Impedimetric blood pH sensor based on MoS₂–Nafion coated microelectrode

Abstract

Ion selective electrodes are a good alternative to glass-based pH sensors to replace traditional bulky pH sensors with compact and robust point of care analyzers. This is especially important for healthcare applications, as the pH-sensing element of hand-held blood gas analyzers to aid blood gas analysis in remote or mobile medical set ups. In this work, we have presented an impedimetric pH sensor consisting of parallel coplanar electrodes of aluminum coated with molybdenum disulfide (MoS₂) nanoparticles and Nafion as a permselective layer. This sensor showed a sensitivity of 6193 Ω pH⁻¹ and a response time of 4 s on testing with Sorensen’s phosphate buffers at different pH values, used as a physiological buffer. A preliminary study of the sensor with capillary blood shows that its estimated pH lies close to that of the venous blood pH, thus showcasing good characteristics to be a potential candidate for the pH-sensing element of a point of care analyzer.

---

1 Introduction

Measurement of the pH value is a very useful technique due to its wide range of applications in a variety of fields such as food and beverage quality testing, soil examination, pharmaceutical research, cell culture research in biotechnology etc.^1–3 Similarly, in healthcare, blood pH monitoring for arterial blood gas (ABG) analysis is a versatile diagnostic and prognostic tool to assess disease severity for a large variety of disease states ranging from cardiopulmonary derangements and poisoning to renal failure, and in a wide range of clinical settings varying from regular wards to critical care setups like Intensive Care Units (ICUs) and Operation Theaters (OTs).^4,5 Apart from that it also helps to determine the best treatment strategy to keep up with the pace of changes in the disease dynamics when done in a real-time manner, unlike the intermittent analysis done by current ABG analyzers.⁶ But presently the ABG analysis stands out as a cumbersome and skill-based technique involving a relatively costly, bulky and maintenance oriented setup, possible only in a tertiary care facility.

Barring the continuous intra-arterial blood gas analysis done in a surgical setup,^7,8 the traditional blood gas analysis protocol followed in wards and ICUs is time consuming, thereby missing out on the benefits of real-ime sensing. Also the full utility of this technique is undermined by the inherent drawbacks of intra-arterial blood sampling, such as the expertise required, the pain associated with repeated arterial punctures, the risk of direct intra-arterial infection and the intermittent nature of sampling.^9–11 There have been reports that venous blood and arterialised capillary blood correlate well with arterial blood regarding pH and pCO₂ values and can be mathematically arterialised for other blood gas values.^12–15 Since, at present, venous and capillary blood sampling is simpler and easy to repeat, it can be used to circumvent the drawbacks associated with traditional arterial blood sampling and develop a point of care (POC) blood analyzer similar to blood glucose sensor.

The reason for the bulkiness of traditional pH sensors is due to their traditional cumbersome design consisting of an electrolyte buffer filled glass electrode, which is an ion selective electrode, integrated with an Ag(s)/AgCl(s) reference electrode and measuring the pH potentiometrically.¹⁶ Recently, glass based ion selective electrodes are being replaced with permselective membranes like ionophore doped polymer matrices or conducting polymers to overcome the bulkiness issues with glass and to improve the sensitivity.^17–19 Nafion, a conducting polymer, is commonly used as a proton selective membrane, especially for proton exchange membranes (PEMs) in fuel cells and pH sensors.^20,21 The Nafion layer acts as a selective permeable membrane for H⁺ ions due to its high proton conductivity relative to the conductivity of other cations.²²

Nafion has hydrophobic (hydrocarbon backbone) domains surrounding hydrophilic domain (polar sulfonate side group), where sulfonate groups enclose the water molecules and the hydrated protons to form clusters. These clusters are in turn connected to other such clusters by nano-range aqueous channels, as given by the cluster-network model.²³ The proton conduction occurs in Nafion by two mechanisms, surface diffusion, which occurs near the sulfonic acid groups and bulk diffusion which includes a vehicle-type mechanism where H₃O⁺ carries the proton as it migrates through the medium in the aqueous channels and a Grotthuss type mechanism where the proton transfer occurs by “hopping” from one site to another via formation and breaking of hydrogen bonds as well as reorientation of the proton environment.^20,24 So, based on this model of nano-aqueous pores in a Nafion membrane, Choi et al.²⁴ has given the overall proton conductivity (σ_H⁺) as

where ε_i is the porosity of membrane, τ is the tortuosity factor, F is Faraday’s constant, R is the gas constant, T is temperature, , and are the diffusion coefficients for the surface, Grotthuss, and en masse mechanisms, respectively, and C_H⁺ and are the concentrations of protons participating in the bulk and surface phases, respectively.

Apart from good selectivity towards H⁺ ions, a pH sensor should also have high enough sensitivity especially if it is used for healthcare applications. Some research groups have tried to improve sensitivity by utilizing tungsten trioxide (WO₃)²⁵ or zinc oxide (ZnO)²⁶ nanoparticles-based potentiometric pH sensors and tin dioxide (SnO₂) based field effect transistors (FET) as pH sensors.²⁷ Another such material is molybdenum disulfide (MoS₂), a biocompatible material consisting of 2D stacked layers of covalently bonded transition metal and dichalcogenide atoms.²⁸ Transition metal dichalcogenides (TMDs) like MoS₂ shows a change in oxidation state under the influence of pH change.²⁹ Also MoS₂ has pristine surfaces (without out-of-plane dangling bonds), which reduce interface traps and so it has better electrostatic control at the interface.²⁸

Here we have investigated the pH sensing capability of an impedimetric pH sensor based on micro coplanar parallel ion selective electrodes integrated with MoS₂ nanoparticles (NPs), and have tested it for capillary blood pH sensing. The sensor was fabricated by deposition of aluminum on glass as a cost effective substrate followed by drop casting of an MoS₂ NPs suspension and later Nafion drop casting. It was then tested using physiological buffers of different pH and conductivities followed by testing with capillary blood. This sensor was shown to provide a potential alternative as a point of care, real time blood pH analyzer due to its good sensitivity, fast response time and small analyte volume requirement (0.5 μL). Furthermore, the use of inexpensive fabrication techniques and small volumes of chemical materials ensure the cost effectiveness of the sensor.

2 Materials and methods

2.1 Reagents and solutions

Piranha solution (H₂SO₄ : H₂O₂ = 1 : 1 (v/v)) and isopropyl alcohol were used as cleaning agents. An electric probe station (Cascade Mircotech, PM 5) was used for electrode gap formation. The thickness of the deposited film was measured by a surface profilometer (Dektak 150). A Leica DM2700 M microscope was used to measure the electrode dimensions. Nafion perfluorinated resin solution (5 wt% solution in lower aliphatic alcohols and 15–20% water) was purchased from Sigma Aldrich for use as the proton permeable layer. For preparation of the MoS₂ NPs suspension, bulk MoS₂ powder (99%, 2 μm in size, Sigma-Aldrich) was dissolved in organic solvent N,N-dimethylformamide (DMF, 99.9%, Merck Millipore). An ultrasound probe sonicator (UP100H, Hielscher) was used for preparation of the MoS₂ NPs suspension. A PANalytical X-ray diffractometer was used to collect the XRD spectrum of different samples. Raman spectra were collected by using a Jobin–Yvon–Horiba spectrometer. Sorensen’s phosphate buffer solutions with different pH values were prepared by mixing 66.7 mM of both sodium dihydrogen phosphate and sodium hydrogen phosphate in different proportions similar to the protocol described by Cutie and Sciarrone, 1969.³⁰ Their conductivity was adjusted by adding suitable volumes of 1 g% (w/w) NaCl solution in deionized water. The pH and conductivity of the prepared buffers were estimated using OAKION PC-2700. Hioki 3532-50 LCR HITESTER was used as the standard impedance analyzer for recording the impedance of the analyte–electrode system.

2.2 Fabrication of pH sensing electrode

2.2.1 Deposition of Al over glass substrate. In this study, a parallel coplanar electrode of aluminum (Al) over a glass substrate was fabricated by the shadow masking technique in a thermal deposition system. Before deposition, the glass substrate was chemically cleaned by piranha solution for 30 min to remove the organic, ionic and metallic impurities. A shadow mask of Al foil was designed containing rectangular shaped windows with dimensions of 1 mm × 17 mm, as shown in Fig. 1(a), and wrapped around the glass substrate. Then the sample was placed in the loading chamber of the thermal vapour deposition unit. At 1.8 × 10⁻⁵ mbar pressure in the chamber, 100 V was gradually applied between the terminals of the tungsten filament for 3 min. By this process, rectangular Al electrodes were fabricated with dimensions (average of three measurements) of 871.3 ± 10.3 μm (width), 17 ± 2 mm (length) and 96.9 ± 17.6 nm (thickness) as shown in Fig. 1(b).

Fig. 1 (a) Shadow mask of Al foil used for physical vapour deposition of Al over a glass substrate. (b) Magnified view of the deposited Al electrode over the glass substrate. (c) Electrode gap creation by scratching the deposited Al using a surgical blade under the probe station. (d) Microscopy image of the created gap in the reflection mode of illumination. (e) SEM image of MoS₂ nanoparticles deposited on the glass substrate. (f) EDS image of a randomly selected nanoparticle. (g) Final fabricated sensor with analyte drop. (h) XRD spectra of bulk MoS₂, MoS₂ NPs, Nafion and the sensing membrane (MoS₂ NPs–Nafion). (i) Raman spectra of bulk MoS₂ and MoS₂ NPs.

2.2.2 Creation of a gap between the electrodes. A gap was created by scratching around the midpoint of a rectangular shaped deposited Al film along the width direction with a scalpel fitted in the electric probe station, resulting in the formation of two electrodes with a uniform but minimum gap, as shown in Fig. 1(c). This technique is followed due to its simplicity without going through costly standard lithography processes to realise the electrodes. The gaps formed had an average dimension of 25.39 ± 4.19 μm with almost smooth and parallel edges, as shown in Fig. 1(d).

2.2.3 MoS₂ NPs synthesis and characterization. MoS₂ NPs were prepared by a simple ultrasonication method. Bulk MoS₂ powders were dissolved in N,N-dimethylformamide by probe sonication performed at 100 W under an ice bath. Thereafter, the suspension was centrifuged at 12 000 rpm to obtain MoS₂ NPs. The MoS₂ drop cast layer was studied by SEM to determine the size of the deposited particles, as shown in Fig. 1(e). Then, the chemical composition of a randomly selected nanoparticle on the glass substrate was validated by Energy Dispersive X-ray Spectroscopy (EDS). The EDS study showed trace amounts of Mo and S (from the nanoparticle), along with Au (from the gold thin film) and other elements like Si, O (from the underlying glass substrate) as seen in Fig. 1(f).

2.2.4 Coating of MoS₂ NPs and Nafion. Wires were connected to the ends of the electrodes with silver paste and 1 μL of MoS₂ NPs suspension was drop-cast on the electrode over the gap region. This was allowed to dry for half an hour and was followed by deposition of Nafion as an ion selective barrier. The Nafion layer was deposited by drop casting 1 μL of the 5% by weight Nafion solution over the gap region, to function as a selective proton-permeable layer. As soon as the first layer dried, another drop was cast over it. The process was repeated 5 times to increase the Nafion layer thickness so as to improve its proton conductivity.³¹ Finally it was allowed to dry for 2 h. The thickness of the sensing membrane of the final fabricated sensors was found to be 6.19 ± 1.37 μm (the average of three different sensor’s thickness). The final sensor is shown in Fig. 1(g).

The XRD spectra for bulk MoS₂, MoS₂ NPs, Nafion and the sensing membrane (MoS₂ NPs–Nafion) is shown in Fig. 1(h). For bulk MoS₂, four prominent peaks are identified at 14.35° (002), 32.57° (100), 39.49° (103) and 49.71° (105) (the peak for the (105) plane is not shown here). For MoS₂ NPs, only one peak was observed at 14.37° (002), which indicates that the exfoliation process successfully produced NPs and also reveals that no other new phases were introduced into MoS₂ NPs. The XRD intensity of MoS₂ decreased after exfoliation, which implies the synthesis of NPs.^32–34 For Nafion, an amorphous peak is observed because of its polymer nature, and for the sensing membrane (MoS₂ NPs–Nafion) the spectrum contains a small peak for MoS₂ NPs at the same position and an amorphous peak for Nafion.

Raman spectra of bulk MoS₂ and MoS₂ NPs exhibit two dominant characteristics peaks as shown in Fig. 1(i). In the case of bulk MoS₂, Raman modes E¹_2g (in plane vibration) and A_1g (out-of-plane vibration) were observed at ∼370.5 cm⁻¹ and ∼397.06 cm⁻¹, respectively, and the corresponding Raman modes in MoS₂ NPs were blue shifted (stiffened) to ∼372.6 cm⁻¹ and ∼398.3 cm⁻¹, respectively. In the literature, it has been shown that the Raman mode E¹_2g blue shifts (stiffens) and A_1g red shifts (softens) with the decrease in the thickness of MoS₂ (from bulk to monolayer).^32,35,36 The different characteristics in our Raman results reveal that for the MoS₂ NPs both the out-of-plane and in plane vibrations are found to be stiffened, instead of stiffening in only the vertical dimension. Therefore, this shows clearly the scaling down of the MoS₂ dimensions in all three directions.³⁶

2.3 Data acquisition and analysis

The two wires of the pH sensing electrode were connected to the fixture of the impedance analyzer. Before measurement, the impedance analyzer was open and short-circuit calibrated. Then, for impedance spectroscopy, a potential of 0.2 V (peak to peak) was applied with a sweep of 150 frequencies between the 50 Hz to 1 MHz settings. First the impedance spectroscopy of the sensor was carried out without the analyte. Then, an analyte drop of volume 0.5 μL was carefully put over the sensing region using a micropipette to record the impedance change with respect to time. The impedance readings were recorded at 1 s time intervals by applying a 0.2 V ac current of 20 kHz frequency to study the time response of the sensor. When the impedance readings stabilized to a relatively constant value with minor fluctuations (indicating equilibration of the analyte with the ion selective layer), impedance spectroscopy of the sensor with the analyte drop was performed. For each frequency, the impedance was measured consecutively 5 times with an accuracy of ±0.08% by the instrument and its average value was considered. The recorded magnitude of the impedance was plotted against frequency, which is also known as a Bode impedance graph, to analyze the pH sensor parameters.

3 Results and discussion

3.1 Effect of analyte conductivity on bare sensor impedance

Initially the bare Al electrode was used to establish the sensitivity of differentiating the pH buffer solutions with different conductivities (i.e. Sorensen’s phosphate pH buffers with conductivity in increasing order corresponding to the pH value, and deionized water) using impedance analysis, after depositing a 1 μL drop of analyte. The impedance was found to correlate with the conductivity of the buffer instead of the pH value, with a greater fall in impedance for increasing conductivity values, as shown in Fig. 2(a). Furthermore, the deposition of MoS₂ NPs was found to increase the resolution of conductivity measurement, as shown in Fig. 2(b). This is probably due to change in its oxidation state with pH, thereby altering the local potential and hence the measured impedance.²⁹

Fig. 2 Comparison of the influence of pH buffers of different conductivities after repeat reads: Bode impedance plot (log–log scale) for (a) a bare electrode and (b) a MoS₂ NPs drop casted electrode. (c) Effect of pH buffer solutions having the same conductivity: Bode impedance plot (log–log scale) for a bare electrode.

The Bode plot of the deionized water shows high impedance values due to the poor conductivity since, apart from H⁺ ion, no other ions are present to contribute to the conductivity. Here in the case of a bare Al electrode, the impedance values depend on the overall conductivity of the analyte and not specifically on the pH of the analyte as all the ions including H⁺ ion, which determine the conductivity, are able to reach the electrode surface and influence the impedance.

Impedance measurements were carried out with different pH buffer solutions having the same conductivity over the bare electrode and the recorded data were used to plot the Bode impedance graph as shown in the Fig. 2(c). This plot shows that impedance curves almost overlap for the buffers with similar conductivity (specially at higher frequency), which implies that the bare electrode was measuring the conductivity rather than the H⁺ ion concentration of the solutions.

3.2 pH sensing by permselective Nafion layer and MoS₂ NPs

Impedance readings were recorded for analyte pH buffer dropped over the Nafion coated electrode, as shown in Fig. 3(a). It shows selective measurement of the conductivity contributed by the concentration of H⁺ ions in the analyte buffer, seen as the drop in impedance from the baseline (i.e. impedance of the native sensor without the analyte). This selective influence of H⁺ ions is indicated by the greater fall in impedance for lower pH values. So, the impedance readings no longer correspond to the conductivity of the pH buffers, which followed a reverse order as seen in the case of bare electrode. Furthermore, in case of MoS₂ drop cast with a Nafion layer, the resolution of impedance separation between the different pH curves appears to be better, as shown by Fig. 3(b). This Figure also shows the repeatability of the results with the same pH buffers.

Fig. 3 Comparison of the influence of pH buffers of different conductivities after repeat reads: Bode impedance plot (log–log scale) (inset: linear fitting of the impedance versus pH values) for (a) a Nafion coated electrode and (b) a MoS₂–Nafion coated electrode. (c) Effect of two pH buffers with the same conductivity: Bode impedance plot (log–log scale) for a MoS₂–Nafion coated electrode. (d) Time plot of impedance change for a Nafion coated electrode as different pH buffers are applied after intervals of drying by soaking with tissue paper. (e) Time plot of the impedance change for the pH 7.15 analyte over a Nafion coated electrode. (f) Time plot of the impedance change for repeated study of the pH 7.38 analyte over a Nafion coated electrode.

From the literature, it has been found that upon increasing the H⁺ concentration up to a certain extent, the membrane conductivity of Nafion membrane increases.³⁷ So, in our experiment the small decrease in pH in the range 5–8, increases the membrane conductivity of Nafion, thereby increasing its dielectric constant. Furthermore, the bandgap of MoS₂ NPs has been found to have a strong dependence on the dielectric environment, with a change from 2.8 eV to 1.9 eV by an increase in the dielectric constant of the environment.³⁸ This is due to dielectric screening of the Coulomb interactions either between electrons and holes, or between different electrons that can affect the binding energies of excitons and trions or the quasiparticle electronic band structures of the MoS₂.³⁹ So in our case, the decrease in band gap energy of the excitons in MoS₂ NPs may be influencing the decrease in impedance with the diffusion of charge carriers, thereby decreasing the impedance values obtained for lowering of analyte pH.

Linear regression analysis was performed to find the best linear fit for impedance vs. pH data for each frequency and the optimum frequency was calculated corresponding to the best linear fit. The analysis of the slope of the linear fit shows that the sensitivity of pH measurement for the electrode with only Nafion coating is 346.68 Ω pH⁻¹ (R² = 0.91), while that for the MoS₂ drop-cast electrode with Nafion coating is 6193 Ω pH⁻¹ (R² = 0.99), as shown in Fig. 3(a) and (b) (inset), respectively. This clearly shows that inclusion of the MoS₂ NPs layer improves the sensitivity of the pH sensor.

In order to rule out the effect of conductivity and confirm that the impedance changes corresponded to pH values, pH buffers with similar conductivity within an acceptable range were used to conduct an impedance spectroscopy study. As seen in Fig. 3(c), the results indicate a similar relation between the impedance fall from the baseline and the corresponding pH value for the two same conductivity pH buffers. This shows minimal interference by other ions in the case of the Nafion coated sensor. The effect of pH change on the same sensor was recorded as a time response plot with the impedance values, as shown in Fig. 3(d). This also follows the same relation as the impedance change with pH. During the fabrication, batch to batch variability was seen for the gap between the electrodes, the Nafion thickness and the degree of its adherence to the substrate, which induces variations in the baseline impedance of the sensor and may cause an error in reproducibility. This can, however, be improved by heat annealing with other fabrication techniques like the hot press method.²¹

3.3 Time response analysis

The time response plot of the sensor was recorded for three pH buffers, and the response and stabilization time were calculated as an average of three readings. The response time (t_R) is the time taken to achieve a 90% change in impedance from the baseline while reaching the equilibrium value after the sensor comes into contact with the analyte drop. It was calculated to be 4 s, which is at least one order better than that for glass-based and paper-based potentiometric pH sensors.^40,41

The stabilization time is considered to be the time taken to reach the average final impedance value after a fall from the baseline impedance on deposition of an analyte drop. This is found to be 8.3 s, which is much better compared to the stabilization time in minutes for an impedimetric pH sensor reported by Mzoughi et al. in 2012.⁴² The time response plot of the sensor at pH 7.15 is shown in Fig. 3(e). The time response plot for the Nafion coated electrode shows good repeatability of the impedance change for the pH 7.38 buffer as shown in Fig. 3(f).

3.4 Preliminary impedance study with capillary blood

Impedance analysis was done for a MoS₂–Nafion coated electrode with 0.5 μL of capillary blood obtained by needle prick of the index fingertip. The pH of the venous blood of the same person was found to be 7.39 after testing in a blood gas analyzer. All the experiments have been carried out in compliance with the relevant laws and institutional guidelines. The current work is part of a pilot study to work on the recommendations of institutional ethical committee (ref no. IIT/SRIC/SAO/2016 dt 15th Feb 2016). In this study, first informed consent was obtained from a healthy volunteer explaining him about all aspects of the experiment. The blood collection was done following ethical guidelines and then measurements have been performed. The venous blood gas analysis lab report and the signed consent form has been properly maintained by the experimenters and proper measures have been taken to maintain participant confidentiality as per the ethical guidelines. The Bode impedance plot for the capillary blood drop shows a similar impedance profile to that obtained for the pH buffers and is found to lie close to the curve for pH 7.38, as seen in Fig. 4(a). This is expected, as the Nafion allows H⁺ ions to permeate preferentially compared to other cations, anions, proteins and other constituents of the blood.²² The capillary and venous blood pH values have been found to have a statistically insignificant difference from arterial blood pH values,^12,43 so they can be used as reliable indicators of arterial blood pH.

Fig. 4 Bode impedance plot (log–log scale) for a blood drop on a MoS₂–Nafion coated electrode, compared with that of other pH buffers. (b) Time plot of impedance change for blood drop over the MoS₂–Nafion coated electrode.

The time response analysis shows that the impedance rapidly falls on contact with blood and is maintained at a constant value for about 4.75 min, after which the impedance raises, indicating blood clotting, as seen in Fig. 4(b). From this, the response time of the sensor for the blood drop is found to be 10 s, which is lower than that for fiber optic and electrode based commercial blood pH sensors (∼30–40 s).^18,44 So, this rapid sensing, faster than the blood clotting time, gives the possibility of heparin free pH-sensing in the POC analysis mode thus avoiding the errors associated with heparin addition in the traditional ABG analysis technique.⁴⁵

3.5 Equivalent circuit analysis

The AC current impedance spectroscopy study of a simple electrode in contact with an electrolyte can be modeled using the Gouy–Chapman–Stern model as seen in Fig. 5(a). This takes into account the potential due to the electrostatic interaction of the immobile counter-ions with the electrode, the charge transfer due to redox reactions at the Stern layer near the electrode surface as well as the contribution to the potential due to the diffusion of ions in the diffuse layer.⁴⁶ This can be simplified into an equivalent Randle circuit model for impedance analysis, as given in Fig. 5(b).

Fig. 5 (a) The electrical circuit equivalent of the sensor modelled as per the Gouy–Chapman–Stern model.⁴⁷ IHP: inner Helmholtz plane, OHP: outer Helmholtz plane, Diff layer: diffuse layer, C_st: capacitance of Stern layer (combination of IHP and OHP), C_dif: capacitance of diffuse layer, Z_Diff: diffusion impedance, R_sol: resistance of analyte solution, R_ct: charge transfer resistance. (b) General Nyquist plot of the Randle circuit with R_s as an electrode resistance, R_ct as a charge transfer resistance, C_dl as a double later capacitor and Z_w as a Warburg diffusion element. (c) Fitted curve for the Nyquist plot of pH 6.2 buffer solution, which is well described by the electrical equivalent circuit diagram (shown in inset).

Nyquist plot of pH 6.2 (as shown in Fig. 5(c)) of analyte over MoS₂–Nafion coated electrode showed that only the transport controlled regime is dominant while kinetically controlled regime is negligible. Transport controlled regime characterizes the finite diffusivity of the redox species (H⁺ ions) in the solution through the ion permeable layer at lower frequencies. Absence of a kinetically controlled regime infers that charge transfer of ions at the surface of electrode can be neglected, indicating that faradaic phenomena i.e. redox reactions at the electrode surface, may have a minor role. So, the impedance in this case is governed by non-faradaic phenomena i.e. the rate of diffusion of ions through the diffuse layer.

On the basis of the above discussion, finally an electrical equivalent circuit (as shown in Fig. 5(c) inset) was chosen to fit the impedance data, which were recorded for different pH buffer analyte drops on the MoS₂–Nafion coated electrode. Data fitting was performed by available fitting software with the randomize and simplex algorithms. The circuit contains the resistor (R₁), constant phase element (CPE) (Q₂) and the restricted diffusion element (M₂) as shown in Fig. 5(c) (inset). R₁ corresponds to the combinations of the electrode and electrolyte resistance, CPE describes the double layer capacitance and M₂ infers the diffusion of ions in analyte. After fitting the impedance data to the electrical equivalent circuit, it was found that diffusion resistance (R_d) and diffusion time (t_d) increased with increasing pH value, as shown in Table 1. R_d and t_d are the parameters of the restricted diffusion element. Fig. 5(c) shows the fitted curve for a buffer with pH 6.2, which is found to match the experimental curve.

Table 1 Variations in diffusion element’s parameters with pH value^a

pH value	Rd/Ω	td/μs	χ^2 value
a Where Rd: diffusion resistance, td: diffusion time.
6.2	5303	7.98	0.065
7.4	8008	16.81	0.046
7.84	11 728	23.2	0.214

---

4 Conclusion

In this work a compact, robust and cost effective impedimetric pH sensor has been developed, which marks a difference from conventional, bulky potentiometric pH sensing. It has been characterized by pH buffers to have satisfactory pH sensitivity (6193 Ω pH⁻¹ for a Nafion coated electrode with a MoS₂ layer), particularly in the range of pH 6.2–7.8, which is suitable for blood diagnostics. Due to the micro-size of the electrodes, a very small volume of analyte (0.5 μL) is required for sensing purposes. So, this sensor can be integrated with microfluidic channels to work with very small analyte volumes, thereby reducing patient inconvenience, and allowing the assessment of multiple parameters from a smaller sample volume.

A preliminary study with 0.5 μL capillary blood, showed a Bode impedance curve lying close to that of the curve for a 7.38 pH buffer. For validation, an equivalence for capillary blood pH was obtained by standard venous blood analysis of the same person, which was found to have a pH of 7.39. Also, its response time was found to be 10 s, which can allow its adaptation for real time blood gas analysis similar to that of intra-arterial continuous blood gas monitoring, but in this case using arterialized capillary or venous blood. So, this can be used to develop a capillary blood based POC sensor as an indicator of arterial blood pH. Furthermore, these types of sensors can be easily integrated with a handheld microprocessor based impedance analysis system, compatible with Android for easy data transfer to a mobile system, for a tele-diagnostic service.

One of the main drawbacks of the study was the variation in the baseline impedance of the Nafion coated electrodes, which made it difficult to compare the readings of different electrodes. This may have been due to the variability in the thickness of the drop cast Nafion. Another limitation of this study was the lack of information on the effect of temperature and humidity change on the pH sensing. This is necessary because the proton conductivity of Nafion is dependent on its hydration level, while pH depends directly on temperature.

Regarding the stability of the sensor for repeat sensing, it is seen that after about three or more readings the sensor impedance reached a very high value with no change upon applying an analyte drop. This indicated that the attachment of the Nafion layer to the substrate weakened. This can be improved by hot press annealing of the Nafion membrane, which provides better attachment and improved proton conductivity due to the ordered structure produced by heat annealing. Still, this developed sensor shows an overall satisfactory sensing efficiency which can be used as a single-use cartridge or strip-based sensor, further justified by its cost-effective fabrication.

Acknowledgements

The authors would like to acknowledge the Indian Institute of Technology, Kharagpur for providing the necessary facilities to fabricate the device and carry out the experiments, and thank all the members of the BioMEMS Laboratory, School of Medical Science and Technology, IIT Kharagpur for their kind support. The authors extend special thanks to Mr Subhrajit Mukherjee for helping with Raman spectroscopy and analysis of the data, and Chemical Engineering Department, IIT Kharagpur for XRD analysis.

References

1. A. Andres-Bello, V. Barreto-Palacios, P. Garcia-Segovia, J. Mir-Bel and J. Martinez-Monzo, Food Eng. Rev., 2013, 5, 158–170.
2. M. Schirrmann, R. Gebbers, E. Kramer and J. Seidel, Sensors, 2011, 11, 573–598.
3. C. Ceccarini and H. Eagle, Proc. Natl. Acad. Sci. U. S. A., 1971, 68, 229–233.
4. E. Burri, M. Potocki, B. Drexler, P. Schuetz, A. Mebazaa, U. Ahlfeld, C. Balmelli, C. Heinisch, M. Noveanu and T. Breidthardt, et al., Crit. Care, 2011, 15, R145.
5. J. J. Park, D.-J. Choi, C.-H. Yoon, I.-Y. Oh, J. H. Lee, S. Ahn, B.-S. Yoo, S.-M. Kang, J.-J. Kim and S.-H. Baek, et al., Eur. J. Heart Failure, 2015, 17, 601–611.
6. S. A. Sasse, P. A. Chen and C. K. Mahutte, Chest, 1994, 106, 187–193.
7. B. E. Smith, P. H. King and L. Schlain, Int. J. Clin. Monit. Comput., 1992, 9, 45–52.
8. M. Ganter, Br. J. Anaesth., 2003, 91, 397–407.
9. A. L. Byrne, M. H. Bennett, N. L. Pace and P. Thomas, Protocols, Cochrane Database of Systematic Reviews, 2013.
10. A.-M. Kelly, Emerg. Med. J., 2001, 18, 340–342.
11. M. D. Davis, B. K. Walsh, S. E. Sittig and R. D. Restrepo, Respiratory Care, 2013, 58, 1694–1703.
12. G. Malatesha, N. K. Singh, A. Bharija, B. Rehani and A. Goel, Emerg. Med. J., 2007, 24, 569–571.
13. S. Rees, M. Toftegaard and S. Andreassen, Comput. Methods Programs Biomed., 2006, 81, 18–25.
14. P. Mollard, N. Bourdillon, M. Letournel, H. Herman, S. Gibert, A. Pichon, X. Woorons and J.-P. Richalet, Respir. Physiol. Neurobiol., 2010, 172, 179–183.
15. A. Banister, Arch. Dis. Child., 1969, 44, 726–728.
16. J. W. Severinghaus, Anesthesiology, 2002, 97, 253–256.
17. J. Steed and P. Gale, Supramolecular Chemistry: From Molecules to Nanomaterials, Wiley, 2012.
18. U. Oesch, D. Ammann and W. Simon, Clin. Chem., 1986, 32, 1448–1459.
19. O. Korostynska, K. Arshak, E. Gill and A. Arshak, Sensors, 2007, 7, 3027–3042.
20. Z. Zuo, Y. Fu and A. Manthiram, Polymers, 2012, 4, 1627–1644.
21. P. J. Kinlen, J. E. Heider and D. E. Hubbard, Sens. Actuators, B, 1994, 22, 13–25.
22. T. Okada, S. Moller-Holst, O. Gorseth and S. Kjelstrup, J. Electroanal. Chem., 1998, 442, 137–145.
23. K. A. Mauritz and R. B. Moore, Chem. Rev., 2004, 104, 4535–4586.
24. P. Choi, N. H. Jalani and R. Datta, J. Electrochem. Soc., 2005, 152, A1548–A1554.
25. L. Santos, J. P. Neto, A. Crespo, D. Nunes, N. Costa, I. M. Fonseca, P. Barquinha, L. Pereira, J. Silva and R. Martins, et al., ACS Appl. Mater. Interfaces, 2014, 6, 12226–12234.
26. A. Fulati, S. M. U. Ali, M. Riaz, G. Amin, O. Nur and M. Willander, Sensors, 2009, 9, 8911–8923.
27. Y. Cheng, P. Xiong, C. S. Yun, G. F. Strouse, J. P. Zheng, R. S. Yang and Z. L. Wang, Nano Lett., 2008, 8, 4179–4184.
28. D. Sarkar, W. Liu, X. Xie, A. C. Anselmo, S. Mitragotri and K. Banerjee, ACS Nano, 2014, 8, 3992–4003.
29. M. Z. M. Nasir, Z. Sofer and M. Pumera, ChemElectroChem, 2015, 2, 1713–1718.
30. A. J. Cutie and B. J. Sciarrone, J. Pharm. Sci., 1969, 58, 990–993.
31. D. K. Paul, K. Karan, A. Docoslis, J. B. Giorgi and J. Pearce, Macromolecules, 2013, 46, 3461–3475.
32. T. P. Nguyen, W. Sohn, J. H. Oh, H. W. Jang and S. Y. Kim, J. Phys. Chem. C, 2016, 120, 10078–10085.
33. D. Gao, M. Si, J. Li, J. Zhang, Z. Zhang, Z. Yang and D. Xue, Nanoscale Res. Lett., 2013, 8, 129.
34. J. Ali, G. U. Siddiqui, K. H. Choi, Y. Jang and K. Lee, J. Lumin., 2016, 169, 342–347.
35. V. Stengl and J. Henych, Nanoscale, 2013, 5, 3387.
36. S. Mukherjee, R. Maiti, A. Midya, S. Das and S. K. Ray, ACS Photon., 2015, 2, 760–768.
37. R. S. Yeo and J. McBreen, J. Electrochem. Soc., 1979, 126, 1682–1687.
38. J. Ryou, Y. S. Kim, S. Kc and K. Cho, Sci. Rep., 2016, 6, 29184.
39. Y. Lin, X. Ling, L. Yu, S. Huang, A. L. Hsu, Y. H. Lee, J. Kong, M. S. Dresselhaus and T. Palacios, Nano Lett., 2014, 14, 5569–5576.
40. E. Veress, J. Assoc. Lab. Autom., 2000, 5, 64–67.
41. M. Novell, M. Parrilla, G. A. Crespo, F. X. Rius and F. J. Andrade, Anal. Chem., 2012, 84, 4695–4702.
42. N. Mzoughi, A. Abdellah, Q. Gong, H. Grothe, P. Lugli, B. Wolf and G. Scarpa, Sens. Actuators, B, 2012, 171–172, 537–543.
43. G. S. Zavorsky, J. Cao, N. E. Mayo, R. Gabbay and J. M. Murias, Respir. Physiol. Neurobiol., 2007, 155, 268–279.
44. R. Wolthuis, D. McCrae, E. Saaski, J. Hartl and G. Mitchell, IEEE Trans. Biomed. Eng., 1992, 39, 531–537.
45. G. J. Ordog, J. Wasserberger and S. Balasubramaniam, Ann. Emerg. Med., 1985, 14, 233–238.
46. V. F. Lvovich, Impedance Spectroscopy: Applications to Electrochemical and Dielectric Phenomena, Wiley, 2012.
47. R. Burt, G. Birkett and X. S. Zhao, Phys. Chem. Chem. Phys., 2014, 16, 6519.

---

Footnote

† These authors contributed equally to this work.

---