Introduction

Electronic waste generation (e.g. mobile phones, solar panels, smart packaging) is predicted to double by 2050 to reach approximately 111 million tonnes per year [1]. Currently only around 20% of this waste is collected and recycled [1], leaving a significant amount of plastic and heavy metal waste to be released into the environment. The use of hybrid electronics has grown significantly over the last decade [2], offering flexible, light, and freeform circuits that can be more easily incorporated into products. This however only compounds the problem of increasing waste by adding functionality to everyday objects with a short lifespan.

Hybrid electronic devices usually comprise a substrate, conductive ink, conductive adhesive, and a surface mount device (SMD). The SMD accounts for most of the cost, whilst the substrate and conductive printed ink (usually silver) account for the majority of the device by mass. Silver inks dominate the conductive ink market due to the high conductivity silver offers, combined with the long-term stability (e.g. against oxidation when compared with unsintered or unprotected copper [3]); however, it is a high cost material of which demand is beginning to exceed supply [4]. Many articles discuss the development of lower-cost silver inks [5], low conductive filler content inks [6] or alternative filler content (e.g. carbon based) conductive inks [7]; however, the development of more sustainable and recoverable silver inks is less often discussed. It should also be noted that the amount of silver per device is small, making it challenging to develop economically viable recovery processes; the use of lower silver content inks [6] therefore makes recollection less appealing potentially resulting in more silver ending up in landfill. Carbon-based conductive additives are often discussed as biodegradable alternatives to conventional conductive inks; however, the conductivity is commonly several orders of magnitude lower than silver-based equivalents [7].

This article demonstrates how better material selection can produce high-performing inks using bio-derived materials, which can lead to the recycling of full devices to recover and reuse the substrate, SMD and silver filler with minimal processing.

Materials and methods

Cellulose acetate butyrate (CAB, Mn 12,000, 30,000 and 70,000 g/mol), ethyl acetate, amyl acetate, heptyl acetate, octyl acetate and benzyl acetate were sourced from Sigma-Aldrich. Silver flakes EA0295 and particles K-1322P and P620-7 were sourced from Metalor.

Silver ink Loctite ECI 1010 and isotropic adhesive Loctite Ablestick CE 3103WLV were supplied by Henkel. Knightbright 1206 SMD blue light-emitting diodes (LED) were sourced from Farnell. Polyethylene terephthalate (PET) substrate ST506 was sourced from Dupont and 90 gsm standard office paper was sourced from Lyreco.

Polymers were dissolved in solvents by roller mixing overnight at 60 rpm at 25 °C. Silver was dispersed into polymer solutions using a Speedmixer DAC400 dual-axis centrifugal (DAC) mixer, at 1500 rpm for 2 min to produce screen inks.

Rheology was characterised using a TA Instruments DHR2 rotational rheometer whereby two methods were used: a two-point viscosity (TPV) scan to quickly screen viscosity at low-shear rate and a shear sweep recovery (SSR) scan to determine the shear thinning nature and extent to which viscosity recovered when shearing was reduced. The TPV scan subjected samples to 1.5 s−1 for 60 s (s), followed by 15 s−1 shear rate for another 60 s. The SSR scan sheared the sample from 1 to 500 s−1 measuring 3 points per decade, followed by the inverse scan. This data was also used to determine viscosity recovery by comparing the viscosity at 2.2 s−1 of the second scan with the viscosity of the first scan as a percentage. In both cases a sand-blasted 40 mm stainless steel parallel plate geometry was used with a 500 µm gap.

Resistance was measured using a Jandel four-point probe, comprising 100-µm diameter pins with a 1-mm tip spacing applying a 60–150 g load, attached to a Keithley 2450 sourcemeter applying a current of 100 mA. Thickness of printed blocks was measured using a Hanatek FT3V-Lab semi-automatic thickness gauge. Voltage was calculated using Ohms law (Eq. 1) [8], and resistivity and sheet resistance are calculated using Eqs. 2 and 3 [9, 10].

$$V = IR,$$
(1)
$$\rho = \frac{{V {\text{CF}} t}}{I},$$
(2)
$$R_{{\text{s}}} = \frac{\rho }{t}.$$
(3)

Note that ρ is the resistivity, Rs is the sheet resistance, V is the voltage, CF is the correction factor, and I is the current. The correction factor used was for a 20 mm × 30 mm rectangle on a non-conducting substrate [9].

Coatings were prepared by blade coating onto PET using 50 µm spacers. Inks were flatbed screen printed manually and using a DEK Horizon with a 195 screen mesh, as well as being NSM rotary screen printed with a 325 mesh. In all cases samples were dried in a box oven at 150 °C for 5 min. Printed tracks were imaged using a Dynoscope digital microscope and a Talysurf white light interferometer with a × 1.25 lens and a 30 s scan. NFC coils were printed onto PET and paper, and 50 µm scotch tape was used as a dielectric layer across the tracks to prevent shorting once a crossover was attached. Crossovers were printed and attached to the relevant parts of the coil with conductive epoxy adhesive, before attaching an LED with conductive adhesive (see Supplementary data, Fig. S1). Coils were tested by placing on the back of a Samsung S21 phone with the NFC activated to carry out a visual check for a functioning LED.

Adhesion was tested using an Elcometer 1542 cross hatch tape tester, where results were visually compared to the ASTM classification. Thermogravimetric analysis (TGA) was carried out to check the composition of formulated inks using a Perkin Elmer Pyris 1 where samples were subjected to a temperature ramp from 20 to 400 °C at a rate of 10 °C/min under nitrogen. Accelerated ageing was carried out by subjecting samples to 85 °C and 85% relative humidity (RH) in a humidity chamber for 1 week.

For recycling, NFC coils were placed into benzyl acetate and agitated using a lab shaker for 20 min. Substrate and LEDs were removed and rinsed with fresh solvent. The dissolved silver ink was agitated for a further 20 min prior to decanting excess solvent, washing with fresh solvent and allowing the remaining silver to dry in an oven at 150 °C for 20 min.

Material selection

Cellulose acetate butyrate was selected as the binder due to the bio-derived nature of the polymer [11]. To determine a suitable set of solvents, Hansen Solubility Parameters (HSP) were used to identify a candidate solvent [12]. CAB had a dispersive HSP (δd) of 22.1, polar HSP (δp) 6.7, and hydrogen bonding HSP (δh) 5.2 [13] which was compared to HSP values for solvents found in a database in software HSPiP [14]. From the list identified benzyl acetate was found to be the least hazardous and most sustainable (δd 18.3, δp 5.7, δh 6.0). Focussing on esters due to their bio-derived nature, four additional solvents were selected with a range of volatilities to enable control over ink drying: ethyl acetate, amyl acetate, heptyl acetate, and octyl acetate.

Results and discussion

A silver ink was developed in a sequential manner using a fail-fast approach—this allowed for quick screening of important parameters using a cellulose acetate butyrate (CAB) binder, ester solvents and silver conductive fillers.

CAB grades were screened in benzyl acetate by dissolving CAB in the solvent at 5–30% by mass. To achieve a viscosity of ~ 1 Pa s for each grade, CAB12K required 20 wt%, CAB30K required 15 wt% and CAB70K required 7 wt% (Fig. S2).

Several grades of silver with differing sizes and morphologies were screened at 70 wt% loading in a resin containing 15 wt% CAB30K in benzyl acetate, to determine the effect on resistivity and low-shear viscosity (Fig. 1a). The smallest spherical particles exhibited the highest resistivity, thought to be due to a high number of particle-particle contacts [15]. The flake grade resulted in the lowest resistivity and desirable low-shear viscosity; the high-aspect ratio nature of flakes potentially (i) increased the likelihood of contact between particles and (ii) resulted in alignment during shearing, therefore reducing resistance to flow. Silver flake loading was adjusted between 60 and 80 wt% to ascertain the optimal loading. Increasing silver loading resulted in an increase in low-shear viscosity, as expected, with all loadings exhibiting a suitable viscosity for a screen ink. Conductivity was assessed using sheet resistance rather than resistivity to allow for understanding of the inks’ behaviour in application and comparison to datasheets; resistivity is an intrinsic property that accounts for thickness, whereas sheet resistance is an application-related property where thickness is not accounted for in the calculation [9, 10]. Figure 1b shows the effect of silver loading on sheet resistance, where above 70 wt% the drop in sheet resistance was not sufficient to justify the additional cost of using an extra 10 wt% silver.

Fig. 1
figure 1

a Effect of silver particle size and shape on resistivity and low-shear viscosity, b Influence of silver loading on sheet resistance and low-shear viscosity, and c Influence of solvent selection on resistivity and low-shear viscosity of silver inks

Full size image

Solvents were screened by replacing benzyl acetate with the several lower boiling point esters to produce various silver inks. Ethyl acetate and amyl acetate-based inks dried during printing resulting in blocked screens (67 µm pore size), making them unsuitable. Heptyl acetate resulted in a higher resistivity and lower low-shear viscosity which suggested incompatibility between the formulation components. Benzyl acetate resulted in optimum resistivity and low-shear viscosity (Fig. 1c), validating the selection of this solvent via Hansen Solubility Parameters.

Using a mixture of polymers with different molar masses can enable a high degree of control over ink rheology. For a conductive screen ink, it is important to have enough polymer in the ink to enable it to bind to the substrate, whilst also having a sufficiently high molar mass to enable entanglement, thought to result in a high degree of viscosity recovery when under zero-shear conditions. Optimisation of the ink’s rheology was carried out by combining stock solutions created previously using CAB12K, CAB30K and CAB70K. Each solution was combined with the other two solutions at a 1:1 ratio, and all three were mixed at a 1:1:1 ratio. Silver was added at a 70 wt% loading and the optimum formulation contained CAB12K and CAB30K, which resulted in 5B adhesion, a resistivity of 3.3 × 10–5 Ω cm and a viscosity recovery of 96% (Table S1). Note that surfactants are often used to stabilise flakes and particles in dispersion [16]; however, the ink was found to be sufficiently stable without the addition of a surfactant.

The final ink was successfully produced at a 10 g scale several times and at a 1.5 kg scale to validate the ink and demonstrate it could be scaled up, results of quality control tests can be found in Table S2. The rheology of the ink was found to match the characteristic shape for a screen ink; significant thixotropy with increasing shear rate and recovery of the inks viscosity following reducing shear rate (Fig. S3).

Following accelerated ageing to assess long-term stability, adhesion remained at 5B classification and resistivity was measured to be 3.5 × 10–5 Ω cm, which was not a significant increase (Fig. S4). High-resolution tracks (Fig. 2a) with low roughness (Fig. S5a) were achieved through flat bed and rotary screen printing (Figs. S5b and S5c); a resistivity within the same order of magnitude as commercial silver ink Loctite ECI1010 was achieved [17] (Fig. 2b).

Fig. 2
figure 2

a Microscopy image of the ink printed using a high-resolution test pattern and b Comparison of resistivity for the developed ink alongside ECI1010 commercially available screen ink printed onto PET

Full size image

Evaluation of the safety of the developed ink determined that LEV was not required for the process of producing the ink, and thinner butyl gloves could be used for longer periods of time when compared to ECI1010 (Table 1). Since LEV can use between 0.7 and 1.2 kW of electricity per hour [18] (comparable to the energy consumption of a microwave [19]) and less stringent glove requirements would result in a factor of 8 less rubber ending up in landfill, the overall outcome is a safer ink with a lower environmental impact.

Table 1 Comparison of safety and economic viability of solvents used in the developed ink and ECI1010
Full size table

As the polymer in ECI1010 was not known and both inks contain silver, the difference in the cost of solvents was considered and found to be negligible (Table 1). In addition, dissolution of the polymer and dispersion of silver were both achieved without much difficulty, therefore processes could be translated to large-scale processing methods, e.g. impeller mixers and three-roll mills. Thus the developed ink should in theory be as scalable and economically viable as ECI1010.

Recycling an NFC antenna

Previous steps demonstrated how a high-performing ink could be developed using more sustainable materials, however such a device would not lend itself to recycling due to the irreversible crosslinking of the adhesive. To improve the green credentials of this device the configuration was simplified to enable the use of ink in place of the adhesive. Whilst the use of an ink will undoubtably reduce the adhesion strength of bonded components, the purpose was to demonstrate feasibility of recycling this sort of system to inform development of recyclable adhesives in the future work.

Following removal of dielectric tape and adhered crossovers, the remainder of the NFC coils (Fig. 3a) were mostly recycled in a single process. The silver ink and LED were removed without difficulty from the PET substrate (Fig. 3c); however, it was not possible for the paper device as ink had penetrated the porous structure. LEDs were found to still function following recovery and could therefore in principle be reused (Fig. 3e), although more detailed lifetime testing would be required to fully validate this. Recovered silver (Fig. 3d) was validated in an ink and coated onto the recovered PET. The resistivity of the recycled silver ink was found to increase by an order of magnitude (Fig. 3b), which would still be of sufficiently low resistivity to enable use in some antenna applications and shows potential for recovering silver flake in this way. However, it should be noted that the appearance of the dried recycled silver ink was grainy as there was likely significant agglomeration of the silver and residual CAB, therefore a lower resistivity may be possible following further processing. Future work should include better characterisation, purification and deagglomeration of the recovered silver flakes.

Fig. 3
figure 3

a Functioning NFC antenna with integrated LED, b Comparison of resistivity for inks containing virgin and recycled silver flake, printed onto PET, c Recycled NFC antennae in benzyl acetate, d Recovered silver flake, and e Recovered LED still functional.

Full size image

Conclusions and future work

This work successfully demonstrates the use of a safer and more sustainable material set to reduce the environmental impact of printed electronics, whilst achieving resistivity comparable to that of state-of-the-art commercial inks. The inks produced could be both flat bed and rotary screen printed without requiring LEV, showing the potential for these materials at industrial scale. In addition, it was demonstrated how device simplification can result in recyclable devices to enable the recollection of high-value elements and components using low-intensity processing methods. Future work will look to improve purification and deagglomeration methods for noble metallic components, and the development of debondable adhesives to enable good adhesion with options for recycling SMD components.