*2.1. Materials*

Commercial pristine SWCNTs TUBALL ™ were provided in powder form by OCSiAl (Luxembourg, Luxembourg) and used as received. These were chosen due to their high aspect ratio, given their diameter of 1.6 ± 0.4 nm and length above 5 μm, according to the manufacturer.

Epikote MGS RIMR426, a low viscosity epoxy resin (500–900 mPa·<sup>s</sup> at 25 ◦C), usually used for fiber impregnation processes, was used together with an amine-based curing agent, Epikure RIHMH433. Both materials were provided by Lange+Ritter GmbH (Gerlingen, Germany). Epoxy resin and hardener were manually mixed in a single step for 120 seconds at room temperature (RT) and subsequently cured at the investigated curing temperature. Bar-shaped specimens were cured in a silicon mold as follows: samples cured at RT were kept in a climatized lab (23 ◦C and 50% humidity) for 24 h and samples cured at 80 ◦C were heated in an oven for 40 min. Disk-shaped samples were heated by placing the RTM set-up on top of a custom-made heating plate at 100 ◦C for 40 min.

#### *2.2. Dispersion of SWNTs in Epoxy*

Given the initial agglomerate state of the CNTs in powder form [5,6], the first mixing step is crucial to guarantee that the CNTs are su fficiently dispersed in the resin and ensure the reproducibility of results. This was accomplished by using a three-roll mill. Comparing to other dispersion techniques (such as ultrasonication or ball milling), it has the advantage of providing more process control in the mixing [7,20]. This is achieved by the fine tuning of the rotation speed of the rolls together with the gaps between them. Moreover, it is a scalable production method because the mixing energy does not depend on the amount of material, in contrast with other dispersion technologies such as ultrasonication, for example, and so it is more relevant for industrial applications.

The used three-roll mill model was the Exakt 80E (EXAKT Advanced Technologies GmbH, Norderstedt, Germany), whose rolls were made of chemically neutral silicon carbide and measure 80 mm of diameter and 200 mm of length. The gap between rolls can be as low as 5 μm, while the maximum throughput capacity is of 20,000 cm<sup>3</sup>/h. As depicted in Figure 1, three rolls were coupled together to operate in alternating current at scaling velocities: the speed was set for the apron roll, while the center and feed rolls were automatically set at three- and nine-times lower speeds, respectively. The suspension was placed in the gap between the feed and center roll, and was then forced to pass through the first gap while experiencing shear forces caused by the speed di fference among the two rolls. The suspension then went through the second and last gap that is usually set to be three-times smaller than the previous gap, undergoing, once again, shear forces and being collected through the use of a sharp blade in direct contact with the apron roll, concluding a full dispersion cycle that, in this work, will be referred to as "pass".

**Figure 1.** Three-roll mill schematic: suspension is fed between the first two rolls, passes through gap 1 and gap 2 and is finally collected completing a "pass" cycle.

Based on the literature [20,21], the chosen apron roll speed was 180 rpm, at which CNT rupture phenomena are minimized, better preserving their aspect ratio. In order to investigate the influence on the electrical resistivity of the final sample in the absence of the electric field, the gaps between rolls were

varied from 60–20 μm to the minimum of 15–5 μm following two di fferent sequences later presented. The overall number of passes was varied from 2 to 8. The aim was to obtain a stable and reproducible electrical response of the samples produced while minimizing their electrical resistivity. Suspensions with 0.001 wt.% and 0.005 wt.% CNTs were obtained by producing 0.01 wt.% masterbatches that were subsequently thinned down to the target concentration.

#### *2.3. Set-Up for Electric Field Application*

A set-up for the simultaneous production of multiple disk samples was developed, inspired by resin transfer molding, depicted in Figure 2. A silicone form for six disk samples of 20 mm diameter and 1.5 mm thickness (3 in Figure 2) was placed between two gold-coated electrodes (2 in Figure 2). Together with another layer of silicone (4 in Figure 2), which served as a sealant, the four layers were placed between two steel plates (1 in Figure 2). The outer plates were fitted with four nuts screwed together and used for clamping the structure. The inner arrangemen<sup>t</sup> was additionally fixed with 3 guiding pins, in order to prevent slipping or warping of the silicone form and thus avoiding the clogging of the narrow flowing channels between the disk molds.

**Figure 2.** Exploded view of the mold used to apply the electric field during resin curing: six disc samples are produced at a time.

After mounting the system together, two hose couplings (5 in Figure 2) were screwed on the top plate, each then connecting a feeding tube (6 in Figure 2) fastened with a tightening nut (7 in Figure 2). One of the hoses was closed with a clamp, the other one connected to a syringe with double sealing ring. With the syringe, a negative pressure is generated in the mold. Subsequently, the epoxy dispersion was mixed with the curing agen<sup>t</sup> and, to avoid air bubbles in the samples, degassed under vacuum in a desiccator. The composition was then filled into a second syringe, connected to the feeding tube and fed to the cavity until it emerged from the other tube. The electrodes were connected to the voltage source through a temperature-resistant silicone coated cable. Finally, the complete mold was placed on top of an aluminum heating plate set to the desired curing temperature. In order to be able to remove the samples from the electrodes, a mold release agen<sup>t</sup> had to be used. The standard mold release agen<sup>t</sup> is usually either silicone- or polytetrafluoroethylene-based, and so in both cases electrically nonconductive. In order to evaluate the e ffect of the electrical conductivity of the interface between nanocomposite and electrode, experiments were performed with three di fferent release agents: conductive silver paint, mechanical grease and carbon black filled grease.

#### *2.4. Electric Field*

A sinusoidal voltage of 60 Vpp at a frequency of 10 MHz was used to generate a sinusoidal alternating electric field strength of 400 Vpp/cm using a function generator Agilent 33250A and a high frequency power amplifier (Tabor 9260). The voltage and the frequency were chosen based on results presented in [22], where the procedure for electric field application is further described.

#### *2.5. Characterization of Electrical Resistivity*

For the investigation of electrical properties of samples produced with the three-roll mill, SWCNT/epoxy dispersions were used to produce bar-shaped samples (5 × 2 × 40 mm) in a silicone mold. For each experimental configuration, the electrical resistivity of six samples was measured using the four-point method. In this method, four contact points are used where the two outer ones lay an electrical current and the resulting voltage is measured by the inner ones.

Disk-shaped samples produced in the mold presented in Section 2.3 were characterized with electrochemical impedance spectroscopy in an IM6 workstation from Zahner-Elektrik GmbH (Kronach, Germany). This means that a measuring setup was used where a small sinusoidal potential of 1 mV with fixed frequency is applied to the sample, the response is measured, and the impedance computed at each frequency. The starting point frequency was set at 3 MHz and then it was gradually decreased to 500 mHz where the impedance assumes a constant value corresponding with acceptable approximation to the through-plane resistance.
