A sensor probe for the continuous in situ monitoring of ammonia leakage in secondary refrigerant systems

Abstract

Ammonia is becoming more widely used in refrigeration systems due to the phasing out of CFCs and HCFCs. However, ammonia is a toxic gas and its leakage from refrigeration systems can lead to human exposure and contamination of refrigerated food stuffs. There is a lack of devices capable of the direct and continuous monitoring of leakage of ammonia into secondary refrigerant systems. Here we demonstrate an ammonia measurement probe for continuous contact monitoring of secondary refrigerants. The probe was based on an ammonia-sensitive film of inkjet printed polyaniline nanoparticles deposited onto an interdigitated electrode array and enclosed behind a polytetrafluoroethylene membrane. When operated impedimetrically, the probe was capable of the detection of ammonia across the industrially relevant range of 0 to 100 ppm from +4 to −15 °C in water and brine. Operation of the probe as a simple threshold alarm without the requirement for temperature monitoring or calibration is illustrated.

---

Ammonia is experiencing a revival in popularity as an industrial refrigerant due to increasing restrictions on the use of chlorofluorocarbons. These restrictions commenced with the signing of the 1987 Montreal protocol which initiated a worldwide phase out of ozone-depleting CFCs.¹ As a consequence, regulations such as those in the US² and EU³ have committed to first introducing new systems which do not use CFCs/HCFCs (e.g., R-22) by 2010 and by 2015–2020 to eradicating all usage or recycling, even in older systems. As a result, ammonia has undergone a significant resurgence in usage in the last decade.^4,5 It is particularly effective as a refrigerant gas as it has a low boiling point and a high latent heat of vaporisation. It can be employed in vapour compression-based refrigeration, air-conditioning and heat pump units. In a typical refrigeration system, the ammonia (or ammonia mixture) is employed as the primary refrigerant and is used to cool a secondary refrigerant such as water, brine or glycol. As a consequence, there is some potential for leakage and contamination into the secondary system,^6,7 with potential hazards of human exposure or contamination of refrigerator contents.^8,9 Toxic effects can be observed at concentrations from 100 ppm. Levels between 100 and 200 ppm can cause eye irritation and exposure to levels of 400 to 1700 ppm may lead to irritation of the airways and coughing. At higher concentrations, ammonia exposure can be fatal.

Ammonia is also flammable between 16 and 25% by volume. There is significant legislation and regulation of the storage, handling and application of ammonia in refrigeration systems. The European Standard for the design and construction of refrigeration systems (EN378:2008 Part 1–4) deals with leakage detection. The charge limit above which gas detection equipment must be installed is >50 kg for ammonia. However, gas detection may be required for smaller charges if practical limits are reached. Hence, there is a need to be able to monitor ammonia leakage in refrigeration units at all times.

Leak detection in ammonia refrigeration systems can be applied to either the refrigerator atmosphere (gas sensors) or to the secondary refrigerant (liquid-based sensors). Generally, for the detection of gaseous ammonia, there exists a wide variety of sensor types including solid-state, electrochemical, optical and photoacoustic systems.^10–13 However, due to the challenging conditions in industrial facilities (temperature and humidity variation, exposure to other gases, etc.) only a few are suitable for refrigerant leak detection. Those available commercially include devices based upon infrared, electrochemical, solid state and charge carrier injection (CI) methodologies.¹⁰ Solid state detectors are typically metal oxide-based and are relatively low cost. However, they suffer from a lack of selectivity, leading to false positives. Infrared sensors are often highly selective and stable but are a relatively bulky and are a higher cost option. CI devices consist of a layer that selectively binds ammonia which then acts as a charge carrier for the sensor. Traditionally, electrochemical sensors for ammonia detection are composed of three electrode systems within a gel electrolyte. These sensors are relatively effective and selective but they suffer from short lifespans as they are gradually consumed during operation. The application of such sensors is in refrigeration spaces, production areas and ammonia cylinder storage and is essential for personnel protection, emergency shutdown and control, room ventilation control and in safety monitoring. However, although atmospheric monitoring is essential from a health and safety perspective, it will only detect the ammonia after it has already been vented from the system. In-line detection of ammonia within the secondary refrigerant would allow for early warning of leakage before atmospheric release, reducing personal exposure and costly damage to the system and refrigerator contents. The MiniCal® III represents one of the only commercial sensors for measuring ammonia in water and brine-based secondary refrigerants at low temperatures and is based on an ion-selective membrane electrode. There has been some research on the development of new materials for aqueous ammonia sensors in recent years based on nanostructured ZnO,^14,15 polyurethane acrylate,¹⁶ polyaniline¹⁷ and beta-Fe₂O₃ nanoparticles.¹⁸ However, few, if any of these have been examined for low temperature applications such as refrigerant monitoring to date.

In this work, we present an in-line probe designed to detect dissolved ammonia in water and brine-based secondary refrigerants. The principle of the sensor was based on measuring the change in impedance of a printed polyaniline nanoparticle (nanoPANI) chemiresistor upon exposure to ammonia. The sensor was mounted in the headspace of a probe, behind a gas permeable polytetrafluoroethylene (PTFE) membrane and could detect an ammonia leakage at a threshold level of 50 ppm at temperatures from +4 to −15 °C without the requirement to calibrate for, or measure temperature.

Experimental

Materials

Polyaniline nanoparticles (nanoPANI) were prepared using a method described previously.¹⁹ In brief, a 40 mL solution of 0.25 M dodecylbenzene sulfonic acid (DBSA) which was then divided into 20 mL solutions. Into the first, 0.36 g of ammonium persulfate (APS) was added and stirred until fully dissolved. Into the second, 0.6 mL of distilled aniline was added and the DBSA–APS solution was added immediately after which the solution was stirred for 2.5 hours. After 2.5 hours, 20 mL of 0.05 M SDS was added to the product (now thick and dark green in colour – indicative of polyaniline in the emeraldine salt form). The product was then centrifuged at 5000 rpm for 30 minutes prior to dialysis in 0.05 M SDS for 48 hours, after which the nanoPANI was ready for use.

Fabrication of the ammonia probe

Interdigitated ammonia sensing electrodes were prepared as previously described.²⁰ An aqueous dispersion of nanoPANI was deposited onto a screen-printed, interdigitated silver electrode using inkjet printing (Dimatix DMP 2831, FujiDimatix, Inc.) and cured at 75 °C for 20 minutes. The sensors were assembled into a bespoke probe assembly which consisted of an outer cylindrical casing, an electrode holder and a membrane holder containing a 0.2 μm circular polytetrafluoroethylene (PTFE) membrane (Z134201, Sigma-Aldrich, Ireland) (Fig. 1a and b). The external probe diameter was 44 mm. The distance between the electrode and the membrane was 6 mm and the volume of the headspace was 1.134 cm³.

Fig. 1 (a) The ammonia probe assembly. The polyaniline-based sensor (1) is held in the sensor holder (2) with the fixing screw (3). The sensor holder then screws onto the PTFE membrane holder (4) and the entire assembly is screwed into the probe casing (5). (b) Assembled ammonia probe and sensor electrode. (c) Schematic of the secondary refrigerant system. The ammonia probe is mounted in a probe holder which allows refrigerant to flow across the surface of the probe. The secondary refrigerant is circulated using a peristaltic pump through a stainless steel heat transfer coil which is submerged in the primary refrigerant bath. An injection port was used to make injections of ammonia. (d) Photograph of the refrigeration system showing probe, holder and heat transfer coil submerged in refrigeration bath.

Refrigeration system

A miniaturised refrigeration system was developed (Fig. 1c and d). Here, a refrigerated water bath containing Thermal HY refrigerant fluid (Julabo Labortechnik GmbH, Germany) acted as the primary refrigerant system. A stainless steel heat transfer coil was immersed within the bath and connected to a chamber which allowed insertion of the probe while allowing secondary refrigerant to flow across the surface of the probe. This was further connected to insulated tubing and a 12 V peristaltic pump with an access valve to allow injection of known quantities of ammonia. The internal volume of the system was 220 mL and was filled with brine (23.3% w/v NaCl, Sigma-Aldrich, 310166 in deionised water). The flow rate in the system was 4.5 mL s⁻¹ (at 12 V) and the linear flow velocity across the probe was 9 cm s⁻¹. The system was allowed to stabilise at the desired temperature prior to measurement. Calibrations were performed using the standard addition method with the volumes of aqueous ammonia injected yielding the desired concentration within the system. An aqueous ammonia stock solution (2.9% w/v) was prepared before each experiment by dilution of ACS grade ammonium hydroxide (Sigma, 320145).

Impedance analysis

The probe was connected to a CH 660 potentiostat equipped with electrochemical impedance analysis. The impedance responses of the probe upon exposure to refrigerants containing ammonia were measured. Quantitative measurements were performed at a fixed frequency of 180 Hz 5 mV rms.

Results and discussion

In order to detect the leakage of ammonia from a primary refrigerant fluid into a secondary system, the ideal configuration would be a probe that was in constant contact with the secondary refrigerant and so was capable of continuous monitoring. Until now, the only system capable of measuring the refrigerant itself was a discontinuous system requiring a more complex sampling system to take a fraction of the flowing refrigerant and measure it ‘in line’.¹³

Polyaniline is a well-known conducting polymer material that responds well to exposure to ammonia due to the de-protonation of the polymer backbone on exposure to ammonia. Previously, electrodes fabricated using polyaniline nanoparticles have been used to evaluate ammonia concentrations in both air using conductimetric methods²⁰ and water using amperometry.²¹ However, the transient response characteristics of ammonia in water as measured using amperometry makes this unsuitable for continuous monitoring applications. While conductivity measurements can be used for continuous measurement, they cannot be performed in contact with liquid samples. Thus, the ideal solution would be to facilitate a means of impedimetric analysis that did not require direct contact with the liquid. Previously, dissolved ammonia has been measured in air through the use of a membrane which is impermeable to water such as polytetrafluorothene (PTFE), but which is permeable to ammonia gas.^16,22,23 In the current device, continuous measurement was also made possible by the use of a PTFE membrane, creating a gas phase headspace between the refrigerant solution and the electrode. In this way, the probe could remain in continuous contact with the refrigerant. However, ammonium could pass through the membrane into the gas phase and be measured by the sensor.

When dissolved in water, ammonia is in equilibrium with ammonium as follows:

NH₃ + H₂O ⇆ NH₄⁺ + OH⁻

Ammonia has a pK_a of 9.3 and so at neutral pH, 99.5% of the ammonia is in the form of ammonium. At a gas–liquid interface, the partial pressure of ammonia in the gas phase is determined by Henry's Law²⁴ which relates the partial pressure of the ammonia in the gas phase to its concentration in solution via the Henry's Law constant, k_H as follows:

p = k_Hc

where c is the concentration of the gas in solution. The Henry's Law constants for ammonia in water from +4 to −15 °C are in the range from 172.77 to 511.88 mol kg⁻¹ bar⁻¹,²⁵ indicating its increased solubility at lower temperatures. Thus, for a given concentration of ammonia in solution, the partial pressure of ammonia in an adjacent gas phase headspace would be expected to be higher at lower temperatures. However, such equilibria are further complicated by the presence of dissolved species such as salts, which may alter pH and water availability according to the Sechenov equation.²⁶ Thus, between a liquid and gas phase of defined characteristics, proportionality is established between the concentration of ammonia in solution and the ammonia in the gas phase. In the event of a leakage of ammonia from the primary (high ammonia concentration) refrigerant to the secondary refrigerant, ammonia leaking into the secondary refrigerant would equilibrate in the gas phase headspace of the ammonia probe. This results in a three phase system in which the ammonia concentration in solution establishes an equilibrium with ammonia in the gas phase, which in turn establishes an equilibrium of ammonia in the conducting polymer film. In such a way, direct and continuous measurement can be performed in a liquid while using the relatively simple measurement of a change in conductivity of the polymer film in response to ammonia.

It has previously been noted that the application of a bias potential to polyaniline for measurements using dc conductivity can significantly alter the electrochemical conductivity of the polyaniline due to either oxidation or reduction.²⁰ To avoid this, perturbation of the film with a small magnitude, alternating current with zero bias potential would maintain the polymer in an unpolarised condition. Thus, an impedance-based measurement of gaseous ammonia in combination with a PTFE membrane was identified as the most appropriate measurement solution with which to exploit the ammonia sensing capabilities of these polyaniline nanoparticle films.

Impedance analysis

Preliminary tests were performed to optimise the impedimetric measurement parameters to be used for ammonia determination. Impedance spectroscopy of the ammonia probe in brine (23.3% NaCl w/v in water) alone and in the presence of 100 mM ammonia is shown in Fig. 2. Impedance spectra were recorded from 0.1 Hz to 100 kHz. In the absence of ammonia, a generally flat response was observed for the absolute impedance (|Z|) across this frequency range (Fig. 2a), with a change in phase angle occurring significantly above 10 kHz, which is as expected for a polyaniline film²¹ and which also demonstrated the effectiveness of the PTFE membrane and probe assembly to exclude moisture from the sensor headspace.

Fig. 2 Impedance spectra of the polyaniline sensor probe. Change of absolute impedance |Z| (a) and phase angle (b) with frequency. Solid line: response in 23.3% NaCl. Dashed line: response in 23.3% NaCl after addition of 100 ppm NH₃. Impedance parameters: ƒ = 0.1 Hz to 100 kHz, E_amp = 5 mV rms.

In the presence of ammonia, the magnitude of |Z| increased by almost two orders of magnitude at frequencies below 1 kHz. Above 1 kHz, the increase in impedance due to the ammonia progressively diminished and a greatly increased capacitive effect was observed, as evidenced by the phase angle shift (Fig. 2b). To avoid measurement of capacitive effects from either the polymer film or the added ammonia, further measurements were performed at a fixed frequency of 180 Hz at which resistance appeared to dominate.

To mimic a circulating refrigeration system, the probe was placed within a miniaturised flow system to monitor for ammonia leaks. The setup allowed the simulation of an industrial refrigeration system with a refrigerated water bath performing the role of the primary refrigeration circuit into which was placed a circulating secondary circuit of defined volume and refrigerant contents. It should be noted that a portion of the secondary refrigerant system was, by necessity, not immersed within the bath. However, the probe holder was designed with a series of S-bends to maximise the portion of the system that was immersed and allow for a greater surface area for heat exchange between the water bath and secondary circuit. Refrigerant temperatures were found to stabilise at the desired value within 45 minutes.

Fig. 3 shows the impedimetric response of the probe to concentrations of ammonia in brine at 4 °C. In terms of the absolute impedance measured, the baseline impedance was approximately 20 kΩ, rising to approximately 150 kΩ at 100 ppm ammonia. This corresponded to an approximate 8-fold increase in impedance over the baseline response taken at 3600 s. 10 ppm resulted in an approximate 2.5-fold increase in impedance. Thus, the ability of the probe to detect changes in ammonia concentration across the required range in appropriate secondary refrigerants at refrigeration temperatures was established.

Fig. 3 Impedimetric response of the ammonia probe to ammonia in the simulated refrigeration system. The refrigerant was 23.3% (w/v) NaCl in water at 4 °C. Values indicate the final concentration of sequential ammonia injections. Impedance was normalised relative to baseline response at 3600 s. Impedance measurement was at 180 Hz, 5 mV rms.

The effect of temperature on sensor baseline impedance

As discussed earlier, temperature has several effects on the system's response to ammonia. It has already been pointed out that the equilibrium of the three phase system will be affected by temperature. In addition, the impedimetric signal of the sensor both in the absence and presence of ammonia will also be strongly influenced by temperature. As it was desired that the probe could operate across a range of typical refrigerant temperatures (+4 to −15 °C), the effect of temperature across this range on the behaviour of the system was investigated. Several approaches can be taken to compensate for changes or differences in operational temperature. One method would be to monitor temperature and include a temperature coefficient in the sensor calibration algorithm. Such a method would allow accurate and quantitative measurement of ammonia. However, our application was focussed on leak detection and not quantification. In this way, complex temperature monitoring would not be desirable. However, any algorithm used to calculate the concentration of ammonia in the system would have to be capable of dealing with a significant variation in responses due to temperature while still being capable of correctly differentiating between the presence or absence of a leak across this range of temperatures. A threshold value of 50 ppm was agreed as an appropriate level at which to trigger a positive result indicative of an ammonia leak. It is assumed that the nature of a leak will result in rapid accumulation of ammonia in the secondary refrigerant which will quickly reach concentrations well in excess of 50 ppm.

To determine the system's suitability for this mode of operation, the response of the sensor in the presence and absence of ammonia across the relevant temperature range was studied. Fig. 4 shows the impact of operational temperature on the normalised impedance response of the ammonia probe in brine following a change from ambient temperature (18 °C) down to the temperatures of interest for refrigeration (+4 to −15 °C). To allow comparison of the effect of different temperatures, the data was normalised with respect to the impedance at 300 s. Following re-stabilisation of the temperature of the system, it can be seen that the sensors moved to new baseline values. It can also be seen that the baseline impedance of the sensors increased with decreasing temperature. This is a well-known feature of conducting polymers in which the conductivity of the film is reduced at lower temperatures.²⁷ The most significant change in baseline impedance, as recorded at −15 °C, was still less than a factor of 1.3. This change was less than that seen for a 5 ppm injection of ammonia at +4 °C in Fig. 3. Thus, a change in temperature between 18 and −15 °C would only bring about a change in impedance that was less than the change in impedance generated for 5 ppm ammonia measured at the highest operational temperature. Taking 50 ppm as the threshold for indication of an alarm, this resulted in a change in impedance in excess of 6-fold. At this concentration, the change in baseline impedance due to temperature would have a negligible effect in that it would not lead to the erroneous indication of a leak. Thus, the probe could be calibrated at a single baseline temperature and operated across the full temperature range.

Fig. 4 Background variation of sensors with temperature (from an initial temperature of 18 ± 1 °C). Experiment carried out in flow system using 23.3% NaCl in water. Normalised impedance is on the same scale as in Fig. 3 to allow comparison with ammonia response. Inset: detail of normalised impedance from 1.0 to 1.4.

The effect of temperature on probe response to ammonia

Although the effect of temperature on the baseline signal response of the probe had been determined, it was also necessary to look at the effect that temperature had on the response of the probe to ammonia. Following baseline stabilisation of the sensors for 1 h, sequential injections of ammonia were again made at +4, 0, −5, −10 and −15 °C and the impedance responses at 180 Hz recorded (Fig. 5). In most cases, the response to ammonia across the different temperature ranges were similar and closely overlapping, with the exception of measurements at −10 °C. The reason for this particular discrepancy is not clear, but is more likely to stem from inter-electrode variations than specific changes to ammonia by the sensor at this temperature. In general, there was a consistent relationship between sensor response sensitivity and temperature across the range of temperatures studied, going from approx. 20 kΩ at 0 ppm, up to approx. 370 kΩ at 100 ppm. However, given that there appears to be significant overlap of responses across the temperature range tested, and given that the partial pressure of ammonia in the gas phase would be expected to increase with decreased temperature, it might be suggested that the sensor's sensitivity to ammonia may be decreasing at lower temperatures, so offsetting this difference.

Fig. 5 Change in impedance of the ammonia probe in response to concentrations of ammonia from 0 to 100 ppm at 4 (●), 0 (○), −5 (▼), −10 (∇) and −15 °C (■). Inset: relationship between impedance change and temperature at different temperatures. The probe was allowed to equilibrate at temperature for 1 h prior to injections.

To further assess whether it would be viable to operate the probe without temperature correction, the impedance data collected across the range of temperatures was averaged and plotted against concentration (Fig. 6). Significant linearity was lost at 100 ppm due to saturation of the polymer film. However, the calibration could be effectively linearised using a semi-log plot with an equation of y = 117.8ln x − 208.2. The averaging of the responses over the temperature range yielded significant deviations. It should also be noted that responses at 5 and 10 ppm ammonia could not be differentiated from zero due to the range of temperatures measured. However, responses at 50 ppm were clearly distinguishable from zero values and across the entire temperature range would correspond to concentrations of 35 to 80 ppm ammonia. Thus, the system would be extremely useful for monitoring leak detection across this range of temperatures without the need for further calibration or monitoring.

Fig. 6 Impedimetric response of the ammonia probe to ammonia concentrations from 1 to 100 ppm across the temperature range of +4 to −15 °C. Impedance measurements were performed at 180 Hz, 5 mV rms. Inset shows linearised plot, y = 117.8ln x − 208.2.

Nonetheless, improvements could be readily incorporated by the addition of defined temperature calibrations, with or without the addition of continuous temperature monitoring.

As briefly stated earlier, several approaches have been taken to the measurement of ammonia. However, only one other technology is suitable for measuring refrigerants. The MiniCal III® is based on ion-selective electrode technology and can also measure in the industrially relevant range of 0 to 100 ppm.²⁸ However, it operates by diverting a sample from the refrigerant system and as a result, requires pumps and valving. The ion-selective electrode cannot withstand temperatures below 0 °C and so the system must also heat the sample before measurement. The current system is a probe-based device which can remain in continuous contact with refrigerant and which can also directly measure across the range of relevant refrigerant concentrations.

Conclusions

An ammonia measurement probe was developed based on an ammonia sensitive electrode fabricated from conducting polymer nanoparticles and placed behind a PTFE membrane. In combination with impedance analysis, this allowed the probe to operate continuously in contact with a refrigerant fluid to monitor ammonia concentrations. The probe could measure concentrations of ammonia from 0 to 100 ppm in a circulating secondary refrigerant containing brine across the temperature range of +4 to −15 °C and a single calibration curve could be used to detect the presence of an ammonia leak. By setting a threshold of 50 ppm, this would allow detection of ammonia leakage across this temperature range at levels between 35 and 80 ppm.

Acknowledgements

This work was supported by Enterprise Ireland under grant number IP-2008-0538.

Notes and references

1. J. A. Mader, J. Staehelin, T. Peter, D. Brunner, H. E. Rieder and W. A. Stahel, Atmos. Chem. Phys., 2010, 10(24), 12161–12171.
2. http://www.epa.gov/ozone/title6/phaseout/22phaseout.html, accessed 28 March 2012.
3. Regulation (EC) no 2037/2000 of the European Parliament and of the Council of 29 June 2000 on substances that deplete the ozone layer, 2000.
4. P. de Larminat, ASHRAE J., 2000, 42(3), 35–40.
5. C. Zhang, J. N. Chen and Z. G. Wen, J. Cleaner Prod., 2012, 25, 96–105.
6. K. Wang, M. Eisele, Y. Hwang and R. Radermacher, Int. J. Refrig., 2010, 33(2), 212–234.
7. N. G. Muralidharan, R. Kaul and K. V. Kasiviswandathan, J. Mater. Eng. Perform., 1996, 5(2), 241–246.
8. F. Karim, F. Hijaz, C. L. Kastner and J. S. Smith, J. Food Sci., 2010, 75(2), T35–T39.
9. M. S. Dworkin, A. Patel, M. Fennell, M. Vollmer, S. Bailey, J. Bloom, K. Mudahar and R. Lucht, J. Food Prot., 2004, 67(6), 1299–1302.
10. R. Moos and D. Schonauer, Sens. Lett., 2008, 6(6), 821–825.
11. B. Timmer, W. Olthuis and A. van den Berg, Sens. Actuators, B, 2005, 107(2), 666–677.
12. A. Elia, P. M. Lugara, C. Di Franco and V. Spagnolo, Sensors, 2009, 9(12), 9616–9628.
13. S. Hiki, K. Mawatari, A. Aota, M. Saito and T. Kitamori, Anal. Chem., 2011, 83(12), 5017–5022.
14. G. N. Dar, A. Umar, S. A. Zaidi, S. Baskoutas, S. W. Hwang, M. Abaker, A. Al-Hajry and S. A. Al-Sayari, Talanta, 2012, 89, 155–161.
15. M. M. Rahman, A. Jamal, S. B. Khan and M. Faisal, ACS Appl. Mater. Interfaces, 2011, 3(4), 1346–1351.
16. S. B. Khan, M. M. Rahman, E. S. Jang, K. Akhtar and H. Han, Talanta, 2011, 84(4), 1005–1010.
17. J. Castrellon-Uribe, M. E. Nicho and G. Reyes-Merino, Sens. Actuators, B, 2009, 141(1), 40–44.
18. M. M. Rahman, A. Jamal, S. B. Khan and M. Faisal, J. Nanopart. Res., 2011, 13(9), 3789–3799.
19. O. Ngamna, A. Morrin, A. J. Killard, S. E. Moulton, M. R. Smyth and G. G. Wallace, Langmuir, 2007, 23(16), 8569–8574.
20. K. Crowley, A. Morrin, A. Hernandez, E. O'Malley, P. G. Whitten, G. G. Wallace, M. R. Smyth and A. J. Killard, Talanta, 2008, 77(2), 710–717.
21. K. Crowley, E. O'Malley, A. Morrin, M. R. Smyth and A. J. Killard, Analyst, 2008, 133(3), 391–399.
22. I. Tarsiche, E. Hopirtean and D. Ciurchea, Meas. Sci. Technol., 1997, 8(11), 1367–1371.
23. J. Junsomboon and J. Jakmunee, Anal. Chim. Acta, 2008, 627(2), 232–238.
24. P. Atkins and J. de Paula, Physical Chemistry, Oxford University Press, UK, 9th edn, 2009.
25. Data on Henry's Law constant for ammonia derived from data compiled by the US National Institute of Standards and Technology [http://webbook.nist.gov/cgi/, accessed 21 June 2012] and based on the following sources: S. L. Clegg and P. Brimblecombe, J. Phys. Chem., 1989, 93, 7237–7238; P. G. Dasgupta and S. Dong, Atmos. Environ., 1986, 20, 565–570; W. L. Chameides, J. Geophys. Res., 1984, 89D, 4739–4755; M. R. Hoffmann and D. J. Jacob, Kinetics and Mechanisms of the Catalytic Oxidation of Dissolved Sulfur Dioxide in Aqueous Solution: An Application to Nighttime Fog Water Chemistry in SO₂, NO and NO₂ Oxidation Mechanisms: Atmospheric Considerations, ed. J. G. Calvert, Butterworth Publishers, Boston, MA, 1984, pp. 101–172; G. Holzwarth, R. G. Balmer and L. Soni, Water Res., 1984, 18, 1421–1427; L. R. Martin, Kinetic Studies of Sulfite Oxidation in Aqueous Solution in SO₂, NO and NO₂ Oxidation Mechanisms, Atmospheric Considerations, ed. J. G. Calvert, Butterworth Publishers, Boston, MA, 1984, pp. 63–100; R. Bone, P. Cullis and R. Wolfenden, J. Am. Chem. Soc., 1983, 105, 1339–1343; G. B. Carter, M. C. McIver and G. J. Miller, J. Chem. Soc. C, 1968, 2591–2592; M. C. Kavanaugh and R. R. Trussell, J. – Am. Water Works Assoc., 1980, 72, 684–692; J. M. Hales and D. R. Drewes, Atmos. Environ., 1979, 13, 1133–1147; T. J. Edwards, G. Maurer, J. Newman and J. M. Prausnitz, AIChE J., 1978, 24, 966–976; R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworth, London, 2nd edn, 1970; L. G. Sillen and A. E. Martell, Stability Constants of Metal-ion Complexes, The Chemical Society, London, 1964; D. W. Van Krevelen, P. J. Hoftijzer and F. J. Huntjens, Recl. Trav. Chim. Pays-Bas, 1949, 68, 191.
26. R. Sing, B. Rumpf and G. Maurer, Ind. Eng. Chem. Res., 1999, 38(5), 2098–2109.
27. S. Kim, J. M. Ko and I. J. Chung, Polym. Adv. Technol., 1996, 7(7), 599–603.
28. http://www.gfg.ch, accessed 28 March 2012.

---