New Potentiometric Screen-Printed Platforms Modified with Reduced Graphene Oxide and Based on Man-Made Imprinted Receptors for Caffeine Assessment

Caffeine is a psychoactive drug that is administered as a class II psychotropic substance. It is also considered a component of analgesics and cold medicines. Excessive intake of caffeine may lead to severe health damage or drug addiction problems. The assessment of normal caffeine consumption from abusive use is not conclusive, and the cut-off value for biological samples has not been established. Herein, new cost-effective and robust all-solid-state platforms based on potentiometric transduction were fabricated and successfully utilized for caffeine assessment. The platforms were modified with reduced graphene oxide (rGO). Tailored caffeine-imprinted polymeric beads (MIPs) based on methacrylic acid (MAA) and ethylene glycol dimethacrylate (EGDMA) were prepared, characterized, and used as recognition receptors in the presented potentiometric sensing devices. In 50 mM MES buffer, the sensors exhibited a slope response of 51.2 ± 0.9 mV/decade (n = 6, R2 = 0.997) over the linear range of 4.5 × 10−6–1.0 × 10−3 M with a detection limit of 3.0 × 10−6 M. They exhibited fast detection of caffeinium ions with less than 5 s response time (<5 s). The behavior of the presented sensors towards caffeinium ions over many common organic and inorganic cations was evaluated using the modified separate solution method (MSSM). Inter-day and intra-day precision for the presented analytical device was also evaluated. Successful applications of the presented caffeine sensors for caffeine determination in commercial tea and coffee and different pharmaceutical formulations were carried out. The data obtained were compared with those obtained by the standard liquid chromatographic approach. The presented analytical device can be considered an attractive tool for caffeine determination because of its affordability and vast availability, particularly when combined with potentiometric detection.


Introduction
Caffeine (1,3,7-trimethyl xanthine) is an alkaloid, commonly used as an ingredient in many foods, beverages, and medicines. It is widely consumed as a psychoactive substance and is commonly found in coffee, tea, chocolate, cocoa, and soft drinks. Depending on the dosage, caffeine can have positive or harmful effects on the consumer [1,2]. Caffeine is a psychoactive drug that is administered as a class II psychotropic substance. It is also types of platforms have been successfully used for rapid analysis of both environmental pollutants [54,55] and biomedical molecules [37,56].
Molecularly imprinted polymers (MIPs) have seen a continuous development as sensing elements in bio-/chemo-sensors since the late 1990s [57]. MIPs are attractive not only for their recognition properties that are close to those of natural receptors and their availability for a wide range of targets but also for their superior chemical and physical stability compared to biological receptors. The field of molecularly imprinted polymer (MIP)-based chemosensors has been experiencing constant growth for several decades [58][59][60]. Since the beginning, their continuous development has been driven by the need for simple devices with optimum selectivity for the detection of various compounds in fields such as medical diagnosis, environmental and industrial monitoring, food, and toxicological analysis, and, more recently, the detection of traces of explosives or their precursors [61,62].
In this work, cost-effective, compact, and portable monitoring potentiometric sensors were manufactured. These sensors have been successfully applied to determine caffeine in different matrices. All-solid-state screen-printed potentiometric electrodes were fully designed, characterized, and proposed. The potentiometric analytical device integrates the indicator polymeric membrane caffeine-ISE with an Ag/AgCl reference electrode and a polyvinyl butyral (PVB) reference membrane. All potentiometric performances of the fabricated screen-printed sensors were investigated and evaluated.
A stock 1.0 × 10 −2 M caffeine solution was prepared by dissolving the definite weight in 100 mL de-ionized water. All caffeine standard solutions (e.g., 1.0 × 10 −7 -1.0 × 10 −2 M) were prepared from the stock caffeine solution after dilution with 50 mM MES buffer solution of pH 5.0. All solutions were stored in the refrigerator.

Preparation of the Imprinted Beads
The imprinted polymers were prepared by mixing 0.5 mmol of caffeine (template) with 3.0 mmol of MAA (functional monomer) in a glass tube and left together for 1 h. 3.0 mmol of EGDMA (cross-linker) was then added and followed by adding 0.3 mmol of BPO (initiator). 15 mL of acetonitrile (porogenic solvent) was added to the mixture followed by sonication for 10 min until complete dissolution of all components. The solution was degassed with Polymers 2022, 14,1942 4 of 14 N 2 gas for 15 min to expel all dissolved oxygen. The solution mixture was placed in an oil bath for 18 h at 70 • C for complete polymerization. The synthesized polymers were washed with methanolic solution till all un-reacted reactants were removed. Using Soxhlet, the templated caffeine molecule was removed from the MIP particles using a mixture of methanol/acetic acid (8:2, v/v). The washing solution was measured spectrometrically at λ max = 250 nm to check the complete removal of caffeine from the MIP particles. The particles were washed several times until there was no further detection of caffeine. The obtained MIPs were dried at ambient temperature before use. Non-imprinted polymers (NIPs) were also prepared in a similar way as the imprinted beads but with the exclusion of caffeine from the procedure.

Sensor Fabrication
The composition of the ion-sensing membrane (ISM) was of a total mass of 114 mg dissolved in 2.0 mL THF. The membrane was prepared via dispersing MIP or NIP particles (12 mg), 2 mg of (rGO), KTpClB (2.0 mg), plasticizer (49.0 mg), and PVC (49.0 mg). For the modified screen-printed electrodes (SPEs), about 5 µL of the ISM cocktail was applied onto the carbon orifice of the SPEs via the drop-casting method and left to dry. The membrane composition of the reference electrode contains 70 mg of NaCl and 78.1 mg of polyvinyl butyral (PVB) dissolved in 1 mL of methanol. A 20 µL of the reference membrane solution was drop cast on the Ag/AgCl ink electrode surface. The solid-state Ag/AgCl reference electrode was integrated with the potentiometric sensor into the screen-printed platform. A schematic representation of the designed sensors is shown in Figure 1. The obtained MIPs were dried at ambient temperature before use. Non-imprinted polymers (NIPs) were also prepared in a similar way as the imprinted beads but with the exclusion of caffeine from the procedure.

Sensor Fabrication
The composition of the ion-sensing membrane (ISM) was of a total mass of 114 mg dissolved in 2.0 mL THF. The membrane was prepared via dispersing MIP or NIP particles (12 mg), 2 mg of (rGO), KTpClB (2.0 mg), plasticizer (49.0 mg), and PVC (49.0 mg). For the modified screen-printed electrodes (SPEs), about 5 μL of the ISM cocktail was applied onto the carbon orifice of the SPEs via the drop-casting method and left to dry. The membrane composition of the reference electrode contains 70 mg of NaCl and 78.1 mg of polyvinyl butyral (PVB) dissolved in 1 mL of methanol. A 20 μL of the reference membrane solution was drop cast on the Ag/AgCl ink electrode surface. The solid-state Ag/AgCl reference electrode was integrated with the potentiometric sensor into the screen-printed platform. A schematic representation of the designed sensors is shown in Figure 1.

MIPs Characterization
The caffeine-imprinted beads were synthesized using methacrylic acid (MAA) as an afunctional monomer and ethylene glycol dimethacrylate (EGDMA) as a cross-linker in the ration 0.5: 3: 3 for the template, monomer, and cross-linker, respectively. In the imprinting process, the interaction between the functional monomer and the templated molecule can be accomplished through: (1) Hydrogen bond formation between the carboxyl group (COOH) in MAA with either the carbonyl group or the tertiary amine in caffeine; (2) A π-π interaction between the π-bonds in the caffeine molecule and carbonyl groups in both MAA and EGDMA. These types of interactions enhance the binding affinity and specificity of the MIP towards caffein recognition [63]. Figure 2 represents a representation pathway for the imprinting process.

MIPs Characterization
The caffeine-imprinted beads were synthesized using methacrylic acid (MAA) as an afunctional monomer and ethylene glycol dimethacrylate (EGDMA) as a cross-linker in the ration 0.5: 3: 3 for the template, monomer, and cross-linker, respectively. In the imprinting process, the interaction between the functional monomer and the templated molecule can be accomplished through: (1) Hydrogen bond formation between the carboxyl group (COOH) in MAA with either the carbonyl group or the tertiary amine in caffeine; (2) A π-π interaction between the πbonds in the caffeine molecule and carbonyl groups in both MAA and EGDMA. These types of interactions enhance the binding affinity and specificity of the MIP towards caffein recognition [63]. Figure 2 represents a representation pathway for the imprinting process. An investigation of the surface morphology of the synthesized polymeric beads was carried out using the scanning electron microscopy (SEM) technique. For the MIP beads, the SEM pictures showed a uniform, regular and semi-spherical shape with an average diameter of 300 nm. These beads were well dispersed in the plasticized PVC membrane. They could reduce the membrane resistance of the sensing membrane and create more recognition sites inside the membrane [63]. As the NIP beads were synthesized in a similar way as MIP beads but without a caffeine template, the morphological structure revealed a more regular, uniform, and spherical shape than MIPs beads. This can be attributed to the presence of caffeine template in the MIP beads. The SEM images for both MIPs and NIPs are presented in Figure 3.

Sensors' Characteristics
The all-solid-state sensors based on either MIPs or NIPs exhibited a potentiometric response towards caffeine within the concentration range of 1.0 × 10 −7 to 1.0 × 10 −3 M caffeine at pH 5 (50 mM MES buffer). The time trace versus the potential response with the corresponding calibration curves is shown in  An investigation of the surface morphology of the synthesized polymeric beads was carried out using the scanning electron microscopy (SEM) technique. For the MIP beads, the SEM pictures showed a uniform, regular and semi-spherical shape with an average diameter of 300 nm. These beads were well dispersed in the plasticized PVC membrane. They could reduce the membrane resistance of the sensing membrane and create more recognition sites inside the membrane [63]. As the NIP beads were synthesized in a similar way as MIP beads but without a caffeine template, the morphological structure revealed a more regular, uniform, and spherical shape than MIPs beads. This can be attributed to the presence of caffeine template in the MIP beads. The SEM images for both MIPs and NIPs are presented in Figure 3. An investigation of the surface morphology of the synthesized polymeric beads was carried out using the scanning electron microscopy (SEM) technique. For the MIP beads, the SEM pictures showed a uniform, regular and semi-spherical shape with an average diameter of 300 nm. These beads were well dispersed in the plasticized PVC membrane. They could reduce the membrane resistance of the sensing membrane and create more recognition sites inside the membrane [63]. As the NIP beads were synthesized in a similar way as MIP beads but without a caffeine template, the morphological structure revealed a more regular, uniform, and spherical shape than MIPs beads. This can be attributed to the presence of caffeine template in the MIP beads. The SEM images for both MIPs and NIPs arre 3.

Sensors' Characteristics
The all-solid-state sensors based on either MIPs or NIPs exhibited a potentiometric response towards caffeine within the concentration range of 1.0 × 10 −7 to 1.0 × 10 −3 M caffeine at pH 5 (50 mM MES buffer). The time trace versus the potential response with the corresponding calibration curves is shown in

Sensors' Characteristics
The all-solid-state sensors based on either MIPs or NIPs exhibited a potentiometric response towards caffeine within the concentration range of 1.0 × 10 −7 to 1.0 × 10 −3 M caffeine at pH 5 (50 mM MES buffer). The time trace versus the potential response with the corresponding calibration curves is shown in Figure 4. For membrane composition optimization, different plasticizers namely o, NPOE, DOP, and DBS were examined. For sensors based on MIPs plasticized in o,NPOE, they exhibited a Nernstian response with a slope of 51.2 ± 0.9 mV/decade (n = 5, R 2 = 0.997) over the concentration range of 4.5 × 10 −6 to 1.0 × 10 −3 M and a detection limit of 3.0 × 10 −6 M. Sensors based on MIPs plasticized in DOP exhibited a Nernstian response with a slope of 43.6 ± 0.5 mV/decade (n = 5, R 2 = 0.999) over the concentration range of 7.7 × 10 −6 to 1.0 × 10 −3 M and a detection limit of 4.0 × 10 −6 M. The sensors' membrane plasticized in DBS exhibited a potentiometric response with a slope of 45.4 ± 1.3 mV/decade (n = 5, R 2 = 0.999) over the concentration range of 8.0 × 10 −6 to 1.0 × 10 −3 M and a detection limit of 4.5 × 10 −6 M. For sensors based on NIPs, they exhibited a sub-Nernstian response with a slope of 28.1 ± 0.9 mV/decade (n = 5, R 2 = 0.997) over the concentration range of 4.5 × 10 −5 -1.0 × 10 −3 M, and a detection limit of 1.7 × 10 −5 M. This can confirm the existence of binding sites in MIPs' particles in the ion-sensing membrane. All the analytical features and the potentiometric response of the proposed sensors are summarized in Table 1.  Table 1.

Transduction Mechanism
In this work, both MWCNTs and rGO were introduced together as solid-contact materials during the manufacturing of the presented sensors. They can be considered as high surface area nanostructured materials that exhibit their ion-to-electron transduction properties, forming the electrical double layer at the polymeric ISE membrane/solid contact interface. The overall reaction includes mainly three equilibrium charge transfers at three boundaries or interfaces. At first, the reaction at the E-conducting substrate and solidcontact interface exhibits the electron transfer reaction. Secondly, an electrical doublelayer (Edl) is formed between cations or anions coming from the ISE membrane and the electrical charges (either electrons or holes) that were formed in the porous structure of the solid-contact material. The potential at the E-conducting substrate and solid contact interface is very small (E ≈ 0). So, smost EDL capacitance-based SC materials are considered

Transduction Mechanism
In this work, both MWCNTs and rGO were introduced together as solid-contact materials during the manufacturing of the presented sensors. They can be considered as high surface area nanostructured materials that exhibit their ion-to-electron transduction properties, forming the electrical double layer at the polymeric ISE membrane/solid contact interface. The overall reaction includes mainly three equilibrium charge transfers at three boundaries or interfaces. At first, the reaction at the E-conducting substrate and solidcontact interface exhibits the electron transfer reaction. Secondly, an electrical doublelayer (E dl ) is formed between cations or anions coming from the ISE membrane and the electrical charges (either electrons or holes) that were formed in the porous structure of the solid-contact material. The potential at the E-conducting substrate and solid contact interface is very small (E ≈ 0). So, smost E DL capacitance-based SC materials are considered highly electronic conductive. At the solid contact/ion-selective membrane interface, there is no charge transfer reaction, but the exhibited potential was declined to double-layer capacitance as shown in Figure 5. Thirdly, an interfacial phase-boundary potential is formed at the interface between the attached ion-selective membrane and the aqueous solution. Thus, the exhibited potential E B confirms the Nernstian response toward the desired ion and emphasizes the thermodynamical potential of double-layer capacitance (C dl )-based solid-contact ISEs.
highly electronic conductive. At the solid contact/ion-selective membrane interface, there is no charge transfer reaction, but the exhibited potential was declined to double-layer capacitance as shown in Figure 5. Thirdly, an interfacial phase-boundary potential is formed at the interface between the attached ion-selective membrane and the aqueous solution. Thus, the exhibited potential EB confirms the Nernstian response toward the desired ion and emphasizes the thermodynamical potential of double-layer capacitance (Cdl)-based solid-contact ISEs.

Method Validation
Reliability, quality, and consistency of results are important features and evidence of analytical method validation that ensure results fitting for the same method under the same set of conditions and control parameters [64]. There are a number of ways to classify the method validation as the following:

Limit of Detection (LOD) and Linearity
The LOD parameter is defined as the lowest concentration noise with a certain degree of confidence. There are several ways to distinguish LOD depending on the signal-tonoise ratio. But in the case of potentiometric analysis, LOD can be determined from the cross point of the lines fitted to the linear segments of the emf vs. log ai curve as shown in Figure 4. The LOD is not a robust or rugged parameter and can be affected by minor changes in the analytical system (e.g., temperature, matrix effects, purity of reagents, instrumental conditions). Linearity definition exhibited from obtained signals, which are directly proportional to the concentration of analyte in the sample. The performance characteristics of the MIP/o,NPOE sensors exhibited a wide and linear dynamic range between 1.0×10 −3 -4.5 × 10 −6 M with near-Nernstian slopes of 51.2 ± 0.9 mV/decade. The calibration plot with regression equation was found to be Y (mV) = 51.2 log [caffeine] + 131.6 with a correlation coefficient of 0.999 between the standard caffeine concentration and the potential measured in triplicates (n = 3). The detection limit of the presented sensor (DL) was found to be 3.0 × 10 −6 M.

Reproducibility and Repeatability
The reproducibility (between-run or inter-assay variation) and repeatability (withinrun or intra-assay variation) were evaluated for the proposed sensor. A standard caffeine sample (10 μg/mL) was measured to carry out these tests. The intra-and inter-day precision was below 0.59 and 0.84%, respectively. This confirmed the agreement between the results obtained by measuring the caffeine reference sample under different conditions with different sensor assemblies and different mV/meters at different times.

Method Validation
Reliability, quality, and consistency of results are important features and evidence of analytical method validation that ensure results fitting for the same method under the same set of conditions and control parameters [64]. There are a number of ways to classify the method validation as the following:

Limit of Detection (LOD) and Linearity
The LOD parameter is defined as the lowest concentration noise with a certain degree of confidence. There are several ways to distinguish LOD depending on the signal-to-noise ratio. But in the case of potentiometric analysis, LOD can be determined from the cross point of the lines fitted to the linear segments of the emf vs. log a i curve as shown in Figure 4. The LOD is not a robust or rugged parameter and can be affected by minor changes in the analytical system (e.g., temperature, matrix effects, purity of reagents, instrumental conditions). Linearity definition exhibited from obtained signals, which are directly proportional to the concentration of analyte in the sample. The performance characteristics of the MIP/o,NPOE sensors exhibited a wide and linear dynamic range between 1.0×10 −3 -4.5 × 10 −6 M with near-Nernstian slopes of 51.2 ± 0.9 mV/decade. The calibration plot with regression equation was found to be Y (mV) = 51.2 log [caffeine] + 131.6 with a correlation coefficient of 0.999 between the standard caffeine concentration and the potential measured in triplicates (n = 3). The detection limit of the presented sensor (DL) was found to be 3.0 × 10 −6 M.

Reproducibility and Repeatability
The reproducibility (between-run or inter-assay variation) and repeatability (withinrun or intra-assay variation) were evaluated for the proposed sensor. A standard caffeine sample (10 µg/mL) was measured to carry out these tests. The intra-and inter-day precision was below 0.59 and 0.84%, respectively. This confirmed the agreement between the results obtained by measuring the caffeine reference sample under different conditions with different sensor assemblies and different mV/meters at different times.

Trueness, Bias and Recovery
The meaning of trueness relates to the systematic error of a measurement system that is considered a closeness of agreement between the average of an infinite number of replicates measured quantity values and a reference quantity value. Bias definition is the agreement between the mean value of replicate measurements and the true value of the measured quantity [64].
Polymers 2022, 14, 1942 8 of 14 Trueness and bias of the proposed sensor were examined by using six replicate measurements of 10 µg/mL caffeine as an internal quality control sample. Their values were calculated as in Equations (1) and (2): where X is the mean of test results obtained for the reference sample and µ is the true value given for the reference sample. The obtained trueness and bias were found to be 99.3 and 0.7%, respectively. The recovery (%) of spiked caffeine samples was tested using six replicate measurements of spiked 1.5 µg/mL caffeine. The recovery (%) was found to be 99.4%.

Ruggedness and Robustness
The term ruggedness in analytical methodology is considered the degree of reproducibility of obtained results of the same sample under varying test conditions. On the other hand, robustness is defined as the stability of the method against small variations of the intrinsic method parameters and variability of the sample matrix [64]. The potentiometric features of the proposed sensor were examined over a wide range of variable pH values [e.g., 2-10]. The test was performed using two concentrations of caffeine (10 −3 and 10 −4 M). The pH of the test solutions was adjusted after adding small aliquots of HCl and/or NaOH. The obtained potentiometric response of the presented sensor was recorded with its corresponding pH value of the test solution. Since the pKa of caffeine is 14.0, therefore caffeine will be in its cationic form at pH below this pKa value. As shown in Figure 6, it exhibited a different pH influence on the applied sensor properties that revealed the range of its stability over the pH range 4.3-8.5. At pH < 5.3, the potential response quickly declined. This can be attributed to the existence of di-valent or tri-valent caffeine cations in the solution that can decrease the potentiometric response. At pH > 8.5, the potential response quickly declined which can be attributed to the decrease of caffeine cation concentration due to the formation of some non-ionized caffeine. This wide range of stable potential readings revealed that a sophisticated, rugged, and durable potentiometric sensor was investigated. For further potentiometric studies of the applied sensor, a MES buffer of a 50 mM (pH 5) was chosen to be a working pH throughout.

Trueness, Bias and Recovery
The meaning of trueness relates to the systematic error of a measurement system that is considered a closeness of agreement between the average of an infinite number of replicates measured quantity values and a reference quantity value. Bias definition is the agreement between the mean value of replicate measurements and the true value of the measured quantity [64].
Trueness and bias of the proposed sensor were examined by using six replicate measurements of 10 μg/mL caffeine as an internal quality control sample. Their values were calculated as in Equations (1) and (2): where X is the mean of test results obtained for the reference sample and μ is the true value given for the reference sample. The obtained trueness and bias were found to be 99.3 and 0.7%, respectively. The recovery (%) of spiked caffeine samples was tested using six replicate measurements of spiked 1.5 μg/mL caffeine. The recovery (%) was found to be 99.4%.

Ruggedness and Robustness
The term ruggedness in analytical methodology is considered the degree of reproducibility of obtained results of the same sample under varying test conditions. On the other hand, robustness is defined as the stability of the method against small variations of the intrinsic method parameters and variability of the sample matrix [64]. The potentiometric features of the proposed sensor were examined over a wide range of variable pH values [e.g., 2-10]. The test was performed using two concentrations of caffeine (10 −3 and 10 −4 M). The pH of the test solutions was adjusted after adding small aliquots of HCl and/or NaOH. The obtained potentiometric response of the presented sensor was recorded with its corresponding pH value of the test solution. Since the pKa of caffeine is 14.0, therefore caffeine will be in its cationic form at pH below this pKa value. As shown in Figure 6, it exhibited a different pH influence on the applied sensor properties that revealed the range of its stability over the pH range 4.3-8.5. At pH < 5.3, the potential response quickly declined. This can be attributed to the existence of di-valent or tri-valent caffeine cations in the solution that can decrease the potentiometric response. At pH > 8.5, the potential response quickly declined which can be attributed to the decrease of caffeine cation concentration due to the formation of some non-ionized caffeine. This wide range of stable potential readings revealed that a sophisticated, rugged, and durable potentiometric sensor was investigated. For further potentiometric studies of the applied sensor, a MES buffer of a 50 mM (pH 5) was chosen to be a working pH throughout.

Sensors' Selectivity
The various interference attitudes that can present in the matrices during caffeine determination was investigated. The selectivity of the caffeine sensor was evaluated using the method presented by Bakker [i.e., the modified separate solution method (MSSM)] [65]. Different interfering species were tested which can coexist with caffeine in either its pharmaceutical forms or biological fluids. Among these species, there are salts of Na + , K + , Mg 2+ , and Ca 2+ as cationic species. In addition, amino acids (alanine, arginine, and glycine), sugars (glucose and lactose), nicotine, chlorpheniramine, ephedrine, codeine, paracetamol, creatinine, camylofine and aspirin were added. The selectivity coefficients (log K pot caffeine, J ) were calculated and are shown in Table 2. It was shown that there is no significant interference from the interfering ions on the potentiometric response of the presented sensor. The sensors offered high selectivity and high efficiency towards the determination of caffeine in real samples. Table 2. The selectivity coefficients (log K pot caffeine, J ) of the proposed sensor.

Chronopotentiometry and Electrochemical Impedance Spectroscopy (EIS) Measurements
To evaluate the double-layer capacitance and membrane resistance, EIS measurements were carried out. The measurements were performed by using a one-compartment threeelectrode cell using (NOVA 2.0 software; Metrohm Auto lap B.V. Utrecht, The Netherlands) attached with a reference electrode (Ag/AgCl/KCl (3 M) and Pt auxiliary electrode. The applied frequency range starts from 100 kHz to 0.1 Hz using a sinusoidal excitation signal with excitation amplitude of 10 mV. The applied method was carried out by using a solution of 10 −3 M of caffeine in a 50 mM MES buffer, pH 5. As shown in Figure 7, the method exhibited the Nyquist plots (complex plane plots of -Z \\ vs. Z \ ) on the equivalent circuit models. The bulk resistance (R b ) of both the modified and non-modified electrodes were 0.1 ± 0.03 and 0.09 ± 0.0002 MΩ, respectively. The double-layer capacitances (C dl ) were measured at the low-frequency branch (semicircle) for both the modified and non-modified sensors and were found to be 9.3 ± 1.1 and 22.5 ± 1.4 µF, respectively. Short-term potential stability was evaluated via reverse-current chronopotentiometric measurements suggested by Bobacka [66]. The applied current was ± 1nA in both the anodic and cathodic directions for 60 s. The chronopotentiograms for both modified and non-modified sensors are shown in Figure 8. The potential drift (∆E/∆t) was found to be 97.7 and 41.6 μV/s for both the non-modified and modified caffeine sensors, respectively. Short-term potential stability was evaluated via reverse-current chronopotentiometric measurements suggested by Bobacka [66]. The applied current was ± 1nA in both the anodic and cathodic directions for 60 s. The chronopotentiograms for both modified and non-modified sensors are shown in Figure 8. The potential drift (∆E/∆t) was found to be 97.7 and 41.6 µV/s for both the non-modified and modified caffeine sensors, respectively. These data show the incredibly increasing potential stability in the presence of the transducers. The capacitance double-layer [C L = I/(∆E/∆t)] was 10.2 ± 0.5 and 24.2 ± 0.7 µF for both the non-modified and modified caffeine sensors, respectively. All data obtained via the EIS and chronopotentimetric measurements reflect the effect of the lipophilicity of the solidcontact transducing material and the highest double-layer capacitance formed upon the insertion of the rGO layer between the ion-sensing membrane and the electronic conductor substrate. The results confirmed that the rGO based sensor type revealed high potential stability, good conductivity, and high compatibility with the caffeine membrane-based sensor for the determination of caffeine in its matrices. Short-term potential stability was evaluated via reverse-current chronopotentiometric measurements suggested by Bobacka [66]. The applied current was ± 1nA in both the anodic and cathodic directions for 60 s. The chronopotentiograms for both modified and non-modified sensors are shown in Figure 8. The potential drift (∆E/∆t) was found to be 97.7 and 41.6 μV/s for both the non-modified and modified caffeine sensors, respectively. These data show the incredibly increasing potential stability in the presence of the transducers. The capacitance double-layer [CL = I/(∆E/∆t)] was 10.2 ± 0.5 and 24.2 ± 0.7 μF for both the non-modified and modified caffeine sensors, respectively. All data obtained via the EIS and chronopotentimetric measurements reflect the effect of the lipophilicity of the solid-contact transducing material and the highest double-layer capacitance formed upon the insertion of the rGO layer between the ion-sensing membrane and the electronic conductor substrate. The results confirmed that the rGO based sensor type revealed high potential stability, good conductivity, and high compatibility with the caffeine membranebased sensor for the determination of caffeine in its matrices.

Water-Layer Test
The prejudicial effect to the potential stability and lifetime of the proposed sensor was also investigated. The test was applied with both the sensors present and absent in

Water-Layer Test
The prejudicial effect to the potential stability and lifetime of the proposed sensor was also investigated. The test was applied with both the sensors present and absent in the solid-contact materials. The sensors were first immersed in 30 mM of the MES buffer (pH 5) for 30 min and the potential was recorded during this interval of time. After that, they were immersed in 10 −4 M of caffeine for another 30 min. and finally, they were immersed again in 30 mM of MES buffer (pH 5) for a further 30 min. As shown in Figure 9, there is an enhanced potential-stability for caffeine sensors when they were modified with the lipophilic rGO layer. There is a potential drift for the non-modified sensors. This reflects the existence of water-layer formation between the ion-sensing membrane and the electronic-conducting substrate. the solid-contact materials. The sensors were first immersed in 30 mM of the MES buffer (pH 5) for 30 min and the potential was recorded during this interval of time. After that, they were immersed in 10 −4 M of caffeine for another 30 min. and finally, they were immersed again in 30 mM of MES buffer (pH 5) for a further 30 min. As shown in Figure 9, there is an enhanced potential-stability for caffeine sensors when they were modified with the lipophilic rGO layer. There is a potential drift for the non-modified sensors. This reflects the existence of water-layer formation between the ion-sensing membrane and the electronic-conducting substrate.

Caffeine Assay in Different Pharmaceutical Formulations
Caffeine is present in different pharmaceutical dosages like tablets and capsules. For caffeine stock preparation, 10 tablet contents were weighed and the mean weight of the active ingredient was calculated per one tablet. The accurate weight of the powder corre-

Caffeine Assay in Different Pharmaceutical Formulations
Caffeine is present in different pharmaceutical dosages like tablets and capsules. For caffeine stock preparation, 10 tablet contents were weighed and the mean weight of the active ingredient was calculated per one tablet. The accurate weight of the powder corresponding to 0.194 g of caffeine was dissolved in 100 mL 50 mM MES buffer, pH 5 to obtain 10 −2 M caffeine stock. The solution mixture was sonicated for 1 h to ensure complete dissolution of the active ingredient. The solution was filtered and was diluted to several concentrations of caffeine (10 −3 -10 −5 M). After constructing the calibration plot, the potential recorded for these samples was introduced to the regression equation of the calibration curve and the amount of caffeine was then calculated.
As shown in Table 3, the recoveries of caffeine measurements by using the proposed potentiometric method were of the range 95.2 to 106.5%. The results obtained were compared with those obtained by the HPLC standard method [67]. The recoveries of this method were of the range 99.0 to 100.8%. F and t-student tests were utilised for the two methods and revealed no significant difference between them, confirming the successful applicability of the proposed sensor for the determination of caffeine.

Conclusions
A new potentiometric method of caffeine determination in its solutions is presented through screen-printed carbon electrodes, modified with reduced graphene oxide (rGO) which were used for the fabrication of the sensors. Tailored caffeine-imprinted polymers (MIPs) were synthesized using methacrylic acid (MAA) monomer and this has been used as an electroactive receptor for caffeine. The sensors revealed a Nernstian response with a slope of 51.2 ± 0.9 mV/decade (n = 6, R 2 = 0.997) over the linear range of 4.5 × 10 −6 -1.0 × 10 −3 M with a detection limit of 3.0 × 10 −6 M. They exhibited fast detection of caffeinium ions with a response less than 5 s response time (<5 s). EIS and chronopotentiometric measurements were used to evaluate the potential-stability and double-layer capacity of the presented modified sensor. The double layer and potential drift of rGO based sensor were 24.2 ± 0.7 µF and 41.6 µV/s, respectively. The application to caffeine assessment in different pharmaceutical formulations was also successfully carried out. This reflects that the presented analytical device can be considered as an attractive tool for caffeine determination due to its affordability and vast availability, particularly when combined with potentiometric detection.