The cavity within the interior of the valinomycin fits potassium ions, and the ions are held there due to their interactions with the ester carbonyl oxygen atoms of the valinomycin molecule. This is what provides selectivity of valinomycin to potassium and allows the reversible exchange of potassium ions with the analyte solution via complexation between potassium and valinomycin. If the ionophore is, in fact, a neutral-carrier ionophore, then it is necessary to also include a hydrophobic counter ion to the primary ion for selectivity. Ion-to-electron transducer layers, hereafter shortened to ‘transducer layer’, is the component in an ISE that is responsible for arbitrating the build-up of electrical charge or potential from a concentration of ions. There are two broad types of transducer layers. The first is transducer layers that operate on redox activity, and the second is transducer layers that operate on a capacitor-like electric double-layer potential. Transducer layers that operate on redox activity have a liquid or solid electrolyte that separates the ISM from the conductor. Conventional ISEs have a liquid separa-tor that performs ion-to-electron transduction by means of an electrochemical reaction. For example, the ‘conventional’ ISE architecture is an Ag/AgCl wire in a glass tube filled with a fixed concentration of Cl−inner-filling solution that is in contact with the ISM. A schematic of a conventional potentiometric ISE sensor is shown in Figure 4.2. For all-solid-state ISEs without redox properties, ion-to-electron transduction is the result of the electrical double layer forming at the ISM/transducer interface. This interface can be schematically described as an asymmetrical electrical capacitor, in which one side carries a charge in the form of ions, i.e., cations and anions from the ion-selective membrane,drainage pot and the other side is formed by an electrical charge, i.e., electrons or holes in the solid contact.

When a transducer layer is not used and the ISM is coated directly onto the conductor, ion-to-electron transduction behaves similar to a redox-free transducer layer where the charge is dependent on the quantity of charge in the electric double layer. However, sensor drift and erratic responses are more prevalent in ISEs without a transducer layer because a deleterious water layer can form more easily between the ISM and conductor, which will be discussed in greater detail later in Section 4.4. Also, generally, there is a smaller contact area between an ISM/conductor interface compared to a transducer/conductor interface because of the nature of the materials used in each. As we discussed earlier, the surface area of this interface is critical for ion-to-electron transduction. Finally, the ISE can become polarized by the small currents associated with potentiometric measurements. The salt bridge separates the transducer and the bulk solution that is being sampled. In practice, the salt bridge is often contained in nanoporous glass , polymer membrane, hydrophilic gel, or capillary. In conventional REs, a glass frit is most commonly used as the salt bridge. The salt bridge must allow electrical contact between the bulk solution and the Ag/AgCl by ionic conductivity with relatively low resistance. In the case of many solid-state REs, the salt bridge is a salt-loaded polymer membrane, hereafter simply called the salt membrane. The salt membrane of a RE behaves like the salt bridge of an electrochemical cell. It is therefore of utmost importance to load the salt membrane with a surplus of salt so that the RE reaction occurs spontaneously. In the case of the Ag/AgCl electrode, the solvated polymer membrane is doped with KCl, NaCl, or another chloride salt such that a surplus of Cl−anions are available for the reaction at the Ag/AgCl electrode. Because of the solubility of these salts, there is a slow diffusion of the membrane-phase salt into the bulk aqueous solution, meaning that over time, Cl−ions will slowly leech into the sample that is being measured. To characterize the sensitivity of the printed nitrate ISE, its potential was measured against a commercially-available conventional Ag/AgCl electrode in varying concentrations of NaNO3, shown in Figure 4.6A.

Figure 4.6B shows the potential over time for one ISE measured against glass commercial reference electrode in nitrate solutions between 20 mM and 0.05 mM. This ISE reported a stable value after about 30 seconds after a change in concentration. The data from 4.6B can alternatively be plotted versus nitrate concentration on a log scale, as shown by the blue circles in figure 4.6C. The other lines in 4.6C represent the sensitivity for six other ISEs in three batches. The average sensitivity for all seven sensors is -54.1 ± 2.1 mV/decade. The linear range of these sensors was found to be between 0.05 mM and 100 mM. This range is equivalent to 3.1 to 6,200 ppm NO− 3 or 0.7 to 1,400 ppm . This is in good agreement with other nitrate ISEs in literature, which typically exhibit a range of 10−6 M – 10−1 M. Concentrations of nitrate in agricultural fertilizer vary widely depending on crop and soil type as well as fertigation technique, but a few 100’s of ppm would be a high nitrate concentration in fertilizer. In the United States, the Environmental Protection Agency’s drinking water quality standards specify a maximum of 10 ppm NO− 3 , and some studies have shown an increased risk of certain health conditions for water with 5 ppm NO− 3 or greater. The sensors presented here cover concentrations from drinking water to concentrated fertilizer. In Figure 4.6C, the sensitivity curves for different sensors are offset one from another. This offset is the result in different values for E 0 ∗ ISE,i, which is the sum of standard potentials for all boundaries in the ISE. This variation in E 0 ∗ISE,i is common in ISEs and means that each sensor must be individually calibrated before use. E 0 variation has a variety of causes, many of which are summarized in Hu et al.. Properly, E 0 is the potential at ion activity of 1, which is outside of the linear range of the sensors. E 0 values presented here are calculated using the potential at 1 mM NO− 3 concentration. Within one batch of ISEs, the E 0 variation was found to be 12.5 mV. The measurements for one batch were done with each ISE paired with one of five different commercial REs. While nominally identical, the standard potential of these five commercial REs was compared in 1 M KCl solution and was found to vary by up to 11 mV different from each other. This difference in commercial RE performance is consistent with E 0 values obtained within a batch of ISEs.

The batch-to-batch variation is 83 mV over six batches. This significant variation may be due to variation in the membrane drying conditions, sections of crystallized PVC in the membranes, or other minor effects. Another reason for mismatched standard potential is when different materials are used as a conductor. Because the standard potential is calculated as the sum of all standard boundary potentials between the ISE and RE,growing raspberries containers changing the conductor layer material causes an offset in boundary layer potential between the transducer layer and the conductor. Figure 4.7 shows the impact that changing the conductor material has on the overall sensitivity and standard potential of the nitrate ISE. All nitrate ISE conductors were made using the same pattern described in the fabrication section. LIG conductors were prepared according to literature. Carbon conductors were screen printed with Creative Materials 114-34A/B187 solvent-resistant carbon ink. The silver conductors were screen printed with Creative Materials 127-07 extremely conductive ink.REs act as an electrochemical ground, therefore their potential must remain unchanged in varying ionic environments. The precise composition of the printed RE will impact E0 in the Nernst equation. However, because E0 is constant, the offset is easily accounted for in calibration. The performance of printed REs was determined by measuring them versus a commercial Ag/AgCl double junction RE, as in Zamarayeva et. al, and illustrated in Figure 4.9A. First, pristine printed Ag/AgCl electrodes were measured, and the resulting data is shown in Figure 4.9B. The output voltage is unstable since these printed REs lacks a source of chloride ions, which are needed for the Ag/AgCl reversible reaction that keeps the reference potential stable. The surface area and composition of the printed RE were modified by adding a CNT layer and a PVB-NaCl membrane. The characterization is shown in Figure 4.9C. These electrodes used the formulation developed in Zamarayeva for use in chloride-rich environments. REs with a NaCl membrane showed a -18 mV/decade sensitivity to nitrate. The optimized RE composition was achieved with the addition of NaNO3 tothe PVB-NaCl membrane. Cattrall and Zamarayeva et al.have shown that including the ion of interest in the membrane of a RE reduces its sensitivity to that ion. To reduce sensitivity to nitrate, NaNO3 was needed in the membrane; sensitivity data for this electrode is shown in Figure 4.9D. This formulation has a sensitivity of -3 mV/decade, which is a marked improvement over the NaCl membrane alone. The effect of adding the ion of interest to the reference electrode membrane is highlighted in Figure 4.9E, where the NaCl membrane and NaCl+NaNO3 membranes are directly compared. In this figure, potentials are normalized by subtracting the average potential in 1 mM nitrate from the average potential at each concentration, and the potential offsets are plotted versus concentration. The RE whose membrane includes NaCl+NaNO3, represented by red triangles, has a flatter slope which reflects its insensitivity to nitrate concentration. Repeatability across different reference electrodes is shown in Figure 4.9F where voltage vs concentration for five printed REs with the NaCl + NaNO3 + PVB membranes is displayed.

All the printed REs showed stable potential response over three orders of magnitude change in the nitrate concentration.Pairing the printed ISE with a printed RE results in a fully printed sensor that realizes the full benefits of printing: low cost, high-throughput manufacturing, no glass or liquid components, and production in form factors that are suitable for use in field deployments. Figure 4.10A shows the potential over time for a printed ISE measured against a commercial reference in blue and that same ISE paired with a printed reference in green. The E0 value changed, which was expected because the interfaces present in a printed RE are different from those of a commercial RE. For this sample, the fully printed senor’s potential was about 87 mV below the printed ISE-commercial RE pair. Both versions have high sensitivity greater over the range 0.1 mM to 100 mM, response times less than 10 seconds, and hysteresis less than 5%. The sensitivity of these ISEs, when measured against glass REs, were -54.3 ± 2.6 mV/dec, which is near Nernstian and comparable to other nitrate ISEs in literature, as shown in Table 4.4. Printed pairs have a sensitivity of 48.0 ± 3.3 mV/decade for n=4 sensors. The sensitivity of the four sensors, from two batches, is shown in figure Figure 4.10B. The sensitivity of fully printed pairs was about 4 mV/decade less than the sensitivity of ISEs measured against glass references, owing to the slight sensitivity of the printed references themselves to nitrate. Again, E0 variation is considerable, particularly from batch to batch. This is expected given the batch-to batch variability of the ISEs and the sample-to-sample variation of printed REs.Fully printed sensors were measured in high organic matter soil from a field site in California. Six small pots of soil were prepared, and each was watered to saturation with a different concentration of KNO3 solution. The printed sensors were inserted into each pot in turn, and the potential was recorded. Actual NO3 concentration– including background NO− 3 already present in the soil prior to watering and was measured using standard techniques. The relationship between the sensors’ potential and the log of the concentration of nitrate is linear with R2 values of 0.98, 0.99, and 0.87. The average sensitivity is -47 mV/decade, which is remarkably close to their sensitivity in an aqueous solution. These results are promising for the future application of printed ISEs in soil media and will be explored further in Chapter 5. Perhaps the most common cause of drift in ISEs is the formation of an undesired water layer.