A polysaccharide bioprotonic field-effect transistor

Abstract

In nature, electrical signalling occurs with ions and protons, rather than electrons. Artificial devices that can control and monitor ionic and protonic currents are thus an ideal means for interfacing with biological systems. Here we report the first demonstration of a biopolymer protonic field-effect transistor with proton-transparent PdH_x contacts. In maleic-chitosan nanofibres, the flow of protonic current is turned on or off by an electrostatic potential applied to a gate electrode. The protons move along the hydrated maleic–chitosan hydrogen-bond network with a mobility of ~4.9×10⁻³ cm² V⁻¹ s⁻¹. This study introduces a new class of biocompatible solid-state devices, which can control and monitor the flow of protonic current. This represents a step towards bionanoprotonics.

Introduction

Proton (H⁺) transport is important in many natural phenomena¹. Preeminent examples include ATP oxidative phosphorylation in mithochondria², the HCVN1 voltage-gated proton channel³, light-activated proton pumping in bacteriorhodopsin⁴, and the proton-conducting single water file in the antibiotic gramicidin⁵. In living systems, electrical signals are communicated and processed by modulating ionic⁶ and protonic currents⁷. In contrast, the development of computing has mainly focused on devices that control electronic currents such as vacuum tubes, solid-state field-effect transistors (FET), and nanoscale molecular structures^8,9,10,11. Few examples of protonic-based devices exist, and include an ice FET working with AC current¹², and a water bipolar junction transistor¹³.

At the nanoscale, ionic (and protonic) conductivity has attracted increasing interest with the advent of resistive ionic memories¹⁴, memristors^15,16, synaptic transistors¹⁷, and nanofluidics^18,19,20. In hybrid bionanodevices, biological multifunctionality has been added to carbon nanotubes²¹ or silicon nanowires²² with transmembrane proton conductive proteins. Bionanoelectronic devices²³ that can control the current of ions and protons—a more appropriate language than electrons in nature²⁴—are uniquely positioned. In this regard, nanofluidics devices are particularly attractive. However, these require microscopic liquid reservoirs, and current control at physiological concentration is limited to difficult-to-fabricate nanometer channels²⁰.

Recently, conducting polymer ion bipolar junction transistor devices have been demonstrated^25,26. These exploit ion selective membranes as contacts to extract and deliver ions (Na⁺, K,⁺ Ca²⁺) and neurotransmitters in solution. To further this exciting route, biocompatible field-effect devices with H⁺-selective solid-state contacts would allow exclusive interfacing with biological proton-conducting channels.

Here we demonstrate a new class of solid-state H⁺-FET devices that afford electrostatic control of the protonic current (Fig. 1). In our maleic–chitosan H⁺-FET, the flow of protonic current—measured with proton transparent palladium hydride (PdH_x) (ref. 27) contacts—is turned on or off by an electrostatic potential applied to the gate electrode. Protons dissociated from the maleic acid groups move along the hydrated molecule hydrogen-bond network following the Grotthus mechanism² (mobility, μ_H+≈4.9×10⁻³ cm² V⁻¹ s⁻¹).

Figure 1: Schematics and atomic force micrographs of protonic field-effect transistors.

(a) AFM image (false coloured) overlayed onto a schematic of bottom contact back-gated transistor type device. PdH_x contacts (10 μm wide, 8.6 μm spacing) are on the top of a 100 nm SiO₂ capping a p-type (0.001 ohm cm⁻¹) wafer. maleic–chitosan thin-film dimensions 3.5 μm wide, 8.6 μm long, and 82 nm thick. (b) Atomic force micrograph of maleic–chitosan nanofibres after partial dissolution of the polysaccharide in water (scale bar 200 nm). (c) Molecular structure of maleic–chitosan, the degree of substitution (m/n) is 0.85 for these devices.

Results

Device architecture

These H⁺-FET devices comprise a maleic–chitosan nanofibre protonic channel bridging PdH_x contacts (source and drain) on the top of SiO₂ (gate dielectric). Maleic–chitosan (poly (β-(1,4)-N-Maleoyl-D-glucosamine)) (Fig. 1b,c) is a polysaccharide chitin derivative of particular interest for the prospect of bionanoprotonics. Most chitin derivatives are biodegradable, nontoxic, and physiologically inert²⁸. Palladium is chosen as the contact material for its ability to form proton-conducting PdH_x on exposure to hydrogen²⁷. PdH_x affords proton exchange between the contacts and the maleic–chitosan channel without electrolysis. The SiO₂ (100-nm) gate dielectric insulates the maleic–chitosan and the contacts from the Si electrostatic back gate.

Two-terminal measurements

We first probe the maleic–chitosan conductivity in thicker films sandwiched by PdH_x contacts, to verify that this H⁺-FET architecture indeed measures protonic current. When a bias is applied between the contacts, the current increases with relative humidity (and polysaccharide hydration level) (Fig. 2a). Hysteresis also increases with humidity, potentially due to increased charge accumulation/depletion at the contacts. Measurements using hydrogen-depleted Pd or Au contacts—both proton blocking—record a considerably smaller current (Supplementary Fig. S1). It is important to note that both Au and Pd are significantly better electronic conductors than PdH_x. We thus confirm that PdH_x contacts effectively measure the protonic current in this material. Increase in protonic conductivity with hydration level is common in other biological macromolecules such as collagen²⁹, cellulose³⁰ and keratin³¹. A higher level of water absorption creates more proton-conducting hydrogen-bond chains (HBC) that serve as proton wires² for Grotthus type transfer to occur (Fig. 2b). In maleic–chitosan, we propose that the protons responsible for the current (as hydronium ions in the HBC) originate from the polysaccharide maleic acid groups, some of which are deprotonated (Fig. 2c). The protonic conductivity in chitin (Supplementary Fig. S1), which has no maleic groups, is significantly lower than in the maleic–chitosan derivative.

Figure 2: Proton conductivity and mechanism.

(a) Conductivity as a function of humidity for a maleic–chitosan thin film (1 cm²×1 cm², 300 μm thick) sandwiched between PdH_x electrodes. (b) Grotthuss mechanism for H⁺ conduction along the hydrogen bonds formed by water and polysaccharide polar groups. (c) A four-monomer segment of maleic–chitosan. The intra- and inter-molecular hydrogen bonds as well as the hydrogen bonds between the water of hydration and the polar parts of the molecule form a continuous network comprised by hydrogen-bond chains. The maleic group donates a H⁺ to the hydrogen-bond network and forms an H₃O⁺ (hydronium) ion.

H⁺-FET measurements

In a H⁺-FET type device, the source-drain protonic current, I_ds, recorded as a function of drain-source bias, V_ds, is modulated by changing the potential of the back gate electrode, V_gs (Fig. 3). As expected for a semiconducting FET with predominantly positive charge carriers, a negative V_gs results in a higher source-drain current for the same V_ds, whereas a positive V_gs almost turns I_ds off. In most materials, protons have to overcome activation energy in the range of 0.1 to 1 eV for current to flow^29,31,32. This activation energy may be associated with producing a H⁺ OH⁺⁻ pair or with transport of H⁺ along the HBC. This behaviour is often referred to as protonic semi-conductivity. However, in highly conducting proton wires, such as in gramicidin, the activation energy associated with transport is very small⁵. In the maleic–chitosan pro-FET, the linear I_ds dependence on V_ds for low V_ds (and V_gs=0 V) indicates that protons do not need to overcome any appreciable energy barrier for I_ds to flow (Fig. 3a). This suggests that highly efficient proton wires are formed in the material.

Figure 3: Protonic transistors transfer characteristics.

(a) Plot of I_ds as a function of V_ds for different V_gs (RH 75%). For device with PdH_x contacts shown in Figure 1. (b) Plot of channel proton density (n_H+) as function of V_gs. Points are n_H+ derived from I_ds–V_ds data (a) assuming a mobility of (4.9±0.5)×10⁻³ cm² V⁻¹ s⁻¹ from n⁰_H+=8.9×10¹⁷cm⁻³ at V_gs=0. Line is (C_g=gate capacitance per unit area, t=device thickness). From simulations C_g×area=1.79×10⁻¹⁴ F. (c,d) Schematics of protonic transistor channel charge carrier n_H+ modulation for negative (c) and positive (d) gate voltage.

H⁺-FET model

A simple model can be used to describe the H⁺-FET. I_ds is proportional to V_ds through the conductivity of the channel σ=e n_H+μ_H+ (e=proton charge, n_H+=number of free protons per unit volume, and μ_H+=proton mobility). For V_gs=0 V, we assume that all the protons available for conduction originate from the maleic acid groups in the polysaccharide (pKa=3.2) (Supplementary Methods). From this assumption, we estimate . Using , the proton mobility can be calculated from the slope of I_ds at V_gs=0 V (Supplementary Methods). The calculated value μ_H+≈4.9×10⁻³ cm² V⁻¹ s⁻¹ is consistent with the mobility of H⁺ in diluted acidic solutions.³³ For a negative V_gs, a positive charge is induced onto the channel because the maleic–chitosan and the gate create a capacitor with the SiO₂ as the dielectric. The charge per unit area induced onto the channel is directly proportional to V_gs through the gate capacitance per unit area, C_g. To create that charge, excess protons are injected into the maleic–chitosan via the proton transparent contacts. The contribution of these excess protons to the conductivity can then be estimated as an increase in n_H+ (higher I_ds) by a factor that is equal to −(V_gsC_g/et) (t=device thickness) (Fig. 3c). At the same time, a decrease in n_H+ (smaller I_ds) is expected for a positive V_gs (Fig. 3d).

The data in Figure 3b corroborates this description. This plot depicts n_H+ estimated from the I_ds–V_ds experimental data (Fig. 3a) as a function of V_gs. These n_H+ values are compared with , which is the contribution to n_H+ expected from the excess charge induced onto the channel by V_gs. Given that this model involves several assumptions and approximate estimates of the maleic–chitosan channel volume, this data is in good, even fortuitous, agreement. Measurements performed on a thicker device show a significantly smaller I_ds modulation from V_gs, as predicted by our description (Supplementary Fig. S2). It is important to note that this kind of electrostatic gating of a protonic (ionic) solution, unlike in nanofluidic devices, can occur for devices thicker than the Debye length (k⁻¹). From n⁰_H+, the maleic–chitosan ionic strength is ~1 mM, which results in k⁻¹~ 10 nm. Even for the Figure 1a H⁺-FET, that is significantly thicker than the Debye length (t ~8 k⁻¹), V_gs modulates I_ds by changing the proton density of the channel. This is a unique property of having proton transparent contacts and reduces fabrication constraints.

Device simulations

We produce two-dimensional plots of the proton charge density n_H+ at different V_gs (Fig. 4a–c) by solving drift diffusion and Poisson equations in the context of a semiconductor device simulator, where the hole mobility is replaced by proton mobility. From these plots, it is clear that, for a negative V_gs, the proton density is drastically increased at the polysaccharide–dielectric interface to form a highly conducting region. The thickness of this region (~10 nm) is in agreement with the estimated Debye length (Fig. 4a). Plots of current density show that the majority of I_ds flows in the same area (Supplementary Fig. S3). At the same time, a positive V_gs considerably depletes the channel from protons reducing n_H+ and consequently I_ds (Fig. 4c). An analogous V_gs effect on the charge density and current density distribution has been observed in thin-film organic transistors³⁴. These transistors have geometry and electrostatics similar to our H⁺-FET. From these plots, it is also clear that in significantly thinner devices the expected charge modulation will be higher. Such devices may be fabricated with individual maleic–chitosan nanofibres and should offer a greatly improved on-off ratio.

Figure 4: Protonic Transistor Simulations.

(a–c) two-dimensional plots of device n_H+ (log scale in cm⁻³) as a function of V_gs (V_ds=0.3 V). (d) Simulated I_ds data (lines) for small V_ds corresponding to device data shown in Figure 3a (only −10, 0, 10 V_gs shown for clarity). This data was generated using n⁰_H+=8.9×10¹⁷ cm⁻³ and a corresponding μ^s_H+=5.5×10⁻³ cm² V⁻¹ s⁻¹, which is close to the one estimated from the device. The experimental data (dashed line) is shifted to I_ds=0 for V_ds=0 to compensate for hysteresis.

We also estimate I_ds as a function of V_ds for different V_gs (Fig. 4d). The simulations are in good agreement with the experimental data at low V_ds, while the nonlinear dependence of I_ds on V_ds for higher V_ds is not reproduced. We have assumed the contacts to be completely proton transparent. A small contact barrier (neglected in our model) may cause charge accumulation or depletion at the maleic–chitosan PdH_x interface. This charge, in turn, may reduce the strength of the electric field at the contacts, thus effectively limiting the current. When the bias is reversed, the charge should be released and cause hysteresis in the I_ds–V_ds dependence. Hysteresis is indeed observed when V_ds is ramped from 0 to 1 V and then back in the same measurement (Supplementary Fig. S4). The causes for hysteresis and the non-linear behaviour of I_ds for high V_ds are still not fully understood and merit further investigation.

Discussion

These experiments demonstrate, for the first time, electrostatic gate modulation of source-drain current in a solid-state biopolymer based thin-film protonic (H⁺) field-effect transistor. PdH_x contacts allow for indiscriminate measurement of the protonic current. A voltage applied to the electrostatic gate controls the maleic–chitosan channel proton density and the channel conductivity. Current modulation occurs even for devices thicker than the Debye length. This allows for current control in highly concentrated electrolytes at any length scale. In the future, several nanostructured biological and organic materials can be measured in these devices. The demonstrated ability to control protonic currents in nanostructured biocompatible solid-state devices (bionanoprotonics) may open exciting opportunities for interfacing with living systems. These opportunities include biomedical applications where protonic currents are important such as the in-vivo study and stimulation of proton selective ion-channels. To this end, H⁺-FETs architectures with optimized performance in physiological solutions will be required.

Methods

Maleic–chitosan nanofibres

Maleic–chitosan nanofibres were prepared following previously published procedures³⁵. The maleic–chitosan hydration level was determined with a thermogravimetric analyser (TA Instruments, model 2050).

Two-terminal devices

Cu contacts (1 cm²×1 cm²) coated with 50 nm e-beam evaporated (Balzers PLS 500) Pd (5 nm Cr adhesion layer) were used to sandwich 300 μm films for relative humidity (RH)-dependent measurements. Maleic–chitosan films were prepared from a 5 ml of 3.0 wt% maleic–chitosan aqueous solution after drying for 5 h. Chitin films were prepared from a 5 ml of 0.5 wt% chitin/HFIP solution after drying for 1 h.

H⁺-FET

Devices were fabricated on p-type Si (Addison Engineering, B, ρ=0.001 ohm cm⁻¹) with thermally grown silicon oxide (100 nm). Photolithography and lift-off was used to define the contacts. Pd metal (50 nm) with a 5-nm Cr adhesion layer was deposited via e-beam evaporation (Balzers PLS 500). After dialysis and freeze drying, maleic–chitosan was prepared in a DI water solution. This procedure eliminates any salt and thus potential salt effects on the conductivity. To make a polysaccharide-based device, 2 μl polysaccharide solutions of concentration ranging from 0.02 mg ml⁻¹ to 0.2 mg ml⁻¹ were carefully drop cast on top of the patterned silicon wafer and blown under N₂ flow. Devices were mounted on a chip, and wire bonded.

Electrical characterization

Measurements were performed with a semiconductor parameter analyser (Agilent 4155C). An environmental chamber was used (5% H₂ or 100% N₂) with controlled RH monitored with a traceable hygrometer (Fisher Scientific, ±0.1% error). During pro-FET measurements at 75% RH and 5% H₂, devices with no connections were monitored to have at most noise current. This was done to ensure that the measured device current was from the maleic–chitosan channel and not from water condensed on the top of the SiO₂.

Atomic force microscopy

Atomic force microscopy (AFM) was employed to measure device dimensions. Tapping mode AFM was performed on a Veeco Multimode V with a Nanoscope IV controller using Veeco Probes Sb-doped Si cantilevers (ρ=0.01–0.025 Ω-cm, k=40 N/m, ν~300 kHz).

Simulations

Electrical properties of the structure were obtained by solving Poisson's equation together with the electron and hole (in this case, proton) continuity equations throughout a CAD tool (ATLAS, Silvaco). We defined a silicon (Si) based metal–oxide–semiconductor transistor with the same dimensions as the experiment. We used a 8.6-μm long, 3.5-μm wide, 82-nm channel with a rectangular cross-section sitting on the top of 100-nm SiO₂ gate dielectric (ɛ_sio2=3.9). Pd (Φ=5.1 eV) source and drain were modelled as ohmic protonic contacts (no barrier) to the material. We replaced the properties of silicon with those of the channel material. Maleic–chitosan ɛ_mc=10, and n₀=8.9×10¹⁷ cm⁻³ from our estimate. Mobility was varied to fit the data. In addition, we assigned the mobility of electrons to a value that is orders of magnitude smaller than that of protons. This minimizes the role of electrons in the electrical response of the device. To calculate the gate capacitance, we first obtained the interface charge at different gate biases; whereas the two other contacts were kept at zero bias. The capacitance was then estimated by the simple equation of ΔQ/ΔV, where ΔQ and ΔV represent the variations of the interface charge and gate, respectively.