Novel approach to utilise highly conductive but electrochemically unstable current collector materials in textile supercapacitor electrodes

Metal-based materials, such as silver or copper, are highly desired as current collector materials for flexible energy storage due to their excellent electrical properties but lack the long-term operational electrochemical stability. Herein we report a method to prevent the corrosion of such materials, while fully exploiting their electrical properties. This was achieved by covering the current collector with an electrochemically stable conductive carbon-based layer. The barrier layer allows the flow of charge between the electrically conductive elements of the textile composite electrodes, while protecting the current collector from contacting the electrolyte. The areal power and energy densities obtained after 1000 bending cycles were 29.88 and 0.01 mWh cm−2, respectively, with no evident degradation. Additionally, patterned current collectors were designed to deposit lower quantities of ink, without detriment to electrochemical performance. After 1000 bending cycles, the textile composite supercapacitors (TCSs) having 50% less current collector material demonstrated an areal power and energy density of 28.08 and 0.01 mWh cm−2, respectively. The proposed strategy is essential in enabling the utilisation of highly conductive metal-based inks, improving the rate capabilities and long-term operation of wearable energy storage devices, while maximising specific power and energy densities of TCSs, and decreasing the manufacturing cost.


Introduction
Wearable electronics have attracted attention due to their wide range of potential applications ranging from body-worn entertainment, protection, and sensing, to communication and medical purposes [1,2]. In order to operate, energy supplying devices are necessary, which are required to be lightweight, flexible, and to possess high energy and power performance [3].
Challenges identified in the development of textile composite electrodes (TCEs) stem from the high electrical resistances [4]. Carbon-based textiles are the most-used conductive substrates [5][6][7][8][9][10][11][12], however the sheet resistance is much higher than that of metal foils, leading to high ohmic losses in TCEs [13,14]. In order to increase the electrical conductivity of the substrates, electrically conductive metallic yarns can be made into fabric current collectors [15], or physico-chemical changes can be made to non-conductive fabrics [16,17], affording the required conductive features. On the other hand, highly conductive materials can be deposited on fabric through printing or coating technologies [18], thus bestowing the electrical conductivity required. In order to maximise the electrochemical performance of TCEs, the current collector should possess high electrical conductivity, electrochemical stability, and be able to withstand intensive flexural stresses.
The range of conductive inks typically employed in printed electronics include carbon or metallic materials. Conductive carbon inks comprise a dispersion with carbon blacks [19,20], graphene [21][22][23], carbon nanotubes [24] or a mixture with the three allotropes [25][26][27], with bulk resistivities ranging from 0.01 and 71.0 Ω cm. Compared to conductive carbon-based inks, metallic inks containing silver (Ag) or copper [28][29][30], possess superior electrical properties, achieving a bulk resistivity of 0.2 mΩ cm. Ag inks have better conductivity and stability than copper-based, but the high cost limits its use on larger scales. On the other hand, copper is a more accessible metal and also possesses high electrical conductivities. However, copper is readily oxidised under air, and has a tendency to form an insulating layer of oxide, which limits its widespread in printed electronics [29]. Although Ag possesses excellent electrical properties and is more stable compared to copper, it has been discarded in electrochemical systems due to the undesirable reactions that lead to decomposition [31][32][33], thus impacting on the lifetime expectancy. Despite of these circumstances, Ag conductive inks were selected as current collector materials in this study, to investigate the degradation mechanism occurring in textile composite supercapacitors (TCSs). For comparison, printed graphene current collectors were introduced. The utilisation of printed Ag current collectors demonstrated an areal power density 144 times greater compared to TCSs with graphene-based current collectors (Ag | AL: 7.22 mW cm −2 ; Gr | AL: 0.05 mW cm −2 ).
Despite the superior electrochemical performance, the long-term stability was compromised due to corrosion of Ag. In order to shield the printed Ag current collectors from undergoing undesirable electrochemical decomposition reactions, and to fully utilise the electrical properties of Ag in TCSs, a printable, flexible, conductive and electrochemically stable carbon-based BL was introduced. It was found that 3 BLs were required to completely mask the printed Ag current collector. Upon bending 1000 times, the TCEs (3 BL) demonstrated stable rate performances, reporting an areal power density of 29.88 mW cm −2 .
The conventional manufacturing process of printed planar-type supercapacitor electrodes involves the deposition of an active layer on the current collector. However, it may be advantageous to reduce the amount of current collector material, in order to increase the energy and power densities of the devices, while minimising the cost. In this regard, the versatility of screen printing was exploited by means of reducing the Ag content in the printed current collectors by 50%, through the design of novel patterns. The modified TCE's architecture with patterned current collectors preserved the electrochemical properties of TCSs, while resulting in a drop of manufacturing costs, and an increase in specific power and energy densities. After 1000 bending cycles, the TCSs with 50% reduced Ag current collectors demonstrated an areal power density of 28.08 mW cm −2 .
To our knowledge, the strategies herein presented were not reported elsewhere. These approaches pave the way to the integration, protection and cost-saving of highly conductive printed metal-based current collector materials in TCSs in an easy and scalable way.

Preparation of hydrogel
For the preparation of the hydrogel, 1 g of poly(vinyl alcohol) (MW 89 000-98 000, 99+% hydrolysed, Merck) was added to 10 ml of triple-distilled deionised water and heated to 80 • C until the solution became clear. Then, a previously prepared solution of 1 ml of H 2 SO 4 (95.0%-98.0%, Merck) and 10 ml of triple-distilled deionised water was added and the mixture vigorously stirred for an additional 30 min. Lastly, the hydrogel was allowed to rest for 24 h at room temperature.

Fabrication of TCEs with different printed current collectors
The fabrication of textile supercapacitor electrodes was carried out printing a layer-by-layer composite system using an ATMA AT-45 PA semi-automated screen-printing machine (Atma Champ Ent. Corp., Taoyuan City, Taiwan), with a pre-defined flooding and printing speed of 50 mm s −1 and 300 mm s −1 , respectively. Prior to printing, a breathable polyester microfiber fabric (Pennine Outdoor, Lancaster, United Kingdom) was washed and ironed to remove any dirt or grease, and to smooth the surface by removing wrinkles, respectively. To minimise surface roughness and solvent absorption, an interfacial layer of polyurethane emulsion (Smart Fabric Inks, Southampton, United Kingdom) was screenprinted on the fabric and cured for 40 min in an ultraviolet exposure unit (LV202E MEGA). To investigate the effect of current collector materials in the electrochemical performance of TCSs, graphene ink (Sun Chemical Ltd, Swansea, United Kingdom) or Ag paste (Sun Chemical Ltd, Swansea, United Kingdom) were deposited on top of the interfacial layer, and dried for 1 h in a box oven set at 130 • C. Then, one layer of active layer ink (5 × 5 cm 2 ), comprised of acrylic printing medium (Daler Rowney, System 3) (46.33%w/w), butyldiglycol (CAS: 112-34-5, Merck, ⩾99%) (28.15%w/w), 2-phenoxyethanol (CAS: 99 122-99-6, Merck, ⩾99%) (4.97%w/w), activated carbon Kuraray YP-50 F (Kuraray Co., Ltd, Tokyo, Japan) (19.84%w/w), Ketjenblack EC-600JD (Nouryon, Amsterdam, The Netherlands) (0.31%w/w), functionalised multiwalled carbon nanotubes (Cheap Tubes Inc., Cambridge Port, United States of America), with outer diameter ranging from 20-30 nm, inner diameter ranging from 5-10 nm, a length ranging from 10-30 µm, and a degree of COOH-functionalisation of 1.2%, was screen-printed and dried at 115 • C for 30 min. Finally, a layer of hydrogel was deposited and further vacuum-impregnated at room temperature for 60 min. The mesh specifications and printing parameters are found in tables S1 and S2 in the supplementary information, respectively. The thickness variation of the printed current collector materials can be found in table S3.

Fabrication of TCEs with BL
To evaluate the effectiveness of the BL, one, two or three layers of carbon conductive ink CHSN8032 (Sun Chemical Limited, Swansea, United Kingdom) were printed on top of the Ag current collector and dried at 130 • C for 60 min. Finally, one layer of the active layer ink was deposited and dried, followed by a screen-printed layer of hydrogel. Prior to printing, the carbon conductive ink was milled using an EXAKT50I three roll mill (EXAKT Advanced Technologies GmbH, Norderstedt, Germany), with both back and front rolls displaying a 15 µm gap, at a speed of 500 rpm. After three passes, the paste was transferred to the screen, and further printed. The thickness variation of the printed textile composite current collector components can be found in table S4. In order to investigate the electrolyte-blocking effect, additional composite electrodes with carbon conductive ink were printed on transparent PET (polyethylene terephthalate-Melinex ® 339, DuPont Teijin Films, United Kingdom).

Fabrication of TCEs with patterned current collectors
To evaluate the impact in the electrochemical performance by decreasing the amount of Ag in the TCEs, patterned current collector stencils were used, which allowed the deposition of 10, 30 and 50% less Ag (figure S1). The patterns were printed on top of an interfacial layer, followed by three layers of carbon conductive ink, a layer of active layer ink and lastly a layer of hydrogel.

Flexural endurance
The flexural strength of textile composite current collectors, translated by changes in resistance, as well as variations in the electrochemical performance after bending the TCSs, were performed with an in-house built cyclic bending station, as reported elsewhere [34]. Both textile composite electrodes and supercapacitors were subjected to a bending angle of 120 • , at a rotational speed of 100 RPM.

Material characterisation
Characterisation of the TCEs and post-reaction changes of the printed Ag current collectors was carried out employing scanning electron microscopy (SEM). X-ray powder diffraction (XRD) samples were mounted in Kapton tape, and the patterns were recorded using a Bruker D8 Discover x-ray Powder diffractometer. The XRD patterns were measured using a primary monochromatic high intensity Co K-α (λ = 1.79 Å) radiation and a Lynxeye detector with an acquisition time of 40 min with an incident angle (2θ) between 15 and 75 • . The raw data was peak-matched using EVA software and PDF-1997 database. SEM images were taken on a Jeol 7100F field emission gun SEM (FEG-SEM). The accelerating voltage used for all SEM images was 5 kV, using a working distance of 10 mm. Energy dispersive x-ray spectroscopy (EDX) measurements were performed with the same Jeol 7100F FEG-SEM at 20 kV, and software processing performed using Oxford Instruments AZTEC.
Resistance measurements of textile composite current collectors subjected to bending were performed with a digital multimeter (Keithley DMM6500, Ohio, United States of America). Sheet resistance measurements of the textile composite current collectors were recorded at room temperature with a four-point probe (Jandel HM21, Leighton Buzzard, United Kingdom).

Electrochemical characterisation
Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were carried out using a potentiostat model PGSTAT203N (Metrohm Autolab, Utrecht, The Netherlands) at room temperature.
CV measurements were performed at scan-rates 5, 10, 50 and 100 mV s −1 , at an operating voltage between 0 to 0.8 V. For all CV experiments, three scans were recorded, and only the third scan was selected. EIS was carried out at 0 V in the frequency range from 0.01 to 100 kHz with the amplitude of 10 mV.
The areal capacitance C A (F), maximum areal energy density E A (Wh) and maximum areal power density P A (W) were calculated according to equations (1)-(3), respectively, where A is the printed active layer geometrical area (cm 2 ), −Z ′ ′ stands for the imaginary part of impedance (Ω) at 0.01 Hz, ω is the angular frequency at 0.01 Hz (rad s −1 ), Z is the electrochemical impedance (Ω) at 0.01 Hz, ∆V the operating voltage of the cell (V), and R i is the equivalent series resistance (Ω).

Effect of printed current collector material on the electrochemical performance
To demonstrate the influence of the current collectors' electrical properties on the electrochemical performance, TCEs comprised of graphene and Ag-based composite inks were selected as current collector materials. As represented in figure 1(a), a typical TCE is comprised of a polyurethane-coated fabric, followed by a flexible layer of conductive material, which in the context of this study was graphene (Gr | AL) or Ag (Ag | AL), and an active layer. The introduction of multiple layers of functional inks on fabric resulted in its warpage, hence permanently deforming the textile electrodes (supplementary information, figure  S2). This effect is attributed to thermal stresses occurring at the vicinity of the upper printed layer, resulting in shrinkage and hardening at the corners during cooling [35]. To address the issue, a pre-layer of flexible hydrophilic polyurethane coating was required to overcome the surface roughness and pilosity of the fabric substrate, thus allowing the construction of composite textile supercapacitors (supplementary information, figure S3).
To investigate the electrical properties of the printed current collector materials and the number of printed layers required to achieve viable electrical conductivities, the sheet resistance was measured as a function of the number of printed layers, and is shown in figure 1(b). The graphene-based current collectors (Gr) demonstrate sheet resistances in the ohmic range, while the Ag-based current collector (Ag) displays sub-ohmic sheet resistances. A current collector comprised of three printed layers of graphene and Ag show sheet resistances of 38.82 and 0.01 Ω □ −1 , respectively.
After printing two layers of Ag ink (0.02 Ω □ −1 ), the decay in sheet resistance was only 19%. Thus, to maximise the specific energy and power densities and minimise the final cost of the TCEs, it was established that printing two layers of Ag would be sufficient.
In order to mimic a real-life scenario where the TCSs are subjected to flexural stresses, the TCSs were bent 1000 times, in intervals of 100 cycles, and their electrochemical properties assessed. To investigate the presence of redox reactions in the TCEs, CV was performed at a low scan rate (5 mV s −1 ) to minimise the capacitive current arising from the electric double layer formation, while allowing Faradaic reactions to take place (supplementary information, figure S4). In figure 1(c) are shown the CVs of the TCSs employing different printed current collector materials. The skewed shape of (Gr | AL) demonstrates low storing properties and high internal resistances. On the other hand, (Ag | AL) displays a quasi-rectangular signal, typically encountered in electric double layer capacitors (EDLC) supercapacitors. However, at both cathodic and anodic potentials, two redox peaks were identified, around 0.31 V and 0.08 V, confirming the presence of a Faradaic-type reaction. Additional CVs of the TCSs at a range of bending cycles between 0 and 1000 are found in figure S5.
The performance parameters of (Gr | AL) and (Ag | AL), translated by their power and energy densities, were investigated as a function of the number of bending cycles, and are reported in figure 1(d). No significant changes in both areal power and energy densities were found in (Gr | AL) and (Ag | AL) after 1000 bending cycles. Nevertheless, (Ag | AL) demonstrates a sudden decrease in performance after 800 bending cycles. Such result could be expected since after sufficient bending cycles, the hydrogel further permeates into the bulk of the electrode, thus increasing the contact between the hydrogel and the printed Ag current collector. At this point, more Ag is readily available, accelerating the undesirable corrosive reactions, thus changing the initial properties of Ag. Nevertheless, the utilisation of Ag, instead of graphene, is advantageous since it possesses higher electrical conductivities, which facilitates the electrochemical process. As demonstrated, after 1000 bending cycles, (Gr | AL) exhibits an areal power and energy density of 0.14 mW cm −2 and 0.15 µWh cm −2 , while (Ag | AL) possesses higher areal power and energy densities 7.22 mW cm −2 and 2.90 µWh cm −2 , respectively.
To further investigate the cause of performance decay in (Ag | AL), EDX and XRD were performed on the printed Ag current collector 30 d after initial assembly. In figure S6 is shown a surface topographical SEM image of the printed Ag current collector combined with its respective elemental mapping for silver (Ag), sulfur (S) and oxygen (O) signals. On the SEM micrograph are identified two distinct areas, on the left a rather smooth region, whereas on the right crystal structures could be identified. Elemental analysis performed on the crystal structures demonstrate strong signals due to the presence of elemental Ag, S and O (supplementary information, figure S7). These results suggest that the hydrogel may have reacted with the printed Ag current collector. To further investigate the nature of the crystal structures, XRD analysis was performed (supplementary information, figure S8). Peak matching further confirms the presence of silver sulfate (Ag 2 SO 4 ) in the sample analysed (00-027-1403). Additionally, characteristic peaks for Ag were also detected (00-004-0783). XRD analysis confirms the formation of Ag 2 SO 4 crystal structures. It is believed that the sulfate anions (SO 2− 4 ) present in the electrolyte reacted with the Ag forming the structures rich in the before mentioned elements. According to Grishina and Rumyantsev, the peaks observed in the cyclic voltammograms are caused by the electrochemical formation and reduction of Ag 2 SO 4 [36].
The results clearly demonstrate the need to employ highly conductive materials as current collector to fully exploit the electrochemical system. However, the imminent contact between the electrolyte and the electrochemically sensitive current collector triggers detrimental reactions that damage the electrochemical performance in prolonged operation.

Effect of BL
In light of the previous findings, it was necessary to search for a method to prevent the contact between the electrolyte and the electrochemically sensitive current collector materials. Herein, a flexible conductive carbon layer was incorporated between the printed Ag current collector and the active layer to serve as a physical barrier for the electrolytic species but also to electrically conduct charge between the active layer and the current collector. To establish the number of layers of the conductive carbon paste required to cease the contact of the electrolyte and the printed Ag current collector, a study was conducted where TCEs comprised of one (1 BL), two (2 BL) and three (3 BL) BLs were fabricated (figure 2), and further investigated in regards of their current collector shielding effect. To show the structure of the TCEs, cross-sectional imaging was performed on electrodes without the BL and with 3 BLs (supplementary information, figure S9).
As previously demonstrated, the electrical properties of carbon-based inks are a detrimental factor for the electrochemical performance, thus it is essential to investigate the impact of the BLs in the electrical properties of the TCEs. In figure 3(a) the changes in sheet resistance are shown after printing 1, 2 or 3 BLs on top of an electrically insulating interfacial layer, and on top of 2 Ag layers. The changes in sheet resistance of the BLs demonstrate a decrease of 4.33 Ω □ −1 for each printed layer. However, when deposited on top of Ag, sub-ohmic sheet resistances are obtained, achieving stable sheet resistance readings in all prints (28.67 ± 1.27 mΩ □ −1 ).
In an e-textile context, the flexural properties of the TCEs are of utmost importance since the TCSs will be subjected to continuous bending stresses. In this regard, the changes in electrical resistance of TCEs comprised of the BL element was assessed in a range of 1000 bending cycles, and is reported in figure 3(b). Upon bending, the textile composite current collectors demonstrate a quasi-linear increase in electrical resistance, 0.18 (1 BL), 0.34 (2 BL) and 0.23% (3 BL) increase of electrical resistance per 100 bending cycles. Although increases in the TCS's resistance are inevitable, it is important to highlight that the initial electrical resistances of the TCEs are in the sub-ohmic range, 0.130 (1 BL), 0.113 (2 BL) and 0.131 mΩ (3 BL), evidencing marginal increases in resistance after bending. The negligible increase of resistance with bending suggests that the mechanical properties of the textile composite current collectors were preserved.
To evaluate the effectiveness of the BLs in protecting the printed Ag current collectors, TCSs were fabricated, and CV was performed to identify possible redox reactions. In figure 3(c) are represented the voltametric profiles of TCSs comprised of TCEs with 1, 2 and 3 BLs, after 1000 bending cycles. For comparison, the voltametric signal of a TCS with no BLs was included. Additional CVs of the TCSs at a range of bending cycles between 0 and 1000 are found in figure S10. The voltammograms demonstrate a quasirectangular shape, characteristic of a EDLC supercapacitor, however redox peaks could be identified in all TCSs except the (3 BL). The amount of additional charge generated from the Faradaic reaction observed in the cyclic voltammograms indicates the oxidation and reduction of Ag 2 SO 4 , and is intimately related to the exposure of Ag towards the hydrogel. It was found that the additional charge generated decreases linearly with the number of printed BLs (supplementary information, figure S11). These results confirm that 3 BLs are required to achieve the full protection of the printed Ag current collector. Further investigations were carried out to evaluate the shielding effect provided by the BLs, through visual inspection of the back of printed composite supercapacitors on transparent PET. In figure 3(e) are shown images recorded from the back of composite supercapacitors, previously subjected to 1000 bending cycles and a resting period of 30 d. As observed, printed composite supercapacitors without a BL demonstrated strong signs of corrosion, as depicted by the changes in the texture and colour on the back of the printed electrodes. These results further suggest that the bending motion forces the hydrogel to contact the metallic Ag current collector, triggering its corrosion. On the other hand, no changes were visually identified on composite supercapacitors with 1, 2 and 3 BL. Moreover, no delamination or cracking could be observed on the printed Ag current collectors, supporting the previous claim that the mechanical properties of the textile composite electrodes were preserved, as noted by the negligible variation of sheet resistance with bending.
To investigate the influence of bending on the electrochemical performance of the TCSs, both areal power and energy densities were determined and are On the other hand, the areal energy density of the TCSs demonstrated a gradual increase, due to the continuous impregnation of the hydrogel into the bulk of the active layer due to bending.
The introduction of 3 BLs permitted the full utilisation of the Ag's electrical properties, as well as preventing its degradation, as schematised in figure S12. Additionally, the TCSs demonstrated stable power performances upon bending, and also an increase in the areal energy density. The architecture adopted for constructing the TCEs paves the way for the utilisation of alternative materials that possess excellent electrical conductivities, but electrochemically sensitive, as current collector materials.

Effect in patterning the current collector on the electrochemical performance
Although Ag possesses appropriate electrical properties, it is considered an expensive material. To decrease the amount of Ag in the TCEs, the current collector was redesigned to allow a reduction of 10, 30 and 50% of deposited ink compared to the initial architecture as shown in figure 4(a). The design mimics a net-like pattern, allowing an equal collection and distribution of charge during the electrochemical process. To our knowledge, the method herein developed was not reported elsewhere. However, considering the importance of the current collector, several publications have focused on the engineering of the electrode component. For instance, Kiruthika and Kulkarni designed a smart electrochromic supercapacitor through the electropolymerisation of PANI on a templated gold mesh [37]. Park et al developed a large-area flexible supercapacitor by in situ polymerisation of PANI on three dimensional nanopillars vacuum coated with gold and titanium [38]. Wang et al investigated knitting strategies employing polypyrrole-coated cotton in order to maximise the current collection [39]. Sun et al manufactured large area textile supercapacitors, with in situ polymerised PANI on rGO-coated on polyester textile, serving as 'peripheral nerves' , with printed Ag current collector grids functioning as 'trunk nerves' . The developed architecture aimed towards the efficient electron collection and transportation [40]. The methods reported entail elaborate and time-consuming procedures that incur high manufacturing costs. In this study, we leverage the versatility of screen printing to reduce the Ag content in printed current collectors through innovative patterns.
Electrochemical measurements of the TCSs with patterned current collectors were performed to understand the impact in the reduction of Ag amount on the electrochemical performance. To simulate intensive bending conditions, the TCSs were bent in a range of 10 000 cycles. Figure 4(b) shows voltametric profiles run at 5 mV s −1 of the TCSs with 90, 70, and 50% of Ag in the TCSs. The CVs demonstrate a quasi-rectangular shape, found in EDLC supercapacitors, while no redox peaks could be encountered, even after 10 000 bending cycles. These results suggest that the 3 BLs printed on top of the patterned Ag current collectors sustained the stresses without forming cracks or fractures, thus completely preserving the Ag from the harsh bending conditions. However, the skewing of the CVs demonstrates an increase in the TCS's internal resistance. As previously reported, the TCEs are subjected to a rise in electrical resistance upon bending, thus it would be expected an increase in internal resistance after intensive bending.
In figure 4(c) are reported the changes in areal power and energy densities of the TCSs in the range of 10 000 bending cycles. All TCSs demonstrate a gradual decrease in areal power density upon intensive bending, achieving the lowest performance after 10 000 bending cycles: 10.11 (50%), 7.40 (70%), and 10.48 mW cm −2 (90%). On the other hand, the areal energy densities achieve their maximum after 1000 bending cycles, experiencing a decrease thereafter. The decrease in areal energy density is intimately related to the losses in power, thus a drop in stored energy could be expected. These findings strongly suggest that the electrochemical performance of the TCSs can be preserved after reducing the amount of current collector material, whilst under intensive bending conditions.
Although the introduction of patterned current collectors in TCEs demonstrated performances as good as non-patterned current collectors, one might think that the decrease in amount of Ag and the reduced contact due to the presence of empty holes would result in a decline in electrochemical performance. The reason for this lies in the designed patterns that mimic a highly conductive net-like structure, which allow a uniform flow of charge. Upon depositing the BLs, the holes are filled with electrically conductive particles, thus forming a continuous pathway for the collected charge.
The patterned current collectors introduced in the TCEs enabled the utilisation of the Ag's excellent electrical properties, but also a decrease in the final cost of the manufactured TCSs. Ag is the most expensive material present in the TCS, thus, a 50% decrease in amount would result in a significant decrease in manufacturing costs.

Conclusion
Herein we report the effect that current collector materials possess in the electrochemical performance of all-printed TCSs. The introduction of Ag demonstrates an areal power density of 7.22 mW cm −2 , while graphene-based materials report 0.14 mW cm −2 , after 1000 bending cycles. The disparity between power densities owes to the electrical properties of the materials. However, Ag readily reacts with the electrolyte, causing changes in the initial properties, thus resulting in a decay of the TCS's performance in the long-term. In order to protect the printed Ag current collector, a conductive and electrochemically stable carbon-based BL was printed, located between the current collector and the active layer. The introduction of three layers allowed the full protection of the current collector and utilisation of the Ag's excellent electrical properties, reporting an areal power density of 29.88 mW cm −2 after 1000 bending cycles. Nevertheless, as Ag is an expensive material, it should be minimised in the manufacture of TCEs. Accordingly, patterned Ag current collectors with a net-like structure were designed to reduce the amount of Ag across the range 90% and 50%. Although less Ag was employed (50%), the layered structure of the TCEs enabled a full utilisation of the Ag current collector, thus achieving areal power densities of 28.08 and 10.11 mW cm −2 after 1000 and 10 000 bending cycles, respectively. Moreover, by reducing the amount of Ag in the current collector, the final cost of manufacturing would drop. In summary, the strategies here reported pave the way for the utilisation of highly conductive, but electrochemically unstable printed metal-based current collector materials, with a significant reduction in the manufacturing costs.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.