Covalent organic framework nanofluidic membrane as a platform for highly sensitive bionic thermosensation

Thermal sensation, which is the conversion of a temperature stimulus into a biological response, is the basis of the fundamental physiological processes that occur ubiquitously in all organisms from bacteria to mammals. Significant efforts have been devoted to fabricating artificial membranes that can mimic the delicate functions of nature; however, the design of a bionic thermometer remains in its infancy. Herein, we report a nanofluidic membrane based on an ionic covalent organic framework (COF) that is capable of intelligently monitoring temperature variations and expressing it in the form of continuous potential differences. The high density of the charged sites present in the sub-nanochannels renders superior permselectivity to the resulting nanofluidic system, leading to a high thermosensation sensitivity of 1.27 mV K−1, thereby outperforming any known natural system. The potential applicability of the developed system is illustrated by its excellent tolerance toward a broad range of salt concentrations, wide working temperatures, synchronous response to temperature stimulation, and long-term ultrastability. Therefore, our study pioneers a way to explore COFs for mimicking the sophisticated signaling system observed in the nature.


Permselectivity evaluation
For investigating the ion transport property of nanochannels, the ion current was recorded by CHI660E with a homemade electrochemical cell. The voltage was scanned with a step of 0.01 V s -1 using Ag/AgCl electrodes. The K + /Clselectivity of TpTag-COF/PAN was evaluated by determining the reversal potentials with different KCl concentration gradients. The X-intercepts (Vr) of the I−V plots represent the average reversal potentials. The permselectivity was determined according to the Goldman-Hodgin-Katz equation. The permselectivity of other electrolytes were evaluated following the same procedures as that of KCl.

Thermoelectric response of TpTag-COF/PAN
The thermoelectric response was recorded by a homemade setup. A micro-ceramic heater (Zhuhai Huiyou Electronics, China) was employed to regulate the solution's temperature. A direct current power (HSPY-60-5, Hanshengpuyuan, China) was connected to the heater to control the heating rate and temperature range. A pair of temperature microsensors (PT100, Tenghui Wenkong Instruments, China) were immersed in two solutions to measure the real-time temperature, which was recorded by a temperature sensor (THMA temperature recorder, Tenghui Wenkong Instruments, China). The measurement accuracy and working range of the temperature sensor are ± 0.1 K and −173.15 − 473.15 K, respectively. The transmembrane potential was synchronously collected by a CHI660E electrochemical workstation using two Ag/AgCl electrodes. The time resolutions of temperature recorder and transmembrane potential are both 1 s.

Theoretical Derivation of Thermoelectric Response
The schematic model shown above is used to derive the theoretical thermoelectric response of TpTag-COF/PAN. Two chambers, designated as cis and trans, contain systematic electrolyte (KCl) with an activity of and . In this work, KCl was used as the electrolyte. Therefore, the subscripts, + and -, refer to K + and Cl -, respectively. Accordingly, the temperature of the solutions is and .
Given a quasi-steady state of ion transport through the membrane from one solution to another, the following equation can be considered, According to the variation of Gibbs free energy of a system ( ), Where , , , , and are the Gibbs free energy, entropy, temperature, volume, and pressure of the solution. and are the chemical potential and the chemical amount of species . Therefore, a temperature change will drive the transport of ions in the opposite direction of the temperature gradient.
In the cis or trans chamber, the system is held at constant temperature and pressure, is presented as follows, Where is the molar number of species i. , ( = , ) is the chemical potential of ions, which is given by, Where , ⊖ , , , and are the standard chemical potential, the activity, and the charge valence of ions in chamber. and are the gas constant and the Faraday constant, respectively. ф is the inner potential of chamber. is the absolute temperature.
Note that the variation of the activity with temperature is neglected here, due to a small temperature change (10 K, maximum) in this work. Considering the electroneutrality condition in two solutions of From above equations, the following one can be derived, Then we can define the transmembrane diffusion potential (ф ) as, Ion transport number, also called the transference number, is the fraction of the total electrical current carried in an electrolyte by a given ionic species , For a steady flow of charge through a surface, the current can be calculated with the following equation, where is the electric charge transferred through the surface over a time , is the charge valence of ions, is the elementary charge, is Avogadro constant, and is the molar number of species i.
From above two equations, we can get the following equations, According to Supplementary Equation 12, ф is a function of both temperature and salt activity in two solutions.
The schematic model b shows the equivalent circuit of experimental system, in which the open-circuit potential ( ) is measured with two silver/silver chloride (Ag/AgCl) electrodes immersed in two solutions. It is the sum of thermoresponsive transmembrane potential (ф , the direction from cis to trans is defined to be positive), the difference of redox potentials of two AgCl/Ag electrodes ( ) and the voltage drops across the membrane ( ) and in solution ( ), where is the ionic current, is the internal resistance of nanochannels membrane, and is the solution resistance. Considering that the ionic current measured is close to 0, we can reasonably neglect the contribution of drop. arises from the dependence of AgCl/Ag electrode potential on the activity of Cland temperature, Combining Supplementary equations 12, 13, and 14, we obtain the following equation, In this work, we studied the thermoelectric response for three cases, in the absence and presence of activity gradient across nanochannels. In the first case, there is no activity gradient across nanochannels (namely = = ). Both ф and are equal to zero at the initial state ( = ). Upon changing the temperature of one solution (the temperature of another solution remained unchanged), they will vary and the magnitudes can be derived from Supplementary  equations 12 and 15, respectively, where ∆ ( ∆ = − ) is the magnitude of temperature change, namely the immediate temperature difference between two solutions separated by nanochannels.
In the second case, when the temperature of solution in cis chamber was changed (the temperature of solution in trans chamber remained unchanged), both ф and will change in the same way.
In the third case (namely ≠ ), there exist nonzero ф and at the initial state. If assuming the initial temperature of two solutions is 0 , they are associated with the activity gradient and expressed as, In the similar way, when the temperature of solution in trans chamber was changed (the temperature of solution in cis chamber remained unchanged), both ф and will change. Their net variations relative to the initial state are dependent on both temperature and activity gradients and can be expressed as, Where is the electrolyte activity in the solution where the temperature changes.

Molecular Dynamics Simulation
For the simulation of the ion permeation through TpTag-COF, three layers of 3.78 nm × 3.27 nm TpTag-COF (4x4x3) were sandwiched by two solutions along the z-direction. One contains 2000 water molecules, 200 K + ions, and 200 Clions, and the other contains 2000 water molecules. A periodic boundary condition was applied to the x-y direction, and the length of the z-axis of the simulation box was 12.0 nm. TpTag-COF was optimized by the Universal Force Field (UFF) force field with the charge equilibrium (QEq) method 1 . The Simple Point Charge (SPC) model was used to describe water molecules 2 . The Optimized Potentials for Liquid Simulations All Atom (OPLS-AA) model was used to describe K + ions and Clions 3 . All MD simulations were performed by the package of Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) and visualized by OVITO software. The atoms of TpTag-COF were frozen during the simulation. The system was progressed to reach the energy minimization using the steepest descent approach before a 1 ns NVT simulation was performed (NVT: constant particle number, volume, and temperature). The initial velocities of K + , Cl -, and H2O were assigned based on the Maxwell−Boltzmann distribution at 300 K and maintained by the Nose-Hoover thermostat. The simulation was run with a time step of 1 fs. where and 0 are the vacuum and relative permittivity, respectively, is the Boltzmann constant, is the absolute temperature, is the elementary charge, is the Avogadro number, and is the ionic strength of the solution.