Printable and flexible graphene pH sensors utilising thin film melanin for physiological applications

The application of highly sensitive pH sensors manufactured in volume at low cost has great commercial interest due to an extensive array of potential applications. Such areas include industrial processing, biotechnology and medical diagnostics particularly in the development of point of care (POC) devices. A novel printable electrochemical pH sensor based on graphene and pigment melanin (PGM), was designed and produced by using a screen printing process that enables up scaling for potential commercial application. We demonstrate a highly sensitive pH sensor (62 mV pH−1  ±  7) over a pH range from 5 to 8, with high stability and superior performance when compared with a number of existing devices and making it suitable for physiological applications.


Introduction
The measurement of the pH of a solution is of universal importance and is used extensively across a range of disciplines (e.g. chemistry, biology, environmental science), both in the laboratory as well as in the field [1][2][3][4][5]. As such, multiple techniques have been developed to measure the pH [6]. The most common applications for pH determination are based on potentiometry, usually using a glass electrode [7,8]. Film electrodes and ion selective membranes are also used to measure pH potentiometrically as are ion selective field effect transistors [9,10]. pH may also be measured using a variety of ductometric [11] and optical [12] methods including measuring colour changes in pH sensitive indicator dyes. This wide variety of ways to measure the pH is indicative of its universal importance i.e. its presence as a variable in many different kinds of systems.
However, there are drawbacks to the various approaches. Many of these methods are subject to sensors with a limited lifetime or their sensing time does not comply with the speed at which the pH of a system is changing, or the active material of a pH sensor is easily contaminated or the loss/degradation of the sensing substances occurs [13]. The ubiquitous glass electrode, as an example, is fragile and requires time to build up a hydrate layer before usage. Furthermore, the ability to optimise it is challenging since remodelling options are limited [13].
As a result, there is a continuing active interest in finding new methods for measuring pH. Recent studies show many material options (organic and inorganic) that can be used as ion sensitive layers (see table 1 for examples). Ideally, a pH sensor needs to have the following properties if it is to be used widely: 2 Z Tehrani et al The purpose of this paper is to demonstrate a commercially viable, robust and sensitive pH sensor that rivals that of 3D graphene with HfO 2 & IrO 2 /RGO (table 1) but utilising readily available materials.

Screen printing
One of the ways to achieve low cost manufacturing to produce in a commercial setting is via a screenprinting process. The added benefit of this process is that it opens new avenues for other kinds of sensors.
Screen-printing is an attractive technique for the production of sensors, because the technology is well established and allows for rapid production. Screen printing is often used for the fabrication of electrodes due to the low cost and simplicity of the technique [28,29].
Furthermore, the base electrode material of interest for this application is based on carbon, since it has been studied extensively and is often used in electrochemistry due to its electrical conductive properties, low density and low thermal expansion [30]. Specifically, we target graphene as the main electrode due to its reported use as an electronic material [31], sensor in bioelectronics [32,33] and in applications based on its high carrier mobility and high carrier concentration. Also it has, biocompatibility, atomic thickness, electrochemical stability and mechanical reliability [34].
However, as will be shown below, graphene on its own is not sensitive enough in this specific application. Instead, this study explores the possibility of using a melanin derivative as the pH sensitive medium in combination with a printed graphene (PG) working electrode. The graphene electrode has been chosen due to it good electrical conductivity and the ability to control its printed topography on which melanin is spin coated to form a very thin layer.
What makes melanin particularly attractive as an active electrochemical sensing material is its reported sensitivity as a function of pH (~ −50 mV pH −1 ) [27,61]. The electrochemistry is due to the hydroxyl groups of quinone moieties and the interconversion between the various oxidative states [46,47] (figure 2). Furthermore, melanin films have been used previously as an active layer in a pH sensing EGFET producing a higher sensitivity to pH than indium tin oxide [27]. Additionally, the material is cheap to synthesise and is very robust. However, it has one flaw, it is partially soluble at moderate to high pH.
As a consequence, we turn to a melanin derivative referred to as DMSO melanin (figure 3). DMSO melanin or D-melanin, is melanin that is synthesised in DMSO instead of water, the standard solvent [64,65]. This material has several advantages over standard melanin. These include ease of spin coating of both hydrophobic and hydrophilic surfaces and stability in aqueous solutions of differing pH, including alkaline solutions [66]. The former property means that DMSO melanin can be spin coated on substrates which require minimal preparation and the latter property makes it very attractive for pH sensing.   [35]. The various differing oxidative states for the monomer are also given, since as melanin is formed under oxidative conditions, these other monomers also become incorporated into the polymer. Melanin is an unusual quinone system in that its quinones can undergo tautomerization to the quinone imine [63] (d) the final reaction is the (de)protonation of the quinone imine.

Materials for screen printing
Polyethylene terephthalate (PET) with a thickness of 175µm (HIFI Melinex 339) was used as the substrate. This film is heat stabilized to provide high temperature resistance and hence dimensional stability during processing [68]. This was used as is. Screen printed electrodes were prepared by first printing conductive silver (SunTronic ref: AST6025) or silver/silver chloride (Gwent ref: C2130809D5). This highly conductive silver polymer ink was selected for the printable refence electrode. It has been developed for a wide range of electronic applications [69] and has the following characteristics necessary for this application: compatibility with carbon inks, high stability and long-term storage, excellent adhesion to PET substrates after sintering. Carbon and graphene ink mixture (Gwent ref: C2171023D1) were used as a working electrode and finally a layer of insulating ink (Gwent ref: D214011D5) to protect the electrode.

Synthesis of melanin
Melanin was synthesized from 3,4-dihydroxyphenyl-DL-alanine (DL-Dopa; Sigma-Aldrich, ⩾98%) with dimethyl sulfoxide (DMSO; PA, Vetec, 99.9%) following a literature procedure [67]. 1.50 g of DL-Dopa and 0.93 g of benzoyl peroxide (Vetec, 75.0%-80.0%) were dissolved in 200 ml of DMSO and kept under magnetic stirring for 58 d at room temperature (≈27 °C.). Afterwards, the reaction solution was concentrated to 1/4 of the initial volume by increasing the temperature to 140 °C. Once it cooled down to room temperature, 150 ml of acetonitrile (Synth, 99.5%) was added to the concentrated solution to allow the melanin to be separated from synthetic byproducts. After two days, the solution was centrifuged at 2500 rpm for 15 min and the precipitate dried in an oven at 90 °C. The black/dark brown powder obtained was used to make a 30 mg ml −1 solution for film fabrication.

pH-sensor fabrication 2.3.1. Fabrication of electrodes by screen-printing
The components of the sensor were printed using a R29 series screen printer from Reprint (UK) under ambient air conditions. Basically, during the screenprinting process, the ink is applied on the substrate through a screen imaged according to the sensor layer requirements. The substrate was cleaned using isopropyl alcohol (IPA) before printing the bottom and top electrodes of the sensor in a stack configuration. The printing of electrode comprises four steps that include silver/silver chloride printed as a reference electrode, carbon printed as a current electrode and graphene as a working electrode, thin film melanin covered the graphene layer and finally some parts of the electrode were covered with an insulator ink. The design of the PGM sensor is shown in figure 4.
There are many process settings that can influence the printed layer, deposition thickness, uniformity and topography, where final parameters settings are summarised below.
The different parameters are the following: 50 mm · s −1 forward speed, 10 mm · s −1 reverse speed, 5.0 front squeegee pressure, 2.0 mm print gap and no snap-off distance (0.3 mm · s −1 snap speed) figure 1S in supplementary information (SI) (stacks.iop.org/ TDM/7/024008/mmedia) for details of printing. Through a comprehensive experimental programme, this was optimised to achieve print to print consistency, thus assuring a path to full scale manufacturing. When these were established, they were found to deposit layers having consistent thickness and appropriate topography. Screens made of polyester having appropriate thread count and thickness for each ink layer were used and the associated oven temperature for the drying regime is summarised in table 2.
For the final step, 30 mg of DMSO melanin was dissolved in 1 ml of anhydrous DMSO (Sigma-Aldrich, ⩾99.9%) which was stirred for 1 hour at 50 ºC and then filtered with a 0.45 µm Hydrophobic PTFE filter (Cole-Parmer). This solution was then spin coated using two steps: first step, at 1000 rpm for 60s; and the second step at 4000 rpm for 30s to form the final PGM.
Thus, a very thin film of melanin was deposited on the screen-printed graphene electrode (PGM) by spin coating and the sensor was ready for testing.

Characterisation techniques
The layers that formed the electrodes was characterized for their thickness and topography using different  [67]. In addition to the standard moieties indicated in figure 1, many moieties are believed to form methylated sulfonate groups at the reactive sites of the initial starting monomer with various degrees of methylation.
techniques to check the quality of the printed layers. These techniques include atomic force microscopy (AFM), white light interferometry (WLI) and scanning electron microscopy (SEM). These were applied at the different stages of fabrication.
AFM was carried out using a JPK NanoWizards II. (Dimension-3100 Multimode, Bruker, Billerica, MA, USA), using a non-contact AFM tip with resonant frequency, spring constant and tip radius of 320 kHz, 40 N m −1 and 8 nm respectively operated on AC mode.
WLI was performed using a Veeco Wyke NT9300 optical profiling system.
Scanning electron microscopy (SEM; Ultra-High-Resolution FE-SEM S-4800, Hitachi) was carried out at 2.5 kV acceleration voltage and a 9.8 mA emission current. The magnification was ×20 k and working distance was 10.8 mm.
X-ray photoelectron spectroscopy (XPS) measurements were taken with a Kratos Axis Supra XPS system with a monochromatic Al Kα x-ray source, with an emission current of 15 mA.
The Fourier-transform infrared (FTIR) measurements were obtained between 4000 and 400 cm −1 , at room temperature, this technique has been employed to analyse D-Melanin in the sensor (SI figures 3S and 4S).
Raman mapping measurements were taken with an Renishaw system utilising a 532 nm excitation laser with approximately 10mW of power on the sample, before and after melanin deposition (SI figure 5S).
Electrochemical analysis was performed using the advanced potentiostat (PGSTAT-302 from Autolab, Metrohm Autolab, Runcorn, UK) with the scanning voltage in the range of 0.0 V-0.8 V for evaluating the electrochemical performance of PG and PGM electrodes. The standard 3-electrode system was used for the electrochemical evaluation, where printed graphene/melanin was used as the working electrode, Ag/AgCl as the reference electrode and printed carbon as the counter electrode. To test pH measurement, sensors were connected to the Ana-Pot from Zimmer and Peacock ltd set at a current range of 100 nA and measured for 600s in different buffer solutions (SI figure 9S).
Electrical measurements were performed using a 4-probe 'EverBeing probe station' with a 2636B Keithley Unit.

Characterization of graphene-melanin electrode: surface morphology
The layers that formed the electrodes fabricated using screen printing was characterized for their thickness and topography. WLI was used to measure the thickness and roughness of different parts of the sensors (figure 5).
As a benchmark, a comparison of thickness and roughness between the in house printed graphene/carbon (PG) electrode with a commercial screen-printed electrode (Metrohm Drop Sense DRP-110GPH) is shown in at table 3. Five measurements were taken at three rows and three columns in three different sheets of printing resulting in a total of 90 measurements collected (for each layer: silver and graphene) the results are shown in at table 3. Where the standard deviation is noted to be very small, thus confirming dimensional consistency (SI figure 16S).
The roughness of the commercial sensor electrode (2.1 µm) is rougher than the PG (1.2 µm see table 3). However, the PG sensor has a more uniform roughness and the smoother surface is advantageous for the complete coverage of the electrode by a very thin melanin layer (figures 12S-14S) The SEM images show the uncoated printed graphene (PG) and the melanin coated graphene (PGM). The small particles (carbon black) and the flake like materials are present in both of the images taken before and after melanin coating took place. This is expected as the melanin molecules are extremely small and therefore not detectable by our SEM. These images show the surface structural characteristics of the working electrode and how they appear to be unchanged  visually at the SEM level following melanin coating. This shows that the amount of melanin that is required for the electrode to become functionalised and therefore sensitive to pH is extremely small ( figure 6). The effect of surface morphology of thin film of melanin was also explored using AFM, figure 7.
The surface roughness of PG substrates before and after melanin deposition (figure 7) shows clear changes in surface topography. The Root Mean Square roughness of 102.1 nm before deposition (figure 7(a)) increases to 292.5nm after deposition ( figure 7(b)).

Characterization of graphene-melanin electrode: surface chemistry
Raman mapping measurements were taken with a Renishaw inVia system using a 532nm laser, before and after melanin deposition. The ratio between the G-peak, at 1580 cm −1 , and the D-peak, at 1350 cm −1 , were mapped out using a custom MATLAB script (SI figure 4S).
As seen in (figure 8), the intensity of the D-peak is significantly higher, resulting in the D-peak/G-peak intensity ratio increase from 0.0744 to 0.0914. The D-peaks are most likely from the additional carbon compounds that are found in the carbon ink and from the application of melanin. The Raman system detects these additional carbon compounds, especially after melanin application, resulting in an increased signal in the D-band [70].
The intensity ratio between the 2D-peak and G-peak in monolayer graphene is approximately 2 [71]. However, this ratio decreases to approximately 0.398, in the printed graphene electrode, denoting that there are multiple layers of graphene present on the electrode [72].
We obtained high-resolution XPS measurements for N1s spectra as a reference for the chemical composition of the blank surface (figure 9, red curve). We then did a comparison test by obtaining data for d-melanin alone and then a melanin-coated electrode. In figure 9, it can be seen that melanin has only one broad nitrogen-peak ranging from 397 to 403 eV related to amine groups in its aromatic structure (figure 1), in agreement with previous studies [67] while the graphene ink also has only one nitrogen peak at 408.2 eV from nitrate compounds. When the graphene electrode is coated with a melanin layer however, both nitrogen peaks can be observed. This indicates the presence of the melanin on the graphene. The small shift towards lower binding energies found for the peak at 408 eV, when melanin is present, is probably related to structural and/or electronic interaction between melanin and graphene [73]. (SI for atomic concentration of carbon and both nitrogen components in table 1S).
We performed Square Wave Voltammetry (figure 10) to characterise the electrochemistry of the electrodes. In this type of curve, the potential of the working electrode can be found with the maximum/ peak of the curve. As can be seen, there is a clear difference between the electrochemical behaviour of just the bare electrode and afterwards when it is coated with D-melanin. There are shifts in current and voltage peaks, clearly showing that the D-melanin was deposited with good electrical contact, this test was carried out on multiple samples (table 4).

Four-point resistance measurements
IV curves using four probes revealed that there was a change in the resistance across the surface of the blank screen-printed electrode following melanin  deposition. Five blank PG and five PGM were tested. Each electrode was tested at five different positions on the electrode to account for any heterogeneity in film deposition. Five measurements were taken at each position resulting in a total of 250 data points collected (125 PG, 125 PGM) the results are shown in ( figure 11). The data clearly indicates that the melanin is present by the way it modifies the resistance measured.

Potentiometric measurements
Having shown the characterisation of the electrodes, we now go on to demonstrate its sensing capabilities. The electrodes' potential (PG and PGM) was measured as a function of time in reference buffers pH4, pH7 and pH10, to see stability of the device over time. The difference between the voltage recorded with each reference buffer is presented in ( figure 12).
As can be seen, the PG electrode does not show any pH dependence at low pH. At high pH there is a dependence, but the electrode voltage decays over time. In contrast, the PGM (D-melanin coated) electrode shows a clear and significant change as a function of pH. In addition, and importantly, the potential is stable over time, showing a robust and instantaneous response to the pH.

Measured potential relationship to pH
The potential of the PGM electrodes were recorded continously in solution as the pH was changed ( figure  13(a)) to demonstrate that the sensor can detect changes in the pH in real time. As can be seen, there are clear changes in the potential at every pH change. The sensitivity of the devices 35 ± 12 mV pH −1 over the pH range from 2 to 12 (SI figure 9S).
We did a further real time monitoring test by checking the hysterysis of the electrode behaviour. We did a test of the pH response going from pH 4 to 10 and then from 10 to 4 ( figure 13(b)). As can be see, the potential measured at the given pH and shows little hysterysis.
The above test were done to demonstrate the sensitivity and robustness of the devices across a wide range of pH. However, we are particualrly interested in   applying these devices for physiological measurements since there is a high demand for measurement of pH in medical settings. As such, we narrowed the range of pH to be studied to between pH 5 and 8. The open circuit potential (OCP) is measured using the potentiostat. No voltage is applied. The voltage potential that is measured is the difference between the working electrode and the reference electrode, this difference changes based upon the pH of the solution that the sensor is immersed in. The counter electrode balances the charge on the working electrode.
We obtained a sensitivity curve across this range by selecting the pH with a reference buffer and then measuring the potential ( figure 14(a)). The sensitivity value we obtained is 62 ± 7 mV pH −1 (table 5), an excellent result rivaling those of 3D graphene with HfO 2 & IrO 2 /RGO [19,20] (table 1).
The major advantage of our sensor, in contrast to the above mentioned sensors is that it was made from commonly avaiable materials coupled to a commerically viable low energy process for fabrication. We now move onto testing of the stability and repeatability of the PGM electrode at physiological pH values. Five PGM electrode were tested using a solution at pH 5 for 10 min followed by pH 5, pH 6, pH 7 and pH 8 solutions. As can be seen, a consistent change in the  measured potential was seen across all the replicates ( figure 14(b)), indicating that there is consistency across the fabrication process.

Mechanism of action
We now turn to the potential mechanism of operation for the DMSO melanin pH sensor. The first thing to note is that the pH sensitivity that has been recorded is in line with previous pH/E data published on standard melanin [27,61]. This initial observation is surprising at first, since one would expect a difference between standard melanin and DMSO melanin which contains additional sulfonated moieties (figure 3). However, the pH/E behaviour is most likely due to the reduction potentials of the one electron redox reactions for the quinone/semiquinone and semiquinone/ hydroquinone pairs [47]. Both these reactions together yield the comproportionation reaction ( figure 2(a)). This contrasts with our previous suppositions that the pK a s of the various constituents are responsible for the effect [73]. Instead, both the the pK a s and the positions of the redox potentials is what drives the redox chemistry and ultimately making our device a sensitive pH sensor.
With the above, it is apparent that the DMSO melanin electrochemically acts as a standard melanin, i.e. the standard comproportionation reaction is active. Considering that not all moieties in DMSO melanin will be sulfonated, there will then still exist a melanin fraction to undergo the electrochemistry. In contrast, the sulfonated moieties may in turn only be spectators electrochemically. Overall, it appears that DMSO Figure 13. pH sensing performance of PGM (a) at extreme pH ranges 2-12, (b) the PGM electrode tested for its ability to recover following exposure to a high pH solution. Figure 14. Sensing results and sensor characterization, (a) the slope was calculated using weighted least squares regression to an uncertainty of two times the standard error with each solution indicated that the sensor remains highly sensitive to pH changes following exposure to pH levels for physiological environment (b) stability and repeatability of the pH sensor. Data from the electrode was reanalysed using the standard error ×2 in order to produce the error bars. melanin has the same sensitivity as standard melanin due to similar redox chemistry, but with the additional advantages of material processibility and stability. The downside would be the presence of fewer standard melanin moieties. However, since the potential is what matters and not the total current, this is not a great problem and further device engineering should enhance the accuracy.

Preliminary tests with clinical samples.
Since we are particulary interested in applying these sensors in physiological conditions, we have performed preliminary tests of the sensors on blood plasma samples obtained from healthy volunteers since plasma is commonly used as a clinical sample type. To test the sensors, plasma was pipetted onto the surface of the sensor and the OCP was measured over a 10 minute period. Following exposure to plasma, the sensors were rinsed with distilled deionised water and measured with pH 7.2, pH 7.3 and pH 7.4 buffers (five PGM sensor were tested) to confirm their correct functioning ( figure 15). The PGM sensor continued to exhibit sensitivity to the different pH buffers after being exposed to blood plasma, showing that the sensor is robust enough to be considered as a candidate in the development of applications where exposure to a clinical plasma sample is desired. This is an example curve for one of the five sensors that were tested ( figure 14).

Conclusions
We have produced printed graphene electrodes with high reproducibility and good electrical conductivity using a novel screen printing process. The fabricated printed electronic sensor exhibited improved performance and properties when compared to a commercial sensor (DRP-110GPH) specifically with regards to reproducibility and homogeneity. Furthermore, in using the active material DMSO melanin, a derivative of the pigment melanin (PGM), and which lends itself to the low energy fabrication, we are able to create a sensitive pH sensor. The sensor itself demonstrates high levels of both accuracy and precision within the physiologically relevant pH levels between pH 5 through to pH 8 (62 ± 7 mV pH −1 ).
The above research provides the foundation for the later development and optimisation of the pH sensitive properties of the electrode to further increase its accuracy and sensitivity. This suggests that the sensor will be well suited for applications that involve biological application and may also have potential applications within a clinical setting. The sensor is sensitive enough to detect clinically relevant changes in blood plasma pH which the preliminary test started.
In conclusion, a novel and highly sensitive pH sensor is demonstrated based on printed graphene coated with a thin film of melanin derivative, DMSO melanin. In short, we have produced a pH sensor that has the 'best of both worlds': low cost and reliable (i.e. commercially viable) while being very sensitive.