Inkjet-printed Ti₃C₂T_x MXene electrodes for multimodal cutaneous biosensing

To inkjet-print MXene electrodes, our first goal was to optimize the aqueous suspension of MXene to fulfill the printing requirements. The suitability of inks for printing is assessed by using the inverse of Ohnesorge number ($Z = 1/Oh = \sqrt {\gamma \rho r} /\eta$), where $\gamma$ is the surface tension, $\rho$ is density, $r$ is the nozzle radius, and $\eta$ is the viscosity [40]. The aqueous MXene solution has a low viscosity of 1.5 cP and high surface tension of 64 mN m⁻¹, which leads to a Z value of 26.5. The printability of MXene as is was therefore found to be inadequate. Concentrations of MXene outside of 2–4 mg ml⁻¹ range did not eject from the printer nozzle. Because the viscosity of MXene within this concentration range did not change, the printing concentration of the MXene ink was fixed at 2 mg ml⁻¹. Note that the average lateral dimension of the MXenes flakes is determined to be 3.2 µm (figure S1 (available online at stacks.iop.org/JPMATER/3/044004/mmedia)), which were then sonicated before use to reduce the flake size to 500 nm. To render the aqueous MXene solution printable and reduce the Z value, we chose to screen the effect of a variety of surfactants added in the formulation. Figure 1(a)–(c) shows photographs of MXene suspensions comprising selected surfactants (top panel), sodium dodecyl sulfate (anionic), cetyltrimethylammonium bromide (cationic), and saponin (nonionic) respectively, as well as microscope images of the suspensions (bottom panel). We observed that negatively charged surfactants caused a branched-like structure, while positively charged surfactants induced visible precipitation. These effects are attributed to the strong interactions (attractive or repulsive) between the charged surfactant molecules with the negatively charged groups of MXene. In contrast, nonionic surfactants were miscible with the MXene solution and produced homogenous films. We thus screened a variety of nonionic surfactants to find the best performing additive, namely, Triton X-100, Tween 20, saponin, Pluronic F-127, and F-68. Each MXene/surfactant mixture was optimized for different surfactant concentrations and substrate treatment to obtain the best performing formulation. While the majority of these surfactants improved the rheological properties of MXene inks, they also reduced the conductivity of resulting films. Figure 1(d) displays the surface tension of each ink and the conductivity of the corresponding drop-casted film. We set a surface tension threshold of 50 mN m⁻¹, above which the ink was not processable (i.e. high Z values). Among these inks that exhibited the required surface tension, the suspension of MXene with saponin produced films with the highest conductivity (831.87 ± 76.06 S cm⁻¹). Although there was a reduction in the conductivity compared to pristine, saponin-free MXene film (drop-casted, conductivity = 1407.74 ± 224.42 S cm⁻¹), this formulation offered a significant reduction in surface tension (40.19 ± 1.54 mN m⁻¹), which is an acceptable trade-off. The MXene ink formulation with saponin was found to have a Z = 15.6. The optimized ink was ejected with ease, as shown in movie S1. The MXene/saponin formulation can be printed on a wide range of substrates with different mechanical properties and surface characteristics such as paper, glass, and as well as a PEDOT:PSS film (figure 1(e)–(g)). The adherence of the printed MXene films on these substrates were tested using a sticky tape test. MXene had strong adhesion to glass (5B classification) and weaker interactions with paper (0B classification) (figure S2).

Zoom In Zoom Out Reset image size

As we will use PEDOT as the flexible carrier for the printed MXene film in biosensing applications, we first evaluated the conductivity of the films printed on PEDOT coated glass substrate. We found that there is a linear trend of film thickness and the sheet resistance determined using a four-point-probe configuration (figure 2(a)). The decrease in sheet resistance of the system with MXene film thickness suggests that each layer of printed MXene is similar to its successive layer in its electrical performance. The system of three layers printed MXene on PEDOT:PSS was found to have a conductivity of 162.2 ± 24.2 S cm⁻¹. This value is lower than that shown in figure 1(d). This may be due to the method we use to estimate conductivity which also takes the conductivity of the PEDOT layer underneath into account. Printed films may also have a decrease in conductivity due to small gaps between successively printed layers. These layers of printed MXene film could be identified in the cross-sectional scanning electron microscope (SEM) image of the film printed with five layers directly on silicon wafer or on PEDOT:PSS coating (figures S3 and S4). Figure 2(b) shows the electrochemical impedance spectrum of such a MXene film printed on the flexible PEDOT substrate. The capacitance of the system was calculated to be 3.76 ± 0.3 mF leading to an areal capacitance of 29.9 ± 2.5 mF cm⁻² (figure S5).

Zoom In Zoom Out Reset image size

Next, we evaluated the shelf-life stability of MXene films in air and in vacuum over 50 d. The films were deposited on paper and allowed to dry at 60 °C. Once the films dried, two PEDOT films were connected on either end to form a two-electrode probe. The films were then placed under vacuum to remove any excess water on the surface. Figure 2(c) shows that the printed films had only a small increase in their resistance over 50 d when left at room temperature in air (R/R₀ = 1.19 ± 0.09). MXene films have thus a remarkable shelf-life, which can be further enhanced with vacuumed packaging. We also compared the stability of printed films to the film that did not contain saponin. This saponin-free film had air stability similar to the films with saponin, suggesting that the presence of saponin does not have a particular effect on the electrical stability of the film in air. The following experiments aimed to evaluate the electrical performance of printed MXene films under deformation. We recorded the electrical resistance of the MXene film printed on the flexible PEDOT substrate while the film was subject to compressive load/unload cycles, causing a maximum buckling amplitude of 2 cm (figure 2(d)). A schematic of the mechanical test setup in figure 2(d) is shown in figure S6. This mechanical load was applied up to 1000 cycles at a frequency of 0.4 Hz. The films showed a negligible increase in their resistance while bending (figure 2(e) and 1(f)), validating their suitability for the intended on-skin use where the electrodes may be exposed to repeated contractions.

Since the printed MXene will directly interface the skin and its surface will be used for biofunctionalization, we carried out SEM and X-ray photoelectron spectroscopy (XPS) to examine surface physicochemical properties. The SEM images shown in figures 3(a) and 3(b) compare the surfaces of a drop-casted, saponin-free MXene film with that of a printed MXene film. For the saponin-free MXene film, the individual MXene flakes are distinct and easily distinguished, and they layer on top of each other, creating a dense film. For the printed MXene film, we see a similar morphology, but we also observe additional circular features (marked with red arrows in figure 3(b)) that surround the edges of each flake. Similar features could be identified in atomic force microscopy (AFM) images, which also reveal a higher root-mean-squared (RMS) roughness for the printed MXene film compared to the saponin-free one (37.46 nm vs. 28.74 nm) (figure S7). We sought after the origins of these features in the printed films through XPS experiments. In the high-resolution XPS spectra of the C 1s region recorded for the printed MXene film, we detected a distinct peak at 285.8 eV (figure 3(c)). This peak was absent in the saponin-free, drop-casted film and attributed to the C–N bond present in the saponin structure. Moreover, the relative intensity of the C–O component was higher for the printed film compared to the saponin-free film, accompanied by a decrease in the ratio of C–Ti to C–C bond intensity. These differences arise from the contribution of C–O and C–C bonds in saponin molecules on the surface. In the O 1s XPS spectra of all MXene films (figure S8), we also note the presence of –OH groups on the surface, independent of the presence of saponin in the film.

Zoom In Zoom Out Reset image size

Considering that XPS investigates the composition of the surface of the film up to ca. 10 nm in-depth and the features observed in SEM and AFM images, we conclude that printed MXene film has saponin aggregates on its surface. These aggregates seem to bind preferentially to the edges of MXene flakes, which are composed of C–C bonds that attract the hydrophobic heads of saponin and are otherwise hardly observed on the surface of the flakes. To investigate the biocompatibility of this particular chemical composition, we carried out an in vitro cell viability test by growing epithelial cells on printed MXene films. We found that the cells adhere well to MXene film and adapt a similar morphology to cells that are adhering to glass. Moreover, the majority of cells remain viable after 48 h in contact with the film (figure 3(d)).

After verifying the composition of the printed MXene surface and its compatibility with biological tissue, we next sought to understand whether saponin is only accumulated on the surface of printed MXene films and investigate the molecular composition of the bulk. Figure 3(e) shows the atomic distribution measured from wide XPS spectra obtained with etching treatment throughout the film. As we examined the film composition from the surface on towards the substrate, we found that the surface coverage ratio of Ti increases rapidly as we etch the film for 1 min, then remains constant after 2 min while there is a continuous reduction in N content, and no N atoms can be observed after 3 min. The analysis suggests that saponin is indeed accumulated predominantly on the surface, and the bulk of the film is formed of MXene only. Ti 2p spectra of the film after iterative etching evidenced that the TiC peak goes down and the TiO₂ peak goes up the farther we etch into the film (figure 3(f)), indicating the oxidation of MXene. We reason that the oxidized content in the bulk arises because of trapped water molecules inside the film, presumably between the flakes. Nevertheless, this printable MXene formulation is highly conductive (831.87 ± 76.06 S cm⁻¹) while the shelf-life stability, mechanical stability, surface functional groups, and biocompatibility highlight the potential of the material for cutaneous biosensing.

The conventional electrodes for recording ECG signals are the commercially available, gel integrated Ag/AgCl electrodes. The gel serves to lower the impedance of the Ag/AgCl at the skin contact, hence improving the signal-to-noise ratio (SNR) of the recordings. Despite their good performance, typical problems with gel electrodes are associated with their durability (as gel evaporates over time) and the fact that the adhesive causes skin rashes and discomfort for some patients. We hypothesized that the printed MXene film could measure ECG without the need for a gel thanks to its conformability, high conductivity and sufficient capacitance in aqueous media. However, printed MXene films are too thin to form freestanding films on their own. A popular approach for attaching electronic materials directly on skin is via a water dissolvable tattoo paper [9, 14]. MXene could be printed on tattoo papers and then transferred on the skin using a water-soluble sacrificial layer (figure S9(a)). The tattoo-transferred, skin attached MXene platform exhibits excellent adherence and conformability on the skin. However, the integration of the tattoo with the acquisition system is often an undermined challenge, requiring adhesives or metallic interconnects that must be placed between the tattooed film and the skin, which severely deteriorates the integrity of the electrode as we observed for this platform as well (figure S9(b)).

To address the issue of connection of the MXene interfacing the skin with the acquisition system, we used PEDOT:PSS as a conductive flexible substrate. Mecerreyes et al have recently introduced a biocompatible cross-linker to stabilize PEDOT:PSS films in water [41]. This cross-linker, divinyl sulfone (DVS), also allows to detach PEDOT:PSS from the substrate and acts as a plasticizer promoting the chain mobility in the film and thus rendering the film flexible [42]. Hence, the freestanding, DVS modified PEDOT:PSS is an excellent substrate for our skin conformable electronics platform. We printed MXene (on a square-shaped area of 10 × 10 mm) on the back-side of conductive and flexible PEDOT substrates (25 × 25 mm), and insulated the remaining geometry by printing a dielectric layer (recall scheme 1). The PEDOT film acted as both a substrate and a conductive line to the acquisition system connected with an Au contact with ease, as illustrated in figure 4(a). Figure 4(b) is a photograph of the MXene film printed on the stand-alone, flexible PEDOT which adheres and conforms well around the finger and can be easily applied to any other parts of the body to measure electrophysiological signals. In our measurement set-up, MXene is the working electrode, and the reference and counter electrodes are commercial Ag/AgCl electrodes (figure 4(c)). These measurements used Ag/AgCl counter and reference electrodes to make sure that we only compare the working electrode which is the electrode of interest, and remove any additional differences that might occur due to a change in the reference or counter electrode material. Despite the absence of a gel between the electrode and the skin, the printed MXene electrode showed lower impedance on the skin over a broad frequency range (from 0.1 Hz to 10 000 Hz) compared to Ag/AgCl electrode (figure 4(d)), which we attribute to the enhanced electrode conformability. The MXene electrodes were also reusable, evidenced by the similar impedance values measured over 3 d of excessive use (figure 4(e)). The combination of MXene with PEDOT allowed intimate interfacing with the skin and an easy connection to the acquisition system, envisaged to reduce the motion artifacts arising from external connectors and prevent the failure caused by metal adhesives. Figure 4(f) shows a trace of ECG signals that the platform on the wrist of a volunteer recorded. When we placed both commercial and MXene electrodes on the skin to measure ECG simultaneously, both electrodes showed similar ECG signal profiles, as shown in figure 4(g), with comparable SNR values (SNR_MXene = 7.1, SNR_Ag/AgCl = 6.7). As we gently removed the electrode from the skin and placed it back, the measurements recovered rapidly with no loss in the signal fidelity, showing the robustness and user-friendly application of the electrode (movie S2). The air stability, reusability, gel-free application, and easy removal and re-attachment of MXene electrodes printed on flexible PEDOT substrates highlight the advantages of this technology compared to the conventional, single-use, wet Ag/AgCl electrodes.

Zoom In Zoom Out Reset image size

The hydrophilicity associated with the surface species of the MXene is expected to allow these printed electrodes to be easily functionalized with biorecognition units. Biofunctionalized printed MXene films can then be used for the detection of disease-associated biochemical molecules present in the sweat, creating a highly modular sensing system designed for specific applications. Our first application concerns the detection of metal cations from sweat. Abnormalities of the metal cation content of sweat can be an indicator for various dysfunctions related to dehydration, fatigue, and kidney or heart failures [43]; hence their analysis is a routine part of clinical evaluations [44]. Na⁺ is one of these target cations, which are typically detected using ion-selective electrodes relying on ion-selective membranes (ISMs) [45]. To render our MXene electrode sensitive to Na⁺, we drop-casted a Na⁺ ISM on top of the printed pattern (recall scheme 1, middle). The sensitivity of the sensor to Na⁺ was evaluated by recording its open circuit potential (OCP) vs. an Ag/AgCl electrode while changing the concentration of Na⁺ stepwise from low to high molarities in the artificial sweat solution. Figure 5(a) shows the linear behavior of MXene sensors (with a slope of 40 mV per decade) within the physiological range of Na⁺ in sweat [46], and their good device-to-device reproducibility. When exposed to solutions containing potassium (K⁺) at its physiologically relevant concentrations [46], the same MXene electrodes exhibited an irregular and less significant change in OCP with respect to an increase in K⁺ content (figure S10), showcasing the specificity of the ISM.

Zoom In Zoom Out Reset image size

Our second MXene biosensor example involves the functionalization of the printed electrode with an antibody (recall scheme 1, right). Interferon gamma (IFNγ) is a pro-inflammatory cytokine that is directly involved in the immune response of the body to pathogens [47]. Increased levels of IFNγ have been related to several diseases such as tuberculosis [48], HIV [49], and, more recently, to the presence of the highly infectious virus, SARS-CoV-2 [50]. IFNγ is typically measured in serum using colorimetric assays in clinical settings, and there have been only a few reports on point-of-care sensors for IFNγ [51–53]. Recently, the detection of IFNγ as an immune biomarker in sweat has been shown [54], however, no correlations between sweat and serum levels have been made. IFNγ sensing in sweat is nevertheless an exciting alternative for non-invasive and fast diagnosis of many of these conditions. To create the IFNγ biosensor, we functionalized the printed MXene surface with the IFNγ antibody. Thanks to the functional groups on the MXene surface, the protein could physically adsorb therein and remain immobilized upon subsequent washing steps. We preferred physical adsorption over chemical bonding to reduce the impact of covalent bonding on antibody function. As the target protein was captured by the antibody on MXene surface, we recorded a change in the OCP of the electrode. These changes scaled linearly with the concentration of the protein, with a sensitivity of 3.9 mV per decade (figure 5(b)). Notably, without the antibody functionalization, the MXene electrode was not sensitive to the protein (figure S11), evidencing the successful incorporation of the recognition unit. These results demonstrate the multi-modality of printed MXene electrodes in a biosensor platform and suggest that the surface functional groups of MXene can be utilized for further functionalization with different biorecognition units selective to disease biomarkers.

4.1.1. Substrates

4.1.2. MXene preparation

4.1.3. Ink preparation

4.1.4. Flexible PEDOT preparation

XPS analysis was conducted using an AMICUS/ESCA 3400 KRATOS instrument equipped with an achromatic Al Kα x-ray source (1468.6 eV). The source was operated at a voltage of 10 kV and a current of 10 mA. The elemental narrow scan region was acquired with a step of 0.1 eV. The pressure in the analysis chamber was 10^–7 Pa during measurements. The obtained spectra were calibrated to the reference C 1s at 284.8 eV. The spectra were deconvoluted using Gaussian and Lorentzian methods, and background subtraction was carried out using the Tougaard method. For XPS etching, the Kratos AXIS XPS Supra instrument was equipped with a Ga⁺ ion source, which was used to etch the surface sample. Sputter-etching was performed every 1 min, and the wide scan and high-resolution spectra were obtained subsequently.

SEM images were obtained using a Merlin SEM enhanced with a Gemini II column. Surface images were obtained at 5k×, 10k×, 30k×, and 50k×, with an electron high tension of 5 kV, probe current of 120 pA. A working distance (WD) of 5.2 mm was used for saponin-free MXene films and 4.7 mm for MXene with the surfactant. AFM images were obtained with a Veeco Dimension 3100 Scanning Probe System. Bruker RTESPA 300 probes were used to probe the film surface. All images were post-treated using Gwyddion software.

The conductivity measurements were done on MXene printed on a PEDOT:PSS-covered glass, which was prepared by spin-coating PEDOT:PSS/DVS at 1500 rpm for 45 s on glass wafer then drying at 140 °C for 2 h. Sheet resistance was measured with a four-point probe using a Jandel RM3000 + Test Unit with an Rµ 100 probe. The thicknesses of the same MXene samples used for the sheet resistance measurement were determined using a Tencor profilometer. The force applied was 1 mg, and the sampling frequency was 50 Hz. Surface tension measurements were conducted using a Krüss Drop Shape Analyzer DSA100. The surface tension was measured with the pendant drop mode. First, the device was calibrated with water until the surface tension was measured to be 72 ± 2 mN m⁻¹. The measurement was then carried out by suspending a drop from the syringe without dropping it. The droplet was made to be as large as possible. The measurement was repeated three times. The mechanical test was performed by applying compressive load/unload on the film specimen (comprising three layers of MXene printed on a flexible PEDOT substrate, while both were supported by a thin polyimide layer), causing a buckling mode, while the resistance of MXene film was measured. Universal testing machine Instron 5944 (2 kN load cell) was used to apply a maximum displacement of 20 mm (under compression) at the speed of 500 mm min⁻¹. The cyclic loading was performed up to 1000 cycles, sufficient to gauge the mechanical durability of the film. The dimension of the film specimen was 110 mm length (gauge length = 60 mm), 25 mm width, 100 µm thickness. Chronoamperometry on the films was done with a potentiostat (PalmSens 4) at −0.1 mV by connecting a wire to the printed MXene film on one side, and to the PEDOT substrate on the other side. The electrochemical impedance spectrum was recorded using a potentiostat (Metrohm) in PBS with a saturated Ag/AgCl electrode as a reference and Pt wire as a counter electrode. Capacitance was then calculated using the following equation considering the frequency of AC pulses (f) and the imaginary part of impedance ($Z’'$):

$$\begin{equation}C = \frac{-1}{{2\pi fZ’'}}.\end{equation} \tag{ 1 }$$

MXene was partially printed (three layers) on clean coverslips and sterilized with ethanol 70% for 30 min. HeLa cells were cultured in Dulbecco's modified eagle media supplemented with 10% fetal bovine serum, 2 × 10⁻³ M Glutamine, and 1% PenStrep (Invitrogen). Cells were seeded on a coverslip and incubated at 37 °C for 48 h. A mixture of Calcein AM (green)/Propidium Iodide (red) (Sigma Aldrich) was added for 5 min at 37 °C to the cells to evaluate the biocompatibility of the printed ink. HeLa cells were imaged with a fluorescent microscope (Nikon). Green color represents the living cells, and red labels are for the dead cells.

Two able-bodied volunteers (two males, aged between 25 and 40 years) participated in this study with informed consent acquired before conducting the experiments. All the volunteers participated in the ECG recordings and the measurement of the electrode–skin impedance. All protocols and procedures were approved by the direction of research of the KAUST Institutional Biosafety and Bioethics Committee. A potentiostat (PalmSense 4) was used to obtain the electrical impedance spectra of electrodes in contact with the skin of the volunteer. Gel Ag/AgCl electrodes (Covidien) were used as counter and reference electrodes. To connect the MXene electrode to the acquisition system, a gold electrode on polyimide was glued to the back of the electrode (to the flexible PEDOT substrate) using silver paste. The working and counter electrodes were placed 3 cm apart. The impedance spectra were acquired between 0.1 Hz and 10 kHz with a V_DC of open-circuit conditions, and V_AC of 0.01 V. The ECG signal was acquired using a Shimmer3 system. The electrodes were attached to the left arm (printed MXene electrode, three layers), right arm (Ag/AgCl electrode), and left leg (Ag/AgCl electrode). All data were filtered using the MATLAB Signal Analyzer through a high pass filter at 1 Hz with 0.85 steepness and a 60 dB attenuation. SNR was calculated using the equation below:

$$\begin{equation}SNR = 20{\text{ }}\log \left( {\frac{{RM{S_{signal}}}}{{RM{S_{noise}}}}} \right),\end{equation} \tag{ 2 }$$

where RMS_signal is the RMS value of ECG data filtered with a 1 Hz high-pass filter; RMS_noise is the RMS value of the raw data minus RMS_signal.

The ion sensors were fabricated by printing MXene (three layers) on the flexible PEDOT substrate then drop-casting the ion-selective membrane on the MXene pattern. The ISM was prepared by mixing Na⁺ ionophore VI 71739 (Sigma), bis(2-ethylehexyl) sebacate (DOS), sodium tetrakis[3,5-bis(trifluoromethyl)phenyl] borate (Na-TFPB), high-molecular-weight polyvinyl chloride, valinomycin (potassium ionophore), and sodium tetraphenylborate (NaTPB) in tetrahydrofuran. The active area was defined by a PDMS well (4 mm diameter). Before starting the measurement, the MXene electrodes were conditioned in a 10 mM aqueous NaCl solution overnight. The OCP measurements were performed using a potentiostat (Biologic VSP-300, Science Instruments) in a two-electrode configuration with a saturated Ag/AgCl in 3 M KCl as the reference electrode. The electrolytes to be tested were placed on the ion-selective MXene electrodes immediately before measurement, and the OCP was recorded with different electrolytes concentration. The measurements were done in artificial sweat, which had the following composition: 100 × 10⁻³ M NaCl, 5 × 10⁻³ M NH₄Cl, and 10 × 10⁻³ M MgCl₂.

The sensor was prepared by printing MXene (three layers) on a flexible PEDOT substrate and incubating the active area defined by a PDMS well (4 mm diameter) for 30 min with the IFNγ antibody (Abcam) in 10 × 10⁻³ M PBS. The surface was then incubated with 10 g l⁻¹ of bovine serum albumin (BSA) in 10 × 10⁻³ M PBS for 1 h, acting as a blocker molecule. Each concentration of the recombinant IFNγ protein (Abcam) was sequentially added to the artificial sweat from low to high (100 fg ml⁻¹ to 100 ng ml⁻¹). The electrodes were exposed to each solution for 30 min before rinsing with PBS three times. OCP was then measured in 10 × 10⁻³ M PBS using a two-electrode setup as described above for the ion sensor. In the negative control experiments without the IFNγ antibody functionalization, the electrode was only incubated with BSA for 1 h before the measurements.