Low-power/high-gain flexible complementary circuits based on printed organic electrochemical transistors

The ability to accurately extract low-amplitude voltage signals is crucial in several fields, ranging from single-use diagnostics and medical technology to robotics and the Internet of Things. The organic electrochemical transistor, which features large transconductance values at low operation voltages, is ideal for monitoring small signals. Its large transconductance translates small gate voltage variations into significant changes in the drain current. However, a current-to-voltage conversion is further needed to allow proper data acquisition and signal processing. Low power consumption, high amplification, and manufacturability on flexible and low-cost carriers are also crucial and highly anticipated for targeted applications. Here, we report low-power and high-gain flexible circuits based on printed complementary organic electrochemical transistors (OECTs). We leverage the low threshold voltage of both p-type and n-type enhancement-mode OECTs to develop complementary voltage amplifiers that can sense voltages as low as 100 $\mu$V, with gains of 30.4 dB and at a power consumption<2.7 $\mu$W (single-stage amplifier). At the optimal operating conditions, the voltage gain normalized to power consumption reaches 169 dB/$\mu$W, which is>50 times larger than state-of-the-art OECT-based amplifiers. In a two-stage configuration, the complementary voltage amplifiers reach a DC voltage gain of 193 V/V, which is the highest among emerging CMOS-like technologies operating at supply voltages below 1 volt. Our findings demonstrate that flexible complementary circuits based on printed OECTs define a power-efficient platform for sensing and amplifying low-amplitude voltage signals in several emerging beyond-silicon applications.

Complementary metal-oxide-semiconductor (CMOS) field-effect transistors, made from silicon (Si), have been the workhorse of the integrated circuit (IC) industry since the 1980s, in part due to their low power consumption. Dennard scaling 1 , the CMOS scaling law, states that the supply voltage for each new CMOS generation is reduced by 30%, and the power consumption subsequentially reduces by 50%.
After decades of development, the latest 7-nm-node CMOS process reaches a supply voltage of 0.75 V 2 .
Today, the Si-CMOS technology is heavily explored in Internet of Things (IoT) applications, serving as low-power outposts that record physical sensor parameters (e.g., motion, light, temperature), communicate over long distances, and harvest and store energy for its operation 3 . Expanding IoT modules with flexible, soft, or large-area chemical sensors and actuators only possible off-Si, enables a circuit technology that can amplify and route signals, facilitating signal compatibility and low-cost integration between Si-technology and embedded devices. Further, for many IoT and bioelectronic applications (e.g., (bio-)chemical sensors and neuronal interfacing), the on-site technology is preferably realized without Si-chips to enable many different form factors, proximity, elasticity, and signal transduction, tailor-made for the actual chemical/biological environment. Also in this case, a lowpower/voltage, high-performing, and flexible circuit technology operating at the site of stimulation or sensing is needed to record and transfer signals at high signal-to-noise performance.
Unlike field-effect transistors, organic electrochemical transistors (OECTs) typically operate at less than 1 V and consist of a conducting polymer channel and a gate connected by a common electrolyte 4 . When a voltage is applied to the gate, ions from the electrolyte enter the bulk of the channel material to compensate for the injected charge carriers in the oxidized or reduced organic semiconductor, thus modulating the channel conductance. Since OECTs operate at low voltages and exhibit transconductance values that are orders of magnitude higher than their (organic) field-effect transistor counterparts, they are ideally suited to sense and amplify low-amplitude voltage signals 5 . To date, OECTs have been used to construct digital circuits [6][7][8] , sensors of biological, physical, and chemical signals [9][10][11] , and neuromorphic computing devices 12,13 .
While the large OECT transconductance can translate a small voltage into a sizable current, postsignal processing often necessitates further current-to-voltage conversion. To address this need, a handful of high-gain OECT-based voltage amplifiers have been developed [14][15][16][17][18][19][20]  In this work, we report a different approach to reduce power consumption, which takes advantage of CMOS technology. Si-based electronics have already shown significant improvements in power efficiency afforded by CMOS circuits compared to unipolar technologies, wherein only one type of transistor is used 21 . In order to develop CMOS-like OECT technology, both p-type and n-type OECTs working in enhancement mode are required 20 . Printing technologies, like ink-jet printing and screenprinting, are compatible with organic electronic materials and devices to enable flexibility, conformability, large-scale integration, and cost efficiency 8,22 . However, despite prior reports of largescale circuits based on all-printed unipolar OECTs 6 , printed complementary OECT amplifiers have not yet been reported.
Here, we report a printed, flexible single-stage voltage amplifier based on a pair of complementary OECTs, which operates at low voltages (0.3-0.7 V) with less than 2.7 µW power consumption. Both ptype and n-type OECTs operate in the accumulation mode, with maximum transconductance of ~0.20 mS and threshold voltages of less than |0.25| V, enabling low voltage operation. Electrochemical bulk doping of the channel materials is confirmed by in-situ spectroelectrochemistry measurements. The driving strength and the operating voltage of both p-type and n-type OECTs are well balanced and enable the development of single-stage complementary inverters having voltage gains of up to 26 V/V, static power consumption as low as 12 nW, and excellent noise margin (89%). With a DC offset at the input, the inverter operates as a pull-push amplifier which is able to sense AC voltage signals as low as 100 µV with a gain of 30.4 dB. This ability to detect small voltage signals with very low power consumption yields a power-normalized gain of up to 169 dB/µW, which is over 50 times greater than state-of-the-art amplifiers based on OECT technologies. The DC voltage gain can be further increased by cascading double inverters into a two-stage amplifier that reaches 193 V/V, which is the highest gain among emerging CMOS-like thin-films technologies operated at low voltages (sub-1-volt) to date. Our voltage amplifier based on printed complementary OECTs offers a power-efficient solution for autonomous, conformable, wearable, and portable sensors. In addition, the low operation voltage (0.3 V) offers the possibility of being (self-)powered by light, heat, wireless power, triboelectricity, and other power sources that can only provide low voltage and/or limited power supply, thus opening the way to battery-free wearable electronics.

Printed complementary organic electrochemical transistors
Complementary electronic circuitries require the development of both p-type and n-type enhancementmode OECTs. Here, we use polythiophene functionalized with tetraethylene glycol side chains (P(g42T-T)) and poly(benzimidazobenzophenanthroline) (BBL) as the p-type and n-type semiconducting polymers, respectively (Fig. 1a). While neither P(g42T-T) nor BBL are soluble in water/alcohols, they can be dispersed in alcohol solvents in the form of nanoparticle inks, which are better suited for largescale printing. The BBL nanoparticles are obtained by solvent exchange from BBL-methanesulfonic acid (MSA) solution to isopropanol (IPA) under rapid stirring, whereas the P(g42T-T) nanoparticles can be obtained by solvent exchange from its chloroform solution to IPA. The nanoparticle sizes of BBL and P(g42T-T) are 28 nm and 21 nm, respectively, as measured by dynamic light scattering (DLS, Supplementary Figure 1). The BBL and P(g42T-T) nanoparticle dispersion inks in IPA are printable by both screen printing and inkjet printing. We chose a poly(sodium-4-styrene sulfonate) (PSSNa, Fig. 1a) based hydrogel as the printable sodium electrolyte for the n-channel OECT. D-sorbitol and glycerin are employed to enhance sodium conductivity and the stability of the PSSNa-based hydrogel. We chose polyquaternium-10 (PQ-10, a non-toxic quaternized hydroxyethyl cellulose chloride, Fig. 1a) based hydrogel as the printable chloride electrolyte for the p-channel OECT. Both PQ-10 and PSSNa hydrogels have similar or even higher ion conductivity than 0.1 M NaCl aqueous solution (Supplementary Figure 2). The all-printed OECT-based circuits are fabricated through the combination of screen printing and inject printing (Fig. 1b). First, carbon electrode and silver electrode layers are screen printed sequentially on a flexible A3 sized polyethylene terephthalate (PET) substrate. The electrochemically inert carbon electrodes are in direct contact with the polymer semiconductor, while the silver underlayer reduces the electrode resistance. An insulating layer is used to pattern the channel and gate regions. 6 The silver/silver chloride (Ag/AgCl, 100 nm thick) gate layer and polymer semiconductor layer (20 nm for P(g42T-T) and 250 nm for BBL, to match their ON-state channel conductances) are sequentially deposited by spray-coating. Finally, the polymer hydrogel electrolytes (PQ-10 or PSSNa) are screen printed to finish the fabrication of the electrochemical circuits (Fig. 1c).
For each OECT, the channel length/width (L/W) is 200 μm/2 mm, and the Ag/AgCl gate region is 2×2 mm 2 . When implemented as OECT channels, both P(g42T-T) and BBL show relatively low threshold voltages 23,24 , ensuring that a low supply voltage is required for operation. The device structure of the printed complementary OECTs are depicted in Fig. 1b-d. Here, three layers, including a carbon layer for drain/source contact, a silver layer for interconnects, and an insulating layer to define the gate and channel windows, are sequentially screen-printed on a polyethylene terephthalate (PET) substrate (Supplementary Figure 3). Semiconductor layers (P(g42T-T) and BBL) are then successively spraycoated through a shadow mask to form the transistor channels. A layer of Ag/AgCl is deposited at the silver gate electrodes and the electrolyte hydrogels (PQ-10 and PSSNa) are finally deposited to bridge the gate and the channel. More details of device fabrication can be found in the Methods section. The printed OECTs have a side-gate configuration, which is optimal for low-cost and large-scale manufacturing, as this outline requires fewer processing steps compared to top-gate configurations.

Electrical characteristics of printed p-/n-type OECTs
In a three-terminal (drain, source, and gate) OECT device (Fig. 1b), a bias voltage between gate and source (VGS) drives ions from the electrolyte into the channel and results in the modulation of the conductivity between drain and source. For enhancement-mode OECTs, the drain-to-source current (IDS) under a voltage bias (VDS) can be approximately expressed as: 5 where µ is the charge carrier mobility, CV is the volumetric capacitance, W, L, and d are the channel width, length, and thickness, respectively, and VT is the threshold voltage. Unlike conventional fieldeffect transistors, the channel thickness d contributes to the conductivity as the accumulation of charges occurs throughout the entire bulk of the semiconductor layer.  Robust operation of CMOS-like inverters requires a balance between the driving strengths of the ptype and n-type transistors to maximize noise margins. For instance, the channel width of the PMOS (ptype MOS) transistor in a silicon CMOS inverter is typically advisable to be three times larger than that of the NMOS (n-type MOS) transistor to achieve balanced driving strengths. Because of the volumetric bulk conductivity in OECTs, we are able to achieve a well-balanced p-/n-type OECT operation by using the same channel W and L, but different p-/n-type semiconductor film thickness, so that the same transistor footprint design can be applied regardless of the difference in electron/hole mobility. This OECT feature has no equivalent in other organic and/or inorganic FET technologies, and simplifies substantially the device manufacturing protocol.

In-situ spectroelectrochemistry of OECTs
To understand the operation of P(g42T-T) and BBL OECTs, we measured absorption spectra of the polymer films on FTO inside a three-electrode electrochemical cell. We For P(g42T-T) the bleaching of the ground state is especially notable, as the extent of the bleaching is larger in magnitude than the ground state absorption of the same sample measured in air. It is important to note that since P(g42T-T) is doped in air, the undoped baseline in the spectroelectrochemistry measurements has a larger ground state absorption. Nevertheless, most of the P(g42T-T) ground state absorption is bleached, illustrating that a majority of P(g42T-T) segments are doped at the highest applied voltages. The bleaching is accompanied by a broad polaronic absorption band extending well into the IR region. The polaronic absorption reaches a maximum at VGS = -0.3 V, followed by a decrease in the polaronic absorption while the ground state further bleaches slightly at more negative potentials. This is attributed to increased bipolaron formation at more oxidative potentials, as recently shown for P(g42T-T). 26,27 Unlike P(g42T-T), the bleaching of BBL ground state absorption does not have the same shape as the pristine ground state absorption shown in Supplementary Figure 8. This is due to overlapping positive polaronic absorption bands, indicating that the fraction of polaronic BBL cannot be directly derived from the amplitude of the ground state bleach. In addition, BBL has multiple positive polaronic absorption bands with maxima at 400, 720, and 865 nm. The 720 and 865 nm bands are close to each other in energy and form one broad absorption band at the largest applied gate voltages. This band reaches a magnitude that is a third in intensity of the pristine ground state absorption, which is more intense than previously published spectra of molecularly-doped BBL films. 25 We can conclusively say  The switching threshold of an inverter (VM), defined as when Vin equals Vout, is an indication of the balanced driving strengths of p-type and n-type transistors, and ideally, the threshold is equal to VDD/2.
The switching threshold is shown in Fig. 4b, and it closely follows the ideal behavious at all VDD.
Another benefit of balanced p-type and n-type OECTs is to obtain large high and low noise margins (NMH and NML), defined as follows: where VIH and VIL are the input voltages HIGH and LOW at the operation points of /# $%& /# '( = −1 at the voltage transfer characteristics (VTC) of the inverter (Fig. 5b). Here, the NMH and NML are listed in Supplementary Table 1 and the total NM with respect to VDD is up to 89 % (VDD = 0.7 V).
The voltage gain Av in units of V/V and decibels (dB) is defined as: As shown in Fig. 4c

Printed OECT-based push-pull voltage amplifiers
Although typically implemented as digital devices, CMOS inverters can be used as analog amplifiers.
A major advantage of using CMOS inverters as amplifiers in the push-pull configuration (Fig. 5a), as opposed to active load and/or current source load configurations, is that these devices benefit from the summation of the transconductance of both transistors 21 , as shown in Eq. (5). The voltage gain of a OECT-based CMOS-like inverter can be calculated as: Our push-pull amplifiers with printed complementary OECTs show a gain of 16 (V/V) (24 dB) when amplifying a 10 mV (amplitude) sinusoidal signal with a DC offset of 0.28 V (Fig. 5b), which is consistent with the DC gain of the inverter shown in Fig. 4c. When amplifying signals as low as 100 µV, the voltage gain reaches 29 (V/V) (29.2 dB), as shown in Supplementary Figures 13-14. The maximum amplification observed is 33 (V/V) (30.4 dB) with a 200 µV input signal (Fig. 5c). We evaluated the gain of the voltage amplifier with respect to input frequencies ranging from 0.06 Hz to 5 Hz with a 10 mV input signal. The detailed waveforms at each frequency are shown in Supplementary   Figures 13 and 14.
These OECT-based single stage push-pull amplifier can serve as a building block for more complex amplifiers. Here, we demonstrate a two-stage amplifier to obtain a high voltage gain by cascading a number of single-stage amplifiers, as shown in Fig. 5d.  to 50 times higher than that of state-of-the-art OECT amplifiers [17][18][19] (Fig. 6a). In addition to existing OECT technologies, we have also compared our results with voltage amplifiers based on electrolytegated thin-film transistors [28][29][30][31] and organic field-effect transistors 32-36 (details can be found in Supplementary  2 41 . Assuming that a self-powered device has an area of 1 cm 2 dedicated to carry the power supplier, the power that can be provided by such energy supplier is shown in Fig. 6a.
This shows that our printed voltage amplifier has huge potential for self-powered devices.
As we demonstrated, the voltage gain of the printed amplifier can be further increased by cascading multiple stages together. With two-stage printed amplifiers, we report DC gains of 193 V/V, which, to the best of our knowledge, are the highest not only among the OECT technologies 19,20 but also among other sub-1V amplifiers based on emerging CMOS-like technologies like CNTs 42,43 , graphene 44 , and 2D transition metal dichalcogenide 45-50 (see Fig. 6b and Supplementary Table 3  (for BBL) nanoparticles were generated. The P(g42T-T) and BBL nanoparticles were respectively collected by centrifugation (5000 rpm, 30 min) and washed by IPA for 6 times until neutral. The neutral P(g42T-T) and BBL nanoparticles were re-dispersed in IPA to obtain a dispersion inks (about 0.006 mg/mL for P(g42T-T) and 0.1 mg/mL for BBL). The preparation of the electrolyte hydrogels was done as follows. PQ-10 (2.5 g) was added to water (7.5 mL) in a 20 mL reaction vessel and was then sealed and stirred at 150 °C for 4 hours until a uniform, bubble-free highly viscous hydrogel was formed. The PQ-10 hydrogel was then cooled down to room temperature before use. PSSNa (4.0 g), d-sorbitol (1.0 g), and glycerol (1.0 g) were added to water (4.0 mL) in a 20 mL reaction vessel and the mixer was sealed and stirred at 150 °C for 4 hours until a uniform, bubble-free highly viscous hydrogel was formed.
The PSSNa hydrogel was then cooled down to room temperature before use.

Data availability
The data that support the findings of this study are available from the corresponding author upon request.