on a glass substrate prepatterned with 4 gold electrodes.b) Sketch of the morphology of PEDOT:PSS with PEDOT electronic conductor in a matrix of hygroscopic ionic conductor PSS.c) Percolation path of PEDOT chains that are stacked in aggregate.d) Chemical structure of positively electronically charged PEDOT and negatively ionically charged PSS.

Figure 2 .
Figure 2. a) Freezing response versus time under a cooling/heating cycle for a film of 1:2.5 PEDOT:PSS + 0.1%GOPS on a glass substrate.Blue curve shows resistance, orange curve is temperature and grey solid curve is the optical reading (grey value in the camera analysis).b) Upon heating, the absorption of heat by the melting of ice on the glass surface from the ceramic appears as temperature rise delay by a time dt.c) Upon cooling, the temperature drops but at the freezing point heat is produced on the glass surface, heat passes through the glass and reaches the thermometer with a delayed dt.

Figure 3 .
Figure 3. a) Sketch of the measurment set-up of a PEDOT:PSS film coated or not by a oil drop; and its resistance behavior upon a cooling/heating cycle.

Figure 4 .
Figure 4. a) The wedge corrected GIWAXS image recorded for PEDOT:PSS at 0 °C and b) the 1D scattering patterns obtained during cooling from 10 °C to 0 °C and back, at a temperature step of 1 °C.Scattering patterns are presented after applying all necessary normalizations and background subtraction.c) Zoom at the low q values of the 1D scattering patterns of PEDOT:PSS during cooling from 10 °C to 0 °C.

Figure 5 .
Figure 5. a) Resistance (blue) and dR/dt (black) versus time under freezing/melting cycles for the device depicted in b): a film of 1:2.5 PEDOT:PSS + 0.1%GOPS on glass substrate with Au electrodes.c) Resistance (blue) and temperature (orange) versus time under freezing/melting cycles for the device depicted in d): a film of 1:2.5 PEDOT:PSS + 0.1%GOPS on a prepeg substrate with graphene electrodes.Photographs of the two devices are displayed in Figure S1, Supporting Information.

Figure 6 .
Figure 6.Atomic force microscope images of a-d) PEDOT without GOPS and (e-h) PEDOT with GOPS for a,e) pristine films, and films that have been exposed to variations of temperatures between 25 °C and −10 °C for b,f) 10 cycling, c,g) 20 cycling and d,h) 40 cycling.

Figure 7 .
Figure 7. Evolution of the roughness measured by AFM for both PEDOT films for 0, 10, 20, and 40 cycles of temperature variations.

Figure 8 .
Figure 8. Sketch of the mechanism of freezing detection by resistance jump.