The detection of gaseous ammonia using an optical evanescent wave sensor coated with a PVC film containing a chromogenic calixarene (nitrophenyl azophenol calix[4]arene) is described. On addition of varying concentrations of ammonia gas, the absorbance maximum shifts to 500 nm which is characteristic of deprotonation of the chromogenic group. The sensitivity of the sensor to ammonia is shown to vary depending on whether the free ligand or lithium-complex is used in the membrane. Intermediate sensitivities can be generated by varying the mole ratio of lithium to calixarene. The response time of the lithium complex measured at 520 nm to a 5ppm to 50 ppm change in ammonia concentration is fast (several minutes to steady state) but the signal is slow to return to baseline when the gas is replaced by nitrogen.

The determination of ammonia gas is important in clinical, environmental and industrial process analyses. Ammonia is recognised as one of the primary irritants to humans and concentrations of 25 ppm in air have been recommended as threshold limit value for human exposure.

We now present results obtained with a new optical sensor for gaseous ammonia based on a calix[4]arene to which a nitrophenylazophenol chromophore is attached. This compound belongs to a family of chromogenic calixarenes which display a dramatic change in absorption upon complexation with lithium and to a lesser extent with sodium and potassium, in the presence of an appropriate base. An extremely rapid concentration dependent colour change from yellow to red, corresponding to a shift in the absorbance maximum from 380 to 520 nm, occurs upon the addition of lithium perchlorate to this calix[4]arene in the presence of tridodecylamine. No colour change is observed in the absence of base.

Figure 1: Structure of the nitrophenylazophenol calix[4]arene and the reactions involved in the colour generation.

 In this approach, reactions which could occur can be summarised as:

L¾ COH_(m) + M⁺_(m) LM^+¾ COH_(m) (1)

(1) Complexation of a metal ion by the ligand (L)^$ in the absence of a base. In this case there is no colour change as the chromophore (COH) is not deprotonated.

L¾ COH_(m) + B_(m) L¾ CO- _(m) + BH⁺_(m) (2)

(2) Deprotonation of the chromophore by a base (B). This will cause a colour change independent of complexation which is to be avoided if one is designing a system for determining a target metal ion. On the other hand, if the base is the target species, then this colour change is desirable.

Where;

 In contrast to simple acid-base indicators, these ligand-based systems offer the additional possibility of subsequent complexation of the protonated base by the ligand, which may lead to selectivity in observed response to volatile bases, as a ‘best-fit’ mechanism would be involved, implying that bases with differing sizes or shapes could be selectively distinguished.

L¾ CO- _(m) + BH⁺_(m) L BH^+¾ CO- _(m) (3)

Furthermore, the metal ion complex can be predicted to have a more acidic nature that the equivalent free ligand.

LM^+¾ COH_(m) + B_(m) LM^+¾ CO- _(m) + BH⁺_(m) (4)

In this case, the cation complex is deprotonated by the base, generating the colour change in the process. The presence of the cation in close proximity to the chromophore renders the loss of the proton more facile due to stabilisation of the resulting phenolate anion, with the consequent effect of increasing the apparent acidity of the labile proton. Clearly, this system is much simpler than that based on the use of chomoionophores which must be in the protonated form as only a single reagent is required (chromophore is attached to the ligand) instead of three (ligand, protonated indicator and counter ion). On the other hand, it requires more emphasis on synthesis, as combined chromophore-ligand is required for every target species.

However, it is the potential of this compound to detect basic gases as defined by equations (2) and (4) above, and in particular, the possibility of varying the sensitivity by doping the film with metal ions which have differing effects on the chromophore acidity, which prompted us to investigate the effect of ammonia gas on a PVC membrane which incorporated the nitrophenylazophenol calix[4]arene. Both the free ligand and metal complexes were examined as possible ammonia sensors. In contrast to the approach employed in a previous investigation, which demonstrated the viability of using this compound as a sensitive visual indicator for trimethylamine, we have adopted a membrane-coated optical fibre configuration in this study.

Figure 2: Schematic diagram of the instrumental set-up used for the optical fibre measurements.

Typical responses to a range of ammonia concentrations obtained with optical fibres coated with the calixarene-doped PVC are shown in Figure 3 for the free ligand (a) and the Li-complex (b) in the wavelength range of 400 to 750 nm. All spectra were recorded relative to a scan of the coated fibre in nitrogen. In Figure 3 (a) the shift of absorbance maximum to 500 nm characteristic of deprotonation of the azophenol nitrophenyl chromophore occurs on the addition of varying concentrations of ammonia gas in the range of 5 ppm to 500 ppm. Clearly, the absorbance increases with the concentration of ammonia but the values are quite small, (the addition of 500 ppm leads to an absorbance of 0.12).

Figure 3: (a) Optical responses of the free ligand to gaseous ammonia. A, 500 ppm; B, 200 ppm; C, 100 ppm; D, 50 ppm; E, 20 ppm; F, 5 ppm; G, N₂. (b) Optical responses of the lithium complex to gaseous ammonia. A, 100 ppm; B, 50 ppm; C, 20 ppm; D, 10 ppm; E, 5 ppm; F, 2 ppm; G, N₂. A scan of the coated fibre in nitrogen is used as a reference.

Figure 3(b) shows the absorption spectrum of the optical fibre coated with a PVC membrane containing the ligand and lithium perchlorate at 1:10 mole ratio. The formation of the Li-complex has greatly increased the sensitivity to gaseous ammonia as expected, with the absorbance at the l _max (520 nm) reaching 1.2 on exposure of the optical fibre to 100 ppm NH₃.

From these results, it is clear that the sensitivity of the membrane to ammonia can be varied by changing the active species from the free ligand to the Li-complex, and that intermediate sensitivities can be generated through varying the Li-to-ligand mole ratio, or by changing the metal ion from lithium to potassium. The very large response obtained with the Li-complex to relatively low levels of ammonia (e.g. 0.7 absorbance at 10 ppm ammonia in Figure 3b) demonstrate that this sensor could easily detect this gas at low ppm levels. Furthermore, we predict that with appropriate optimisation, detection levels may well be sub-ppm. However, there is a shift in the l _max from around 500 nm for the free to 520 nm for the 1:10 mole ratio Li-Complex at higher ammonia concentrations (see Figure 3). We are currently investigating whether this shift is due to deprotonation of a second azophenol nitrophenol group on the calixarene.

 Figure 4: The response time of the optical fibre coated with the lithium complex (1:10 Li-ligand mole ratio) to a step change in concentration from 5 ppm to 50 ppm ammonia.

 Figure 4 shows the time response of the lithium complex coated optical fibre represented by the absorbance measured at 520 nm. The response from 5 ppm to 50 ppm ammonia is of the order of a few minutes but the signal is extremely slow to return to baseline when the ammonia is replaced by nitrogen. The recovery times between measurements are very slow (up to one hour) and are concentration dependent. However, it is well known that the diffusion of ammonia in plasticised PVC films is slow compared to silicone films (the permeability coefficient of silicone for ammonia is ca. 8500 compared to 3.7 for PVC). We are therefore confident that the response time of the sensor can be significantly improved by using silicone based films rather than PVC-plasticised films. Given the very sensitive responses of the 1:10 mole ratio Li-ligand membrane and the use of an evanescent wave approach, there is clearly scope for improving response times without unduly affecting sensitivity by reducing the film thickness to around 100 nm, as this is the approximate distance of penetration of the evanescent wave into such films. We also intend to investigate planar waveguides as device substrates for further work, rather than coated fibres, as characterisation and optimisation is simpler (e.g. film thickness can be more easily controlled and measured).

In terms of storage, the ligand is very stable, and batches have been successfully used over a period of 18 months so far. Possible applications of this system include the distribution detection of ammonia in large refrigeration units using coated fibres or the development of low-cost LED-compatible compact ammonia sensors. It should also be possible to couple ammonia detection with enzyme catalysed reactions to enable a broad range of optical biosensors for important target species such as biogenic amines to be developed. The fact that the linear range and sensitivity can ‘tuned’ by varying the identity and amount of dopant cations introduces an additional flexibility not normally available in these measurements.