Selective Deionization of Thin-Layer Samples Using Tandem Carbon Nanotubes–Polymeric Membranes

Herein, we investigate the selective deionization (i.e., the removal of ions) in thin-layer samples (<100 μm in thickness) using carbon nanotubes (CNTs) covered with an ionophore-based ion-selective membrane (ISM), resulting in a CNT-ISM tandem actuator. The concept of selective deionization is based on a recent discovery by our group (Anal. Chem.2022, 94, (21), , 7455−745935579547), where the activation of the CNT-ISM architecture is conceived on a mild potential step that charges the CNTs to ultimately generate the depletion of ions in a thin-layer sample. The role of the ISM is to selectively facilitate the transport of only one ion species to the CNT lattice. To estimate the deionization efficiency of such a process, a potentiometric sensor is placed less than 100 μm away from the CNT-ISM tandem, inside a microfluidic cell. This configuration helped to reveal that the selective uptake of ions increases with the capacitance of the CNTs and that the ISM requires a certain ion-exchanger capacity, but this does not further affect its efficiency. The versatility of the concept is demonstrated by comparing the selective uptake of five different ions (H+, Li+, Na+, K+, and Ca2+), suggesting the possibility to remove any cation from a sample by simply changing the ionophore in the ISM. Furthermore, ISMs based on two ionophores proved to achieve the simultaneous and selective deionization of two ion species using the same actuator. Importantly, the relative uptake between the two ions was found to be governed by the ion–ionophore binding constants, with the most strongly bound ion being favored over other ions. The CNT-ISM actuator concept is expected to contribute to the analytical sensing field in the sense that ionic interferents influencing the analytical signal can selectively be removed from samples to lower traditional limits of detection.

Instrumentation to control the actuator-sensor system.Electrochemical experiments were performed using a PGSTAT302 Autolab potentiostat (Metrohm Nordic AB) operated using the Nova 2.1.5software on a PC.The EMF for the potentiometric sensors was recorded using a high input impedance (1015Ω) EMF16 multichannel data acquisition device, Lawson labs EMF16 Interface (Lawson Laboratories, Inc.).The pH was measured using a 914 pH/Conductometer from Metrohm (6.0228.000).For screen printed electrodes, either a Dropsens DRP150 was used for mono-ionophore based ISMs (actuators and potentiometric sensors), or a Dropsens DRP1110 electrode for two-ionophore experiments (potentiometric sensors).

Calculations and additional comments on experimental results
Estimated time for ion uptake.The time of diffusion  over a certain distance  in one dimension can be estimated by Equation S1.
Where  is the diffusion coefficient of the phase.The diffusion coefficient is be estimated to be ca.10 -5 cm 2 /s for an aqueous phase, 1 and ca. 10 -8 cm 2 /s for a membrane phase. 2Using Equation S1 for a thin-layer sample (aqueous) of 75 µm, the time of diffusion for an ion through it becomes 2.8 s.For the membrane phase used in this this work (ca 200 nm of thickness) the time becomes 20 µs.Thus, diffusion through the membrane phase is not expected to limit the rate of the ion uptake process.
Differential Capacitance measurement and The Point of Zero Charge.The point of zero charge (PZC) was studied between -0.8 and 0.8 V in acetonitrile with TBAPF6 as background electrolyte with the signal amplitude of 10 mV (Figure S6).The capacitance Cd was then calculated from the angular frequency (2, where f is the frequency; 15 Hz) and the imaginary component of the impedance, given in Equation S2.
This revealed the U-shape curve observed in Figure S6 which is characteristic for a purely capacitive mechanism. 3e PZC indicates the potential where the surface charge is zero, where at higher-and lower potentials the surface charge is positive and negative, respectively. 3This gives an indication that the surface charge is not limited for negative polarization for storage of cations in the double-layer, but indicates the possibility to charge the CNTs positively for anion uptake.This has indeed been utilized without membranes in desalination technologies aimed for drinking water applications, 4 and could likely be explored in the future for selective uptake of anions in analytical applications using ISMs based on anion-selective ionophores (such as for Cl - uptake).
The energy of ion-transfer.The energy for transferring an arbitrary ion from one phase [1] to another [2] is given by Equation S3. 5 where Δ + is the difference in solvation energy between the two phases, z is the valence charge, e the elementary charge,  1 the permittivity of free space, a the ion radius and  ' and  $ the dielectric constants of the respective phases.
Examining Equation S3 in the case of ion-transfer from water to the ISM phase, the energy of transfer is positive ( ' >  $ , as water has a higher dielectric constant than DOS and PVC) 5 .Additionally, the energy of transfer increases by the square of the charge and increase by ' 0 for decreasing ion-radius.Thus, the energy of transferring Ca 2+ and Li + into the CNTs are expected to be larger than that of larger, monovalent ions (such as K + and Na + ). S-4

Figure S6 .
Figure S6.Differential capacitance measurements of the CNTs in 0.1 M TBAPF6 / acetonitrile solution.The measurements were conducted at 15 Hz, from 0.8 to -0.8 V vs the Ag/AgCl pseudo reference electrode.At this frequency, the contribution of any faradaic processes is negligible within the potential window (i.e., only capacitive contributions are expected in the differential capacitance).

Figure S7 .
Figure S7.Examples of the dynamic uptake of Ca 2+ for different concentrations of the cation exchanger in the membrane.The gray area represents the time over which the potential was applied to the actuator (-0.4 V versus the OCP, 120 s).Experiments were performed in 1 mM CaCl2 solution with 10 mM MgCl2 as the background electrolyte.

Figure S8 .
Figure S8.Uptake of Li + and Ca 2+ using an ISM with two ionophores (M XIII) at an applied potential of a) -0.2 V versus the OCP and b) -0.4 V versus the OCP.The sample was 1 mM LiCl/1 mM CaCl2 with 10 mM MgCl2 as the background electrolyte.

Table S1 .
Membrane compositions.All the membranes were prepared with 1:2 mass ratio of PVC:DOS.

Table S2 .
Calibration parameters obtained from the different potentiometric sensors, where the corresponding sensitivities and intercepts are given as the average with standard deviation for seven different sensors (n=7).
a mV/pH,