Integrating Conjugated Polymers with Bacteriorhodopsin to Realize Quasi Dual‐Gate Organic Field‐Effect Transistors

Four conjugated polymers are integrated with a bacteriorhodopsin D94N‐HmBRI in organic field‐effect transistors (OFETs) with the device architecture of doped Si bottom gate/SiO2 bottom electric/semiconducting polymer/Au electrodes/D94N‐HmBRI top dielectric. Photosensitivity of D94N‐HmBRI in the devices is validated. It is proposed that light activates the conformational change of all‐trans retinal to 13‐cis retinal in D94N‐HmBRI. The D94N‐HmBRI layer then applies momentary dipoles to the semiconductors, affecting threshold voltage, resulting in photosensitivity. Film‐thickness variation, capacitance determination, and grazing‐incidence X‐ray scattering are performed to support the proposed working principle. Quasi dual‐gate OFETs with a highly doped silicon wafer as the bottom gate and light as the top gate are thus realized by combining conjugated polymers with D94N‐HmBRI.


Introduction
In dual-gate field-effect transistors, a semiconducting layer is sandwiched between two dielectrics to which two gate electrodes bond. [1]In 2005, Cui and Liang applied the concept of dual gate to organic transistors to attain dual-gate organic field-effect transistors (DG-OFETs). [2]In their devices, pentacene was the semiconducting material, thermal SiO 2 was the bottom dielectric, self-assembled SiO 2 was the top dielectric, the bottom-gate electrode was a highly doped silicon wafer, and the top-gate electrode was aluminum.Self-assembled SiO 2 was fabricated from sequential deposition of poly(diallyldimethylammonium chloride), polystyrene, and colloidal SiO 2 .They concluded that the DOI: 10.1002/aelm.202300278DG-OFET had better device performance by adjusting two-gate biases than a single-gate OFET.In 2005, Gelinck, Veenedaal, and Coehoorn clarified the working principle for DG-OFETs, stating that the improved performance of DG-OFETS resulted from nonconstant threshold voltage. [3]In 2009, Park and Salleo fabricated a sensing device on the basis of a DG-OFET architecture with a highly doped silicon wafer as the bottom-gate electrode, SiO 2 as the bottom dielectric, polythiophenes as semiconductors, and AlO x as the top dielectric.AlO x was deposited by atomic layer deposition. [4]When H 2 O adsorbed on AlO x , it formed a layer of dipoles on AlO x , applying negative surface potential to AlO x as a result of the electronegativity of O of H 2 O.The negative surface potential resulted in a positive threshold-voltage shift, leading to a significant current increase at an appropriately selected gate voltage.The electric field induced in the top AlO x dielectric arose from the adsorption of a polar molecule, such as water, rather than a genuine top gate.In our opinion, it might be regarded as a quasi DG-OFET.7][8] Palazzo, Torsi, and co-workers reported OFETs with a purple membrane containing bacteriorhodopsin. [9,10]In a bottomgate-top-contact (BGTC) configuration, the purple membrane was deposited on the top of a SiO 2 dielectric.The authors drew conclusions that photons triggered the photocycle of protein in the purple membrane, generating protons being injected into poly(3-hexylthiophene) (P3HT), bringing about a current increase.
In this study, we use a high expression level purified bacteriorhodopsin protein from Haloarcula marismortui, D94N-HmBRI, instead of a purple membrane. [11,12]Apart from D94N-HmBRI being an integral and seven-transmembrane protein, it is a light-driven outward proton pump owing to several critical residues and possessing a retinal binding with lysine at helix seven by Schiff base.After a specific light wavelength activates D94N-HmBRI, the structural form of retinal is isomerized from all-trans to 13-cis, converts several intermediate states, and eventually returns to all-trans form.In this series of structure isomerization, a proton is released and uptakes through all-trans retinal, Asp85, proton releasing group, and Asp96. [13]our various conjugated polymers P3HT, PDPP4T, PffBT4T, and PNDI2T are integrated with D94N-HmBRI in OFETs with the device architecture of doped Si bottom gate/SiO 2 bottom Scheme 1.Chemical structures of P3HT, PDPP4T, PffBT4T, and PNDI2T.P3HT: poly(3-hexylthiophene).PDPP4T: poly [2,5- electric/semiconducting polymer/Au electrodes/D94N-HmBRI top electric.Photosensitivity of D94N-HmBRI in the devices is demonstrated.It is proposed that light functions as a top gate to activate the conformational change of all-trans retinal to 13-cis retinal in D94N-HmBRI.The D94N-HmBRI layer then applies momentary dipoles to the semiconductors, affecting threshold voltage, resulting in the photosensitivity.Film-thickness variation, capacitance determination, and grazing-incidence X-ray scattering (GIXS) are employed to support the proposed working principle.Quasi DG-OFETs with a highly doped silicon wafer as the bottom gate and light as the top gate are realized by integrating conjugated polymers with D94N-HmBRI.

Plasmid Construction, Protein Expression, and Purification
The gene of D94N-HmBRI was constructed in pET-21b (+) and described in detail in the previous study. [11]A single colony of transformed E. coli C43(DE3) cells was inoculated in LB medium supplemented with 50 μg mL −1 of ampicillin and incubated at 37 °C overnight.For large-scale protein expression, a 1:50 (v/v) dilution of overnight culture was cultured to a fresh LB/ampicillin medium and incubated at 37 °C.When the D 600 of the culture reached 0.4-0.6,isopropyl -d-1-thiogalactopyranoside (final ) and broken by ultrasonic processing (M-4000; Misonix).For the separation of the membrane fraction, total cell-extract centrifugation was performed at 18 000 rpm for 15 min at 4 °C (Hitachi CR-21, R20A2).Then, the supernatant was centrifuged at 48 000 rpm for 1 h at 4 °C (Hitachi CP 80WX).
The sediment was dissolved in buffer B (buffer A supplemented with 2% DDM (n-dodecyl--d-maltoside) for at least 12 h at 4 °C, followed by centrifugation at 18 000 rpm for 45 min at 4 °C (Hitachi CR-21, R20A2) to separate the detergent soluble fraction.Solubilized protein was purified by affinity purification using the Ni-NTA (Ni 2+ -nitrilotriacetate) method.After the detergentsoluble solution containing 20 mm imidazole was incubated with Ni-NTA agarose at 4 °C for 4 h of slow nutation, it was transferred to a chromatography column and washed with buffer C (buffer A with 0.02% DDM and 50 mm imidazole).The target protein was eluted with buffer D (buffer A with 0.02% DDM and 250 mm imidazole).Purified protein was concentrated and exchanged into buffer E (4.0 m NaCl, 50 mm MES, 0.02% DDM, pH 5.8) using a protein concentrator (Millipore, Amicon, cutoff size of 30 kDa) for long-term storage.For further application on the device, the purified protein was dialyzed in 0.1 m NaCl, 0.008% DDM, and pH 5.8.

Device Structure
D94N-HmBRI can be blended into a semiconducting layer or incorporated as an additional layer above or below a semiconducting layer in an OFET.Solution processing is normally used to fabricate a polymer semiconducting layer.Tolerance of D94N-HmBRI to organic solvents is thus considered a significant factor in determining an appropriate device configuration.The color of an aqueous solution of D94N-HmBRI was purple, while the purple color vanished in organic solvents, such as MeOH, tetrahydrofuran (THF), dimethylformamide, acetone, and MeCN (Figure 1a).D94N-HmBRI exhibited an absorption maximum ( max ) at 568 nm in water (Figure 1b).Subsequent to the addition of methanol, the  max significantly decreased in intensity.
The  max at 568 nm virtually vanished in a solution of water and THF.The Schiff-base linkage in D94N-HmBRI might dissociate in THF or methanol to yield free retinal, accounting for the reduction in the absorbance at 568 nm.
The results suggested that D94N-HmBRI would denature in the presence of organic solvents.Contact of D94N-HmBRI with an organic solvent is inevitable when it is blended into a semiconducting layer and incorporated below a semiconducting layer.Therefore, a BGTC configuration is used in this study to avoid contact of D94N-HmBRI with organic solvents (Figure 2).D94N-HmBRI was incorporated above a semiconducting layer by dropcasting.It is noted that in previous studies carried out by Palazzo, Torsi, and co-workers, the purple membrane containing bacteriorhodopsin is deposited below a semiconducting layer.The device configuration between the current and previous studies is different.

Photoactivity of D94N-HmBRI in a Thin Film
Photoactivity of D94N-HmBRI in a solution has been demonstrated. [38]Photoactivity of D94N-HmBRI in a thin film has to be confirmed.The  max at 568 nm was observed for a D94N-HmBRI film, being not exposed to external light except for the spectrometer source light.When a daylight lamp was switched on, the absorption intensity at 568 nm decreased appreciably in company with the increase in the intensity at 412 nm (Figure 1c).Similar spectral change was detected when green light was applied whereas red light gave marginal variation (Figure S4, Supporting Information), suggesting that green light was responsible for the photoactivity.On account of previous studies, the absorption band at 568 nm was assigned to the bR state with all-trans retinal and the absorption band at 412 nm resulted from the M state with 13-cis retinal. [39]A thin film of D94N-HmBRI lacking retinal did not exhibit apparent absorption bands and photoactivity (Figure S5, Supporting Information), indicating that the spectral variation should stem from retinal.Time-dependent UV-vis spectra of D94N-HmBRI in a thin film subsequent to green-light illumination are depicted in Figure 3. D94N-HmBRI after the illumination would return to its ground state in about 20 s.The spectral change was reversible and regulated by the switch of green light or daylight, being suitable for further applications.
On the other hand, photocurrent of D94N-HmBRI in an aqueous solution was determined (Figure 4a).Details are included in the Supporting Information.According to previous studies, [38,40] the positive photocurrent indicated that D94N-HmBRI upon illumination conducted proton pumping outwardly of the protein and the negative photocurrent suggested that D94N-HmBRI received proton from the surroundings.D94N-HmBRI in a film exhibited similar photoresponse that in a solution (Figure 4b), revealing that the behavior of D94N-HmBRI in a film is similar to that in a solution.

OFETs of P3HT, PDPP4T, PffBT4T, and PNDI2T
OFETs with the BGTC configuration were fabricated (Figure 2).Details related to device fabrication are described in the Supporting Information.Transfer characteristics of illustrative devices with P3HT as the semiconducting layer in the saturated region are depicted in Figure 5a.Drain current (I DS ) at a drain voltage (V DS ) of −80 V and a gate voltage (V GS ) of −80 V in the dark or upon illumination was used to quantify the photosensitivity of a device (Equation ( 1)). [41]The light source was a green laser with the wavelength of 532 nm and the light intensity of 4 kW m −2 .Figure 5b revealed that the variation in the I DS was insignificant after 30 s of illumination.Therefore, the illumination time was set to be 30-40 s for all devices.Devices with D94N-HmBRI exhibited more apparent current change after illumination than those without D94N-HmBRI (Figure 5a).On the basis of Equation (1), for P3HT, devices with D94N-HmBRI gave an average photosensitivity of 1.13, whereas those without D94N-HmBRI furnished an average photosensitivity of 0.81 (Table 2).[44] Excitons are electron-hole pairs, which could affect I DS .Nevertheless, devices with D94N-HmBRI still gave greater photosensitivity than those without D94N-HmBRI, indicating the additional effect brought by D94N-HmBRI.Statistical  analysis was carried out.The p-value in null-hypothesis testing was calculated to be 0.00005 between two sets of data, suggesting the difference was statistically significant.The enhancement of photosensitivity by the introduction of D94N-HmBRI was supported.
On account of the OFET equation in the saturation region, the change of I DS might result from the variation carrier mobility (μ) or threshold voltage (V T ).Variation of μ or V T against the time of light exposure is illustrated in Figure 5c, revealing that V T was susceptible to the light exposure and μ was not.Subtracting V T in the dark from V T upon illumination yielded ΔV T .For P3HT, mean ΔV T was 20.28 V for devices with D94N-HmBRI and devices without D94N-HmBRI delivered a mean ΔV T of 12.32 V (Table 2).The significant difference between two groups of data was supported by the low p-value of 0.000007.
Mean photosensitivity and ΔV T for OFETs of PDPP4T, PffBT4T, and PNDI2T are summarized in Table 2. Significant differences in the photosensitivity and ΔV T between the devices with and without D94N-HmBRI were found for PDPP4T and  PffBT4T.Furthermore, the photosensitivity and ΔV T of a device with PffBT4T as the semiconducting layer and D94N-HmBRI as the top dielectric were examined for four cycles of light and dark.The V t shift and I DS shift can be observed for the four cycles (Figure 6), indicating that the device photosensitivity is reproducible.

Working Principle
A hypothesis that the device photosensitivity comes from the change of conductivity of D94N-HmBRI upon illumination is discussed in the first place.The conductivity of D94N-HmBRI in a thin film was determined to be ≈4 μS cm −1 (Figure 7).The conductivity was low.Moreover, it did not vary much upon illumination.The hypothesis was not supported by the results.
As mentioned in the Introduction, Park and Salleo reported the fabrication of a quasi DG-OFET with doped Si as the gate electrode, SiO 2 as the bottom dielectric, polythiophenes as semiconductors, and AlO x as the top dielectric. [4]When H 2 O adsorbed on AlO x , it formed a layer of dipoles on AlO x , applying negative surface potential to AlO x as a result of the electronegativity of O of H 2 O.The negative surface potential resulted in a positive ΔV T .ΔV T in a DG-OFET has been formulated in Equation ( 2). [3,4,7] top and C bottom are the capacitance of the top and bottom gate dielectrics, respectively.V G,top and V T,top are the gate voltage and threshold voltage resulting from the top gate and V G,top − V T,top is the top-gate bias.Equation ( 2) indicates the strong coupling of ΔV T with C top and the top-gate bias.It is proposed that the role of D94N-HmBRI would resemble that of AlO x in the quasi DG-OFET.After light absorption, the D94N-HmBRI layer would exert momentary dipoles to the semiconductors, bringing about ΔV T and the resulting photosensitivity.
On the basis of Equation ( 2), greater C top can give larger ΔV T .For a plate capacitor, the capacitance is inversely proportional to the film thickness.Therefore, one can expect that increasing the thickness of D94N-HmBRI should reduce ΔV T and the corresponding photosensitivity.In P3HT devices, when the thickness of D94N-HmBRI was increased to around 200 nm, the ΔV T and photosensitivity were 16.61 V and 0.80, respectively,  being smaller than the values with thinner D94N-HmBRI (Table 2).

Capacitance
Equation (2) indicates that ΔV T is associated with C top and the top-gate bias.It is challenging to estimate the top-gate bias.Nevertheless, the capacitance of D94N-HmBRI is experimentally accessible.A device with the Au/D94N-HmBRI/Au/ITO/glass substrate configuration was submitted to an inductance, capacitance, and resistance (LCR) meter.In the device, the thickness of D94N-HmBRI was around 200 nm.The capacitance of D94N-HmBRI at a frequency of 500 kHz was determined to be 1.18 μF in the dark and that under the illumination of a green laser (intensity: 4 kW m −2 ) was 1.30 μF.LCR measurements revealed that the capacitance of D94N-HmBRI was increased after illumination, being in resonance with observed ΔV T .When D94N-HmBRI was exposed to light, conformational change of all-trans retinal to 13cis retinal gave rise to 1) structural deformation of D94N-HmBRI and 2) release of proton from D94N-HmBRI.Both would generate momentary dipoles.The momentary dipoles should be responsible for the rise in the capacitance.It was envisaged that the momentary dipoles were not strong in magnitude, indicating that OFETs are capable of sensing subtle dipole change, paving a way for delicate sensing applications.

GIXS
GIXS was conducted at the national synchrotron radiation research center, Taiwan.A diffraction signal at q z = 1.8 Å −1 in the out-of-plane direction was detected for a thin film of D94N-HmBRI prepared by drop-casting, suggesting that D94N-HmBRI possessed a certain degree of anisotropy in a thin film (Figure 8a).In contrast, the diffraction signal at q z = 1.8 Å −1 was not detected for a thin film of D94N-HmBRI without retinal (Figure 8b).The results indicated that the anisotropic signal at q z = 1.8 Å −1 might be associated with retinal.It is envisioned that the momentary dipoles of the D94N-HmBRI layer triggered by the conformational change of retinal upon illumination might be anisotropic, as well.The anisotropic momentary dipoles could be advantageous to enhancing the ΔV T and photosensitivity of the devices.

Figure 1 .
Figure 1.a) Color of D94N-HmBRI in various solvents.b) UV-vis spectra of D94N-HmBRI in water, a solution of water and THF, and a solution of water and methanol and c) UV-vis spectra of D94N-HmBRI in a thin film in the dark and under daylight-lamp illumination.

Figure 3 .
Figure 3. a) Time-dependent UV-vis spectra of D94N-HmBRI in a thin film subsequent to green-light illumination (The light source is a green laser with the light intensity of 4 kW m −2 and the wavelength of 532 nm.) and b) recovery of the absorbance at 568 and 412 nm after the illumination.

Figure 4 .
Figure 4. Photocurrent of D94N-HmBRI a) in a solution and b) in a film upon illumination for 850 ms (green shade) and in the dark for 950 ms.

Figure 5 .
Figure 5. a) Transfer characteristics of illustrative devices with P3HT as the semiconducting layer in the saturated region.Black: Without D94N-HmBRI in the dark, blue: without D94N-HmBRI upon illumination, red: with D94N-HmBRI in the dark, and green: with D94N-HmBRI upon illumination.b) Transfer characteristics of a device with P3HT (the semiconducting layer) and D94N-HmBRI (the top dielectric) against the time of light exposure.c) Variation of μ (hollow square) or V T (solid square) of a device with P3HT (the semiconducting layer) and D94N-HmBRI (the top dielectric) against the time of light exposure.

Figure 6 .
Figure 6.The V T and I DS shift of a device with PffBT4T (the semiconducting layer) and D94N-HmBRI (the top dielectric) for four cycles of light and dark.The lower data points were measured in the dark.The upper data points were measured under illumination from a green laser for 30-40 s.

Figure 7 .
Figure 7.The conductivity of a D94N-HmBRI thin film upon cycling of illumination.Green rectangles represent continuous illumination from a green laser for 45 s.The illumination was repeated three times.

Figure 8 .
Figure 8. 2D GIXS images of a) D94N-HmBRI in a thin film and b) D94N-HmBRI without retinal in a thin film.

Four
conjugated polymers P3HT, PDPP4T, PffBT4T, and PNDI2T in conjunction with D94N-HmBRI are used.BGTC devices with the architecture of doped Si bottom gate/SiO 2 /polymer/Au electrodes/D94N-HmBRI top electric are fabricated.D94N-HmBRI is deposited at the last step to avoid contact with organic solvents.Photoactivity of D94N-HmBRI in a thin film is confirmed and green light triggers the photoactivity.P3HT devices with D94N-HmBRI give photosensitivity.The enhancement of photosensitivity by introducing D94N-HmBRI is supported by statistical analysis.The photosensitivity depends on the threshold-voltage shift ΔV T rather than μ.Moreover, PDPP4T, PffBT4T, and PNDI2T are examined.Significant differences in the photosensitivity and ΔV T between the devices with and without D94N-HmBRI are found for PDPP4T and PffBT4T.The proposed working principle states that light functions as a top gate to activate the conformational change of all-trans retinal to 13-cis retinal in D94N-HmBRI.The D94N-HmBRI layer then applies momentary dipoles to the semiconductors, affecting ΔV T , resulting in photosensitivity.The devices are thus considered quasi DG-OFETs with doped Si as the bottom gate and light as the top gate.ΔV T in a DG-OFET has been estimated.It is associated with C top and the top-gate bias.The LCR measurements reveal that the capacitance of D94N-HmBRI is increased after illumination, being in resonance with ΔV T observed in the devices.GIXS indicates that D94N-HmBRI has a certain degree of anisotropy in a thin film.The anisotropic momentary dipoles could be advantageous to enhancing the photosensitivity of the devices.Overall, our study demonstrates that D94N-HmBRI acts as a top electric integrating with conjugated polymers to realize quasi DG-OFETs.