The ECA investigated in this study consists of an epoxy resin, a polar liquid and silver flakes (x_50,3 app. 11 µm in diameter, thickness: 100 nm), which were provided by Protavic International (France). PEG 200 [average molecular weight: 190-210 Da, glass transition temperature (10 K·min^-1): -89 °C, viscosity (25 °C): 0.05 Pa·s] was purchased from Sigma-Aldrich (Germany).

ECAs with silver contents ranging from 1 vol% to 20 vol% were prepared. A low-filled ECA containing 5 vol% silver and modified with PEG 200 was investigated in more detail. These formulations also contain 0.5 vol% of a polar liquid. PEG 200 was added separately at concentrations varying between 0 and 10 vol%, relative to the polymer matrix. For electrical characterization, samples were cured on a hot plate at 150 °C for 2 min. Samples for mechanical testing were cured at 150 °C for 10 min in an oven (vacutherm, Heraeus, Germany). The longer curing time in the oven accounts for the slower heat transfer compared to direct hot-plate curing.

Volume resistivity was measured by 4-point-measurement with a 2450 SourceMeter (Keithley, USA). Three ECA lines of 4 mm width and 0.1 mm thickness were applied onto glass substrates using a doctor-blade through a stencil, cured, and measured three times each at a current of 100 mA. Mean values $$ \bar{x} $$ and standard deviations σ were calculated, according to Eqs. 1 and 2, where x refers to the value, i to a single measurement, and n to the total number of measurements using Microsoft Excel.

(1) $$ \bar{x}=\frac{\sum_{i}^{n} x_{i}}{n} $$

(2) $$ \sigma=\sqrt{\frac{\sum_{i}^{n}\left(x_{i}-\bar{x}\right)^{2}}{n-1}} $$

Contact resistivity measurements were adapted from^[11]. An ECA line was cured on a circuit board featuring ten silver fingers. Joint resistance R_J, ECA bulk resistance R_B, and finger resistance R_F were measured using a 4-point-measurement. Contact resistivity ρ_c was calculated according to Eq. 3^[11]:

(3) $$ \quad \rho_{c}=\left(R_{J}-R_{B}-R_{F}\right) / 2 \cdot A $$

where A denotes the contact area. Mean values and standard deviations were derived from 28 data points.

Yield stress was measured using a rotational rheometer (MARS, ThermoFisher, Germany) equipped with a plate-plate geometry. Shear stress was increased from 0.1 Pa to 1,000 Pa in 81 logarithmic steps. The resulting deformation was plotted against shear stress in a double-logarithmic diagram. The yield stress was determined as the intersection point of tangent fits to the regimes of elastic deformation and viscous flow^[26]. Each sample was measured three times, and mean values and standard deviations were calculated.

Lap-Shear strength was measured according to^[27] using carbon-steel as the substrate. The bonding area was app. 15 mm². Three measurements per formulation were performed using a tensile testing machine (TA.XT plus Texture Analyser, Stable Micro Systems, UK) at a crosshead speed of 0.2 mm·s^-1. For the determination of Young’s Modulus, ten specimens per formulation were prepared in accordance with DIN EN ISO 20753, type A22^[28] and pulled at a crosshead speed of 0.01 mm·s^-1. For both parameters, mean values and standard deviations were calculated.

Microstructural analysis of uncured and cured ECAs was conducted using an optical microscope (Leica DMRX, Leica, Germany) at 100× magnification. To quantify the particle area coverage and the mixing quality, microscope image sections with dimensions of 1000 × 1250 pixels were analyzed. The microscope images were converted to black and white using a threshold value of 125 (greyscale range: 0-255), where white pixels represent silver particles and black pixels correspond to the epoxy resin, polar liquid and PEG. The homogeneity of the particle/polymer mixture was determined using the mixing quality parameter according to Eq. 4^[29]. A total of 100 subsections (100 × 125 pixels) were evaluated. A mixing quality of 1 indicates perfect mixing, whereas a value of 0 corresponds to complete phase separation.

(4) $$ \quad { mixing } \ { quality }=1-\sqrt{\frac{\sum_{1}^{n}\left(p_{i}-\bar{p}\right)^{2}}{n-1}} $$

where pⁱ denotes the fraction of white pixels in subsection i and $$ \bar{p} $$ the average fraction. The characteristic length of the polymer-rich regions (black areas) was determined by placing nine horizontal and nine vertical lines on each image. The number of consecutive black pixels was counted and converted into a physical length (µm) in order to quantify the degree of heterogeneity of the particle/polymer mixture. Thereby, a minimum of 10 consecutive white pixels was classified as a single particle. Smaller clusters were assigned to the polymer phase. For both mixing quality and characteristic length analyses, four images per ECA formulation were analyzed.

Differential scanning calorimetry (DSC; DSC 214 Polyma, Netzsch, Germany) was used to determine the glass transition temperature T_g of the components and ECA formulations. Samples were cooled to -130 °C and subsequently heated to -20 °C at a rate of 10 K·min^-1 for two cycles. T_g was determined from the step-change in the heat flow during the heating cycle as the temperature corresponding to half of the enthalpy change.

For contact angle determination, silver particles were compressed into flat discs. A droplet of secondary liquid (polar liquid and PEG) was set on the particle surface within the surrounding main liquid (polymer and PEG). A shadow image of the drop was recorded and analyzed using the Drop Shape Analyzer (DSA, Krüss, Germany) software. Drop shape analysis was performed five times per liquid combination, and mean values and standard deviations were calculated. Interfacial tension was determined using the pendant drop method. Sample turbidity was characterized by measuring the optical density at app. 600 nm using a photometer (Implen OD600, Implen, Germany). Each sample was measured three-times relative to a reference sample.

The components of a commercially available ECA, consisting of an epoxy resin, silver flakes, and a small amount of polar liquid, were employed in this study. In a previous study^[13], the silver flake volume fraction was increased from 5 vol% to 20 vol%. Even at a low filler content of only 5 vol%, the cured ECA exhibited a volume resistivity below 10^-2 Ω·cm [Figure 1A], which is sufficient for application in solar cell shingling. Increasing the filler content led to a further reduction in volume resistivity. At higher filler contents (10-20 vol%), the yield stress of the uncured ECA remained nearly constant at (56 ± 8) Pa [Figure 1B]. However, the yield stress of the low-filled ECA containing only 5 vol% silver flakes was significantly lower, at (5 ± 3) Pa. Such a low yield stress is insufficient for screen printing or dispensing processes. Moreover, suspensions containing micron-sized, high-density particles with such low yield stresses are prone to sedimentation and phase separation. Nevertheless, the electrical performance at low filler content indicates that ECAs containing 5 vol% filler or less may be accessible, provided that adequate paste stabilization is achieved.

As shown in Figure 2, the addition of polyethylene glycol with an average molecular weight of 200 Dalton (PEG 200) enhances the yield stress of such a low-filled ECA and its lap-shear strength in the cured state. For an ECA containing 5 vol% silver flakes, the addition of 1 vol% PEG 200 (relative to the epoxy resin) increases the yield stress from (5 ± 3) Pa to (32 ± 1) Pa [Figure 2A]. Upon further addition of PEG 200, the yield stress decreases again and vanishes at 10 vol% PEG 200. A similar non-monotonic trend is observed for the lap-shear strength [Figure 2B]. In the absence of PEG, the ECA exhibits a lap-shear strength of (17 ± 4) MPa, which increases to a maximum of (28 ± 3) MPa at 3 vol% PEG 200. Even at a high PEG 200 content of 10 vol%, the ECA retains a comparatively high lap-shear strength of (24 ± 2) MPa. Notably, the Young’s Modulus of an ECA containing 5 vol% silver is not affected by the addition of PEG and remains constant at an average value of app. 1,380 MPa [Supplementary Figure 1]. Low-filled ECAs containing 1-5 vol% PEG 200 thus meet the requirements for yield stress and lap-shear strength for shingled solar cell interconnection.

Remarkably, volume and contact resistivity remain unaffected by the addition of PEG. The volume resistivity remains merely unchanged at app. (0.4 ± 0.04) × 10^-2 Ω·cm [Figure 3A], while the contact resistivity is (0.03 ± 0.004) mΩ·cm² across all formulations, independent of PEG 200 content [Figure 3B]. Therefore, PEG 200 can be used to tailor yield stress and lap-shear strength without compromising the low electrical resistivity required for the intended application.

For the ECA containing 5 vol% silver flakes, the maximum yield stress is observed at 1 vol% PEG 200, whereas the maximum in lap-shear strength occurs at 3 vol% PEG 200. Therefore, a PEG 200 content corresponding to half the particle volume was selected for subsequent experiments. To determine the minimum required silver content, the silver volume fraction was further reduced in steps of 1 percentage points. Formulations containing 3 vol% silver flakes or more exhibit a yield stress. At 4 vol% particles (with 2 vol% PEG 200), the yield stress reaches (12 ± 3) Pa, thereby exceeding the lower limit required for application [Figure 4A]. Figure 4B depicts the corresponding volume resistivity. Formulations without a yield stress are electrically insulating. At 3 vol% silver, the volume resistivity decreases to (2 ± 0.4) × 10^-2 Ω·cm, while at 4 vol% silver flakes it meets the target value for solar cell shingling of 1 × 10^-2 Ω·cm.

Since PEG is known to be hygroscopic and moisture plays a critical role in solar cell operation, moisture uptake was evaluated for both uncured and cured ECA samples. No measurable moisture content was detected in the uncured sample, and the mass of the cured samples remained constant over one month during storage under standard laboratory conditions. For all samples, including 3 replicates per formulation, variations were within the measurement uncertainty of the balances. These results indicate that no significant water uptake occurs in the investigated ECAs containing PEG.

Beyond initial performance, ECAs must withstand stresses encountered during work life. In shingled solar modules, these stresses primarily arise from changing environmental conditions, such as temperature fluctuations and therefore, thermomechanical stresses due to mismatched coefficients of thermal expansion of the different materials in the PV module, as well as moisture exposure. On ECA level, four specimens each were subjected to simplified thermal cycling and damp heat testing. After aging, the electrical performance of the ECA containing 4 vol% silver and 2 vol% PEG varied only within the measurement uncertainty compared to the initial values. At module level, stress testing is conducted according to IEC 61215-2. Most common are the thermal cycling test (MQT 11), in which modules are cycled between -40 and 80 °C, as well as the damp heat test (MQT 13), involving storage at 85 °C and 85% rel. humidity. Modules produced with ECAs containing the same components as those used in this study, but in different concentration, exhibited a competitive initial module efficiency of (19.6 ± 0.2)% and limited relative efficiency losses of (1.0 ± 1.4)% after 500 h of damp heat exposure and (2.4 ± 1.0)% after 200 accelerated thermal cycles^[30]. Therefore, comparable long-term performance can be expected for the ECAs developed in this study.

In comparison with commercially available ECAs dedicated to or suitable for PV applications, the ECA containing 4 vol% silver exhibits comparable or superior electrical and mechanical performance despite its very low silver filler content. Specifically, it achieves a yield stress of (12 ± 3) Pa, a volume resistivity of (1.0 ± 0.6) × 10^-2 Ω·cm, a contact resistivity of (0.04 ± 0.02) mΩ·cm², a lap-shear strength of (20 ± 3) MPa, and a Young’s Modulus of (517 ± 70) MPa. Reference data presented in Figure 5 were obtained from product datasheets of commercial silver-based ECAs [e.g. Henkel (Germany), NagaseChemteX (US) and Polytec PT (Germany)] and from literature reports^[31]. Recent studies focusing on alternative conductive fillers, such as copper^[32-35], were not considered here, as the present work emphasizes filler reduction rather than substitution.

In conclusion, PEG 200 effectively stabilizes low-filled ECAs against de-mixing or sedimentation by inducing a yield stress in the range of 10-100 Pa, thereby enabling processing via screen-printing and dispensing. The corresponding cured ECAs meet all relevant electrical and mechanical requirements for solar cell shingling. Thus, PEG 200 enables a significant reduction in silver consumption for solar cell interconnection, allowing the use of ECAs containing as little as 4 vol% silver instead of conventional systems with 20-30 vol% silver.

PEG 200 has previously been reported to partially remove the organic fatty acid layer, which forms on the silver surface during particle processing^[17], resulting in a decreased volume resistivity. However, in the present study, no such decrease in resistivity was observed. Instead, thermogravimetric analysis indicates PEG 200 absorbs onto the silver surface (see Supplementary Figure 2), partially replacing rather than removing the fatty acid layer. Notably, the molecular weight of PEG 200 is comparable to that of commonly reported fatty acids typically on the particle surface, such as stearic acid (284 g/mol) or oleic acid (282 g/mol). Consequently, while the chemical composition of the organic surface layer may change, its overall thickness is likely preserved, and a reduction in resistivity is therefore not expected.

PEG 200 primarily enhances the mechanical properties of the ECA in both the uncured and cured states. To further investigate this effect, light microscopy images of the uncured and cured samples were taken. No structural changes were observed upon curing, indicating that the particle arrangement is largely preserved during crosslinking. Representative micrographs of cured samples containing 5 vol% silver flakes and varying PEG 200 content (0-10 vol%) are shown in Figure 6. The original greyscale images were converted into binary images using a threshold value of 125 (greyscale range: 0-255). An overlay of the original and processed images is provided in the Supplementary Materials [Supplementary Figure 3]. In the binary images, black pixels represent the epoxy resin, including the polar liquid and PEG where applicable, while white pixels represent the silver particles. The average area fraction of white pixels of app. 32% exceeds the actual particle content of 5 vol%. This discrepancy can be attributed to the preferential alignment of flake-shaped particles parallel to the sample surface, and to some extent to the optical depth of field of the microscope lens. The two-dimensional images capture features within a depth of several micrometers rather than an infinitesimally thin plane, thereby giving the impression of a higher particle loading. From the images, a morphological evolution in particle distribution is evident with increasing PEG 200 content. The initially homogeneous particle distribution transitions into a progressively heterogeneous microstructure. This change is quantitatively reflected by an increase in the average characteristic length of polymer-rich regions (black areas) from (31 ± 1) µm to (59 ± 4) µm [Figure 7A], indicating a microscale phase separation induced by PEG addition. Consistently, the mixing quality parameter decreases from (67 ± 4)% to (48 ± 3)% [Figure 7B], confirming progressive structural coarsening as the PEG 200 content increases from 0 vol% to 10 vol%.

While the morphological analysis reveals a monotonic coarsening of the microstructure with increasing PEG 200 content, both yield stress and lap-shear strength [Figure 2] exhibit non-monotonic behavior. Regarding the mechanical strength of solid composite materials, Gordon et al.^[36] demonstrated that the cohesion of composites strongly depends on particle distribution and orientation. The increasing microheterogeneity with increasing PEG 200 content may lead to an enhancement in lap shear strength. This effect is, however, counteracted by the increasing PEG 200 content itself, which does not participate in the curing reaction. Therefore, PEG 200 does not contribute to the cohesive strength, contrarily, it substitutes up to 10 vol% of the cured polymer matrix. Consequently, the cohesive strength decreases at high PEG contents, resulting in an optimum at 3 vol% added PEG 200.

The increase in yield stress upon addition of PEG [Figure 2A] indicates the formation or reinforcement of a strong, uniform particle network. At higher PEG contents, however, this network appears to transition into a structure of weakly bound particle agglomerates, resulting in a reduction of both strength and yield stress. Such a transition from well-distributed particles to a strongly bound, uniform particle network and subsequently to a network of weakly bound particle agglomerates is typical for capillary suspensions^[37] and is consistent with the observed microstructural evolution in the investigated ECAs.

Capillary suspensions^[25] are ternary solid-liquid-liquid systems in which a small amount of an immiscible secondary liquid is added to a binary suspension. The secondary liquid induces the formation of a sample-spanning, so-called percolating particle network, resulting in a strong yield stress and pronounced shear thinning behavior. When the secondary liquid preferentially wets the particles (three-phase contact angle < 90°), strong pendular bridges form between the particles and accordingly this is called the “pendular” state. The strength of a “pendular” state suspension is governed by the capillary force acting on each particle bridge, which depends on the interfacial tension between the main and secondary liquid phase and the three-phase contact angle^[38]. A large fraction of preferentially wetting secondary liquid may lead to so-called spherical agglomeration, i.e. the formation of large compact agglomerates, which at high enough particle loading self-assemble into a weakly bound percolating network, hence resulting in a decrease of yield stress compared to a uniform pendular network^[39]. In systems in which the bulk liquid preferentially wets the particles (three-phase contact angle between 90° and app. 145°), particles assemble around droplets of the secondary liquid, shielding it from the main liquid, forming clusters that reduce the free energy of the system. These particle clusters self-assemble into a percolating particle network, referred to as the “capillary” state, and again result in a strong yield stress and pronounced shear thinning of the suspension. The cluster configuration and hence the yield stress of the suspension primarily depend on the three-phase contact angle and the amount of secondary liquid^[25].

A necessary requirement for the formation of a capillary suspension is the presence of a distinct, immiscible secondary liquid phase. However, DSC results [Supplementary Figure 4] show that PEG 200 is miscible with both the epoxy resin and the polar liquid, as evidenced by single glass transition temperatures in their respective mixtures, located between the T_g of the pure components. In contrast, epoxy resin and the polar liquid are non-miscible, as confirmed by a finite interfacial tension of (7.7 ± 0.2) mN·m^-1 between the phases.

DSC results for ECAs containing 5 vol% silver flakes and varying PEG 200 contents are depicted in Figure 8A. The curves exhibit a glass transition at -56 °C, corresponding to the epoxy resin (lower diagram in Figure 8A). This T_g decreases with increasing PEG 200 content, confirming partial dissolution of PEG 200 in the resin. Additionally, a second glass transition appears for the ECAs containing PEG 200 at app. -85 °C, the glass transition temperature of PEG 200 (upper diagram in Figure 8A). The relatively small step height of this lower T_g reflects the low mass fractions of PEG (0-6.5 wt%) and polar liquid (0.4 wt%). The presence of the glass transition between -80 and -90 °C is further confirmed at higher heating rates, where the signal becomes more pronounced [Supplementary Figure 5]. As the fraction of polar liquid remains constant, the increasing intensity of the low-temperature glass transition must be attributed to PEG 200. It is therefore concluded that PEG partially contributes to the secondary liquid phase volume. Quantitative analysis suggests that app. 17% of the PEG 200 in the ECA containing a total of 5 vol% PEG contributes to the secondary liquid phase. For the ECA with 10 vol% PEG 200, this fraction increases to 26%. This corresponds to secondary liquid volume fractions of 0.8 and 2.6 vol%, resp., in addition to the 0.5 vol% polar liquid. An increase in secondary liquid volume fraction was confirmed by determination of the absorbance at 600 nm of samples containing epoxy resin, polar liquid and different volume fractions of PEG 200. The absorbance is a measure for turbidity and increases slightly with increasing PEG 200 content [Figure 8B], indicating an increase in the number and/or size of dispersed secondary liquid droplets.

To assess whether the formation of a capillary suspension is possible and to identify which state of capillary suspension may form in the ECA, the three-phase contact angle of a PEG/polar liquid mixture including 62 vol% PEG 200 and 38 vol% polar liquid (corresponding to the composition in the ECA with 5 vol% PEG as determined from DSC) in epoxy resin (main phase) against pressed silver flakes was determined. The resulting contact angles which represent mean contact angles expected to be present in the investigated system are shown in Figure 9A. The contact angles of 126° ± 1° and 124° ± 2°, for 0 vol% or 5 vol% PEG 200 in the epoxy resin, respectively, indicate the formation of a capillary suspension in the “capillary” state. For comparison, the three-phase contact angles of drops of pure polar liquid were determined as 145° ± 5° and 140° ± 5°, depending on the PEG 200 content in the epoxy resin. According to^[25], a critical upper limit of the contact angle exists beyond which the formation of particle clusters enclosing the immiscible secondary liquid is no longer energetically favorable, i.e. a “capillary” state type capillary suspension cannot form when the contact angle exceeds this critical value depending on cluster geometry and secondary liquid droplet volume. Although the considerations in^[25] refer to a system of monodisperse spherical particles, we assume that such a critical contact angle value also exists for the flake-shaped particle suspensions investigated here. The high contact angle of the pure polar liquid may prevent the formation of a capillary suspension, despite the non-miscibility of the epoxy resin and the polar liquid. The combination of polar liquid and PEG 200 therefore appears to be necessary to obtain a strong capillary suspension in the “capillary” state. Indeed, no comparable morphological changes were observed in systems containing only increased polar liquid content or PEG 200 in the absence of the polar liquid (data not shown).

Figure 9B shows the interfacial tension between different mixtures of polar liquid and PEG 200 and epoxy resin. Obviously, the addition of PEG 200 to the secondary phase leads to a drastic drop in interfacial tension and hence should weaken a capillary suspension network^[25]. While a mixture of polar liquid and PEG 200 is needed to reduce the three-phase contact angle and enable the formation of a sample spanning particle network controlled by capillary forces which is characterized by a strong yield stress, the capillary force decreases with decreasing interfacial tension, i.e. increasing PEG 200 fraction in the polar liquid/PEG mixture and accordingly the yield stress decreases [Figure 2A]. Finally, the PEG/polar liquid mixture including 84 vol% PEG 200, as calculated for the ECA containing 5 vol% silver and 10 vol% PEG 200, does not form stable droplets in the epoxy resin (with or without PEG 200), as evidenced by pendant drop experiments. The liquid pulls threads and the phase surface is crumply [Supplementary Figure 6], i.e. there is no finite interfacial tension anymore. This explains the disappearance of a yield stress in the ECA containing 10 vol% PEG 200 [Figure 2A].