Introduction

Transparent conducting oxides (TCOs) such as indium tin oxide (ITO) are critical materials for electronic and optoelectronic devices, providing the transparency necessary for displays, solar cells, sensors, and various user interface devices. However, state-of-the-art TCOs have two major drawbacks – they are comparatively expensive due to the need for vacuum deposition and they are traditionally considered to be brittle materials1 with limited suitability for flexible, lightweight systems. The minimization of film stresses via strain-tolerant serpentine structures has shown the promise of transparent metal oxides as active materials for wearable sensors, but these approaches have the drawback of requiring complex patterning and yielding low areal density circuits2. Recent works have also demonstrated complementary approaches of fabricating on ultrathin substrates3 to minimize the bending strain of metal oxide thin films and using multilayered organic-inorganic hybrid structures to relieve bending-induced mechanical stresses4. Studies of the mechanics of TCOs such as ITO have revealed that minimizing the TCO film thickness and reducing the process temperature both lead to enhanced mechanical reliability5.

An emerging class of ultrathin liquid metal-derived6 two-dimensional (2D) oxides7 could overturn the longstanding limitations of transparent conducting oxides, offering a platform for high-performance wearable electronics based on inorganic, wide-bandgap materials. The field of 2D oxides has demonstrated, for example, the formation of ultrathin materials with layered polymorphs (e.g. β-TeO2) just 5–40 Å thick8, the potential for assembling these materials into heterostructures with unique electronic properties9, a tendency to favor highly crystalline phases compared with oxides printed from solution10, and the ability to yield conductors with extremely high transparency (>99%) and enhanced mechanical flexibility11. In one significant recent example, Kong et al. showed controlled recession of a linear liquid metal meniscus to print 2D a-GaOx with ultrahigh conductivity and exceptional mechanical reliability12. However, the 2D oxide field faces several ongoing challenges limiting its impact. For example, there is a demand for advancing large-area uniformity while precisely controlling the liquid metal meniscus – the key feature for modulating the printed film thickness and controlling the oxidation kinetics. Widely variable thicknesses are reported in the literature for a single material such as In2O3 or Ga2O3 due to the variable dynamics of manual methods based on blade coating13,14, dispensing from a syringe15, and squeeze printing between two flat substrates16. Additionally, there is a need to introduce extrinsic dopants to control and enhance electronic properties7 since the precursor metal alloy constituents compete for representation in the surface oxide17. If these challenges can be addressed, 2D oxides could make a significant impact on the field of flexible electronics.

Herein we report the first demonstration of high-speed automated liquid metal printing over large areas with Å-level precision control over Cabrera Mott oxidation kinetics. We apply this method to fabricate flexible, two-dimensional (2D) indium tin oxide (ITO) transparent electrodes from the surface oxides of liquid indium tin alloys at temperatures as low as 140 °C on flexible polymer substrates. Our detailed materials characterization of these films reveals that by controlling crystallization and doping of 2D ITO it is possible to achieve superlative conductivity and high transparency at unprecedented process speeds as well as outstanding mechanical resilience, including high scratch resistance and enhanced bending strain. Finally, using these materials, we report the first demonstration of 2D oxide-based bioelectrical measurements, showing an efficient multimodal sensing approach for combining electrocardiography (ECG) and pulse plethysmography (PPG) utilizing the high transparency of liquid metal-printed ITO.

Results and Discussion

Liquid metal printing of 2D Indium tin oxide

We have developed an automated platform for liquid metal printing (Fig. 1a) that precisely controls the Cabrera-Mott oxidation kinetics of liquid metals. This is the first report of robotic, wafer-scale liquid metal printing of 2D ITO (Supplementary fig. 1), replacing previous manual “touch printing” methods11. Our system can vary the printing speed over a 200-fold range, from 0.1 cm/s to 20 cm/s, allowing us to print uniform films over areas greater than 100 cm². The precursors for liquid metal printing in the present study are hypoeutectic, indium-rich In-Sn alloys, though we note that the automated method can be extended to a multitude of alloys for printing various 2D oxides. Indium and tin form a eutectic mixture at 52–48 wt. % Sn with a melting temperature of approximately 117 °C (Supplementary fig. 2). The concentration of Sn in the alloy determines the melting temperature, with hypoeutectic compositions (less than 48% Sn) lowering Tm (Fig. 1b) significantly below the Tm of pure In. This allows us to print 2D ITO films at extremely low deposition temperatures, making the process compatible with thermally sensitive substrates such as polyethylene terephthalate (PET), polyethylene naphthalate (PEN), polyimide (PI), and glass (Fig. 1c).

Fig. 1: Kinetically-controlled liquid metal printing, CM modeling, AFM, TEM.
figure 1

a Schematic of automated liquid-metal printing of 2D ITO with controlled printing speed and film thickness. b Dynamic scanning calorimetry showing melting temperature vs. Sn concentration in the In-Sn alloy system. The inset shows the molten In-Sn alloy (20 at. % Sn). c Photographs of 2D ITO films printed on glass and flexible polymer substrates. The scale bar is 1 cm. d Large area (30 cm2) film of 2D ITO (left) with corresponding thickness map of 20 cm long 2D ITO strip (200 °C, 7 at. % Sn). e Measured thickness of 2D ITO and 2D In2O3 films printed at 260 °C vs. printing speed with Cabrera-Mott growth kinetics equation shown. f Histogram of measured thicknesses for 1, 2, and 3-layers 2D ITO films shown in the inset. g AFM phase image of a single layer 2D ITO film. h HRTEM image of a single layer ITO film. Insets show crystallites of c-ITO in multiple orientations. i Selected area electron diffraction (SAED) pattern for single layer 2D ITO film of the same HRTEM area. All the films are as deposited, with no post annealing.

Full size image

The automated process gives us the ability to model the kinetics of liquid metal surface oxidation and explore the inherent limits to the thickness uniformity across large areas, assessing the fundamental repeatability of this synthetic method. Figure 1d shows a printed 2D ITO film covering approximately 10 × 10 cm, with a thickness of 5.8 nm and a standard deviation of just 4 Å. The thickness is measured using spectroscopic UV reflectometry (see Materials and Methods and Supplementary fig. 3) and confirmed with atomic force microscopy (AFM) (Supplementary Fig. 4). An essential feature of our method is that it offers the ability to print films of programmed thickness by modulating the printing speed. Figure 1e shows the thickness of 2D oxide printed at 260 °C at speeds ranging from 0.1 cm/s to 20 cm/s for both undoped In2O3 and ITO films (7 at. % Sn). These results show that, at high speeds, the films asymptotically approach a thickness of approximately 2.4 nm in the case of ITO and 2.5 nm in the case of In2O3. Reducing print speed (<1 cm/s) allows the printing of thicker single layers approaching 5–6 nm. This control over the 2D oxide thickness as a function of speed stands in contrast to the recent report by Kong, et al., which showed a constant thickness vs. speed, albeit in a comparatively low speed regime (0.001–1 mm/s). To understand the physics determining the 2D oxide film thickness (tox), we consider the Cabrera Mott oxidation kinetics model shown below (Equation 1), where the speed of the roller (v) is treated as the inverse of time (t) for the oxide growth on the liquid metal meniscus. In this approximation, A and B are fitting constants, though we refer the reader to the literature showing their derivation from the oxygen diffusion length, Mott potential, and temperature18,19:

$$\frac{1}{{t}_{{ox}}}=A-B\log \left(\frac{1}{v}\right)$$
(1)

While the field of 2D oxides has postulated that the growth is governed by Cabrera Mott kinetics, this is, to our knowledge, the first study to model the kinetic dependence of 2D oxide thickness on the rate of the liquid metal meniscus deformation. This is critical because the electronic and optical properties of 2D oxides are highly thickness-dependent10,20,21, for example, due to the quantum confinement-induced bandgap widening in 2 nm thick InOx close to the Bohr radius. We further investigated the oxidation kinetics during printing at various temperatures, and the experimentally measured film thickness aligned well with the oxidation kinetics model described in Equation 1 (Supplementary Fig. 5).

To assemble thicker 2D oxide films and heterostructures, we can stack these layers to print films of 2–100 nm total thickness. Figure 1f shows a histogram of thicknesses for single, double, and triple ITO layers, indicating the ability to freely control the thickness via multilayer assembly. To put the speed of Cabrera Mott oxidation in perspective, we highlight that each of these layers is printed in less than 2 s. We note that the equivalent growth rate of these films is orders of magnitude higher than what is currently achievable with nanoscale growth methods such as Atomic Layer Deposition (ALD)22,23. We also highlight the repeatability of our automated liquid metal printing process, which, over the course of ten prints executed on ten different substrates, varies by less than 6 Å, signifying a highly reproducible and reliable process for the growth of 2D ITO (Supplementary fig. 6).

Materials characterization of liquid metal printed 2D ITO

Materials characterization of the liquid metal printed 2D ITO films reveals their high crystallinity and large-grained nature. AFM image in Fig. 1g shows 2D ITO exhibits grain sizes ranging from 24 to 67 nm, with an average of 40 ± 11 nm (the histogram is shown in Supplementary Fig. 7). These large grains are notable, given that the film is just 6 nm thick. This is consistent with our previous reports of superlattices of 2D In2O3 that were shown by HRTEM to exhibit large platelike grains9. For the 2D ITO in the present study, we have also observed that the grain size is substantially larger than that of sol-gel-derived ITO films12,13,24,25. Based on other similar reports in the 2D oxide field7, we expect that this could be a unique feature of the liquid metal interfacial reaction. One possible explanation for the high crystallinity of our 2D ITO is the total elimination of residual organic species through our solvent-free synthesis method. Moreover, the AFM images demonstrate smooth interfaces, with an average roughness of only 0.2 nm, as shown in Fig. 1g. Such roughness number is comparable to or even smoother than some sputtered ITO films26,27. HRTEM images (Fig. 1h) obtained via printing of these films onto TEM grids also show the high crystallinity of 2D ITO films. The appearance of Moiré Fringes in the 2D ITO HRTEM images indicates the overlap of stacked crystalline grains produced in a single print. This result could suggest that both the top and bottom interfaces of the liquid metal meniscus contribute to their surface oxide layers during the printing process. The ability to achieve such highly crystalline morphologies at low temperatures in such a rapid process is unique compared even with vacuum-deposited films that crystallize on longer time scales, require films above a critical thickness, and generally need post-annealing at higher temperatures28,29,30. The enlarged view in the inset of Fig. 1h displays well-defined lattice fringes corresponding to the (400), (222), and (413) lattice planes of cubic In2O3. The selective area electron diffraction (SAED) data, taken from Fig. 1i show a diffraction pattern typical for the cubic phase of ITO, matching reference spectra31,32. We note that there is a lack of considerable amorphous signal at low angles, indicating the nearly full crystallization of the 2D ITO film printed at 260 °C.

The potential for extrinsic Sn doping of the 2D ITO films was investigated by varying the In-Sn alloy composition and measuring the resulting ITO film via XPS. Figure 2a shows the Sn at. % extracted for films printed at 180 °C and at 240 °C. Sn doping concentrations in the 2D ITO match closely to the Sn concentration in the parent alloy. This corresponds well with previous literature11,33. Figure 2b shows the oxygen O1s peaks and their decomposed constituent peaks, including stoichiometric M-O bonding, oxygen-deficient M-O bonding, and M-OH bonding for 2D ITO printed with varying Sn concentrations. The percentage of oxygen-deficient M-O bonding peaks at ~ 532 eV increases from 24.6% for 5 at% Sn doping to 31.6%. for 20 at. % Sn doping in the parent alloy. This matches previous literature34,35 and could correspond to the higher free carrier concentration of these heavily doped 2D ITO films due to a higher oxygen deficiency concentration. We also observe that with high-temperature deposition, such as at 240 °C, the XPS signal from M-OH peak bonds (~534 eV) weakens, while the stoichiometric M-O peak (~530 eV) signal increases by, compared to the film deposited at 180 °C. (Supplementary Fig. 8).

Fig. 2: XPS, XRD, and electrical characterization of 2D ITO film.
figure 2

a Comparison of Sn concentration in oxide measured through XPS to Sn metal concentration (at. %) in the alloy, printed at 180 °C and 240 °C with a print speed of ~1cm/s. b XPS O1s analysis of ITO films from 5, 10, and 20 at. % Sn doping, printed at 260 °C, with no post annealing and a print speed of ~1cm/s. c XRD spectra for a single layer ITO film printed with the alloys of 1, 7, and 20 at. % Sn at 260 °C (top half) and XRD spectra for single layer ITO films vs. printing temperature (bottom half), printed with the alloys of 7 at. % Sn, with no post-annealing and a print speed of 0.5 cm/s. d Conductivity of a single layer ITO film vs. Sn concentration (at.%) in the alloy, printed at 260 °C with no post-annealing and a print speed of 0.5 cm/s. e Conductivity of a single layer ITO film vs. deposition temperature, printed with 7 at. % Sn concentration in the alloy, no post-annealing, and a print speed of 0.5 cm/s. f Conductivity measurements of unannealed, single-layer ITO films printed at various temperatures as a function of printing speed.

Full size image

Figure 2c shows the XRD spectra of as-deposited 2D ITO films with varying tin doping levels (1 at. %, 7 at. %, and 20 at. % Sn) and varying process temperatures (170 °C, 200 °C, 260 °C, 290 °C). These show peaks corresponding to a mixture of (222) and (400) orientations, indicating the presence of the cubic In2O3 structure30. Notably, at 7 at. % Sn doping, the 2D ITO film exhibits preferential growth in the (100) orientation, with the (400) peak showing higher intensity than the (222) peak. With high doping, such as 20 at. %, the crystalline peaks exhibit attenuation, likely due to the heavy Sn doping impeding In2O3 grain growth. Figure 2c (bottom half) also displays XRD spectra for 2D films printed with 7 at. % Sn in a Sn-In alloy at various temperatures, again revealing the presence of (222) and (400) peaks, with the latter notably more intense. Remarkably, even at a modest processing temperature of 170 °C, a distinct crystalline (400) peak is evident. These findings suggest that the typical range of electrically optimal Sn doping in liquid metal printed films consistently favors the growth of ITO in the (400) orientation, a parallel to reports of sputtered ITO36.

Electrical and electromechanical properties of liquid metal printed 2D ITO

The electrical properties of 2D ITO were extensively characterized as a function of Sn doping (Fig. 2d) and print temperature (Fig. 2e), both of which have a large impact on the conductivity. Increasing Sn doping from 1 at. % to 7 at. % results in a substantial increase in the conductivity of 2D ITO. The electrical conductivity peaks at approximately 1340 S/cm at 7 at. % Sn for a deposition temperature of 260 °C. For heavier Sn doping approaching 20 at. %, the ITO films begin to exhibit slightly lower conductivity. Previous studies at lower temperatures11 reported that Sn concentrations above 10% resulted in the amorphization of 2D ITO. The higher print temperatures and the automated printing process of our present study could contribute to maintaining the crystalline cubic ITO phase even at high Sn concentrations, leading to high conductivity. For a film with a thickness of ~6 nm and without any post-annealing, the conductivity is superior to that of previous reports of printed ITO and comparable to even post-annealed vacuum-deposited ITO28,29,30,37,38,39,40,41. Table 1 shows a comparison of the conductivity and post-annealing conditions for thin ITO deposited by printing (upper half of Table 1) as well as vacuum-based processing (lower half of the table). These literature results also reflect the general trend that ITO deposited by vacuum methods exhibits lower conductivity when the film thickness is scaled to allow for ultratransparency. The parity achieved here between our liquid metal printed 2D ITO and sputtered films speaks to the unique physics of Cabrera Mott oxidation and the process control facilitated by our automated methods. While a single ultrathin layer of ITO can achieve high conductivity, it is also possible to stack multiple layers to result in lower sheet resistance, making the films suitable for various optoelectronic device applications. Supplementary fig. 9 illustrates the relationship between the sheet resistance of ITO films and the number of printed layers, showing results for single, double, and triple layers. For example, stacking three layers under identical parameters—260 °C, 0.5 cm/s, and 7 at. % Sn in the precursor In-Sn alloy—achieves a sheet resistance of 300 Ω/sq. The kinetics of oxidation also influence the 2D ITO film conductivity. Figure 2f compares the conductivity of ITO films printed at different speeds and temperatures. As print speed increases, conductivity decreases since slower printing speeds result in thicker films, which promote larger crystallite sizes and, consequently, improved conductivity. This holds a similar trend for an extensive range of deposition temperatures (180 °C–260 °C). It is essential to highlight that the sheet resistance of ITO films printed on different substrates, such as glass and plastic substrates, exhibits a similar trend to those printed on SiO2 when the print temperature is varied (Supplementary Fig. 10).

Table 1 Summary of reported performance of printed and vacuum-deposited ultrathin indium tin oxide (ITO) as a function of thickness and annealing temperature
Full size table

Mechanical and functional properties of flexible 2D ITO for bioelectrode applications

The utility of TCOs in user interfaces such as touch screens and displays could offer an opportunity for integrating new biosignal measurement functionality into wearable technology if the associated challenges of mechanical flexibility of TCOs can be overcome. We have characterized the mechanical properties of 2D ITO specifically to address the viability of wearable biosignal measurements. Wearable TCOs must pass rigorous mechanical testing, including abrasion, tape, and bending tests, to ensure they can withstand the physical demands of continuous use. These tests simulate conditions such as repeated friction against the skin and bending around curvilinear surfaces, ensuring the electrodes maintain functionality and integrity under mechanical stress. The clinical standard for biopotential and bioimpedance measurements involves using wet electrodes such as Ag/AgCl, though these gels can be uncomfortable, can dry out over time, and may cause irritation42, rendering them unsuitable for long-term ECG signal acquisition. Previous research has demonstrated thin Au films43, PEDOT: PSS44, and Ag NW for flexible bioelectrodes45 due to their flexibility, but these materials have limitations to their mechanical and thermal stability.

In this work, we established 2D ITO as a potential dry electrode because it is highly flexible, transparent, and abrasion-resistant compared to traditional dry electrodes. Traditional sputtered ITO films have been extensively examined and studied for use as biosensors for various purposes46 but have not been shown as non-invasive epidermal bioelectrodes for measuring biopotential signals. Here we show that 2D ITO printed on polymer substrates is more strain-tolerant than sputtered flexible ITO. Figure 3a shows the normalized resistance change after 100X bending cycles at 1% tensile strain for sputtered and liquid metal-deposited ITO films. The resistance change is 5X higher in the case of sputtered ITO compared with the liquid-metal printed film. One potential factor limiting the sputtered ITO film flexibility could be the potential for inducing residual stresses from the high-energy growth method47, since these residual stresses have been documented to reduce sputtered ITO’s mechanical reliability under cyclic bending strain48. Another potential hypothesis is that the mixed amorphous and crystalline phases in 2D ITO could provide a better balance of mechanical flexibility and electrical performance49,50. To test this hypothesis, we measured strain response for films printed at 180 °C and 260 °C. As shown in Supplementary Fig. 11a, the 180 °C ITO film demonstrates greater strain tolerance, with a smaller resistance increase under strain, likely due to its higher amorphous content enhancing flexibility. In dynamic bending tests over 100 cycles (Supplementary Fig. 11b), the 180 °C film also shows superior durability, with lower resistance change than the rapidly degrading 260 °C film. These results indicate that lower-temperature deposition improves both flexibility and mechanical stability. This makes 2D ITO films more suitable for applications where the electrodes need to endure bending and flexing, such as wearable electronics and flexible biomedical sensors. As shown in the SEM images in Fig. 3b, the sputtered ITO films exhibit significant fractures transverse to the bending direction, whereas the 2D ITO films remain intact, demonstrating superior mechanical bendability.

Fig. 3: Mechanical characterization of 2D ITO electrodes.
figure 3

a Normalized resistance change vs. bending cycles for sputtered ITO (45 nm) and liquid metal printed 2D ITO (12 nm) at 1% tensile strain. Scale bars are 50 µm. b SEM images for 2D ITO film and sputtered ITO film after 100 cycles of tensile strain. c Normalized resistance change from scratches with various pencil leads with increasing hardness on PEDOT: PSS (100 nm) and 2D ITO film (23 nm). d Normalized resistance change vs number of tape test applications for adhesion test for 2D ITO (23 nm) and sputtered Au (15 nm). e Skin-electrode impedance versus frequency for 2D ITO film and commercial Ag/AgCl gel electrodes, Inset shows the configuration of the ITO electrode impedance setup with ITO electrode as working electrode (WE) and commercial Ag/AgCl gel electrode as reference electrode (RE) and counter electrode (CE). f Spider chart summarizing the relevant properties of various dry bioelectrodes.

Full size image

The mechanical stability of 2D ITO was also characterized via abrasion and adhesion tests, which effectively simulate real-world conditions by evaluating the response to potential debonding from substrates during skin contact and the capability to repeatedly apply the films to the skin51,52,53. We assessed the abrasion resistance according to the ASTM standard by using pencil leads of increasing hardness to determine the hardness at which the film could be visibly scratched and to measure changes to its resistance (details in the Materials and Methods section). As shown in Fig. 3c, our 2D ITO films demonstrated abrasion resistance that is 4X higher than that of the control PEDOT: PSS film, a common dry bioelectrode material. The control PEDOT:PSS film was scratched by the softest lead (2B) available, whereas one of the hardest leads (2H, 4X stronger than 2B) did not scratch the 2D ITO film. This scratch resistance illustrates a unique advantage of ceramic bioelectrodes relative to softer alternative materials. Figure 3d illustrates the adhesion capabilities of the 2D ITO film compared to a sputtered gold film on polyimide substrates. A peel test (details in the Methods section) was conducted to observe how resistance changed with repeated tape test applications at the same locations. After the tape tests, the gold film was nearly completely delaminated, indicated by a sharp increase in normalized resistance after five applications. In contrast, the ITO film maintained its adhesion and did not de-bond, showing only a small change in normalized resistance.

Characterization of the electrode-skin impedance for 2D ITO was performed to assess its suitability for various biopotential measurements. Impedance testing ensures that dry bioelectrodes make consistent and adequate contact with the skin, which is crucial for accurately measuring bioelectric signals with a high signal-to-noise ratio54. We conducted electrode-skin impedance tests on a 2D ITO film and compared it to clinical standard Ag/AgCl gel control electrodes using a three-electrode setup (details in the Material and Methods section). As shown in Fig. 3e, the 2D ITO electrode exhibits comparable skin-contact impedance to Ag/AgCl per unit area from 1 Hz to 100 kHz. For comparison, the 2D ITO electrodes also show better skin-contact impedance (∼95 kΩ/cm²) compared to PEDOT: PSS values reported in the literature (169-194 kΩ/cm²)55,56, as well as gold electrodes (305 kΩ/cm²)56. This low contact impedance makes the 2D ITO electrodes promising candidates for dry bioelectrodes in biopotential measurements. Figure 3f shows a spider chart of various dry bioelectrodes’ performance in terms of flexibility, transparency, conductivity, adhesion, bio-compatibility, and wear resistance. As highlighted in this chart, 2D ITO excels in its transparency and its mechanical resilience to abrasion while offering sufficient conductivity for accurate biopotential measurements. Another key factor for the practical use of ITO electrodes is the long-term stability of their electrical properties. We measured and compared the sheet resistance of single- and multilayer-printed ITO films immediately after printing and again three months later. As shown in Supplementary fig. 12, the single-layer films exhibit a 2.5x increase in resistance, while the multilayer films show a 2x increase. Despite this rise, the resistance remains well within the acceptable range for biosignal acquisition, such as ECG, making these films suitable for bioelectrode applications.

Multimodal biosignal acquisition

ECG measurements using flexible 2D ITO electrodes were performed by contacting the index fingers on both hands (as described in Materials and Methods). The left inset of Fig. 4a shows a zoomed-in view of the P wave, QRS complex, and T wave of a single period of the electrocardiogram signal. We performed a Pearson correlation analysis comparing the measured heart rates from both the standard gel electrodes and the 2D ITO electrode. As shown in Fig. 4b, the analysis revealed a high positive correlation of 0.98 between measurements from both the 2D ITO and gel control electrodes under variable conditions (resting and light exercise). This strong correlation signifies the reliability and practical potential of the 2D ITO dry electrodes. Moreover, we employed our highly conductive 2D ITO films to perform electromyography (EMG) measurements for tracking two different hand gestures (open and closing a hand). Figure 4c shows the placement of the electrodes on the forearm and the measured EMG signal overlaid with the corresponding time divisions for open and closed hand gestures.

Fig. 4: Multimodal measurement of biosignals with 2D ITO electrodes.
figure 4

a ECG signals are recorded with a 2D ITO electrode; the left inset shows a representative peak showing the PQRST parts of the peak, right inset shows the single lead ECG setup at the fingertips. b Pearson correlation between heart rate measurements with 2D ITO and gel electrodes for two different physiological conditions. c EMG signal for two kinds of hand gestures are captured with 2D ITO electrodes; the inset schematic shows the ventral placements of the electrodes on the forearm. d Transmittance vs. wavelength for 1 layer of ITO on PI printed with 7 at. % Sn at 260 °C (baselined with the transmission of the bare PI substrate). e A schematic of the placement of the transparent ITO on the wrist area underneath the PPG. f Simultaneous measurements of PPG and ECG.

Full size image

To leverage the multifunctionality of these transparent flexible bioelectrodes, we next demonstrated a set of multimodal heart rate measurements using both biopotential and optical methods. This is possible because 2D ITO is highly transmissive to the source wavelengths utilized in PPG (e.g. 530 nm and 940 nm), as shown in Fig. 4d, with an average transmittance above 95%. This allows these electrodes to be co-located with light-based PPG sensors without hindering light transmission from the PPG LED or the collection of reflected light by the PPG photodetectors. The high transmittance of 2D ITO would also allow vertical integration of these bioelectrodes with displays in future wearable systems. Figure 4e shows a schematic of the dual measurement of both ECG and PPG signals. Such integration enables simultaneous ECG and PPG measurements from the same site, combining the detailed electrical activity captured by ECG with the vascular blood volume changes detected by PPG. The advantage of performing these measurements simultaneously for cardiovascular monitoring is that ECG can be utilized to validate and correct motion artifacts commonly seen in PPG signals57,58. As shown in Fig. 4f, heart rate measured using PPG (78-83 BPM) during light activity matches the precise measurements from electrical monitoring with the 2D ITO electrode (80-82 BPM). Additionally, as seen in Fig. 4f, the peak of the PPG coinciding with the T peak of the ECG indicates temporally synchronized cardiac cycle phases. An additional advantage of these synchronous measurements could be the ability to detect anomalies, such as arrhythmias, that are impossible to perceive via PPG measurements alone. This dual measurement capability, facilitated by the transparency of ITO electrodes, provides a more robust and holistic approach to cardiovascular health monitoring while also minimizing the area needed for multimodal measurements in a miniaturized device form factor.

In summary, we present flexible 2D TCOs fabricated at ultralow temperatures using vacuum-free Cabrera Mott oxidation of liquid hypoeutectic In-Sn alloys, demonstrating wafer-scale films with angstrom-level thickness control via an automated, kinetically driven approach. Our synthesis rapidly (at up to 20 cm/s) produces smooth, ultrathin 2D ITO with unprecedented conductivity (> 1300 S/cm) comparable to vacuum-deposited films. Surface morphology and structural characterization confirm the effectiveness of Sn-doping, revealing the high crystallinity of these 2D oxides and the large, plate-like grains formed in the liquid metal reaction environment. We observe that a significant result of the ultrathin nature of 2D ITO and the liquid metal printing process is enhanced bending strain tolerance, superior scratch resistance, and low contact impedance for 2D ITO when used as wearable bioelectrodes. Finally, leveraging the conductivity and transparency of 2D ITO, we enable simultaneous, multimodal measurements via electrocardiography (ECG) and photoplethysmography (PPG). These findings represent a significant improvement in the performance of printed metal oxides and introduce a promising new material for multimodal biometrics.

Experimental methods

Alloy preparation

The In-Sn alloys were prepared by melting In (Luciteria, 99.995%) and Sn (Luciteria, 99.995%) pellets in a graphite crucible at 300 °C for 2 hours in an inert nitrogen atmosphere glovebox (<10 ppm O2, <10 ppm H2O) to minimize surface oxidation. Alloys containing 1, 2.5, 5, 7, 10, 20, and 30 at. % Sn was prepared by mixing the respective amounts of Sn with In.

Automated 2D TCO synthesis and deposition

2D indium tin oxides (ITOs) were deposited by rolling molten In-Sn alloy droplets along a substrate (Si/SiO2, glass, PET, or PEN) on a hotplate using a silicone roller controlled by a 3-axis inline gantry robot (Fisnar 5300N). The deposition process was conducted at speeds from 0.1 to 20 cm/s and temperatures from 140 to 290 °C (printing at 140 °C was specifically enabled by high Sn-concentrations, as detailed in Fig. 1b). Prior to deposition, the target substrates were treated with approximately 10 seconds of atmospheric plasma using a Plasma-Etch 1000 W system supplied with 30 LPM compressed dry air to promote the 2D ITO film adhesion. Two dummy substrates placed before and after the target substrate were used to allow the deposition process to reach equilibrium and produce a uniform and continuous metal oxide film deposition. After deposition, the residual liquid metal on the surface of the oxide film was removed with a squeegee while still on the hotplate and again once the sample had cooled to room temperature.

Thin film fabrication

For the abrasion tests, the control PEDOT: PSS films were prepared by spin-coating a 1 wt. % PEDOT: PSS solution in H2O onto a SiO2 (300 nm)/Si substrate at 6000 RPM for 1 minute, followed by drying on a hotplate at 100 °C. An Anatech LTD Hummer 6.2 sputtering system was utilized to deposit a 15 nm-thick gold film on a soda lime glass substrate for the peeling test.

Materials characterization

A Differential Scanning Calorimeter (DSC) (Discovery DSC 250, TA Instruments) was used to measure the melting temperature of the Sn-In alloys with various Sn compositions with a ramp rate of 10 °C /min under N2 flow. X-ray photoelectron spectroscopy (XPS) was conducted using a Kratos Axis Supra XPS at approximately 10−9 Torr on three layers of 2D ITO films printed on 100 nm SiO2 substrates. Elemental analysis of the 2D ITO films was performed by comparing the Sn 3d, In 3d, and O1s peaks. Optical microscope images were captured with a Keyence VHX-7100 microscope. UV-Vis spectroscopy was conducted using a DeNovix DS-11 FX+ spectrophotometer to measure the printed ITO films’ absorbance spectra (270–800 nm) on glass substrates. AFM was performed using an AIST-NT instrument in tapping mode to measure film thickness and grain morphology. High-resolution transmission electron microscopy (HRTEM) was carried out with a Thermo Scientific Talos F200i instrument. Samples for HRTEM imaging were prepared by liquid metal printing of 2D ITO (7 at. % Sn) directly onto TEM grids (Carbon Square Mesh, Cu, 300 Mesh, UL, EMS) at 260 °C, with excess liquid metal removed using a silicone squeegee. X-ray diffraction (XRD) was performed using a Rigaku UltraX Cu-anode diffractometer (Cu Kα radiation at 40 kV, 300 mA, λ = 0.154 nm) with a scanning rate of 0.5° per minute on single printed layers of 2D ITO on 300 nm SiO2 substrates. Grain size analysis was conducted via AFM phase imaging and confirmed through HRTEM images. Scanning electron microscopy (SEM) was performed using a Thermo Scientific Helios 5 CX tool.

Electrical characterization

Sheet resistance was measured using a four-point probe at room temperature in air.

Thickness characterization

Films for measuring thickness were printed on SiO2/Si substrates (300 nm SiO2). The exact SiO2 thickness was measured to facilitate the modeling of the reflectance spectrum of the 2D ITO films on SiO2 spectroscopic reflectometry (F3-sX, Filmetrics) from 380 nm – 1050 nm to extract the thickness of the films. These thicknesses measured by reflectometry were confirmed via AFM line scans of films patterned by wet etching. Multilayer films > 20 nm thick were also measured via stylus profilometry (KLA Tencor D-500) to confirm the reflectometry measured thickness.

Mechanical characterization of flexible 2D ITO films

Bending resilience measurements were performed on 2D ITO films deposited onto 60 µm thick polyimide substrates at 260 °C and sputtered ITO onto 175 µm PET substrate. The films were measured after the substrate was bent to 1%, 1.25%, and 1.5% of tensile strain until 100 bending cycles were achieved. The film hardness test was conducted using the ASTM D3363 standard, which entails scratching thin films, such as 2D ITO (printed with 7 at. % Sn doped at 260 °C) on silicon and spin-coated PEDOT, using pencil leads of varying hardness. The films were imaged with optical microscopy, and changes in electrical resistance were observed after each iteration of abrasion with the specified pencil leads. Abrasion was applied with a force of 7 N, an angle of around 45°, and a speed of approximately 0.5 cm/s. An adhesion test for the 2D ITO film (printed with 7 at. % Sn doped at 260 °C) on polyimide and sputtered Au on glass was performed using Kapton tape (Uline). The tape was removed at a speed of approximately 1 cm/s and at an angle of ~90° to the tested film.

Transmittance measurements

Transmittance measurements (380–950 nm) for the 2D ITO films on a polyimide substrate were conducted using a Vernier Go Direct SpectroVis Plus spectrophotometer. A polyimide substrate was used for baseline calibration.

2D ITO bioelectrode characterization

ECG measurements were conducted by wrapping printed ITO electrodes that are printed with 7 at% Sn doped at 260 °C on polyimide in a single lead setup (Fig. 4a). As a control, a gel electrode (3M-2238 Electrode) was placed on the forearm. The heart rate was calculated using the equation 2, shown below:

$${\rm{Heart}}\,{\rm{rate}}\,({{\rm{beats}}\,{\rm{per}}\,{\rm{minute}}})=60/{{\rm{RR}}}$$
(2)

Where RR is the time interval in seconds between two consecutive R peaks of the measured ECG signal. Both measurements were performed using a Vernier EKG Sensor (Go Direct). EMG measurements were performed by placing these ITO electrodes on polyimide on the forearm (Fig. 4c) and using the Vernier Go Direct system for data collection. A Polar Verity sensor was used to measure the PPG from the wrist. The transparent ITO electrode on polyimide, which measures the ECG signal, was placed between the PPG LED and the wrist. This setup allows simultaneous measurements of both PPG and ECG.