Research ArticleENGINEERING

Nonthrombogenic, stretchable, active multielectrode array for electroanatomical mapping

See allHide authors and affiliations

Science Advances  19 Oct 2018:
Vol. 4, no. 10, eaau2426
DOI: 10.1126/sciadv.aau2426


High-precision monitoring of electrophysiological signals with high spatial and temporal resolutions is one of the most important subjects for elucidating physiology functions. Recently, ultraflexible multielectrode arrays (MEAs) have been fabricated to establish conformal contacts with the surface of organs and to measure propagation of electrophysiological signals with high spatial-temporal resolution; however, plastic substrates have high Young’s modulus, causing difficulties in creating appropriate stretchability and blood compatibility for applying them on the dynamically moving and surgical bleeding surface of the heart. Here, we have successfully fabricated an active MEA that simultaneously achieves nonthrombogenicity, stretchability, and stability, which allows long-term electrocardiographic (ECG) monitoring of the dynamically moving hearts of rats even with capillary bleeding. Because of the active data readout, the measured ECG signals exhibit a high signal-to-noise ratio of 52 dB. The novel stretchable MEA is carefully designed using state-of-the-art engineering techniques by combining extraordinarily high gain organic electrochemical transistors processed on microgrid substrates and a coating of poly(3-methoxypropyl acrylate), which exhibits significant antithrombotic properties while maintaining excellent ionic conductivity.

Multielectrode arrays (MEAs) have been used to investigate the position of active/inactive cells, propagation of neural signals, and/or networking among multiple neurons (1, 2), as well as in the diagnosis of diseases (37) by measuring biological signals at multiple points. The earliest reported and widely used system included in vitro MEA made on flat glass to measure cellular excitement in cultured myocardium (8) and neuronal cells (9), and the propagation of signals from heart (10) and brain slices (11). More recently, the noninvasive in vivo MEA was developed on flexible plastic foil. This system has the ability to contact soft and moving living tissue (3, 12).

To apply flexible MEAs to complex structures within the body, it is necessary to further increase the flexibility of devices (13). For example, conformal contact on wrinkled brain surfaces can be realized by reducing the thickness of the device to less than 2 μm (13, 14). As a result, it has been shown that electrocardiogram (ECG) measurements can be conducted at the skin near the heart by passive MEAs on 3-μm-thick polyimide substrates (15, 16). Furthermore, accurate ECG measurements on dynamically expanding and contracting hearts require a high degree of device stretchability, which can be provided by grid patterns on an ultrathin polyimide substrate (1719). For example, the ECG measurements on heart tissues have been measured at 5 dB, with a passive array of MEA on a polyimide substrate, which has a thickness of ≤5 μm (20).

By putting active elements in each cell, active arrays of MEAs have been realized with local signal amplification and multiple addressing (1, 2, 2123). These functions are very important to allow scaling to high spatial-temporal resolution measurements far beyond the area accessible in passive addressed systems (24, 25). It has been reported that an active MEA made of Si field-effect transistors (FETs) with a thickness of 10 μm was partially mapped and used in ECG measurements on the hearts of pigs (2). By replacing such an active MEA with organic active elements of organic electrochemical transistors (OECTs), which have 100 times higher transconductance than Si FETs (26, 27) and organic FETs (OFETs) (23), the thickness can be further reduced to 2 μm, with a more smooth contact (21). The ultraflexible nature of the active array of MEA is demonstrated by the smooth contact to muscle cells, cerebral cortex, and measured electromyography (EMG) and electrocorticography (ECoG) (21, 22).

However, a stretchable and blood-compatible, active MEA has not yet been realized owing to two remaining bottlenecks. First, degradation of the device occurs due to blood clots from surgical bleeding. This occurs on flexible substrates of high Young’s modulus, such as polyimide or parylene with high process compatibility. The stimulation of platelets, which is the reason for blood clots, easily occurs in flexible substrates with less elasticity. The low Young’s modulus rubber-based passive MEAs, which showed high stretchability and biocompatibility, have been developed (28); however, it is difficult to develop active MEAs on low Young’s modulus substrate because the complicated circuit requires process-compatible substrates for multiple-layer interconnection. Second, it is difficult to make high-performance active elements that show stretchability. Normally, the stretchable active element is made of an organic material exhibiting a low mobility of 2 cm2/Vs and a high driving voltage of 80 V (29, 30). However, for measuring biological signals, the active element requires high amplification factors and low drive voltages.

Here, we succeeded in fabricating an ultrathin, stretchable grid-patterned active OECT matrix and measured the distribution of ECG signals with a signal-to-noise ratio (SNR) of 52 dB by directly contacting the dynamically beating heart of a rat. The stretchable 4 × 4 active OECT array was fabricated with a total thickness of 2.6 μm and a high transconductance (gm) of 1 mS on average with the parylene honeycomb grid substrate. The device was entirely coated with 100-nm-thick poly(3-methoxypropyl acrylate) (PMC3A) for antithrombotic properties by maintaining excellent ionic conductivity. The electrical properties and the response time of OECT were changed less than 2% even after PMC3A coating. The feasibility of 4 × 4 ultrathin, stretchable, antithrombotic active OECT arrays was demonstrated by mapping an ECG from the heart surface of a rat. The artifact noise caused by dynamical moving did not appear in recorded data due to high conformability of the grid substrate. Because of its antithrombotic property, it is capable of stable measurements over long periods (e.g., over a period of 1 hour) even in an environment where bleeding occurs.

Materials, mechanics, and design strategies

The stretchable MEA consisting of OECTs and grid substrates is fabricated on a 1.2-μm-thick parylene substrate (Fig. 1A). The total thickness is as low as 2.6 μm, including the top encapsulation of a 1.2-μm-thick parylene layer, resulting in high flexibility and conformability with stretchability. The fabrication process and methods are shown in fig. S1. The cross section of a stretchable OECT is shown in Fig. 1B. The active layers of a thin poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) system and the wirings are realized on a honeycomb grid substrate. The honeycomb grid substrate was chosen due to its mechanical stability and structural stretchability, which has been well investigated by experiment and simulation (31, 32). The outermost layer is PMC3A, which has high blood compatibility (33). To maintain antithrombogenicity of PMC3A, the thickness should exceed 80 nm (33, 34) but should be suppressed for mechanical stability (14). The coating of PMC3A was performed by the dipping method with a total thickness of 100 nm. The circuit diagram of a stretchable MEA is shown in fig. S2. The cross-talk evaluation of this circuit was performed by a previous report (22). Figure 1 (C and D) shows a photograph of the 4 × 4 OECT array and a magnified view when it is stretched. This stretchability is essential when the device contacts dynamically moving biological surfaces. A 4 × 4 array of stretchable OECTs is fabricated with 3-mm spacing (fig. S3A). The scanning electron microscopy (SEM) image of OECT channels on the honeycomb grid is shown in fig. S3B.

Fig. 1 Stretchable OECT array on honeycomb grid substrate.

(A) Image of the stretchable electrophysiology OECT array on a honeycomb grid parylene substrate. (B) Cross section of the stretchable OECT array. (C) A 4 × 4 stretchable OECT array when intrinsically stretched. Scale bar, 3 mm. (D) Magnified view of 4 × 4 stretchable OECT array when stretched. Scale bar, 500 μm.

Before biological experiments, the electrical characteristics and mechanical stretchability of the device is systematically investigated. The channel width and length of PEDOT:PSS are both 10 μm. Figure 2A shows the electrical performance of stretchable OECT. The curve of drain current (Ids) and gate voltage (Vgs) and the measurement setup are shown in fig. S4. Figure 2B shows the statistics of the transconductance of the array. The average transconductance is 1.1 mS, which is 100 times higher than that of Si FETs and sufficiently large to measure the ECoG or EMG signals of rats (21, 22). The mechanical durability of the extension strain was performed with one stretchable OECT (fig. S5). The device was free-standing in the phosphate-buffered saline (PBS). Figure 2C shows normalized Ids-Vgs curves of OECT when an extension strain was applied from 0 to 15%. At 15% extension strain, the change in transconductance (gm) and Ids at Vgs = 0 V was 0, 5, 10, and 15%. The cyclic test was then carried out, applying an extension strain of 15% (Fig. 2D). The gm and Ids values of stretchable OECT were observed to decrease by 7%, showing excellent mechanical durability of the honeycomb grids. There was a difference in Ids-Vgs and gm-Vgs curves due to series resistance (35) from different wire design and variation between badges (36).

Fig. 2 Electrical and mechanical characteristics of stretchable OECTs.

(A) Ids-Vgs curve of OECTs in 4 × 4 stretchable OECT array (fig. S3). The channel width and length were 10 and 15 μm, respectively. (B) Distribution of transconductance in a 4 × 4 stretchable OECT array. (C) Drain current (Ids) − gate voltage (Vgs) curve of single OECT (fig. S5) after 0, 5, 10, and 15% stretching. Inset: Microscope image of single OECTs, which were intrinsically 10% stretched. Scale bar, 200 μm. The channel width and length were 10 and 15 μm, respectively. (D) Normalized value of Ids and transconductance (gm) at Vg = 0 V after 15% stretching with the number of cycling.

The blood compatibility of PMC3A was conducted using platelet adhesion method (fig. S6). First, platelet suspensions were made by extraction from human blood. The OECTs or film samples were prepared with or without PMC3A coatings. PMC3A coating was performed by dipping the sample into 1% (w/v) PMC3A solution. The thickness of PMC3A was approximately 100 to 120 nm, as measured by a surface profiler (DektakXT). The water contact angle of PMC3A was confirmed as 44° (fig. S7), which is consistent with previous work (33). The PMC3A is known to exhibit ionic conductivity; therefore, the channel layer of OECTs made of PEDOT:PSS can be controlled appropriately by the gate bias, although the polymeric layer covers the channel layer entirely. We confirmed ion conductivity by measuring site impedance through the PMC3A-coated gold in PBS, as shown in fig. S8. This result is consistent with the previous study in which its ionic conductivity was explained by structural characterizations (33, 34): In water, PMC3A consists of the polymer-rich phase that aggregates each other and the polymer-poor phase (33, 34) that connected continuously and formed ionic conductive paths. The aggregated surface of PMC3A was observed clearly in our specific device structure by atomic force microscopy (AFM) measurement (fig. S9). In addition, fig. S8 shows that OECTs with PMC3A coating have lower impedance than those without. This is because methanol, which is the solvent of PMC3A, improves the impedance of PEDOT:PSS (37).

Blood compatibility evaluation

The Ids-Vgs curve of OECTs was measured, with a width of 50 μm and a length of 50 μm, before and after PMC3A coating (fig. S10A). The gm values of the OECTs before and after PMC3A coating are as high as 3.6 and 3.5 mS, respectively, at a gate voltage of 0 V. Note that the change in gm is only 2%. The response times (τ) of the OECTs before and after PMC3A coating are measured by applying a gate voltage pulse with a duration of 1 ms (fig. S10B). τ is evaluated by fitting experimental data with an exponential function. The τ values of OECTs before and after PMC3A coating were 60 μs, with a gate voltage pulse of 100 mV. This indicates that the dipping process of PMC3A does not change the electrical properties of OECTs. Also, the long-term stability of OECT coated with PMC3A was confirmed by measuring transconductance for 10 hours (fig. S11).

Figure 3A shows the SEM image of the polyethylene terephthalate (PET) film (up) and PMC3A film (down) after 1 hour of platelet adhesion. The average amount of platelet on samples is shown in Fig. 3A. The morphology of the adhered platelet, which is one of the indices expressing the degree of platelet activation, is divided into three (38): the original round shape (type I), the dendritic shape with a few protruding pseudopods (type II, early activated stage), and the spread-dendritic shape with many pseudopods (type III, intermediate activated stage), as shown in the inset of Fig. 3B. The averaging of platelet number was done from SEM images of films (fig. S12). The adhered platelet on the channel of OECTs ensures higher impedance between channels of OECTs and the electrolyte, resulting in the degradation of τ of OECTs (39). After platelet adhesion with adhesion times of 0.1, 0.2, 0.5, 1, and 2 hours, the τ values of OECTs with or without PMC3A coating were measured by applying a gate voltage pulse with a duration of 50 ms (Fig. 3C). The τ value was degraded from 60 μs to over 50 ms after 30 min of platelet adhesion. τ values above 50 ms indicate that the sampling rate will be lower than 20 Hz, making it unusable for detecting ECG signals (~100 Hz) (26). On the other hand, the τ value of OECTs with PMC3A coating only changes from 60 to 81 μs after 2 hours of platelet adhesion (Fig. 3D), indicating excellent blood compatibility, which enables long-term ECG signal monitoring.

Fig. 3 Blood compatibility evaluation of PMC3A coating on OECTs.

(A) SEM image of PET- and PMC3A-coated film after 1 hour of platelet adhesion. Scale bar, 10 μm. (B) Number of adhered platelets on PET, PEDOT:PSS, and PMC3A films after 1 hour of platelet adhesion. (C) Comparison of normalized ΔIds of OECTs without PMC3A coating with a Vgs pulse width of 30 ms after 0.1, 0.2, 0.5, 1, and 2 hours of platelet adhesion. ΔIds is defined as [IdsIds(Vg=0V)]. The normalization of ΔIds was conducted by maximum ΔIds at 100 mV of Vgs before platelet adhesion [ΔIds(Max at 0h, Vg=100 mV)]. (D) Comparison of the normalized ΔIds of OECTs with PMC3A coating with a Vgs pulse width of 30 ms after 0.1, 0.2, 0.5, 1, and 2 hours of platelet adhesion. ΔIds is defined as [IdsIds(Vg = 0 V)]. The normalization of ΔIds was conducted by the maximum value of ΔIds at 100 mV of Vgs before platelet adhesion [ΔIds(Max at 0h, Vg=100 mV)].

In vivo demonstration

The feasibility of the stretchable and blood-compatible OECT array has been performed by ECG measurements on the heart surface of rats. The 4 × 4 array of stretchable and blood-compatible OECTs is attached to the exposed surface of the heart to measure physiological signals (Fig. 4A). The conformal contact between the device film and the surface of the heart is achieved by the honeycomb holes on ultrathin substrate (movie S1). Figures 4 (B and C) shows the ECG signals measured by stretchable OECTs without PMC3A coating. Each experiment was performed with different rats. The beating rate was 420 and 330 beats/min, respectively, as shown in Fig. 4 (B and C). The SNR is calculated to be 41 and 52 dB for the PMC3A noncoated OECT and the coated OECT at 0 min after attachment, respectively. The results indicate that the performance of the OECT already degrades right after attachment to the heart surface. The signal peak was not confirmed in the data from the PMC3A noncoated OECT at 30 min after attachment. However, the SNR of the PMC3A-coated OECT was 51 dB at 30 min after attachment, which was almost the same value recorded right after attachment.

Fig. 4 In vivo recording on the heart of a rat.

(A) Photograph of the stretchable OECT array on the heart of a rat. Scale bar, 5 mm. (B) Recorded ECG by stretchable OECT array without PMC3A coating. (C) Recorded ECG by stretchable OECT array with PMC3A coating. (D) Image of the positioning of the 4 × 4 OECT array on the heart of a rat. (E) ECG signals from multiplexed 4 × 4 OECT array. The drain voltage of each channel was −0.3 V. (F) Time delay of ECG signals. Data were changed by transconductance of each OECT.

Values of SNR above 50 dB can be attained because of conformal contact and the millisiemens order of transconductance. The millisiemens order of transconductance can be obtained due to water permeability of PEDOT:PSS. Conventionally, the transconductance of in vitro gating transistors increases with the surface area of active material. However, the transconductance of OECTs increases with volume because of water permeability of PEDOT:PSS. Hence, the transconductance of the OECT is shown to go up to approximately 100 times higher than that of the Si FET (36).

Furthermore, we have demonstrated the mapping of ECG signals with a 4 × 4 stretchable and blood-compatible OECT array. The device with 3-mm spacing was placed on a heart surface covering the right and left ventricle areas (Fig. 4D). By designing the load impedance to be 0 ohm, the cross-talk in OECT array can be substantially suppressed, as demonstrated in a previous research (22). The reading circuit is shown in fig. S13. Setting load impedance to 0 ohm was achieved using an operational amplifier with the rectangular pulse of input, which is connected to the ground. The result of the spatial distribution of the recorded ECG outputs using the 4 × 4 OECT array as shown in Fig. 4E shows good agreement with previous results. Spatial voltage maps of all nodes at four sequential time points appear in Fig. 4F and movie S2. The anatomical signals show different shapes based on the location of the sensors. The developed stretchable and blood-compatible OECT array successfully recorded spatial-temporal distribution of ECGs on heart surfaces of rats with multiplexing. The heartbeats of rats were in the normal range at 310 beats/min (40). The ECG waveforms at C1, C2, D1, and D2 in Fig. 4E were recorded with a strong right ventricular potential, and the ECG waveforms at C3 and D3 showed the left ventricle potential. The measured signals correspond with the ECG waveform, which is obtained by the standard 12-lead method (41). The recording ECG on the body surface is a result obtained via various tissues such as muscle and skin; however, this device can directly measure the myoelectric potential of myocardium with high sensitivity (42). The amplitude of measured T wave was smaller than that of the human ECG due to the membrane potential of rats (43).

The high SNR (52 dB) can be achieved in this study for the following two reasons. First, we have succeeded in making full use of high transconductance of OECTs in the order of millisiemen, which is higher (by a factor of 10) than that of the single crystal Si FET driven in the electrolyte (36) in the presence of surgical bleeding. The device degradation induced by the formation of blood clot can be substantially suppressed by the PMC3A layer. Simultaneously, high ionic conductivity can be maintained despite the coating of PMC3A. Second, the motion artifact noise can be suppressed because of the high conformability of the microgrid structure. It is noteworthy in Fig. 4 (C and E) that the baselines of measured traces are almost constant, showing that the device can adhere to the target during the motion of the heart (21).


Device fabrication

PVA [10 % (w/v)] was spin-coated at 1000 rpm on glass or Si as a sacrificial layer. Parylene diX-SR was deposited as an ultrathin substrate of the device by SCS Labcoater. The honeycomb grid was patterned by photoresist ZPN 1150 with an MA6 mask aligner. The sample was etched using a reactive ion etcher (RIE) for making holes. Then, the remaining photoresist was washed out using acetone. The 70-nm-thick Au for the source and drain electrode was deposited on the honeycomb grid by the lift-off process. Then, 600-nm-thick parylene diX-SR was deposited as a passivation layer. The via-hole was made by photoresist and RIE etching processes. The second layer of 70-nm-thick Au was patterned for interconnection by the lift-off process. Then, RIE was performed for making the contact pad and active channel. For the PEDOT:PSS layer, 20 ml of an aqueous dispersion (Clevios PH 1000) was mixed with 5 weight % (wt %) ethylene glycol and 0.1 wt % dodecyl benzene sulfonic acid additives to enhance the conductivity and film-forming properties. GOPS [(3-glycidyloxypropyl)trimethoxysilane] (1 wt %) (Sigma-Aldrich) was used to cross-link the film for stable operations in aqueous conditions. Furthermore, the active layer of PEDOT:PSS was spin-coated at 2000 rpm. Then, the device was annealed at 140° for 1 hour for cross-linking of PEDOT:PSS. Finally, the channel was patterned using orthogonal resist (OSR 5001) and RIE. The final device was delaminated from the glass by dissolving PVA into the water.


All characterization of the OECTs was performed using solutions of PBS as the electrolyte and an Ag/AgCl wire (Warner Instruments) as the gate of the OECT electrode. The I-V characteristics of the OECTs and OFETs were measured with a semiconductor parameter (4155C, Agilent). The Waveform Generator/Fast Measurement Unit (WGFMU) module of B1500 (Keysight) was used in measuring the drain currents, while the other WGFMU module was used in measuring the gate voltages of the OECT to determine the time response. A multifunction generator was used to generate the sinusoidal waves.

Preparation of polymer substrates and human platelet adhesion test

The PMC3A-coated substrate was prepared using a spin-coating method. PMC3A was dissolved in methanol, and the polymer solution was prepared at 0.2% (w/v). The polymer solution was spin-coated twice onto PET substrates (ϕ = 14 mm, thickness = 125 μm) using Mikasa Spin Coater MS-A100 at the consecutive rates of 500 rpm for 5 s, 2000 rpm for 10 s, a ramp up for 5 s, 4000 rpm for 5 s, and a ramp down for 4 s and was then dried. The human platelet adhesion tests were performed according to a previously reported procedure (31). Platelet-rich plasma (PRP) and platelet-poor plasma (PPP) were obtained from human whole blood using a two-step centrifugation, and then a plasma solution containing platelets (4 × 107 cells/cm2) was prepared by mixing the PRP and PPP solutions. The plasma solution (200 μl) was placed on each polymer substrate, and the substrate was incubated for 1 hour at 37°C. After 1 hour, each polymer substrate was rinsed with PBS, and then the platelets that had adhered on the polymer substrates were fixed by immersion in 1% glutaraldehyde in PBS for 120 min at 37°C. Finally, each polymer substrate was rinsed with PBS and Milli-Q water. The platelet adhesion number was evaluated using SEM.

In vivo evaluation

Animal experiments were conducted in accordance with the guidelines of the Animal Experiment Committee at the University of Tokyo. The rat (male, 12 weeks old, 268 to 312 g) was anesthetized using 2 to 2.5% isoflurane mixed with air, and the skull of the rib was incised to expose a heart surface. The ECG signal was measured by connecting a semiconducting parameter analyzer (the WGFMU module of B1500, Keysight) or developed multiplexing circuit (fig. S13) to the attached OECT array at a sampling rate of 2 kHz. The measured data were obtained from a single trace without averaging. The filtration of the noise above 1 kHz was obtained from the following formulaEmbedded Image(1)where x, ts, and f−3dB denote the raw data, the interval time, and the cutoff frequency, respectively.

The SNR was calculated after applying the filter. The signal region was set to the maximum fluctuation, and the noise region was set at the middle of two signals during 50 ms. The SNR was calculated by the maximum peak of signal and root mean square of the noise region (27).


Supplementary material for this article is available at

Fig. S1. Fabrication process for manufacturing the stretchable active OECT array.

Fig. S2. Circuit diagram of stretchable OECT array.

Fig. S3. The schematic of a 4 × 4 array of stretchable OECTs.

Fig. S4. The measurement Ids-Vgs curve of OECTs in stretchable MEA.

Fig. S5. The schematic of device for evaluation of stretchability.

Fig. S6. Process of platelet adhesion test.

Fig. S7. The water contact angle of PET, PEDOT:PSS, and PMC3A film.

Fig. S8. Impedance measurement of PMC3A film.

Fig. S9. The surface morphology of PMC3A measured by AFM (Keysight 9500 AFM).

Fig. S10. The comparison of device characteristic before and after PMC3A coating.

Fig. S11. The long-term stability of OECTs coated by PMC3A.

Fig. S12. The SEM image of three samples (left, middle, and right) after platelet adhesion at 1 hour.

Fig. S13. The circuit design of multiplexing recorder.

Movie S1. The electroanatomical mapping on a heart surface using active MEA on ultrathin honeycomb grid substrate.

Movie S2. The electroanatomical mapping on a heart surface by multiplexing.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank N. Matsuhisa, D. Kim, S. Park, S. Lee, and H. Lee for the discussion on fabrications and M. Nishinaka for technical assistance. We extend our gratitude to J. Kim for a fruitful discussion on the machine technique. Funding: This work was supported by JSPS KAKENHI grant number JP817H06149. W.L. was supported by the Japan Society for the Promotion of Science through Program for Leading Graduate Schools (MERIT). Author contributions: W.L., S.K., G.G.M., T. Yokota, M.T., and T.S. designed research. W.L. and S.K. fabricated materials and devices. W.L. and S.K. analyzed in vitro data. W.L., S.K., and Y.J. performed blood compatibility evaluation. W.L., M.N., Y.J., I.S., and M.S. measured electrophysiology signal. W.L., Y.I., and T. Yambe analyzed in vivo data. W.L. and T.S. wrote the manuscript. T.S. supervised the project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article