Alessandro Chiolerio*a,
Paola Rivolob,
Samuele Porroa,
Stefano Stassiab,
Serena Ricciardib,
Pietro Mandraccib,
Giancarlo Canavesea,
Katarzyna Bejtkaa and
Candido Fabrizio Pirriab
aCentre for Space Human Robotics, Istituto Italiano di Tecnologia, Corso Trento 21, 10129 Torino, Italy. E-mail: alessandro.chiolerio@iit.it
bApplied Science and Technology Department, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy
First published on 23rd September 2014
Nanostructured polymeric composites are promising materials for the fabrication of piezoresistive devices because they show a huge variation in electrical resistance when subjected to mechanical deformation. Quantum tunneling composites feature a conduction mechanism occurring between the metallic filler and copper particles embedded in a polydimethylsiloxane (PDMS) insulating matrix, and the mechanism is enhanced by the spiky morphology of the particles. PEDOT:PSS electrodes are patterned on either side of the composite by inkjet printing, a technology that allows one-step fabrication processes. The adhesion and spreading of conductive printed ink drops are controlled and enhanced by pre-treating the samples surface in an atmospheric pressure plasma customized system. Because of an extremely high metal to polymer ratio, which results in the different surface and dielectric properties of the composite, conventional plasma conditions are not suitable to allow the control of spreading. The optimal plasma conditions for ink/surface compatibility were found using quantitative comparison based on image analysis and numerical interpretation of the adhesion/roughness properties such as bulging and spread.
Moreover, biomedical and robotic tactile sensors usually need to be flexible and adapt to the curvature of the host's structure and stretchable to cover joints and moving parts to correctly measure the interaction force.6
In previous experimental work we have reported various novel piezoresistive composites based on nanostructured metal fillers dispersed in a silicone matrix as functional material for tactile devices, in which the wide variation of piezoresistance achieved was attributed to quantum tunnelling conduction induced by the spiky particles covered by nanometric tips.7,8 Under un-deformed conditions, the composite material shows insulating behaviour, while after applying strain the electrical conductivity exponentially increases.9,10 In these composites, the conductive particles are separated from each other by a thin gap of insulating polymer, which represents the tunneling barrier. By applying a deformation to the sample, the thickness of the tunneling barrier is reduced, thus increasing the probability of tunneling phenomena and resulting in a huge reduction of the bulk electrical resistance.
Concerning the electrode deposition technique, literature reports several technologies and materials for flexible composites and rubber-based functional materials.11–13 Direct metal deposition allows the fabrication of very conductive electrodes and ensures very low contact resistance values; however, it suffers from a stiffness mismatch between the metallic electrodes and the polymer composite functional material. The differences in Young's modulus and elongation at break could induce the formation of a crack in the stiff electrode material during the stretching or bending of the device, consequently compromising electrical conduction. Moreover, the bonding force between the polymeric-based functional material and the metalized electrode film might be too weak to ensure good adhesion, and peeling off/detachment effects may occur.14–16 Because of an extremely high metal to polymer ratio, resulting in the different surface and dielectric properties of the composite, conventional plasma conditions are not suitable to allow control of spreading. To overcome these problems, this work reports the optimization of the direct printing of polymeric electrodes on the surface of plasma-treated composite materials.
Inkjet printing (IjP) is a versatile manufacturing technique for the direct deposition of electrical contacts on a variety of substrates.17 Several reviews on the new applications of IjP technology report the use of conductive inks based on metallic nanoparticles, which can either be dispersed in a carrier or generated in situ using a chemical reaction.18 Nanoparticle-based inks normally feature extremely high conductivities but require rather high temperatures for sinterization.19,20 Another possibility is represented by polymeric inks filled with graphene; however these are not yet suitable for practical applications as they feature high resistivities or require thermal sintering.21,22 More recently, inks based on conductive polymers were also developed.23 These are predominantly indicated for the fabrication of electrical patterns on substrates that cannot undergo thermal treatments, which is required for most inks based on metallic nanoparticles. In particular, inks based on poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) were developed for this purpose because of their good solubility and ease of processing, and can be used as transparent, conductive coatings with high ductility.24
The adhesion of printed drops to the substrate is not only influenced by the rheological characteristics of the printed fluid, which can be adjusted before and during the printing process, but also by the physical and chemical properties of the substrates surface. PEDOT:PSS patterns show poor adhesion properties and imprecise geometry reproduction when printed on highly hydrophobic substrates such as polydimethylsiloxane (PDMS). Liquid spread is influenced by fluid properties, such as surface tension and viscosity, and by the relative surface energy and roughness of the substrate. Therefore, it is possible to limit and tune ink drops spread by carefully modifying the surface energy of the substrate.25 Moreover, surface morphology and modification can heavily impact the manner in which sessile drops deposit on a substrate by enhancing or limiting their coalescence.26
Surface modification techniques, such as wet chemical treatment, low pressure plasma treatment, and UV irradiation, have been widely applied to different substrates to enhance their adhesion properties and their applicability in functionalization processes.27,28 Among them, low pressure plasma treatments have been strongly exploited because they have been proven to be an effective way of modifying the surface of polymers. Many properties and functionalities can be obtained using plasma treatment, depending on the application: plasma can be used to tailor surface wettability, improve barrier characteristics, adhesion, dye-ability, printability, and oleophobicity.29 Chemical surface modification is initiated by a radical reaction between the plasma species and most of the external layers on the substrates surface, which occurs without the modification of bulk properties. In comparison to traditional low pressure plasma treatments, requiring high vacuum equipment or wet chemical treatments, which often involve the use of non-environmentally friendly chemicals and methods, plasmas operating at atmospheric pressure (APPs) offer an interesting alternative because of in-line process capabilities, relatively low costs, possible high throughput and low requirements of personal and environmental safety. In addition, non-thermal APPs usually work at temperatures in the range of 50–200 °C such that they can be safely employed for polymer modification. Several types of APPs operating in the regime of a non-thermodynamic equilibrium exist. Typical APP systems are corona discharges and dielectric barrier discharges (DBD).29,30 With the purpose of achieving a uniform surface treatment, plasmas homogeneous in time and space are desirable, although they are not always achievable with traditional corona discharges and DBDs. For instance, large plasmas were obtained using a series of individual discharges to form a larger array or using the DBD homogeneous mode, also called atmospheric pressure glow discharge (APG).31,32 This mode is characterized by the absence of streamers and the presence of a uniform glow, which is highly reminiscent of a low pressure glow discharge. Its selective characteristics can be adapted to a variety of treatments for various shapes and types of substrates.
In this work, an APP was tested for the surface modification of highly hydrophobic PDMS-based composite materials, containing a high volume fraction of metallic particles, whose presence causes a strong interaction of the material with the APP metallic electrodes. This procedure allowed the optimization of the surface properties of the composites for the deposition of PEDOT-PSS electrodes by inkjet printing.
A sketch of the complete technological process starting from the preparation of the piezoresistive material up to the electrodes deposition is presented in Fig. 1.
Sample | OCA PEDOT [°] | OCA H2O [°] | OCA CH2I2 [°] | Surface energy [mN m−1] | Dispersive component [mN m−1] | Polar component [mN m−1] |
---|---|---|---|---|---|---|
Cu–PDMS | 50 ± 1 | 122 ± 4 | 90 ± 1 | 13.0 | 13.0 | 0.0 |
Cu–PDMS after Ar-plasma | 22 ± 1 | 16 ± 1 | 45 ± 2 | 70.0 | 22.2 | 47.8 |
Cu–PDMS after 2 h ageing post-plasma | 28 ± 1 | 29 ± 4 | 49 ± 1 | 64.0 | 21.4 | 42.6 |
The Cu–PDMS composite shows the typical water contact angle of bare PDMS of about 120° suggesting that the metallic microparticles are confined in the bulk and do not affect the surface properties. Moreover, the surface energy of the sample is very low (13 mN m−1) because of its scarce interaction with diiodomethane, as reported for pure PDMS in ref. 39. The polar component of the diiodomethane surface tension cannot be ignored, thus preventing its spreading on the highly non-polar PDMS surface. A higher wettability could be reached only with non-polar liquids such as hexadecane (polar component of surface tension is equal to 0 mN m−1). The samples measured just after the plasma process show an abrupt decrease in the water contact angle (down to 16°) and an increase in surface energy (up to 70 mN m−1) typical of highly hydrophilic substrates. Both the polar and dispersive components are strongly modified by the treatment due to the strong oxidation on the surface as Ar plasma is activated in atmospheric air (the former), and structural changes in the PDMS polymeric surface chains (the latter), which are attributed to physical ablation performed by the discharge.40 The increased roughness and surface area, even though not visible to the naked eye, could improve the wettability by a non-polar liquid, such as diiodomethane, when compared with the poor spreading obtained on the untreated smooth substrate.41
Because surface energy may not be the only reason for the improved adhesion of the PEDOT ink after solvent evaporation, although it certainly affects its wettability when the ink is still in the liquid phase, we also evaluated the surface roughness before the plasma processing after the treatment and on aged samples. The images shown in Fig. 3 were used for the roughness evaluation, adapting them to run on a Matlab® routine previously developed to analyze particles and grains.42–44 The results reported in Table 2 indicate that roughness is doubled by the plasma treatment and is not affected by aging because it produces a physical modification of the composite.
Cu–PDMS | Cu–PDMS after Ar-plasma | Cu–PDMS after 2 h ageing post-plasma | |
---|---|---|---|
Roughness [nm] | 383 ± 300 | 716 ± 120 | 716 ± 120 |
The permanence of the effect of plasma treatment was studied by performing aging experiments. The contact angle analyses were repeated on the samples left for 2 h in air. Table 1 shows that the contact angles measured with the standard liquids are slightly affected by aging, resulting in a small decrease in the surface energy (64 mN m−1). Among the several attempted processes, only one (ceramic multi-electrode tool, power = 1100 W, 5 passes) achieved our non-trivial goal, such as a good compromise between the surface spreading enhancement, limited advantageous modification of surface roughness and permanence of the produced effects for a sufficiently time long to allow inkjet printing without the need of a multi-step implant. Considering that the PEDOT:PSS ink has a surface tension of 62 mN m−1 (reported in CleviosTM datasheet), a contact angle similar to water is expected.
It is worth noting that the ink shows a lower contact angle than water on the bare Cu–PDMS composite because of the presence of organic components that make it more compatible with the hydrophobic untreated polymeric surface. As expected, the plasma treatment produced a similar effect on both water and ink, and thus it can be inferred that the treatment acted on the aqueous component of the ink. The PEDOT:PSS wetting of the polymeric/metallic composite can be considered to be stable within 2 h after plasma modification. A longer storage time is known to lead to the recovery of the hydrophobic surface properties of pristine PDMS, even after an oxygen plasma treatment under vacuum.45 In the following paragraphs, it is demonstrated that a decrease of about 20° in the contact angle, i.e. 6 mN m−1 in surface energy, achieved a desirable spreading of ink drops, promoting liquid coalescence, and its adhesion was helped by the improved wettability of the surface. The fulfilment of these requirements is needed for the production of functional printed electrodes.
The best results of surface modification using the metallic electrode tool have been obtained using the 4 passes process at 550 W. The values of contact angle for water and diiodomethane are comparable. However, the use of this setup leads to a less homogenous discharge due to the formation of separated arcs. This possibly caused the motion of Cu microparticle clusters toward the surface, which resulted in the removal of material and the creation of holes and cracks on the surface of the composite. The result of this plasma treatment was clearly observed by FESEM analysis and a topographical view is shown in Fig. 3a. This phenomenon induces a modification of bulk composition in terms of micro-particles distribution that may negatively affect the piezoelectric device performance inducing the formation of electrically non-homogeneous clusters on the sample surface. These phenomena are considerably reduced when using the ceramic tool, which limits particle removal and surface damage, as is visible in Fig. 3b and c.
The bulging phenomenon may be controllable when the APP treatment is performed by the multi-electrode ceramic tool with plasma powers up to 1100 W. In the next paragraph, an operative procedure to correlate an engineering parameter (plasma hardness, defined further in the text) with post-printing features, such as good adhesion, bulging, etc., will be presented. We will see that a low hardness plasma treatment leads to a very low wetting, while a high one leads to high bulging.
The information content of each microscope image has been extracted by numerically assisted image analysis (IA) procedure, as shown for the composite in Fig. 5, where the first row sketches an example of the processing of a sample image, while the second row contains the data extracted from all the images taken on samples treated at 400 W using the metallic electrode. In particular, each histogram frame has been adjusted before conversion from RGB to 8-bit grayscale, and further posterized to a 2-bit valued B/W map (Fig. 5b); then a filter to detect the image edges was applied based on Sobel Gaussian computation. The resulting edge image (Fig. 5c) was imported in the Matlab® environment for statistical analysis. Each edge position, both the upper and lower, has been stored in a sequence of vectors (Fig. 5d) and further studied, extracting its mean edge position, standard deviation and so on. Each edge pixel, identified by a couple of values corresponding to its (x,y) coordinates, has been used to populate a statistic, which is limited to the y-position. Hence, the bin centre is the average y-position of the edge (zero) and the relative frequency is the number of pixels having a certain position over the total number of edge pixels (Fig. 5e). It is clear that a straight edge would result in monodisperse statistics with a very narrow peak centred at the mean value. Conversely, a bulging edge would result in a very disperse and broad peak, which is still centred at the mean value. When the periodicity of the bulging is such that in the same frame we can identify more than one ink circular area, a double peak broad distribution is found. For completeness, the experimental cumulative distribution is also given (Fig. 5f), showing perhaps the strongest evidence of the effects induced on the adhesion between ink and substrate by printing parameters, in this case the droplet density.
Using the profiles represented in Fig. 5d we performed the following statistical analysis shown in the relative frequency plot of Fig. 5e. Three different printing densities (indicated as points per inch, ppi) result in three very different situations, where peak broadening is consistent with the occurrence of bulging from a straight line. On the x-axis, the bin centre is given in pixels; it is straightforward to see that increasing the IjP density, the printed area broadens and the excursion covered on the x-axis is greater. A slight asymmetry of the frequency profiles (horizontal shift with respect to zero) is due to the misalignment of line axis with respect to microscope frame axis but this un-compensated effect is not affecting the following discussion. The cumulative frequency plot of Fig. 5f compares the almost perfectly linear statistics of the sample printed at low density (straight line) with the other cases at a higher printing density, featuring a smoother distribution.
The results of the IA procedure are graphically presented in Fig. 6, showing the efficacy of our numerical procedure in classifying the different plasma treatments. Fig. 6a represents bulging broadness as a function of APP hardness, also giving information on the effective ink volumes as a colour scale (goal function). The plasma treatment hardness (engineering parameter) is defined as the APP power multiplied by the number of passes over the time elapsed because the treatment before IjP and the engineering volume is defined as the track density multiplied by the number of printing passes. Fig. 6a demonstrates that the phase space can be experimentally explored almost entirely by changing the process parameters. A low hardness plasma treatment results in a very low broadness/wetting of the surface, on low jetted volumes, while a high hardness plasma treatment leads results in high bulging broadness on almost every volume.
Fig. 6b represents the Gaussian fit R2 to the relative probability (Fig. 5) as a function of APP hardness and effective volume: the higher this factor, the better the Gaussian fit. Note that there is no theory behind the choice of the Gaussian distribution other than its simplicity. When a simple criterion is applied with success, the corresponding process is also easily controlled. Depending on the relative probability distribution, single Gaussian or double Gaussian fits were performed. Both hard and soft APP treatments produce results whose statistics are not well represented by a simple Gaussian model, while mild treatments, those around 40 Js−2, are better described. The fill factor is defined as the number of pixels containing ink over the total number of pixels belonging to the area to be printed. This number has been computed only for the samples having a defined area to be printed (untreated substrate), while for the other samples, where line broadness results directly from droplet density and is not user-defined, we have given a fixed value of 1. Peak broadness is computed as the Gaussian fit peak width over its height. One of the key objectives of this study is to find an effective experimental procedure to limit ink consumption, and therefore the engineering volume should be as low as possible and the fill factor should be 1. We have observed that even a high number of passes (10) is not enough to approach these values, hence the necessity of an APP treatment. The fit is also useful for a quick evaluation of the fabrication procedure because its broadness should approach zero for perfectly straight lines without printing defects, while lines displaying bulging would result in a broader and double peak distribution.
This type of material is normally used under low voltage conditions because it is intended to be coupled to standard electronic devices (5 V) and low power energy sources.8 However, high electric field conditions were already used to induce the electric ageing of the nanocomposite samples in similar systems.47 Therefore, the high voltage regime was also investigated to potentially induce electromigration and exclude drifts in its response. Fig. 7b shows that in the high voltage regime our system after PEDOT:PSS printing saturates, the standard deviation may be as high as 100%. The saturation observed in the response indicates that there is no threshold for electromigration and that this noise is entirely due to instability toward high charge.
A novel procedure to quantify drop spreading was also developed based on optical microscope image analysis, and a numerical routine used to extrapolate the quantitative data from the bulging degree of the printed track profile. In addition, by quantifying the energy transferred to the substrate by means of the plasma treatment, it was possible to create a phase diagram, which was useful to understand and smartly classify optimal conditions for the improvement of compatibility. Electrical characterization was performed on untreated/treated samples after inkjet printing of electrodes, showing an improved signal and the absence of electromigration.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra06878e |
This journal is © The Royal Society of Chemistry 2014 |