Photothermoelectric Response of Ti3C2Tx MXene Confined Ion Channels

With recent growing interest in biomimetic smart nanochannels, a biological sensory transduction in response to external stimuli has been of particular interest in the development of biomimetic nanofluidic systems. Here we demonstrate the MXene-based subnanometer ion channels that convert external temperature changes to electric signals via preferential diffusion of cations under a thermal gradient. In particular, coupled with a photothermal conversion feature of MXenes, an array of the nanoconfined Ti3C2Tx ion channels can capture trans-nanochannel diffusion potentials under a light-driven axial temperature gradient. The nonisothermal open-circuit potential across channels is enhanced with increasing cationic permselectivity of confined channels, associated with the ionic concentration or pH of permeant fluids. The photothermoelectric ionic response (evaluated from the ionic Seebeck coefficient) reached up to 1 mV·K–1, which is comparable to biological thermosensory channels, and demonstrated stability and reproducibility in the absence and presence of an ionic concentration gradient. With advantages of physicochemical tunability and easy fabrication process, the lamellar ion conductors may be an important nanofluidic thermosensation platform possibly for biomimetic sensory systems.

W ith recent advances in nanofabrication technologies and growing interest in bioinspired smart nanochannels, significant progress has been made in the design and fabrication of artificial nanofluidic channels, mimicking tailored functions of biological ion channels. 1,2 For instance, human skin, possessing various sensory receptors, has inspired extensive research to develop nanofluidic sensory systems having a fast and selective response to different chemical and physical stimuli. Among various types of biological ion channels, thermosensitive transient receptor potential (thermoTRP) cation channels have been of key interest in such nanofluidic applications as bionic thermometer and thermal energy conversion and harvesting. 3,4 The thermoTRP channels in the thermoreceptor cells of the skin convert external thermal stimuli into electrical signals associated with action potentials.
In earlier studies, temperature-sensitive gatekeepers such as elastin-like polypeptide loops or polyNIPAM brushes, which were grafted onto synthetic or protein pores, described crucial roles in controlling the ionic flow in response to temperature changes. 5−10 Most recently, intrinsic nanochannels with ultraconfined pores and plenty of surface charges revealed their potential applications to a variety of external stimuli-responsive ionic sensory systems. 11−14 When the confined space of nanochannels is as small as the Debye screening length near charged walls, an overlapped electrical double layer (EDL) mostly composed of surface charge-compensating counterions can be created inside nanoconfined channels, which enables charge-selective ion transportation in response to external stimuli. 14−17 Especially, under chemical potential driven by temperature or salinity, such charge-selective ion transportation can serve as a driving force to generate electrical potential. In light of this feature, stimuli-responsive artificial nanochannels with surface charges have been explored with a wide range of fluidic conduits such as perforated nanopores, nanoslits, or nacre-inspired lamellae nanochannels. 14,18−21 Focusing on nanoporous lamellae structures composed by multilayered two-dimensional (2D) nanosheets, we explored the cation channels of metal carbide and nitride (MXene) lamellar membranes in pursuit of mimicking thermosensing process of thermoTRP channels. As the lamellar membranes provide densely interconnected and interplanar nanocapillaries with subnanometer dimensions, it is beneficial to detect collective motion of ions in response to electrochemical potential. Additionally, the nanoconfined channels are covered by an EDL, enabling charge-permselective transport. Arising from aqueous etching and exfoliation processes during MXene synthesis, negatively charged surface terminal groups (−O, −OH, or −F) on the MXene sheets contribute to creating subnanometer-sized interplanar spacing in the lamellar membranes and endows the MXene channels with the surface charges. 14,22,23 The MXene-based lamellar membrane makes an enticing ion-selective conductor with potential applications in salinity-gradient energy harvesting and ion-sieving membranes for water purification. 14,24−26 In this study, we investigated ionic thermoelectric transport through Ti 3 C 2 T x MXene-based fluidic channels, especially under a temperature gradient created by light-driven heating. Coupled with inherent photothermal conversion performance of MXene materials, 23,27,28 the subnanometer capillaries in MXene membranes exhibit thermo-osmotic cation flow and consequential thermoelectric response of up to 1.0 mV·K −1 under local sunlight exposure. Even under tiny temperature differences below 1 K, the MXene channels had a very sensitive thermoelectric response. Such nonisothermal voltage is found to be enhanced with increasing cationic permselectivity of channels, which is associated with ionic concentration or pH of permeant fluids. Furthermore, the MXene ion channels revealed stable and reproducible responses to temperature change under light illumination. We envision the MXene cation channels mimicking the biological thermosensation process would offer a promising avenue for many potential applications, including temperature sensing, photodetection, or photothermoelectric energy harvesting. Figure 1a demonstrates the ionic thermoelectric voltage generation through the confined ion channels under a local light irradiation-driven temperature gradient. Transverse fluidic channels in the lamellar MXene membranes offer the cationexchange features enabled by EDL inside the nanoconfined channels. When the channel height is comparable to the electrostatic screening length given by the Debye length, the EDL formed on each interplane can overlap, enabling the strong cation-selective passage under chemical potentials. 14−17 When the fluidic channel is stimulated by localized thermal excitation, a temperature gradient can be created through the permeating fluids. Note that a chemical potential of water decreases linearly with the increase of temperature, and water molecules are prone to moving from higher chemical potential to lower chemical potential. As a consequence of local photothermal heating, a chemical potential along the temperature gradient can create a thermo-osmotic ion transport with concurrent water flux from cold to hot sides. 11,29 In principle, a transport takes place in a direction producing more entropy, and more specifically heat diffuses down, whereas a confined permeant flows against a thermal gradient. 26 Aiming to create such thermodynamic conditions under light illumination, the MXene lamellar structures are exploited, offering outstanding photothermal conversion performance as well as plenty of surface charge groups accommodating hydrated cations.

RESULTS AND DISCUSSION
The Ti 3 C 2 T x MXene sheets were prepared from MAX phase (Ti 3 AlC 2 ) through selective etching of Al using a mixture of aqueous hydrochloric acid and lithium fluoride (Experimental Section and Figure S1). The scanning electron microscopy (SEM) image of synthesized MXene monolayers clearly verifies the presence of planar surfaces with sharp and welldefined edges. Atomic force microscopic measurement shows that a monolayer Ti 3 C 2 T x possesses a thickness of 1.5 to 2.0 nm while further confirming its well-defined planar morphology. The observed larger thickness, taking into account a theoretical monolayer thickness of around 0.98 nm, 30,31 is ACS Nano www.acsnano.org Article most likely associated with the absorbed water or impurities on the Ti 3 C 2 T x sheets. The lamellar MXene membranes were fabricated by a vacuum-filtration method of aqueous-exfoliated Ti 3 C 2 T x sheets with porous polymeric supports. 32−34 The lamellar structure of the MXene membrane was confirmed from a cross-section areal SEM image as shown in the inset of Figure 1b. As a prerequisite for the ion-selective passage, the nanoconfined interplanar channel present in lamellar MXene membranes was verified by the X-ray diffraction (XRD) analysis of fully hydrated Ti 3 C 2 T x lamellar membranes. The interlayer spacing is calculated using Bragg's law, λ = 2d sin θ, where λ is the wavelength of X-ray radiation, d is the d-spacing, and θ is the diffraction angle. As shown in Figure 1b, the diffraction peak (002) at around 5.28°for the fully hydrated membrane indicates a d-spacing of around 1.67 nm, which is larger by around 0.4 nm than that of the ambient-state membrane. This volumetric expansion in hydration is strongly correlated with water molecules accumulated in the lamellae channels possessing abundant surface charge groups. 22−26 The effective interplanar height for the fluidic motions is estimated to be around 0.69 nm, which is large enough for hydrated small ions such as K + or Na + to pass through. 15,35,36 Moreover, high optical absorptivity of the MXene membranes, essential for photothermal heating, was verified from their rapid temperature changes under light, which is further supported by the lower diffuse reflectance of Ti 3 C 2 T x films compared to that of cellulose acetate (Figure 1c). Given that the darkening effect works in the wet state, attributable to the significant decrease of light scattering from liquid to the material interface, the diffuse reflectance of wet MXene membranes can be further reduced. 37 To elucidate the charge-selective ion transport through MXene nanoconfined channels, the transverse ion flux was measured at different equimolar concentrations as well as salt concentration gradients (Figure 2a). The width, height, and length of the singular two-dimensional channel are approximately 3 mm, 0.69 nm, and 10 mm, respectively. Based on the identified interplanar spacing from XRD analysis, the effective transport area is estimated to be 3.27 × 10 −9 m 2 with a porosity of around 41.7% for a 2 μm thick membrane (see Supporting Information for details). Unless otherwise mentioned, all the ion transport measurements were conducted at a membrane thickness of 2 μm. Figure 2b shows the ionic channel conductances (G array ) of the membranes at varying concentrations, and the deviated transport from bulk transport below KCl 1 mol·L −1 confirms the dominant surface charge effect at lower salt concentration. 14−16 Furthermore, a scaling behavior is observed at low salt concentration, which is caused by salt-concentration-dependent variation of surface charges as previously reported. 38,39 The current−voltage (IV) transport was also investigated under a salt concentration gradient (c high / c low ) of 1 to 10 3 at a fixed c low of 10 −4 mol·L −1 (Figure 2c  With an increasing salt concentration gradient, the IV curves show obvious increment in both I diff and E diff . For a quantitative evaluation on the charge selectivity, a cation transference number (t + ) is calculated by the equation t + = 0.5(1 + E diff /E redox ), where E redox is a redox potential arising from the unequal chloride concentration at the two Ag/AgCl electrodes. 15,21,40 For the redox potentials, theoretically predicted values were applied over varying salt concentration gradients of 10 1 to 10 3 . 15,21 As shown in Figure 2d, it approaches 0.97 under a 10-fold difference in salt concentration, close to unity cation selectivity. Note that the transference number is defined as the fraction of the current carried by either the anion or the cation to the total electric current. 15,40 A higher transference number under a 10-fold difference is attributed to stronger cationic selectivity observed in the lower concentration regime. This feature is previously reported from graphene oxide lamellar membranes as well. 15 The salinity gradient-induced osmotic power further supports their charge-selective ion transport through the confined channels with negative surface charges. 40 To introduce a thermochemical potential across the fluidic MXene channels, localized photothermal heating was applied, creating an axial temperature gradient (Figure 3a). The array of transverse ion channels generates a parallel flux of electrical double layers under thermal potential, and their directionality is strongly associated with the light irradiation position. The ionic thermoelectric response under light is characterized with the ionic Seebeck coefficient calculated from an open-circuit voltage through ionic channels by a given temperature difference. 41 The employed polydimethylsiloxane (PDMS) layers exhibiting excellent optical transparency and thermal insulation are beneficial for suppressing the optical scattering or thermal dissipation under light illumination. Particularly, due to the encapsulated structure by PDMS, interferential factors influencing the thermoelectric response could be ruled out, such as thermal evaporative cooling and derived fluidic motion under light. 27,42,43 At a light intensity spanning a range of 74 to 127 mW·cm −2 , the temperature gradient and derived photothermal voltages were investigated across the MXene channels. The KCl concentration is fixed to 5 × 10 −4 mol·L −1 . With increasing the light intensity, a proportional open-circuit voltage (ΔV) is observed across the produced temperature gradient, yielding a positive ionic Seebeck coefficient (Figure 3b). The thermoosmotic ionic flow directing unilluminated to illuminated sides determines the magnitude and polarity of the generated opencircuit voltages. In addition, the MXene channels exhibited a prompt response under light stimulation ( Figure S3). The  (Figure 3c). When the MXene channel is locally irradiated, a stable thermal flux across the channel could be identified over varying light intensities. We ascribe this conductive thermal feature to drive the stable temperature difference across the interlayer water in the transverse cation channels. As shown in Figure 3d, the MXene nanochannels obviously illustrate a proportionate rise in the ionic thermoelectric voltages under elevated light intensity. The channels also produced the different polarity of the thermal voltages, depending on the irradiation position. A prolonged light illumination over 60 s was strictly limited, provoking an osmotic ion diffusion against the photothermal ionic flux ( Figure S4). The resulting ionic Seebeck coefficient is in the range of 0.6 to 0.9 mV·K −1 while revealing thermal enhancement. It is worth mentioning that a photoinduced charge separation can also impact the mobile-ion flux through fluidic nanochannels according to recent studies. 44−47 Particularly, in nanochannels associated with wide band gap semiconducting materials, a concomitantly created transmembrane potential by the charge separation can induce a directional and charge-selective ionic flow across the channel.
To understand the effects of photoelectronic responses of conduits on the permeating ionic flux, a Nb 2 CT x MXene showing a different thermoelectric feature is comparatively investigated ( Figure S5). The Nb 2 CT x membranes exhibit an identical lamellar configuration to Ti 3 C 2 T x membranes, and the highly ordered interlayer spacing was confirmed from the fully hydrated membranes. Furthermore, the Nb 2 CT x membranes demonstrate comparable optical absorptivity to that of Ti 3 C 2 T x . Static thermoelectric measurement of both Ti 3 C 2 T x and Nb 2 CT x reveals their contrasting response under a temperature gradient ( Figure S6). Given that the Ti 3 C 2 T x MXene behaves as a metal-like conductor, 48 such an opposite response from the Nb 2 CT x membrane suggests the dominant presence of a hole charge carrier. The majority carrier in the Nb 2 CT x sheets could be reconfirmed by Hall effect measurements as well as Seebeck coefficient measurements. However, under light illumination, the Nb 2 CT x channel shows no discrepancy in magnitude and polarity of the ionic Seebeck coefficient compared to those of Ti 3 C 2 T x membranes ( Figure  S7). Furthermore, both species Ti 3 C 2 T x and Nb 2 CT x exhibit the same directional flux of cations even under a direct joule-heating-induced temperature gradient ( Figure S8). Hence, we postulate that the photothermal ion transport is dominantly governed by the temperature-gradient-derived chemical potential across the MXene fluidic channels.
More interestingly, the charge selectivity of the ion channels can play a central role in regulation of trans-nanochannel photothermal potential (ϕ m ) under a temperature difference, according to previously suggested nonisothermal ion transport models as 11,29 t where R is the gas constant, F is the Faraday constant, and T i and c i are, respectively, the temperature and the salt concentration at designated reservoirs. To experimentally discern the relationship between cation permselectivity and ionic Seebeck coefficient, we explored the photothermal voltages at various KCl concentrations and under regulated pH. The intensity of irradiated light is fixed to 100 mW·cm −2 while maintaining equimolar concentrations between two electrolyte reservoirs. With decreasing salinity, the ionic Seebeck coefficient obviously exhibits an incremental performance, reaching 1.0 mV·K −1 at a concentration of 10 −4 mol·L −1 (Figure 4a). The Seebeck coefficient, which is inversely proportional to the concentration, can be understood as a consequence of enhanced charge selectivity at lower concentration, as proven earlier. This salinity-related thermoelectric response is previously reported from commercially available cation exchange membranes ( Figure S9). 41 Note that the narrow interplanar channels can exert a compression effect or dehydration on the potassium ions having a similar hydration shell size with the effective channel height. 15,49,50 This implies that for larger cations than a nanochannel, the steric effect with increased hydrated sizes can limit the cationic permeation through the nanochannels, producing lower photothermal voltages. Meanwhile, at an acidic condition of pH 3.34, the suppressed photothermal voltage was observed ( Figure S10). When the pH is decreased, the protonation of the terminal groups on MXene sheets leads to reduced negative surface charges, accumulating less counterions on their interplanar spacing. 14,51 The fewer cations inside channels and the associated decrease of cationic selectivity consequentially yield a lower ionic Seebeck coefficient. The ionic photothermal response is further investigated in the presence of a salt concentration gradient as shown in Figure 4b and c. When the light is irradiated on the membrane near the diluted side, the thermo-osmotic ionic flux from cold ACS Nano www.acsnano.org Article to hot is aligned to the direction of self-diffusive ion flux, as shown in the inset. The ionic Seebeck coefficient from the interpolated output voltages is estimated to be 0.81 mV·K −1 . The observed higher ionic Seebeck coefficient under a salt gradient, taking the applied higher concentration into account, is most likely due to the self-diffusion-driven, lowered energy barrier for cation passage. 52 Additionally, the photothermoelectric response was stable and reproducible on a 1000 s time scale. For practical purposes, we devised a photothermosensitive system with a large-areal transmembrane in contact with an ionic gel electrolyte ( Figure S11). Under light illumination, a stable heat transfer through a quasi-solid-state gel creates a stable temperature gradient across the membrane. The device, despite a limited temperature change across the membranes, obviously displays a light intensity-sensitive response as well as reproducible thermal voltages.

CONCLUSION
In summary, we have employed subnanometer confined ion conductors in lamellar Ti 3 C 2 T x MXene membranes as a bionic thermosensitive platform. The nanoconfined channels covered by an electrical double layer lead to highly charge-selective ion diffusion under the electrochemical potential. Coupled with the inherent photothermal properties of MXenes, the cationselective transport by localized photothermal stimuli is converted to a trans-nanochannel diffusion potential, corresponding to open-circuit voltage over a temperature gradient. A thermoelectric response of up to 1.0 mV·K −1 was obtained under one sun illumination, comparable to that of biological thermosensory channels. Such photothermoelectric responses show not only obvious dependency on the cationic permselectivity of nanochannels, as anticipated from the theoretical consideration, but also its high sensitivity even under a tiny thermal gradient of <1 K. Moreover, the MXene channels exhibited stable and reproducible response in the absence and presence of an ionic concentration gradient. Our work advances a fundamental understanding of the photothermo-osmotic ion transport in MXene, providing an avenue for the application of MXene membranes in biomimetic sensory systems.

EXPERIMENTAL SECTION
Fabrication of MXene Lamellar Membranes. Ti 3 C 2 T x and Nb 2 CT x MXene were synthesized as previously reported in our work. 53 Specifically, the MAX phase Ti 3 AlC 2 etchant was prepared by the method in which 1 g of lithium fluoride (LiF 98%, Alfa Aesar) was slowly added into 20 mL of 9 mol·L −1 hydrochloric (HCl 35−38%, Fisher Scientific) and stirred at 500 rpm for 5 min. All chemicals were used as received without further purification. The 1 g of Ti 3 AlC 2 powder (98 wt %, 400 mesh, Laizhou Kai Kai Ceramic Materials Co., Ltd.) was immersed in the as-prepared etching solution at 40°C for 24 h. The slurry was centrifuged and rinsed several times using deionized water until the pH of the supernatant was around 6. Then, the sediment was intercalated by 10 mL of 1 mol·L − 1 tetramethylammonium hydroxide (TMAOH) at room temperature for 24 h and centrifuged three times at 11 000 rpm for 10 min. After intercalation, the dark green Ti 3 C 2 T x MXene suspension can be collected after 5 min of centrifugation at 3500 rpm using deionized water. The experimental procedure for the synthesis of Nb 2 CT x MXene is the same as above, except that the raw material Nb 2 AlC was etched by 20 mL of 49% hydrofluoric acid. The lamellar MXene membranes were fabricated by filtering specific amounts of an MXene dispersion through a polyvinylidene fluoride (PVDF) membrane (0.22 μm pore size and a diameter of 43 mm). All filtrated membranes were ambient-dried overnight and could be easily detached from the support.
Ion Transport Measurements. The fluidic cell fabricated with a polydimethylsiloxane elastomer was custom-made for the ion transport measurement. The MXene lamellar membranes trimmed into rectangular pieces were embedded in the PDMS prepolymer and curing agent, and after curing the elastomer, two electrolyte reservoirs were carved out to expose two ends of the PDMS-immersed MXene membranes to the electrolyte solution. The MXene nanofluidic devices were then immersed in electrolytes for at least 1 day for fully hydrating the fluidic channels. The ionic transport measurement was performed by taking the current−voltage characteristics across the trans-nanochannels at room temperature, using a Keithley 2400 sourcemeter. The measurements were performed at various KCl concentrations of 10 −4 to 1 mol·L −1 . A pair of Ag/AgCl electrodes was employed to measure the ionic transport across membranes. For specific measurement under the concentration gradient, a pair of commercial salt-bridge electrodes (CHI111P, CH Instruments, Inc.) was applied. The Ag/AgCl electrodes in 1 mol·L −1 KCl that contact the electrolyte reservoir via porous Teflon salt bridges can eliminate the potential generated from the redox reaction on the electrodes under unequal ionic strength. 19,54 For the photothermal reaction, the MXene membranes were irradiated with a Newport 94043A solar simulator with an optical filter for the standard AM 1.5G spectrum, and resulting temperature changes over the MXene membranes were captured by an infrared camera (A600-series, FLIR). Under localized light illumination, the open-circuit voltages corresponding to photothermoelectric ionic voltages were measured across the MXene nanochannels. The thermoelectric ion transport was further investigated using localized Joule heating over the MXene nanofluidic devices. Carbon nanotube−poly(vinyl alcohol) composite films were applied as a Joule heater electrically connected with a DC power supply. 55 Solid-State Thermoelectric Transport Measurement. The static thermoelectric transports of Ti 3 C 2 T x and Nb 2 CT x are comparatively studied using a closed circuit wherein the MXene membrane is serially connected with the Keithley 2400 sourcemeter. Temperature control over the MXene channels was carried out by using a commercial thermoelectric generator module. Under local heating or cooling, the thermoelectric current by diffusion of thermally activated majority carriers was measured across the MXene membranes. More precisely, the Seebeck coefficient of the free-standing MXene films (1 × 1 cm 2 ) was tested in the in-plane direction by a commercial thermoelectric measurement system (RZ2001i, Ozawa Science Co. Ltd.). During the measurement, a dynamic flow of 4% H 2 /Ar was supplied to obtain a reducing surrounding. The conductivity was measured by the four-probe method under thermal equilibrium in each temperature. Hereafter, a small temperature difference along the sample channel was created by local cooling at one side of the sample. Two inner probes are also the Pt thermometers; therefore, the Seebeck voltage (ΔV) and the temperature gradient (ΔT) could be measured simultaneously. According to the formula S = ΔV/ΔT, the Seebeck coefficient could be obtained. In each temperature, five different data points were collected and analyzed. The details were shown in our previous work. 56 Hall Effect Measurement. Carrier concentration and Hall mobility of the MXene membranes were measured by a Hall-effect measurement system (7700A, Lake Shore) using the van der Pauw technique for 1 × 1 cm 2 samples at RT under a magnetic field range of ±5 to ±20 KG by linear sweep with field reversal mode. Ohmic contact was achieved by fast-drying silver paste (Ted Pella) at the four corners of the sample with lead wires from a printed circuit board for the measurement.
Characterizations. A field emission scanning electron microscope (Merlin, Carl Zeiss) was used to characterize the microstructures of the MXene lamellar membranes. The XRD analysis was conducted by using Bruker D8 Advance with Cu Kα radiation (λ = 0.154 06 nm); the step scan was 0.02°over a step time of 2 s. The diffuse reflectance from 200 to 800 nm wavelength was measured using ultraviolet visible ACS Nano www.acsnano.org Article absorption spectroscopy (Cary 5000, Agilent). The zeta potential of the MXene suspension was measured over a range of pH with a Malvern Zetasizer Nano ZS.
Morphologies of the exfoliated Ti 3 C 2 T x MXene sheets, determination of nanochannel dimensions in lamellar MXene membranes, photothermal response of the MXene cation channels, effect of prolonged light irradiation on the photothermal voltage generation, Nb 2 CT x MXene-guided ion channels, solid-state thermoelectric transport of Ti 3 C 2 T x and Nb 2 CT x MXene membranes, photothermal response of Nb 2 CT x MXene membranes, ionic transport under Joule-heating-driven temperature gradient, comparative study with commercially available cation exchange membranes, photothermal voltage generation at lower pH, photothermosensing device (PDF)