Nanocellulose based carbon ink and its application in electrochromic displays and supercapacitors

Conventional electronics have been highlighted as a very unsustainable technology; hazardous wastes are produced both during their manufacturing but also, due to their limited recyclability, during their end of life cycle (e.g. disposal in landfill). In recent years additive manufacturing processes (i.e. screen printing) have attracted significant interest as a more sustainable approach to electronic manufacturing (printed electronics). Despite the field of printed electronics addressing some of the issues related to the manufacturing of electronics, many components and inks are still considered hazardous to the environment and are difficult to recycle. Here we present the development of a low environmental impact carbon ink based on a non-hazardous solvent and a cellulosic matrix (nanocellulose) and its implementation in electrochromic displays (ECDs) and supercapacitors. As part of the reported work, a different protocol for mixing carbon and cellulose nanofibrils (rotation mixing and high shear force mixing), nanocellulose of different grades and different carbon: nanocellulose ratios were investigated and optimized. The rheology profiles of the different inks showed good shear thinning properties, demonstrating their suitability for screen-printing technology. The printability of the developed inks was excellent and in line with those of reference commercial carbon inks. Despite the lower electrical conductivity (400 S m−1 for the developed carbon ink compared to 1000 S m−1 for the commercial inks), which may be explained by their difference in composition (carbon content, density and carbon derived nature) compared to the commercial carbon, the developed ink functioned adequately as the counter electrode in all screen-printed ECDs and even allowed for improved supercapacitors compared to those utilizing commercial carbon inks. In this sense, the supercapacitors incorporating the developed carbon ink in the current collector layer had an average capacitance = 97.4 mF cm−2 compared to the commercial carbon ink average capacitance = 61.6 mF cm−2. The ink development reported herein provides a step towards more sustainable printed green electronics.


Introduction
Carbonaceous materials have many different applications such as pigments, UV-light stabilizers, reinforcing agent and many more.Carbon black, which manufacturing is a well-established process, is one such material that is a common component in many materials and products such as car tires [1], paints [2], plastics and coatings [3].In recent times, focus has been given to the manufacturing of carbon black [4] with adequate properties for use in added value applications such as printed electronics [5,6].
Printed electronics is the field of application where additive manufacturing such as screen, inkjet, gravure, stamp, spray, and many more printing techniques are used in the manufacturing of electronics and electronic components [7,8].Compared to traditional electronics, the most important advantages of printed electronics are its potential to be lowcost, both in the end product such as flexible displays and in its production equipment, its reduced environmental impact and its potential for large area manufacturing.Carbon inks, when considered within the field of printed electronics, have been developed to be utilized as conductors in addition to electroactive materials within functional printed devices.Carbonous materials, unlike some printed metals, cannot be easily corroded and retain good electrical properties with conductivities ranging from approximately 1-100 S cm −1 [9,10].Many existing carbon inks on the market have been successfully used in the development of, for examples, sensors [11], energy storage devices [12] and electrochromic displays (ECDs) [13].Unfortunately, due to the poor dispersibility of the carbon, harsh solvents and thinning agents are often present within the inks to ensure their printability or good properties.Currently available carbon inks are subsequently not very suitable for printed electronic applications having environmental impact and sustainability as their core requirement.Therefore, more environmentally friendly carbon inks that remain dispersed, conductive and possesses good printability are required.
The incorporation of nanocellulose into inks, as binders and/or rheological agents is a promising route towards more environmentally friendly inks.With the main origin being from wood, nanocellulose is a sustainable, renewable and composable nanosized material and has been reported as a tunable platform for the production of a variety of high value products [14].These nanoparticles, exhibit unique characteristics due to their nanoscale size, fibril morphology and large specific surface area [15].In the preparation of nanocellulose, the wood pulp fibers are delaminated to nanofibrils by chemical and mechanical means, reducing the particle width from tens of microns to often single digit nanometers.These nanofibers possess very high specific surface area and aspect ratio, often in combination with high surface charge, generating highly viscous and transparent suspensions.The conversion of cellulose into smaller sized nanofibers allows for bottom-up material assembly resulting in materials with higher mechanical strength, improved optical transparency, lower surface roughness and extra ordinary gas barrier properties [14,16].Combining these advantages with its ability to be modified and its affinity to be composited with many different materials, the interest in nanocellulose in recent years has escalated immensely [17].The entangled fibrillar network of aqueous nanocellulose suspensions have also been shown to assist in dispersing various particles that otherwise are not readily dispersed in water, by entangling them in the network.This gives it the possibility to be used as a dispersant for many conductive materials [18] and to control material formation, from dense films and filaments to porous foams and aerogels, making it highly interesting for electroactive applications [19].
Nanocelluloses have been shown to disperse different carbon materials (i.e.graphene and carbon nanotubes) very efficiently.The nanocellulose fibrils and the carbon particles interact [20] to create stable and homogenous dispersions even at very low concentrations [21,22].Most of the reported works have been focused on the utilization of carbon nanotubes and graphene; these show amazing properties, but their production capacity is still low, and some health concerns, due to their shape, make their applicability still problematic [20,[23][24][25][26].Alternatively, pure carbon has low chemical reactivity and thus low toxicity, nonhazardous isotropic shape reduces the health concern.Carbon blacks are isotropic, micron-sized carbon materials produced at the industrial scale and can even possess food grade qualities, making this a superior choice for a safer and more cost-efficient carbon ink.
Herein, we report the development and characterization of a carbon black/nanocellulose screenprintable ink to be incorporated into printed electronic components.Different grades of nanocellulose modification, more specifically carboxymethylated cellulose nanofibrils (CNFs), together with different mixing strategies (i.e.rotational mixing and high shear force mixing) have been investigated.The best protocol identified was then selected and three different ratios of carbon black to CNF were produced.Characterization of rheology, printability and morphology studies were carried out to investigate the properties and performance of the developed inks.Finally, the optimized carbon black-CNF ink was incorporated into fully printed electrochromic devices and energy storage devices to highlight its functionalities.The development of environmentally friendly carbon inks that have sufficient functionality to replace commercially available and more hazardous carbon inks have great potential, not only in the two applications presented here, but also for the timely push for green electronics such as, energy harvesting, bioelectronics and sensing applications.

Materials
The carbon black (denoted as C, Ensaco 250P) that was used in the development of the C:CNF ink was a gift of the supplier (Imerys); in this document C will be used to abbreviate the carbon black.Active carbon YP50F (Kuraray) and conductive carbon TiMC-ALC65 (Cabot) were used for the manufacturing of the active layer in the coin cell supercapacitors.
2-hydroxylethyl cellulose (2-HEC), glycerol and propylene glycol, used in the formulation of the carbon ink, were obtained from Sigma Aldrich and used as received.
Paper (285 µm thick invercote Albato) and polyethylene terephthalate (PET) (125 µm from Policrom) were used as substrates in the printing tests.Al foils (20 µm, Lifasa) were used as substrates in the manufacturing of the supercapacitors.
Commercially available, semiconducting poly(3,4-ethylenedioxythiophene) polystyrene sulfonate ink, PEDOT:PSS, (Clevios S V4, Heraeus) a dielectric ink (5018, Dupont), carbon inks (7102, Dupont and C2130925D1, Gwent Group), and a Ag ink (Ag 5000, DuPont) for the electrical connection were used as received.The electrolyte ink used for the ECD application was developed in house by RISE Printed electronics.
1-Methyl-2-pyrrolidone (NMP, Scharlau) and 5130 poly(vinylidene fluoride) (PVDF, Solef) were also used in the formulation of the active layer of the supercapacitors.The active ink used in the supercapacitor application was fabricated by mixing NMP, active carbon YP50F (Kuraray), conductive carbon TiMCALC65 (Cabot) and binder PVDF at dry weight ratio of 80:10:10 by using dispermat with a designed 3D printed head.
The fabricated electrodes were assembled in CR2032 coin cells from MTI by using NC separator (Kodoshi) and 1 M solution of tetraethylammonium tetrafluoroborate (Et 4 NBF 4 Sigma) in acetonitrile (Sigma) as electrolyte.

Nanocellulose preparation
The nanocelluloses were prepared by mechanical fibrillation of pre-treated pulp fibers, resulting in a material referred to as cellulose nanofibrils (CNF).The pulp fibers were either enzymatically pre-treated [27,28] or carboxymethylated [29,30], the latter were performed at two different degree of substitutions (D.S.); 0.1 and 0.3 by varying the monochloroacetic acid and NaOH amounts.The D.S. relates to the amount of carboxymethyl groups per monomer unit.Mechanical disintegration was performed by passing the pre-treated fibers' slurries once through a high-shear homogenizer (Microfluidizer M-110EH, Microfluidics Corp.) at high pressure (1700 bar).The CNFs are, in this report, abbreviated as CNF-Enzy (enzymatic CNF), CNF-DS01 (carboxymethylated CNF with DS = 0.1) and CNF-DS03 (carboxymethylated CNF with DS = 0.3).When only CNF is used in plots and tables in the result section it refers to CNF-DS03 as that was the optimum CNF identified in the study.When CNF is used in the text it means either all of the different CNFs or just CNF-DS03.

Co-suspension preparation
Carbon black (C) was first mixed with deionized water and then mixed with the CNF suspensions.
A CNF concentration of 1.5 wt.% was used for CNF-Enzy and 0.7 wt.% for CNF-DS01 and CNF-DS03; these variations in CNF concentrations were made to ensure to have starting materials suspension with comparable viscosity.When studying the best mixing strategy and type of CNF to stabilize the carbon, 50 ml suspensions at a mass ratio of 4:1 (C:CNF) were prepared.Mild mixing (magnetic stirring, 300 rpm, 16 h) was compared to intense mixing (high-speed rotational mixing, 12 000 rpm, 10 min).The larger batches (300-500 ml) were prepared by microfluidization (400 bar, 1 pass).For the larger batches only CNF-DS03 was used, but higher carbon to CNF ratios were prepared, 10:1 and 20:1.When preparing the 10:1 suspension, the CNF suspension had to be diluted before adding the carbon, from 0.7 to 0.4 wt%, not to make the suspension too thick that would not allow it to be passed through the microfluidizer.When preparing the 20:1 the diluted CNF suspension was used, even if more carbon was added it still passed easily through the microfluidizer.The total solid concentrations (wt% in the wet state) were 2.8, 2.6, and 7.8 for the 4:1, 10:1 and 20:1 co-suspensions, respectively.

Ink preparation
The prepared C:CNF water suspension were solvent exchanged from water to propylene glycol using an automated rotary evaporator prior to incorporation into the ink formulation.The system consisted of a rotavapor (Hei-VAP Expert, Heidolph) with integrated a cooling module, a vacuum pump and a control unit which enabled, thanks to the presence of several sensors, to monitor vacuum, to empty the flask collecting the condensate and to fill (from a stock reservoir)/empty (in a collection canaster) the flask where solvent exchange was taking place.The described system enabled the production of ca.1.5 l of the solvent exchanged C:CNF solution.The solvent was exchanged as water is typically a poor solvent for screen printing processing.
Finally the ink was developed by mixing the C:CNF with glycerol and 2-HEC for printability.The glycerol aids in the prevention of cracking and the 2-HEC aids in the binding of the ink in addition to increasing the viscosity of the ink.The optimal ink mixtures were composed of 80 wt.% C:CNF, 16 wt.%glycerol and 4 wt.%2-HEC.The total solid content concentrations (wt% in the ink wet state) were 6.24, 6.08 and 10.08 for the 4:1, 10:1 and 20:1 ink, respectively.The different components of the inks were finally mixed using a speedmixer at 8000 rpms for 10 min.

Ink and printed structure characterization 2.5.1. C:CNF co-suspension stability
The stability of the C:CNF co-suspensions was investigated by centrifuging the 4:1 suspensions for 1 h at 3600 g.Phase separation was evaluated visually and by gravimetrically measuring the concentration of the top and bottom phase.Aliquots were taken from the centrifuged tubes, at least 1 ml was pipetted from the top 5 ml and 1 ml from the bottom 5 ml.Suspensions (1 ml) were also dried on glass substrates to see their film forming ability.

Rheological measurements
Rheological profiles were recorded with a MCR 102 rheometer from Anton Paar.A cone and plate geometry (D: 50 mm, 1 • ) with a gap distance of 0.102 mm was used for all the measurements (environmental condition: temperature 21 ± 1 • C and RH% of ca-45-50).The samples were pre-sheared (1 s 1 in 60 s) before viscosity profiles were produced.

Scanning electron microscopy
A ZEISS Gemini field emission scanning electron microscope (SEM) was used to investigate the microscopic structure of the test structures printed using the developed inks.The SEM chamber was kept at 10 −6 mbar during measurement.

Electrical conductivity
The resistivity and therefore the conductivity of the printed carbon ink was obtained using the experimentally obtained value of the sheet resistance and printed film thickness.The measurement of the sheet resistance was performed using the four-probe method where four electrical probes arranged in a line and having equal spacing between each are used.
The sheet resistance (R s ) of a sample is then calculated by using the equation [31]: where R is the resistance in Ω, t the sample thickness in m and s the probes distance in m (in the reported case s = 0.001 m).F1 and F2 geometrical factors with F1 depends on the geometry of the sample and F2 dependent from the correlation of the sample thickness and the probe distance.In this work a circular sample (0.015 m in diameter) and films with finite area (1 ⩽ d/s ⩽ 40) and with thicknesses below 0.5 (probe distance, s) were used.All this led to the definition of the following equation for the sheet resistance calculation.
R was calculated by the EW-Ece (V) vs I (A) experiment being R the slope of the obtained linear fit.Finally, the resistivity and conductivity were then derived by the R s using the equations below: where ρ is the resistivity in Ω m, R s is the sheet resistance in Ω and t the thickness in m.
where σ in the conductivity in S m −1 and ρ is the resistivity in Ω m.

Devices/demonstrators manufacturing 2.6.1. ECDs
Printing of the developed and commercial inks was performed by screen printing technology using an ERKA semi-automatic screen-printer and screen frames of 120-34 (120 fibers per inch with the diameters of 34 µm) screen mesh.The fabrication of the printed ECDs was performed as following: PET substrates were preheated at 150 • C for approximately 30 min before the first layer was deposited; this step was needed to dimensionally stabilize the substrate.Two layers of a commercial PEDOT:PSS were deposited in rectangle designs and each layer cured at 120 • C for 4 min through a belt oven.Dielectric layers (two layers) were printed on top of the PEDOT:PSS layers and UV cured through the belt over for 1 min.The design of the dielectric layer was used to define (opening in the film) the pixel geometry.Following two layers of the RISE standard commercial electrolyte were printed in the opening of the dielectric and UV cured.And finally, two layers of the developed C:CNF ink with optimal composition were printed to form the counter electrode in the electrochromic device.
Characterization of the ECDs was performed by evaluating their switching behavior.Switching of the ECDs was performed using a potentiostat (Octastat, Ivium) driven by the dedicated software to apply a pulse wave form (3 V to −3 V for 5 s).Cyclic voltammetries were also recorded (3 V to −1 V at a scan rate of 50 mV s −1 ) to further understand the electrochemical behavior of the displays.
Color contrast measurements were performed using a Mercury spectrophotometer from Datacolor.L * , a * b * values were obtained and color contrast calculated using the following equation [32]:

Supercapacitor
The optimized C:NFC (20:1 ratio) ink was tested as current collector into CR2032 coin cell supercapacitors.
Al substrates (Lifasa, 20 µm thick, 12 cm wide), used as the conductor, were bar coated (50 µm bar) either with the C:CNF inks or either with the reference commercial carbon ink (C2130925D1).Coated films were then thermally cured at 120 • C for 2 h.Then, the coating of the active ink was made once more by bar coating using a 150 µm bar.The so prepared samples were placed in the oven at 120 • C for a further 2 h to dry.The electrodes were then cut in a circular shape with a 15 mm diameter by using an EL-cut from EL-cell.The electrodes were inserted into a glove box utilizing an argon atmosphere where the assembly of the supercapacitor was performed.The different components of the supercapacitor cell were: a stainless steel cap with silicone gasket, two electrodes (diameter 15 mm), separator (diameter 17 mm), electrolyte (40 µl of 1 M Et 4 NBF 4 in acetonitrile), two stainless steel spacers and springs to keep the stack under pressure and a stainless-steel cap.
The electrochemical testing was performed with a multichannel potentiostat VMP3 (Bio-Logic) run with the EC-Lab software.Cyclic voltammetry was performed (potential range between 0 and 2.5 V) to investigate side reactions, evaluate the potential window of stability and the device behavior.Galvanostatic measurements (constant current between two potentials, 0 and 2.5 V) were used to calculate the specific capacity, energy, power and equivalent series resistance, ESR (from the ohmic drop) of the developed supercapacitors.
The cell capacitance (C cell ) of a symmetrical capacitor can be calculated according to [33]: where C e = C + = C − are the capacitance of the two electrodes.
The cell capacitance of a supercapacitor can be calculated from the charge-discharge experiment: This can be then used, when combined with C cell equation to obtain the electrode capacity, and subsequently obtain information on the capacity of an investigated material: where I is the discharge current, m the total mass of active materials in two electrodes and dV/dt the slope of the discharge curve between 80% and 40% of the cut-off voltage.
From the total specific cell capacitance of twoelectrode system (C cell ), the maximum energy (E max. ) and maximum power (P max. ) of a supercapacitor cell can be calculated according to: where, V is the cell voltage minus the ohmic drop, and R s is the total ESR of the supercapacitor.

Results and discussion
Development of environmentally friendly and sustainable inks has been investigated for many years now.Sustainable research has become even more relevant in recent years due to the waste from electronics becoming a global problem, and with sustainability becoming a crucial aspect in new technologies.However, formulating and manufacturing inks with more sustainable and environmentally friendly components is not a trivial task.Active component's sedimentation, surface energy mismatching and wetting issues are common problems that require to be overcome in ink optimization.The first step in the ink formulation was the development of a stable C:CNF co-suspension with high solid (C) content.As a first step towards this goal the effect of different CNF grades (CNF-Enzy, CNFDS01 and CNFDS03) and of different approaches of mixing the C and CNF (rotating mixing and microfluidizer) were explored.In this evaluation, a C:CNF dry weight ratio of 4:1 was used.The results of this investigation are summarized in figure 1 and table 1. Clear phase separation was observed (following centrifugation, 1 h 3600 g) when C was stabilized with enzymatic CNF (abbreviated Enzy in figure 1) independently from the mixing approach.Phase separation was also observed when applying mild rotational mixing for the carboxymethylation CNF with lower degree of substitution (DS01, figure 1); on the contrary when microfluidizer mixing was used no visual separation was observed for DS01.No clear separation was recorded, independently from the method used, when C was mixed with CNF DS03 (DS03, figure 1).Casting tests (1 ml of the co-suspensions on glass slides) showed that C:DSO3 co-suspensions formed the film with lower defects/cracks (figure SI1 (available online at stacks.iop.org/FPE/6/045011/mmedia)).
To complement the visual observations, gravimetric analysis was also used to investigate the separation of the samples after centrifugation; the results of this investigation are summarized in table 1.The C:DS01 sample prepared via microfluidizer mixing that visually looked homogenous, after centrifugation had separated significantly.By pipetting 1 ml from the top and from the bottom of the centrifuge tubes and measuring the dry content of the two phases a 52× higher concentration was obtained for the bottom phase.The samples of C:DS03 showed no significant differences between the top and bottom independently from the mixing approach used.The cosuspensions showing the best characteristics was the C:DS03 prepared using microfluidizer; this result was ascribed to the fact that CNF DS03 presents a more homogenous and monodispersed nanofibril network that is expected to allow improved dispersion ability of the carbon particles.Thus, CNF-DS03 and microfluidizer mixing was selected for all future studies.With the most suitable CNF grade and C:CNF mixing approach identified, co-suspensions with different C:CNF ratio (4:1, 10:1 and 20:1) were prepared.Different C loading in the co-suspension were tested with the aim of maximizing the solid content and subsequently maximizing the electrical conductivity of the printed structures.It should be noted that the absolute solid content of CNF in the cosuspension decreases with the increase in C concentration; this was needed to compensate for the thickening effect of CNF.This can be seen in the rheology data (figure 2(A)); even though the total solid content in the 20:1 co-suspension is over double when compared to the 4:1 (2.8 compared to 7.6 wt%), the former co-suspension has a lower viscosity that reflect its lower CNF content (0.4 wt% vs 0.7 wt%).
Due to the poor suitability of water as a solvent in screen printable inks, the exchange of water with a more suitable solvent (propylene glycol, PG) for the C:CNF co-suspensions was successfully achieved.In figure 2(B) the rheological behaviors of the original and solvent exchanged 20:1 co-suspensions are presented.As it can be seen, the two co-suspensions presented too low viscosity to be suitable for screenprinting.Furthermore, the behavior of the PG based co-suspension (initial drop in the viscosity) seems to indicate the presence of aggregates not visible in the water based one.To improve the rheological and  printability properties of the co-suspension, glycerol (printability agent, plasticizer) and 2-HEC (thickening agent) were added leading to a good screen printable C:CNF ink.The rheology of the final ink is also reported in figure 2(B).The developed ink showed a viscosity suitable for screen printing while still retaining its shear thinning behavior.
The printability of the C:CNF inks prepared using C:CNF with different ratios (4:1, 10:1, and 20:1) were tested by printing: (a) fine lines (for evaluation of printing resolution) and (b) large areas (evaluation of printing defects, pin holes).Fine lines experiments on PET substrates were performed using a screen mesh (120 fibers per cm with 34 µm diameter fibers) containing lines of 100 µm in width and with 100 µm spacing arranged in a spiral layout (to evaluate edge resolution and effect of printing direction).Figure 3 presents the images recorded for prints obtained with all the three developed C:CNF inks and, as a reference, a commonly used commercial carbon ink (Dupont, 7102).Figure 3(A) shows that the lines obtained using the 4:1 C:CNF ratio ink were narrower than those printed using the 10:1 and 20:1 inks (figures 3(B) and (C)); this observation is most likely due to the lower solid content of the former ink. Figure 3(C) shows the image of lines printed with the C:CNF 20:1 ink; these showed the best resolution fidelity when compared with other inks including the commercial ink which has spread slightly (figure 3(D)).The printability results in figure 3 indicate the suitability of the developed ink for screen printing application.
Application in supercapacitors require the fabrication of large area structures with little to no defects (pin holes).To verify the suitability of the developed C:CNF inks to produce low-defect, large area films, screen printing of 45 × 45 mm squares on paper substrates was performed using the developed C:CNF inks and the commercial 7102 ink for comparison.Even if reasonably good film uniformity was obtained when a single ink layer was deposited, independently from the C:CNF ratio (figure SI2), the overprinting of a second layer resulted in a film with only minimal defects.This is not surprising since two layers of carbon are usually deposited for applications such as electrochromic displays and energy storage applications [34,35].A comparison of films manufactured for the developed 20:1 C:CNF ink (the best performing among the developed formulations) and the commercial 7102 inks (figure SI3) shows that no significant differences, in terms of defect density, could be appreciated.
Further investigation of the printed inks was performed by SEM (figure 4) to observe the morphology of the printed structures.While the structures manufactured with the developed C:CNF inks (figures 4(A)-(C)) are very similar, those printed using the commercially available carbon ink (7102; figure 4(D)), appear more 'melded' together.It is believed that this fused morphology is due to more effective, yet hazardous solvent and thinners used in the ink.
Porous structure of the printed carbon layers was confirmed by cross section analysis as it can be appreciated in figure SI4.It should be noted that this cross section is of a C:CNF 4:1 sample.
Table 2 presents the conductivity values recorded for the reference commercial inks (C2130925D1 from Gwent and 7102 from Dupont) and the C:CNF 20:1 developed carbon ink (resistance values of all the developed inks can be found in the supporting information, table SI1).As expected, the conductivity of the developed carbon ink is lower than those recorded for the commercial inks.Several factors might be responsible for this result including, for example, lower solid content of the ink, nature and particle size of the carbon used in the ink formulation and percolation path formation in the film.It is important to  note that the function of carbon in many printed electronic applications does not require high conductivity.In the case of the applications (protective layer of metallic current collector in supercapacitor and counter electrode in ECDs) demonstrated herein, the carbon layer only requires minimal conductivity to carry the charge (out-of-plane) through the devices.Therefore, it is expected that the developed C:CNF 20:1 ink should function adequately while allowing more environmentally friendly and safer printed electronic devices.

Electrochromic applications
Within the structure of an all-printed electrochromic device, the carbon layer serves as both a conductive, counter electrode and as a protective layer, from direct contact with the electrolyte, of a metal current collector; position of the carbon layer within the ECDs stack is clearly highlighted in figure 5(A).In this work the use of the current collector was omitted for simplicity.
Only the C:CNF 20:1 ink (the best performing ink among those developed) was evaluated in the all-printed electrochromic displays; the performances of the obtained ECDs were compared with those of ECDs manufactured using commercial carbon ink (DuPont 7102).The ECDs were fabricated by screen printing with the design shown in figure 5(A) with the C:CNF 20:1 and commercial carbon ink screen printed on top of the electrolyte.Two overprints of the carbon layer were deposited to ensure the deposition of a uniform layer with adequate functionality (as described above).In figure 5(B) the images of the manufactured ECDs, in their OFF and ON states, are presented: a clear color difference can be observed, the ON state having a far darker blue color (reduced state of the PEDOT:PSS film).An average color contrast for the devices using the 20:1 C:CNF ink was calculated to be ∆E = 30.08± 1.7 compared to the commercial carbon average color contrast of ∆E = 28.92 ± 1.8, highlighting the developed C:CNF ink performance as a counter electrode for electrochromic devices.The  results of three all-printed electrochromic displays for each type of carbon can be seen in the supporting information (tables SI2 and SI3).
The switching behavior and cyclic voltammetry of the all-printed electrochromic devices incorporating the C:CNF 20:1 developed carbon ink (red) and a 7102 commercial carbon (black) can be seen in figures 5(C) and (D).The behavior of the C:CNF 20:1 ink is similar to those recorded for its commercial counterpart.More significant differences could be seen in the case of the cyclic voltammetry (figure 5(D)).This difference may be due to its lower conductivity resulting in a higher energy consumption in addition to a slower switching speed.Also from the cyclic voltammetry, it appears that the C:CNF 20:1 ink may also experience some electrochemical degradation after −1 V.This is an irrelevant result since the operation voltage for these devices is +3 V to −1 V and therefore this potential degradation can be ignored for this application.
Another important characterization for electrochromic devices is the ability to retain their color after electrical stimuli has been removed (optical memory).Figure SI5 presents the retention time of the device incorporating C:CNF 20:1 carbon vs the 7102 commercial carbon.Unfortunately, the ECDs manufactured with the C:CNF 20:1 carbon, while presenting higher initial color contrast, experienced a faster loss of color than the commercial carbon device.The origin of this is still under investigation.

Supercapacitor applications
Another attractive printed electronics application in which carbon inks are widely used is supercapacitors.Commercial carbon inks have been incorporated in printed supercapacitors to serve as a protective layer for the metal conductors [34].In this application the deposited/printed carbon layer is required to present, in first place, a high uniformity with no defects (pin holes) but also to allow an efficient electrical connection between the active layer and the metal contact.In the work reported herein, the supercapacitors were manufactured accordingly to the protocol detailed in the material and methods section; a schematic of the developed device is presented in figure 6(A).It should be noted that the carbon inks within the supercapacitors were deposited by bar coating showing the versatility of the developed formulation.The supercapacitors were developed with IoT applications in mind.
Supercapacitors integrating electrodes with the 20:1 C:CNF ink or a commercial carbon ink (C2130925D1, Gwent) protective layer were assembled and tested for their capacitance.The difference in the internal resistance of both devices can be seen in Table SI3.Cyclic voltammograms recorded for the devices are shown in figure 6(B).Despite the two devices having the same active layer some significant differences were recorded.The device incorporating the 20:1 C:CNF ink presents a square-like voltammogram, close to ideality, while the device with commercial carbon ink has a more resistive voltammogram.
The charge/discharge curves obtained for the two devices are shown in figure 6(C).These consisted of the following step wise protocol; the measurement begins with applying a constant current (in this case, 0.5 mA cm −2 ) until the cutoff voltage is reached (2.5 V); this is followed by a constant voltage step and finally the discharge at the same current density (−0.5 mA cm −2 ).As it can be seen from table SI4 the supercapacitor fabricated by using the developed C:CNF 20:1 ink outperformed those fabricated using the commercial ink (C2130925D1.Gwent) presenting higher capacity (Average = 97.4mF cm −2 compared to 61.6 mF cm −2 ) and lower internal resistance (Average = 4.38 Ω compared to 11.54 Ω).The differences observed for these devices increased with the current density.This result is unexpected as the conductivity of the 20:1 C:CNF ink is significantly lower than those of the commercial carbon.It is believed that this is because the 20:1 C:CNF ink has a better interfacial connection with the active ink in the supercapacitors, thus generating a lower contact resistance between the two inks, which translates into a lower internal resistance of the system and a higher capacity.

Conclusion
Within this report we have shown the development of a more sustainable and environmentally friendly carbon ink designed for printed electronics.With the combination of a non-hazardous solvent and nanocellulose as the binder material, several carbon inks with different solid contents were developed.The developed carbon inks showed good rheology properties for usage in screen printing technology and comparable printability to commercial carbon inks.The calculated electrical conductivity was much lower than the commercial carbon inks (1000 S m −1 for Gwent carbon compared to 400 S m −1 for the 20:1 C:CNF) which could be explained by the morphology as shown by scanning electron microscopy images in addition to carbon content and conditions.However, the developed C:CNF 20:1 ink was still suitable for all screen printed electrochromic displays and even outperformed the commercial carbon ink in supercapacitor applications.Therefore, this report and the carbon inks developed within are a step forward in the future of sustainable printed electronics.

Table 1 .
The stability C:CNF co-suspension upon centrifugation, showing the concentration at the top, the bottom in the tubes after centrifugation.In the last row the concentration ratio of the bottom and top.

Figure 2 .
Figure 2. (A) Rheology data of the three C:CNF samples in H2O with inset giving solid content values and (B) rheology data of the C:CNF 20:1 samples in H20, solvent changed to PG and when incorporated into a screen printable ink (in PG).

Figure 5 .
Figure 5. Screen printed electrochromic display application incorporating the 20:1 C:CNF developed ink.(A) Schematic showing each layer in the printed device, (B) real world printed electrochromic displays showing the reduced (dark blue) and oxidized (light blue) states.(C) Switching behavior of the developed 20:1 C:CNF electrochromic device compared to the commercial 7102 ink when a voltage of 3 to −1 volts is applied in a square wave and (D) the cyclic voltammetry graph of the developed 20:1 C:CNF electrochromic device compared to the commercial 7102 ink.

Figure 6 .
Figure 6.Bar-coated supercapacitor application incorporating the C:CNF 20:1 ink.(A) CR2032 Coin cell supercapacitor schematic (B) cyclic voltammetry and (C) galvanostatic charge-discharge of the supercapacitors when a commercial carbon ink (black) and the developed C:CNF 20:1 ink (red) are used as the current collectors.Figure (D) presents a zoom in of the discharge cycle.
Figure 6.Bar-coated supercapacitor application incorporating the C:CNF 20:1 ink.(A) CR2032 Coin cell supercapacitor schematic (B) cyclic voltammetry and (C) galvanostatic charge-discharge of the supercapacitors when a commercial carbon ink (black) and the developed C:CNF 20:1 ink (red) are used as the current collectors.Figure (D) presents a zoom in of the discharge cycle.

Table 2 .
4 point probe conductivity measurements comparing commercial Gwent carbon and 7102 Dupont inks and the C:CNF 20:1 developed ink.