Flexibler organischer elektrochemischer Transistor als Biosensor für Organ-on-a-Chip

2.1 Materials

The mechanical properties of the microfluidic structure dictated the material choice for OECT manufacturing. Since chip „body“ is made of PDMS (DOW Silgard), the used materials for OECT must be flexible, too. Besides, the organic electrochemical transistor-based biosensor designed for organ-on-a-chip applications has tobe located in the microfluidic channel in the direct contact to cultured cells. Therefore, organic polymers have tobe applied. As a material for electrodes (source, drain and gate), the organic conductor the Poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) (Osilla Ltd.) is applied [6]. It is a flexible and transparent conductive polymer. Due to its unique combination of conductivity, transparency, ductility, and ease of processing, PEDOT:PSS has become a benchmark material in thin-film electronic fabrication. For the active layer between source and drain the polymer semiconductor Poly(3-hexylthiophene-2,5-diyl) (P3HT, Ossila Ltd) was selected [7]. It is soluble (50 mg/ml) in chlorinated solvents such as chloroform, chlorobenzene, dichlorobenzene and trichlorobenzene. Once thin-film is performed, and the solvent is evaporated, it can be used in biosensors as a bio-semiconductor material. P3HT is known as a highly biocompatible and H⁺, K⁺ ion sensitive material [3]. The issue of the organic conductors and semiconductors is variation of the properties due to the processing and microstructuring. As an example, charge carrier mobility for P3HT (M104) is 0.12 cm²/Vs. After the mixing and spin-coating or printing, mobility can be significantly different, up to ±0.03 cm²/Vs.

2.2 OECT manufacturing and testing

The 3D conceptual arrangement of OECT with P3HT (pink colour) is shown in Fig. 2 in combination with organic conductor poly (3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) (blue colour). The transistors are developed on a flexible, transparent PDMS planar substrate. The structured PDMS layer was aligned and bonded on top. Therefore, it was possible to obtain simultaneously the microfluidic reservoir for electrolyte and isolation for the source and drain electrodes. The structured PDMS layer with a microfluidic channel (1mm width and 100 μm height) was obtained from the SU8-100 on a 4¨ Si master wafer by moulding technology with the following procedure. Two components of PDMS were mixed in ratio 10:1, poured on master and cured over 48 hours. Once the polymer was cured, the layer was released, cut and bonded to the non-structured PDMS with adhesive layer (liquid PDMS) and oxygen plasma surface activation (0.5mbar, 1 min). PEDOT:PSS was spin coated on planar PDMS, soft backed at 60°C over 30 min and structured by IR laser ablation (power 4W, speed 1000mm/s, frequency 50 kHz) to obtain source, drain and gate electrodes. The semiconductor P3HT (with the concentration 5mg/ml dissolved in 1,2,4-trichlorobenzene) was aerosol jet printed on the preheated 60°C PDMS substrate with prepared source, drain and gate electrodes. Measurements of the DC voltage-current relationship (VI) were performed using Keithley 2400 source meter and multi-meter by direct on-wafer probing in three-terminal configuration. The measurement setup is shown on Fig. 3. KCl solution with the concentration of 1 mol was used as an electrolyte and pumped by a pipette in the microfluidic channel. The bonding strength of the two PDMS layers (structured and planar) was analysed by burst and leakage tests. In the first case, into the microfluidic channel with one opening air is pressurised (up to 100 kPa, in steps of 5kPa). In latter, liquid is pumped through a channel with peristaltic pump (over 2h, 60 μl/min). In both cases, the bonding strength defined by a value of the pressure or time of pumping, in the moment bonding fails.

Fig. 3

Photograph of the measurements setup: 1 - source meter, 2 - power source, 3 - probe station, 4 - magnified view on the chip on the probe station with the needles connected to source, drain and gate.

3.1 Proof of concept

We designed, manufactured and tested an organic electrochemical transistor (OECT) based biosensor, as shown in Fig. 4. Dimensions are 3mm × 6.5mm. Only organic polymers were utilized for performing the electrodes and the active layer for OECT. With this structure, ionic concentration can be detected inside of the microfluidic channel, which serves for cell culturing. Therefore, we implemented the developed OECT on flexible substrate into a microfluidic channel (Fig. 4).

Fig. 4

Photograph of the OECT-based biosensor integrated into a microfluidic structure.

Implementation of the transistor structure into microfluidic channel requires positioning and sealing of the two PDMS substrates (one is planar with biosensor structure, the other is the moulded microfluidic channel). By connecting the two layers, a reservoir for the OECT electrolyte was created and the insulation of the source and drain electrodes was ensured. Tight encapsulation of the microfluidic channel was possible only with a liquid adhesive layer, since it had to fill the differences in the height of the substrate and electrodes. With over two hours of pumping the liquid through the device, no leakages could be observed. Also, burst test shows, that structure is stable up to 80 kPa. As can be seen in Fig. 5, the manufactured transistors show stable volt-ampere characteristics having a saturation region. The measured curve was achieved using the electrolyte solution KCL with 1 mol concentration. The calculated curve is obtained by curve fitting using a Bernards-Malliaras (BM) model (see Section 3.2, Eq. (1)) to achieve a good correlation with the measured results.

Fig. 5

Measured (dashed line) and calculated (continuous line) VI characteristics of the manufactured OECT.

3.2 Discussion of influence on drain current

According to the Bernards and Malliaras (BM) model [3], the drain current of an organic electrochemical transistor is given by the following expression

where μ is the carrier mobility, C is the volumetric capacitance, V_g, V_d, V_th are gate and drain and threshold voltages, respectively, and W, T, L are the channel width, thickness and length. Similarly, as in a p-channel MOS field-effect transistor, there are two operating regions: triode V_d > V_g − V_th and saturation V_d < V_g − V_th. For organ-on-chip applications, we are most interested in the saturation region, since the transistor can be operated as an amplifier and the output voltage can be used to sense deviations of ion concentration.

However, one of the major challenges for using OECT as a sensor to characterize biological response is to identify the wanted variation of the drain current due to changes of the ionic concentration of the analyte, because at the same time parameters such as threshold voltage, carrier mobility and channel capacitance also influence the measured current. Hence, this requires a more detailed analysis and calibration prior to evaluation of the measurements.

The variation of the current due to uncertainties in the mentioned parameters can be calculated using derivation of a multivariable function, as shown below

where the partial derivatives of the drain current with respect to such parameters are:

If we put the partial equations back into the original BM model, we obtain the following expression for the sensitivity of the drain current:

For easier interpretation of the results, we renormalize the measurement error to the reference value of the nominal drain current

This shows how sensitive the drain current is to manufacturing tolerances and uncertainties in material properties. Additionally, one can observe that the inaccuracy due to threshold voltage spread can contribute a very large error for small overdrive voltages (V_g− V_th).

3.3 Measurements uncertainty based on manufacturing tolerances

First, as Fig. 5 shows, measured VI characteristic (dotted line) and applied BM model (continuous line) are correlating well. Therefore, we use this measurement for calibration and benchmarking of measured versus calculated parameters. We apply the formula (4) developed in the previous section to estimate how the drain current depends on an increasing channel capacitance (10 F/cm³, 40 F/cm³, 80 F/cm³) and carrier mobility (0.1 cm²/Vs, 0.12 cm²/Vs, 0.14 cm²/Vs), neglecting the dependence on threshold voltage V_th here, since it is less affected by manufacturing tolerances. Due to direct proportionality of C and μ we can expect an increase of the current depending on both parameters. As can be seen in (Fig. 6(a)) the drain current in its saturation scales almost directly with a four or eight times higher channel capacitance. In a similar way the carrier mobility influences the drain current (Fig. 6(b)). Assuming that μ can alter by up to 0.04 cm²/ Vs within the fabrication process, a deviation of I_d by about +40 μA must be expected.

Fig. 6

Measurement uncertainty due to deviation of channel capacitance C and carrier mobility μ.