A low-cost thin layer coulometric microfluidic device based on an ion-selective membrane for calcium determination

Abstract

A prototype of a low-cost and easy-to-use thin layer coulometric microfluidic device based on an ion-selective membrane for calcium detection is described. The microfluidic device was fabricated and consequently assembled with inexpensive materials without using sophisticated and centralized fabrication laboratory facilities. The linear range of the device is found to be 10–100 μM for a 60 s current integration time. Preliminary validations showed that the microfluidic device is suitable for the quantification of calcium in mineral water.

---

Cost-effective microfluidic devices for ion-detection both in clinical diagnostics and environmental monitoring are in great need due to numerous advantages over conventional analytical techniques.^1,2 These state-of-art analytical tools provide rapid analysis of multiple samples, remarkably decreasing reagent volume and power consumption.^3,4 They exhibit a portable design and are very simple to be operated by non-professionals outside of complex laboratory facilities. Additionally, such micro total-analysis systems (μTAS) allow one to significantly reduce the cost of analysis in developed countries and provide new affordable inexpensive analytical tools to emerging markets in developing/undeveloped countries.⁵ Despite multiple efforts the development of an inexpensive, mass-producible, autonomous and simple microfluidic platform for detection of various analytes including ions remains a very challenging task.^6–8

Electrochemical sensing provides a versatile approach for the development of microdevices.^8–10 As inexpensive, easy-to-integrate, simple-to-operate and rugged (durable) analytical tools, electrochemical microdevices have advantages over other microfluidic platforms. For instance, thin layer coulometry is a powerful electrochemical technique which provides a robust platform for potential calibration free and temperature independent detection of ionic species in aqueous samples.^11,12 Current technologies based on liquid ion-selective membranes allow the determination of several ions. In fact, the incorporation of a complexing agent (also called ionophore) in the membrane can tune the selectivity by several orders of magnitude.¹³ Further development of the coulometry technique by introducing a double pulse protocol allows one to suppress interferences from background electrolyte and improves the selectivity of the membrane.¹⁴

Recently, several research groups have developed microfluidic devices for ion detection utilizing the lab-on-a-chip concept.^8,15,16 Unfortunately these micro total-analysis systems for ionic species detection require expensive materials or/and complex centralized laboratory facilities for prototyping and fabrication.^7,8,17,18 Commonly, various lithographic methods and deposition techniques as well as dry and wet etching are utilized for the fabrication of microfluidic structures. Moreover, in order to process liquid samples, the support of external bulky equipment for liquid handling, fluid actuation or sample preparation (e.g. pumps, valves, pipets) is needed.⁵

In this communication, we report on the development and validation of a prototype of a low-cost thin layer coulometric microfluidic device (CMD) based on a calcium-selective membrane as the model system. This device is very simple to fabricate from available inexpensive materials by non-professional personnel. The fabrication of the device can be performed outside of common laboratory facilities. It does not require any support from external equipment for liquid propulsion and can be used where laboratory facilities are nonexistent and resources are limited. We have evaluated the performance of the developed CMD for detection of calcium in sparkling and non-sparkling mineral water. The obtained data were critically compared with a conventional technique, atomic absorption spectroscopy (AAS).

A photograph of the fabricated thin layer coulometric microfluidic device is shown in Fig. 1. The presented CMD was assembled from inexpensive flexible thermoplastic PVC tubing (40 × 10 mm). The device consists of working, counter/pseudo reference Ag/AgCl electrodes with a diameter of 400 μm and an ion-selective membrane made of porous polypropylene (PP) hollow fiber having an inner diameter of 600 μm.

Fig. 1 (a) A photograph of a thin layer coulometric microfluidic device for the detection of calcium ions. The microfluidic device assembled from flexible PVC tubing consists of a hollow polypropylene fiber with an inserted working Ag/AgCl electrode (WE) and counter Ag/AgCl electrode (CE) placed along the inner wall of PVC tubing (top left). The working electrode is coiled with a PVDF fish line (top right). The scale bar is 10 mm. (b) The schematic diagram of subsequent steps during the assembly of CMD.

Working and counter electrodes were made of 640 μm silver wire using a commercially available jewelry drawplate and subsequently oxidized electrochemically in a solution of 10 mM HCl for 3 h. The counter electrode was placed along the inner wall of PVC tubing in a spiral fashion and fixed between two plastic rings with epoxy. The working electrode was coiled with PVDF fish line (Ø 50 μm) in order to avoid direct contact between the electrode and PP hollow fiber.¹² 50 mm PP porous fiber was glued with epoxy to 1/8 in. dia. 15 mm long PEEK tubing and doped with a lipophilic cocktail containing 266.7 mg of plasticizer dodecyl 2-nitrophenyl ether (DDNPE), 30 mg (10 wt%) of electrolyte tetradodecylammonium tetrakis(4-chlorophenyl)borate ETH 500, 0.9 mg (30 mol% relative to ionophore) of cation exchanger salt potassium tetrakis(p-chlorophenyl)borate (KTpClPB) and 2.4 mg (10 mmol kg⁻¹) calcium ionophore IV (ETH 5234).¹³ Subsequently a 5 cm long working Ag/AgCl electrode was inserted into the PP hollow fiber. Prior to the complete assembly of CMD the ion-selective membrane with the inner Ag/AgCl element was conditioned overnight in the solution of 10 mM CaCl₂. The whole device was assembled and glued together with epoxy. Afterwards the outer compartment of the device was filled with a background electrolyte containing 10 mM KCl + 1 mM Ca(NO₃)₂.

For evaluation of the CMD, the experiment described below was performed.^11,14 The microfluidic tubing was filled using a peristaltic pump ISMATEC ISM935C (Glattbrugg, Switzerland) with the sample solution of constant concentration such as 10, 20, 40, 60, 80 and 100 μM of Ca(NO₃)₂ (10 mM KCl was always used as the background electrolyte). The sample volume is defined by the length of PP hollow fiber and is equal to 2.89 (±0.14) μL. The electrochemical measurements were performed with an μ-Autolab System (Metrohm Autolab, Utrecht, The Netherlands) controlled by a personal computer using Nova 1.5 software (supplied by Autolab). After filling the hollow fiber membrane with the sample solution, the open circuit potential (OCP) was measured. Subsequently, a 60 s excitation pulse at OCP plus incremental multipliers of 30 mV for each cycle ranging from 30 to 300 mV was applied and the current was integrated over this time to give the charge. The second background compensation pulse was applied for 60 s and again integrated. Before introducing a new sample the system was regenerated for a period of 120 s at OCP determined at the beginning of the procedure. As suggested, the double pulse protocol (subtracting the charge from the second pulse from that of the first electrolysis pulse) was applied in order to minimize the contribution of the background electrolyte to the total charge associated with the presence of calcium ions.^11,14

Fig. 2a shows the calculated charge for several current responses subtracted from the raw electrolysis (bulk depletion of calcium) and background compensation pulse as a function of the applied potential versus OCP for different calcium concentrations (for more information, see reference).^11,14 Evidently, the transported charge is strongly dependent on calcium concentration within the applied potential window, which is determined by the membrane selectivity. This confirms that the level of calcium ions in the thin layer sample defines the thin layer coulometric readout. The membrane selectivity is estimated from the potential window being ca. 9 orders of magnitude (∼280 mV divided by the Nernstian slope of 29.6 mV). This value is in agreement with previous reported values.¹³ The resulting subtracted charges of both pulses are plotted in Fig. 2b as a function of calcium concentration. At the lower potentials between 30 and 90 mV one observes incomplete electrolysis as the concentration of calcium increases. On the other hand, applying extremely high potentials above 180 mV leads to depletion not only of calcium but also the background electrolyte and results in an over estimation of the charge passing through the system associated with calcium ions. Consequently the charge difference between the first and the second pulse yields linear calibration curves in the case of 120 and 150 mV applied potential relative to OCP within the concentration range of 10 and 100 μM. Accordingly we have performed the calibration of the developed CMD at the potential of 150 mV (Fig. 3).

Fig. 2 (a) Ion transfer charge as a function of the applied potential vs. Ag/AgCl wire for different concentrations of calcium. (b) Ion transfer charge as a function of calcium concentration obtained from chronoamperograms recorded at different potentials.

Fig. 3 Calcium calibration curve recorded at 150 mV. The sample solution contains 0.01–0.1 mM Ca(NO₃)₂ + 10 mM KCl. The outer solution contains 0.1 mM Ca(NO₃)₂ + 10 mM KCl. Error bars represent the standard deviation of three measurements.

For this given applied potential we carried out three measurements for each calcium concentration and the obtained linear fit with the intercept 8.13 (±0.48) μC and slope 0.497 (±0.008) μC μM⁻¹. The theoretical slope is equal to 0.558 μC μM⁻¹. The slope was calculated from the first Faraday's law assuming 100% electrolysis conversion. As one can see the experimental and theoretical values are in good agreement and indicates 89 (±1.5)% recovery. The difference of approximately 10% from the geometrical value is attributed to the length variation of the PP hollow fiber glued to the PEEK tubing, which defines the sample volume.

The developed coulometric microfluidic device was tested for the analysis of sparkling (Henniez, Valser Classic) and non-sparkling (Evian, Valser Silence) mineral water. Before analysis the mineral water samples Henniez, Valser Silence were diluted 100 times and Evian, Valser Classic 200 times with the background electrolyte 10 mM KCl. The sample was aspirated (injected, loaded) by using a disposable syringe. The charge values determined coulometrically were obtained from three measurements with %RSD of 2.1% for Valser Silence and Evian, 0.4 and 0.6% for Henniez and Valser Classic respectively. These values were introduced into the calibration curve previously determined. The estimated concentrations for mineral water samples are summarized in Table 1.

Table 1 Calcium concentration in mineral water samples measured with coulometric microfluidic device and AAS. Standard deviations shown are from three replicate measurements

	Manufacture (mg L^−1, mM)	Coulometry (mg L^−1, mM)	AAS (mg L^−1, mM)
Henniez (semi sparkling)	104, 2.59	103.8 (±0.4), 2.59 (±0.01)	98.3 (±0.5), 2.45 (±0.01)
Valser Silence (non-sparkling)	51, 1.27	46.5 (±0.98), 1.16 (±0.02)	50.1 (±0.15), 1.25 (±0.004)
Evian (non-sparkling)	80, 1.995	76.2 (±1.6), 1.90 (±0.04)	76.4 (±0.08), 1.91 (±0.002)
Valser Classic (sparkling)	418, 10.4	412.9 (±2.5), 10.3 (±0.06)	420 (±3.78), 6.92 (±0.06)

---

The coulometric data thus obtained were compared with the data from conventional AAS. The data show that our method compares favorably. While a dilution was necessary owing to the limiting measuring range, the described approach offers the potential for a rapid and accurate mineral water analysis.

Conclusions

The significant miniaturization of a thin layer coulometric setup is promising for the development of microfluidic devices. We demonstrated that the performance of the developed CMD is adequate for the detection of calcium in sparkling and non-sparkling mineral water in the range of linearity between 10 and 100 μM. The well established palette of selective chemical receptors for a wide variety of ions can be applied for the fabrication of sensors for other ions with the same basic methodology. A simple design and inexpensive fabrication is attractive for emerging markets in developing/undeveloped countries (low-cost, easy-to-make). This platform lends itself to further development and translation into a solid state format compatible with microfabricated disposable microfluidic cartridges (chips). The analytical performance of the thin layer coulometric microfluidic device is potentially suitable for an application in environmental monitoring, food quality control and clinical analysis (point-of-care diagnostics).

Acknowledgements

The authors thank the Swiss National Science Foundation and the University of Geneva for financial support.

Notes and references

1. M. L. Kovarik, D. M. Ornoff, A. T. Melvin, N. C. Dobes, Y. Wang, A. J. Dickinson, P. C. Gach, P. K. Shah and N. L. Allbriton, Anal. Chem., 2013, 85, 451–472.
2. C. D. Chin, S. Y. Chin, T. Laksanasopin and S. K. Sia, Point-of-Care Diagnostics on a Chip: Low-Cost Microdevices for Point-of-Care Testing, Springer, Berlin, Heidelberg, 2013, pp. 3–21.
3. S. Haeberle and R. Zengerle, Lab Chip, 2007, 7, 1094–1110.
4. D. Erickson and D. Li, Anal. Chim. Acta, 2004, 507, 11–26.
5. X. Mao and T. J. Huang, Lab Chip, 2012, 12, 1412–1416.
6. B. Weigl, G. Domingo, P. LaBarre and J. Gerlach, Lab Chip, 2008, 8, 1999–2014.
7. A. K. Yetisen, M. S. Akram and C. R. Lowe, Lab Chip, 2013, 13, 2210–2251.
8. R. D. Johnson, V. G. Gavalas, S. Daunert and L. G. Bachas, Anal. Chim. Acta, 2008, 613, 20–30.
9. Y. Wang, H. Xu, J. Zhang and G. Li, Sensors, 2008, 8, 2043–2081.
10. W. B. Zimmerman, Chem. Eng. Sci., 2011, 66, 1412–1425.
11. E. Grygolowicz-Pawlak and E. Bakker, Electrochem. Commun., 2010, 12, 1195–1198.
12. A. Shvarev, B. Neel and E. Bakker, Anal. Chem., 2012, 84, 8038–8044.
13. E. Grygolowicz-Pawlak and E. Bakker, Anal. Chem., 2010, 82, 4537–4542.
14. E. Grygolowicz-Pawlak, A. Numnuam, P. Thavarungkul, P. Kanatharana and E. Bakker, Anal. Chem., 2012, 84, 1327–1335.
15. Z. Nie, C. A. Nijhuis, J. Gong, X. Chen, A. Kumachev, A. W. Martinez, M. Narovlyansky and G. M. Whitesides, Lab Chip, 2010, 10, 477–483.
16. J. C. Chou, G. C. Ye, D. G. Wu and C. C. Chen, Solid-State Electron., 2012, 77, 87–92.
17. A. Hierlemann, O. Brand, C. Hagleitner and H. Baltes, Proc. IEEE, 2003, 91, 839–863.
18. K. Ren, J. Zhou and H. Wu, Acc. Chem. Res., 2013 DOI:10.1021/ar300314s.

---