Development of PFAS Responsive Electrolyte‐Gated Field‐Effect Transistors Using Conjugated Polymer‐Single‐Walled Carbon Nanotube Complexes Authors: Roozbeh Javad Kalbasi, Mokhamed Ranne, May Ourabi, Benoît H Lessard, Alex Adronov Published: 2026 ABSTRACT ABSTRACTConjugated polymers are effective dispersants of SWNTs which also impart their properties onto polymer‐SWNT complexes. By modifying the polymer sidechain, the properties of the complex can be fine‐tuned for targeted applications. Here, we introduced pillar[5]arenes containing pendant quaternary ammonium groups onto the polymer side chains using strain promoted azide‐alkyne cycloaddition (SPAAC). The polymer effectively dispersed SWNTs, generating a polymer‐SWNT dispersion in aqueous media. Analysis by 19F‐NMR demonstrated that this polymer exhibits electrostatic interactions between carboxylate ions of perfluoroalkyl acids and the pendant ammonium ions. Semiconducting (sc‐) enriched polymer‐SWNT complexes were prepared and subsequently functionalized with the pillar[5]arene side chains. The resulting water‐processable polymer‐SWNT complexes were incorporated in electrolyte‐gated field‐effect transistor (EGFET) devices, which demonstrated strong ambipolar performance and rapid signal stabilization using DI water as the supporting electrolyte. The EGFETs were used as proof‐of‐concept sensors capable of detecting potassium perfluorooctanoate (PFOK) and perfluorooctanoic acid (PFOH), and differentiating them from octanoic acid as a control. This work demonstrates that preparation of functional sc‐enriched polymer‐SWNT complexes enables proof‐of‐concept sensing using EGFET devices as a model platform.Keywords: carbon nanotubes, conjugated polymers, click chemistry, electrolyte‐gated field‐effect transistors, PFAS sensor, Pillar[5]arenes, post‐polymerization functionalization, strain‐promoted cycloaddition Introduction Single‐walled carbon nanotubes (SWNTs) [1, 2] have garnered interest in a myriad of applications due to their exceptional electronic [3], mechanical [4, 5, 6], optoelectronic [7, 8, 9], and structural properties [10, 11, 12]. They have found application in several fields, such as the semiconducting layer of field effect transistors [13], high strength nanocomposites [5, 14, 15], touch screens [16], organic photovoltaics [17], flexible electronics [18, 19, 20], sensors [21, 22, 23], and many other devices [24, 25, 26]. Unfortunately, their application is severely hindered due to inter‐tube π‐stacking resulting in the formation of insoluble bundles of SWNTs, as well as contamination with catalyst particles from their synthesis and the mixture of semiconducting (sc‐) and metallic (m‐) SWNTs that results from standard commercial synthetic processes [27]. Non‐covalent functionalization decorates the SWNT surface without negatively impacting π‐conjugation and electronic properties [28, 29, 30, 31, 32]. This process involves coating the surface of SWNTs with various dispersants, which include conjugated polymers [33, 34, 35, 36, 37, 38, 39], aromatic small molecules [40, 41, 42], DNA [43, 44, 45], proteins [46], and surfactants [47, 48]. In addition to preserving SWNT properties, the resulting polymer–SWNT complexes exhibit dispersant‐derived characteristics such as solubility. Conjugated polymers are highly tunable in backbone structure, side chain, and molecular weight, a versatility that has enabled separation of m‐ and sc‐SWNTs [49], introduction of cleavable side chains capable of altering the conductivity of SWNT‐loaded materials [50, 51], development of pressure sensors [52], and chemiresistive sensing of various analytes [22, 23]. Since many applications depend on side‐chain modification, we developed an azide‐functional polyfluorene that enables efficient post‐polymerization functionalization via click chemistry, allowing polymer–SWNT complexes to be tailored for specific applications [22]. Per‐ and poly(fluoroalkyl) substances (PFAS) are pollutants that can cause adverse health effects at low concentrations, known as “forever chemicals” due to their extreme stability in the environment [53]. They are found in many applications, such as food packaging, clothing, cookware, makeup, firefighting foams, and others [54]. PFAS have also been linked to a multitude of diseases and adverse health effects in humans, which include high cholesterol, reduced response to vaccines, kidney and testicular cancers, decreased reproductive abilities, and liver damage [55]. Their widespread use and environmental persistence have resulted in pervasive human exposure, driving extensive research into PFAS detection and removal from aqueous media, largely based on quaternary ammonium derivatives or fluorinated polymeric moieties [56]. Detection of PFAS at ng·L−1 levels is typically done using LC/MS/MS methods [57], however, these methods require expensive equipment, are time‐consuming, and are not suitable for on‐site use [58]. Consequently, recent research has emphasized low‐cost, portable PFAS sensors for drinking, ground, surface, and wastewater, though most rely on fluorescence‐based responses requiring expensive instrumentation, with limited development of electrically responsive sensors [58, 59, 60]. Recently, pillar[5]arenes (PAs), which are cyclic structures comprised of five aromatic rings linked by methylene bridges, have been investigated in PFAS sensing applications. It has been found that PAs that are decorated with quaternary ammonium groups exhibit sensitivity toward PFAS [61]. In this study, we introduce a sensing platform that leverages the simultaneous presence of CNTs, conjugated polymers, and PAs. We demonstrate that a PA that is mono‐substituted with a dibenzocyclooctyne (DBCO) group is capable of readily undergoing a strain promoted azide‐alkyne cycloaddition (SPAAC) with azide groups present on sidechains of a conjugated poly(fluorene), PF‐N3 . Upon attachment of the PA to the side chain of the polymer, it is possible to further modify it to introduce quaternary imidazolium groups on the rim of the PAs. The resulting polymer can be used to disperse SWNTs in water, producing a polymer‐SWNT complex that was found to be responsive to PFAS. Results and Discussion Preparation of the PFAS‐detecting polyfluorene‐SWNT complex began with the synthesis of DBCO‐functionalized pillar[5]arene 4 (Scheme 1). We began with the commercially available 1,4‐bis(2‐bromoethoxy)benzene, which was reacted with paraformaldehyde in the presence of boron trifluoride diethyl etherate as a catalyst, and 1,2‐dichloroethane as template and solvent to obtain the pillar[5]arene with bromoethyl chains decorating its rim (1). We proceeded to monosubstitute one of the ten bromine groups via Williamson ether synthesis with 4‐nitrophenol and potassium carbonate in acetonitrile to prepare 2. Hydrogenation of 2 in the presence of palladium on carbon (10 wt.%) in methanol and chloroform resulted in selective reduction of the nitro group to the corresponding aniline 3. Subsequent amidation of 3 with DBCO‐NHS resulted in formation of the DBCO‐containing pillar[5]arene 4. Synthesis of DBCO‐Pillar[5]arene. To prepare the PFAS responsive polyfluorene, the previously synthesized poly[9,9‐didodecylfluorene‐alt‐9,9‐bis(6‐azidohexyl)fluorene] (PF‐N3) (Mn = 19 kDa, Đ = 2.1) [62], was coupled with 4 through a strain‐promoted azide‐alkyne cycloaddition (SPAAC) in DCM to yield poly[9,9‐didodecylfluorene‐alt‐9,9‐bis(6‐azidohexyl)fluorene]‐9[Br]‐pillar[5]arene (PF‐PA) (Mn = 31 kDa, Đ = 1.8). The reaction progress was monitored by FTIR spectroscopy, specifically following the disappearance of the azide stretch at ∼2090 cm−1 (Figure 1A). It was found that the azide stretch completely disappeared after 4 h, indicating quantitative conversion. Finally, the PF‐PA polymer was dissolved in imidazole and allowed to stir overnight at reflux to substitute the Br groups on the rim with imidazolium groups, yielding the water‐soluble polymer PF‐PA‐N‐methylimidazolium bromide (PF‐PA‐MIM, Scheme 2). (A) FTIR spectrum of PF‐N3 and PF‐PA showing the disappearance of the azide stretch upon SPAAC coupling. (B) 19F‐NMR spectrum (D2O) of PFOK and PFOK with PF‐PA‐MIM. Synthesis of PF‐PA and PF‐PA‐MIM. Before proceeding with using the polymer to disperse SWNTs, it was necessary to confirm that PF‐PA‐MIM was effective in binding a member of the PFAS class of compounds. This was done by recording the 19F‐NMR spectrum of potassium perfluorooctanoate (PFOK) with and without the presence of PF‐PA‐MIM (Figure 1B). The 19F‐NMR spectrum of PFOK alone consisted of seven unique signals, six in the −110 to −130 ppm region and one at −80.8 ppm. However, in the presence of the PF‐PA‐MIM, it was found that the signals in the −110 to −130 ppm region coalesce, resulting in their appearance as one broad signal. Additionally, the signal at −81 ppm also broadens and shifts further upfield to −82.5 ppm. The broadening and upfield shift of the PFOK peaks after interaction with the quaternary ammonium groups of PF‐PA‐MIM indicate the formation of a complex between PFOK and the cationic pillar[5]arenes. This interaction alters the environment of the PFOK, reducing its mobility and resulting in the observed changes in the spectrum. This effect suggests that the PF‐PA‐MIM polymer displays sensitivity toward PFAS. Dispersions of HiPCO SWNTs were prepared using PF‐PA‐MIM as the dispersant according to literature procedures [63]. This was done by taking 10 mg of PF‐PA‐MIM and 5 mg of HiPCO SWNTs in 10 mL of dH2O and sonicating the mixture using a probe sonicator for 30 min, while cooling the mixture in an ice bath. The resulting black suspension was then centrifuged at 8346 g for 30 min, and the supernatant was carefully removed and filtered over cotton to isolate the polymer‐SWNT complex. To confirm the selectivity of the PF‐PA‐MIM polymer, we first characterized the PF‐PA‐MIM‐SWNT complex using UV/Vis‐NIR spectroscopy. There are three distinct regions in which HiPCO SWNTs exhibit absorption features: the S11 (830–1600 nm) and S22 (600–800 nm) regions correspond to sc‐SWNTs, and the M11 (440–645 nm) region corresponds to m‐SWNTs. As shown in Figure 2A, PF‐PA‐MIM disperses both sc‐ and m‐SWNTs, but clearly has more selectivity toward metallic SWNTs relative to the parent PF‐N3 . This result is not surprising as dispersions prepared in polar solvents have been shown to contain higher concentrations of m‐SWNTs [64]. We further characterized this complex using Raman spectroscopy (Figure 2B,C), which allows additional analysis of SWNTs by using laser excitation wavelengths that overlap with the van Hove singularities present in the density of states for SWNTs. The electronic transitions are affected by the diameter and type of SWNTs, which requires multiple laser excitation wavelengths to analyze the full range of SWNTs. To perform Raman analysis, the PF‐PA‐MIM‐SWNT complex was drop cast onto a silicon wafer and the substrate was irradiated with 532 and 633 nm laser wavelengths. These wavelengths allow separate observation of sc‐ and m‐SWNTs. Using the 532 nm excitation wavelength, a broad G− band at ∼1500 cm−1 was observed, which is indicative of m‐SWNTs. Additionally, the radial breathing mode (RBM) region also showed signals that primarily corresponded to m‐SWNTs (215–310 cm−1). The 633 nm excitation wavelength showed similar results, with signals in the RBM region corresponding to both m‐SWNTs (175–230 cm−1) and sc‐SWNTs (230–300 cm−1). This further suggests that the PF‐PA‐MIM polymer disperses m‐SWNTs to a larger extent than the parent PF‐N3 . (A) UV/Vis‐NIR spectrum of the PF‐PA‐MIM‐SWNT complex. (B) Raman spectrum of the PF‐PA‐MIM‐SWNT complex obtained by exciting at 532 nm. (C) Raman spectrum of the PF‐PA‐MIM‐SWNT complex obtained by exciting at 633 nm. Before proceeding with device fabrication, it was necessary to prepare a SWNT complex that was enriched with sc‐SWNTs, as m‐SWNTs hinder TFT device performance. Additionally, plasma torch SWNTs have been found to be ideal for TFT device applications due to their larger diameter, which has been found to result in higher mobilities and lower contact resistance [65, 66, 67, 68]. To enrich the samples, we employed the strategy of preparing a PF‐N3‐SWNT complex in toluene using the conditions described previously, which have been shown to result in a selective sc‐SWNT dispersion [49]. To further eliminate any trace amounts of m‐SWNTs, the dispersion was treated with SiO2, which has been shown to adsorb m‐SWNTs and enable their removal [69]. This was done by adding 25 mg of SiO2 to the PF‐N3‐SWNT dispersion (10 mL, prepared as described above) and sonicating the mixture for 45 min. The dispersion was left undisturbed for 12 h and centrifuged at 8346 g for 30 min to remove any precipitated SWNTs. The supernatant was collected, and the resulting dispersion was characterized using UV/Vis‐NIR spectroscopy to confirm that the sample was enriched in sc‐SWNTs (sc‐PF‐N3‐SWNT) (Figure 3A). To quantify the enrichment, we calculated the ϕ i values from the UV‐Vis‐NIR absorption spectra pre‐ and post‐silica gel treatment [70]. Generally, the ϕ i value increases with the purity of dispersed sc‐SWNTs, with values above 0.33 indicating a sc‐SWNT purity of >99%. The ϕ i value for the initial nanotube dispersion was 0.35, which is in agreement with previous dispersions of plamsa torch SWNTs with polyfluorene [70]. Upon treatment with silica gel, the ϕ i value increased to 0.41, again indicating enrichment of sc‐SWNTs. (A) UV/Vis‐NIR spectrum of the PF‐N3‐SWNT complex before and after treatment with SiO2. (B) FTIR spectrum of the PF‐N3‐SWNT complex before and after functionalization with 4. To prepare the semiconducting single‐walled carbon nanotubes dispersed with PF‐PA‐MIM, abbreviated as sc‐PF‐PA‐MIM‐SWNT, compound 4 was added to the sc‐PF‐N3‐SWNT dispersion to carry out the SPAAC coupling directly on the SWNT dispersion, and reaction progress was monitored by FTIR spectroscopy (Figure 3B). Upon the disappearance of the azide stretch at 2090 cm−1, the dispersion was collected via filtration through a 0.2 µm pore diameter Teflon membrane and dried over a stream of N2. Imidazole was then added to the dried complex, which was allowed to stir at reflux for 24 h, after which the complex was redispersed by sonicating with a probe sonicator in dH2O for 30 min to yield the sc‐PF‐PA‐MIM‐SWNT complex (Scheme S1). The sc‐PF‐PA‐MIM‐SWNT dispersion was then used to form the semiconductor network in electrolyte‐gated field‐effect transistors (EGFETs). For this, the dispersion was drop‐cast onto octyl‐trichlorosilane‐treated substrates with thermally evaporated CrAu source and drain electrodes, yielding a bottom‐contact device architecture, which was then gated with DI water and a Ag/AgCl reference electrode. The devices demonstrated excellent ambipolar performance, shown in Figure 4A,B, achieving similar currents in p‐ and n‐type operation with I ON/OFF ratios of 103. The n‐type threshold voltage (V T) indicated slightly better performance in this regime, with average values of 0.345 ± 0.030 V, as opposed to the p‐type V T values of −0.445 ± 0.014 V. Similarly, the width‐normalized transconductance (g m, norm) was slightly greater in n‐type operation than in p‐type operation, with median values of 1.92 and 1.40 mS cm−1, respectively. Though SWNT‐based EGFETs tend to demonstrate more pronounced p‐type behavior, the high density of positively charged quaternary ammonium ions on the wrapping polymer leads to n‐doping of the SWNTs, strongly enhancing the n‐type characteristics of the devices. This high ambipolar performance, coupled with the ease of fabrication and water‐processability, renders this SWNT complex a promising option for the development of printed electronics. While other water‐gated SWNT FETs have been demonstrated in the literature [71, 72, 73], ambipolar behavior is sometimes alluded to but only p‐type performance is reported. Ambipolar electrolyte‐gated devices have been reported, though these either rely on additional fabrication steps, such as the addition of a membrane to isolate the SWNTs [74] from the aqueous electrolyte, or employ non‐aqueous electrolytes, such as ionic liquids [75, 76] and ion gels [77, 78]. (A) Output curve showing the ambipolar performance of the sc‐PF‐PA‐MIM‐SWNT devices. Output characteristics were measured using voltages ranging from 0 to −0.8 V for p‐type operation and 0 to 0.8 V for n‐type operation. (B) p‐ and n‐type transfer curves obtained at VDS = −0.5 V and VDS = 0.5 V. (C) Linear plot of the transfer curve obtained at increasing concentrations of PFOH and PFOK added to DI water. (D) Change in current of the sc‐PF‐PA‐MIM‐SWNT devices after exposure to increasing concentrations of PFOH (red), PFOK (blue), and octanoic acid (purple). Error bars represent the standard deviation of the response. n = 4 for PFOH and PFOK, n = 3 for octanoic acid. Prior to the addition of PFOK or perfluorooctanoic acid (PFOH) to the liquid electrolyte, the device current I DS was stabilized through repeated transfer curve acquisitions to ensure minimal drifts in device characteristics during sensing. Though this conditioning can require several hours for EGFETs using other materials, some EG‐SWNT‐FETs have been reported to rapidly stabilize at relatively short time scales. In the case of the sc‐PF‐PA‐MIM‐SWNT devices, similarly quick stabilization was observed in p‐type operation, whereas attempts to stabilize the devices in n‐type proved unsuccessful due to a continuous drift in current (Figure S14). Once stabilized, increasing concentrations of PFOK and PFOH were added to the DI water in the EGFETs and transfer characteristics were re‐acquired to evaluate their response to the analytes (Figure 4C). This was repeated using octanoic acid, a non‐fluorinated counterpart to PFOH, to demonstrate selectivity toward PFAS. The devices demonstrated a turn‐on sensing response to both analytes, stemming from the p‐doping of SWNT by the PFAS anions, seen also in the positive shift in V T (Figure S15). Meanwhile octanoic acid elicited a negative shift in device current (Figure 4D). This was apparent at high concentrations of 10−4 and 10−3 M. While small variations in device current were observed when exposing the devices to PFOK and octanoic acid at the lower bound concentration of 10−5 M, these may indicate that the sensor is not sufficiently sensitive to distinguish natural variations in the electrolyte medium. The limit of detection (LOD) extracted from the 3σ method was high for both PFOH and PFOK detection, corresponding to 7 × 10−4 and 3 × 10−4 M, respectively. This highlights the low sensitivity of this platform, as this is several orders of magnitude higher than other sensors demonstrated in literature. More work remains to be done to improve the overall sensitivity of these devices, which may require more ordered SWNT networks to ensure the entire film is in contact with the analytes. In addition to engineering the device to improve the sensor's performance, additional testing would be required to comprehensively measure its selectivity and ensure that its sensing abilities are not dampened by the presence of interferents in the electrolyte. Though significant improvement is needed to match the performance of current PFAS sensors, this work presents a proof‐of‐concept of integration of PFAS‐sensitive SWNT complexes in electrolyte‐gated transistors. Conclusion In conclusion, we synthesized a novel polyfluorene containing pillar[5]arene side chains bearing quaternary ammonium pendant groups. 19F‐NMR experiments demonstrated that this polymer is able to bind PFOK through electrostatic interactions and can additionally be used to disperse SWNTs in aqueous media. Highly enriched sc‐SWNT dispersions with the PF‐PA‐MIM were produced by first preparing a sc‐SWNT enriched dispersion using PF‐N3 and subsequently functionalizing the dispersion with the pillar[5]arene groups. This sc‐SWNT enriched sample was used to prepare water‐processable EGFET devices, with high ambipolar performance and selective response to PFAS. This work demonstrates the versatility of PF‐N3 as a functionalizable SWNT dispersant, and the ability of the PF‐PA‐MIM derivative to serve as an active component in transistor‐based PFAS sensors. Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Funding This research was funded by the Natural Science and Engineering Research Council of Canada (NSERC) Green Electronics Network (GreEN) (grant number: NETGP 508526‐17) as well as the NSERC Discovery Grant Program (RGPIN‐2025‐03936 to B.H.L and RGPIN‐2023‐04824 to A.A). M.R. is grateful for support through an OGS scholarship. M.O. is grateful for support through an NSERC CGS‐D scholarship. We thank the Ontario Research Fund (Large Infrastructure Fund) and Canadian Foundation for Innovation, CFI# 40178 (HIIT) and CFI# 43247 (SSMART), for support in acquisition and maintenance of the infrastructure needed for EGFET fabrication for this project. Conflicts of Interest The authors declare no conflicts of interest. Supporting Information The Supporting Information is available free of charge at on the 1H‐NMR, UV/Vis‐NIR, and Raman Spectra; synthetic schemes of small molecules and control polymer. Supporting information Supporting File: smtd70909‐sup‐0001‐SuppMat.docx. Contributor Information Benoît H. Lessard, Email: benoit.lessard@uottawa.ca. Alex Adronov, Email: adronov@mcmaster.ca. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request. References Associated Data Supplementary Materials Supporting File: smtd70909‐sup‐0001‐SuppMat.docx. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.