**3.1 Operating principle of the sensor**

An untreated silica PCF is used for the fabrication of the PCFI, its surface is hydrophilic and there fore the adsorption of water vapour on the surface occurs when it is exposed to humid air. Two types of water-vapour adsorption mechanisms occur in sequence at the SiO2-air interface. The chemisorption of water vapour first modifies the SiO2 surface, resulting in a surface with silanol groups (Si-OH). The second type of adsorption, physisorption, occurs on these silanol groups. A schematic illustration of the water-vapour adsorption is given in Fig. 4. At room temperature the physisorption is a reversible function of the relative humidity of the surrounding air, while the chemisorption appears to be irreversible [Voorthuyzen et al., 1987]. So in the succeeding discussion only the physisorption is considered. Awakuni and Calderwood [Awakuni & Calderwood, 1972] investigated the adsorption of water vapour on the SiO2 surface. They measured the amount of adsorbed water as a function of the partial vapour pressure at a constant temperature. It appeared that this so-called adsorption isotherm can be described very well by the BET (Brunauer-Emmett- Teller) adsorption theory [Brunauer et al., 1938].

The evolution of adsorbed water layer structure on silicon oxide at room temperature is demonstrated by David and Seong in [David & Seong, 2005]. They determined the molecular configuration of water adsorbed on a hydrophilic silicon oxide surface at room temperature as a function of relative humidity using attenuated total reflection (ATR) infrared spectroscopy. A completely hydrogen-bonded ice like network of water grows up as the relative humidity increases from 0 to 30%. In the relative humidity range of 30-60%, the liquid water structure starts appearing while the ice like structure continues growing to saturation. Above 60% relative humidity, the liquid water configuration grows on top of the ice like layer. This structural evolution indicates that the outermost layer of the adsorbed water molecules undergoes transitions in equilibrium behaviour as humidity varies. Also it was shown from the adsorption isotherm that the thickness of the adsorbed layer at room temperature starts increasing exponentially above 60% RH.

Tiefenthaler and Lukosz [Tiefenthaler & Lukosz, 1985] have shown that adsorption and desorption of water vapour by the surface of a waveguide changes the effective refractive index (RI) of the guided modes, in their case for a humidity sensor based on an integrated optical grating coupler. In the case of a PCFI a similar adsorption of water vapour changes the effective refractive index (ncl) of the interfering cladding mode propagating in the PCF. Since this adsorption/physisorption is a reversible process, a modulation of the ncl occurs

counterparts, optical fibre humidity sensors offer specific advantages, such as small size and weight, immunity to electromagnetic interference, corrosion resistance and remote operation. A wide range of optical fibre humidity sensors have been reported in the literature. Most of these fibre optic humidity sensors work on the basis of a hygroscopic material coated over the optical fibre to modulate the light propagating through the fibre [Yeo et al., 2008; Mathew et al. 2007, 2011]. A polymer optical fibre has been adapted for humidity sensing [Zhang et al., 2010] without the use of a hygroscopic coating but the fibre is highly temperature dependent and is not suitable for high-temperature applications. An all-glass fibre-optic relative humidity sensor which does not require any special coatings to measure humidity using a reflection-type two-mode photonic crystal fibre interferometer is presented in this section. The spectrum of it exhibits good sensitivity to humidity variations.

An untreated silica PCF is used for the fabrication of the PCFI, its surface is hydrophilic and there fore the adsorption of water vapour on the surface occurs when it is exposed to humid air. Two types of water-vapour adsorption mechanisms occur in sequence at the SiO2-air interface. The chemisorption of water vapour first modifies the SiO2 surface, resulting in a surface with silanol groups (Si-OH). The second type of adsorption, physisorption, occurs on these silanol groups. A schematic illustration of the water-vapour adsorption is given in Fig. 4. At room temperature the physisorption is a reversible function of the relative humidity of the surrounding air, while the chemisorption appears to be irreversible [Voorthuyzen et al., 1987]. So in the succeeding discussion only the physisorption is considered. Awakuni and Calderwood [Awakuni & Calderwood, 1972] investigated the adsorption of water vapour on the SiO2 surface. They measured the amount of adsorbed water as a function of the partial vapour pressure at a constant temperature. It appeared that this so-called adsorption isotherm can be described very well by the BET (Brunauer-

The evolution of adsorbed water layer structure on silicon oxide at room temperature is demonstrated by David and Seong in [David & Seong, 2005]. They determined the molecular configuration of water adsorbed on a hydrophilic silicon oxide surface at room temperature as a function of relative humidity using attenuated total reflection (ATR) infrared spectroscopy. A completely hydrogen-bonded ice like network of water grows up as the relative humidity increases from 0 to 30%. In the relative humidity range of 30-60%, the liquid water structure starts appearing while the ice like structure continues growing to saturation. Above 60% relative humidity, the liquid water configuration grows on top of the ice like layer. This structural evolution indicates that the outermost layer of the adsorbed water molecules undergoes transitions in equilibrium behaviour as humidity varies. Also it was shown from the adsorption isotherm that the thickness of the adsorbed layer at room

Tiefenthaler and Lukosz [Tiefenthaler & Lukosz, 1985] have shown that adsorption and desorption of water vapour by the surface of a waveguide changes the effective refractive index (RI) of the guided modes, in their case for a humidity sensor based on an integrated optical grating coupler. In the case of a PCFI a similar adsorption of water vapour changes the effective refractive index (ncl) of the interfering cladding mode propagating in the PCF. Since this adsorption/physisorption is a reversible process, a modulation of the ncl occurs

**3.1 Operating principle of the sensor** 

Emmett- Teller) adsorption theory [Brunauer et al., 1938].

temperature starts increasing exponentially above 60% RH.

with respect to the ambient humidity values which in turn change the position of the interference pattern accordingly. An increase in humidity causes the shift of the interference pattern of a PCFI toward longer wavelengths and the value of this interference peak shift is exponential with respect to relative humidity [Mathew, 2010]. This shift of the interference peak is mainly due to the adsorption and desorption of H2O molecules along the surface of holes within the PCF, at the interface between air and silica glass. Since the whole device is exposed to humidity the adsorption and desorption of water vapour on the PCF outer surface and on the end face also contribute to the shift of the interference pattern. But considering the field distribution of the interfering cladding mode shown in [Cárdenas-Sevilla et al., 2011; Uranus, 2010] and below the dew point temperature the main contribution to the interference shift is considered to be due to the adsorption of water molecules within the voids of the PCF. The adsorption on the end face mainly causes a shift in the overall power level of the interference pattern.

Fig. 4. Schematic representation of water vapor adsorption mechanisms on an SiO2 surface.

#### **3.2 Experimental characterization of the sensor**

The sensor system is composed of a broadband light source (SLED), a fibre coupler/circulator (FOC), the PCF interferometer or sensor head, and an optical spectrum analyser (OSA) as shown in Fig. 5. The sensor head, as the main part of the sensor system, is composed of a small stub of PCF fusion spliced to the end of a standard SMF. The PCF in the sensor head has a microhole collapsed region near the splicing point and the free end of the PCF is exposed to ambient air. The humidity response of the device was studied at a temperature (25 OC) and at normal atmospheric pressure by placing it in a controlled environmental chamber as shown in Fig. 5. Fig. 6 shows the changes in the reflection spectrum with respect to ambient humidity for a device with L=40.5 mm. The change in the adsorption with respect to ambient humidity changes the effective refractive index of the cladding mode (ncl). The resulting phase change in turn results in a shift of the interference pattern. The curves in Fig. 6 show the position of a zoomed section of the device spectrum at relative humidity values of 30, 60, 80 and 90 %RH. When humidity increases the interference pattern shifts to longer wavelengths and this shift is more significant at higher humidity values. To study the effect of reducing the length of the PCFI a second PCFI was fabricated with a shorter length of 17 mm. Fig. 7 shows the peak shift of the interferometer with respect to humidity obtained for two devices with L=17 mm and 40.5 mm.

It is observed from the Fig. 7 that the sensitivity of the device to humidity decreases as the length of the device decreases. This is due to the fact that for a small device the fibre length

Photonic Crystal Fibre Interferometer for Humidity Sensing 167

for the PCFI with a length of 40.5 mm in these regions are 3.7, 8.5 and 64 nm/%RH respectively and for a 17 mm long PCFI they are 1.7, 3 and 23 nm/%RH respectively. Even though the PCFI with a longer length appears more sensitive, it is likely that increasing the length of the PCFI to a much longer length is not practical because in a longer device the infiltration of water molecules may take too much time. Furthermore, since the propagation loss of the interfering cladding mode is high the fringe visibility will diminish on increasing the length of PCF. Also for a longer device the fringe spacing will be shorter which limits the measurement range of the device. Decreasing the length of the PCFI to a much shorter length is also not suitable because as seen from Fig. 2 & 3 if the length is less than 3.5 mm the fringe spacing will be greater than 100 nm, the bandwidth of a typical SLED spectrum, and therefore not suitable for observing the shift in the interference spectrum. Selecting a shorter length will also result in a reduced sensitivity but that can be improved by infiltrating the microholes with suitable hygroscopic materials. Based on our experimental observations and considering the above explained factors we suggest the best lengths for an

The response of the PCFI to humidity variations is found to be reversible and repeatable with low hysteresis. Under laboratory conditions it is reusable, but humidity is a truly analytical measurement in which the sensor must be in direct contact with the process environment. This of course has implications of contamination and degradation of the sensor to varying degrees depending on the nature of the environment. Possible contamination agents are dust particles and chemical vapours. So a further study of the sensor head contamination in different process environments and the observation of the shift in its response in those conditions are required in order to get a better understanding of the long term stability of our sensor in field applications. In the case of a PCFI based sensor this limitation can be overcome by different ways; a recalibration of the sensor head after a certain period of time and a subsequent reuse of the sensor head during another time interval, or, since the fabrication of the PCFI based sensor head is simple and cost effective, replacing the sensor head or attaching some filters to the sensor head by which it can be protected from contamination or an ultrasonic cleaning and subsequent heating (which will remove the contaminants like dust particles without damaging the sensor head) is

A study of cross sensitivity to temperature reveals that the PCFI based humidity sensor is almost temperature independent. Conventional glass fibre relative humidity sensors require coatings and thus are always temperature dependent and, furthermore, since the majority of such sensors use polymer materials as coatings, they are not suitable for use in hightemperature applications. One significant advantage of the sensor explained here is that the sensor head is made of single material silica. This suggests that apart from low and room temperature applications the PCF interferometer based humidity sensor can also be used in

Dew (condensed moisture) is a problem in the fields of precision electrical devices, automobiles, air conditioning systems, warehouses and domestic equipment, etc. High humidity and condensation can create an environment where the development of mould on the wooden parts can take place and it can also cause corrosion of iron parts. This is a major

efficient humidity sensing to be in the range from 3.5 mm to 100 mm.

another method to make the sensor reusable after contamination.

harsh and high-temperature environments to monitor humidity.

**4. Dew sensor based on PCFI** 

available for interaction between the cladding mode with the adsorbed water vapour is less so the acquired phase difference between the interfering modes will be smaller. Hence the sensitivity to humidity change is less for a device with a smaller length of PCF. It is important to note that the shift of the interference pattern is similar to the thickness variation of the adsorbed layer of water vapour on silica i.e. increases exponentially above 60 %RH [David & Seong, 2005].

Fig. 5. Experimental arrangement for the characterisation of the PCFI with respect to relative humidity.

Fig. 6. Reflection spectrum of a 40.5 mm long PCFI at different humidity values.

The device sensitivity is estimated by dividing the PCFI response to humidity into three regions 27- 60 %RH, 60-80 %RH, and 80-96 %RH. The average sensitivity values observed

available for interaction between the cladding mode with the adsorbed water vapour is less so the acquired phase difference between the interfering modes will be smaller. Hence the sensitivity to humidity change is less for a device with a smaller length of PCF. It is important to note that the shift of the interference pattern is similar to the thickness variation of the adsorbed layer of water vapour on silica i.e. increases exponentially above

Fig. 5. Experimental arrangement for the characterisation of the PCFI with respect to relative

Fig. 6. Reflection spectrum of a 40.5 mm long PCFI at different humidity values.

The device sensitivity is estimated by dividing the PCFI response to humidity into three regions 27- 60 %RH, 60-80 %RH, and 80-96 %RH. The average sensitivity values observed

60 %RH [David & Seong, 2005].

humidity.

for the PCFI with a length of 40.5 mm in these regions are 3.7, 8.5 and 64 nm/%RH respectively and for a 17 mm long PCFI they are 1.7, 3 and 23 nm/%RH respectively. Even though the PCFI with a longer length appears more sensitive, it is likely that increasing the length of the PCFI to a much longer length is not practical because in a longer device the infiltration of water molecules may take too much time. Furthermore, since the propagation loss of the interfering cladding mode is high the fringe visibility will diminish on increasing the length of PCF. Also for a longer device the fringe spacing will be shorter which limits the measurement range of the device. Decreasing the length of the PCFI to a much shorter length is also not suitable because as seen from Fig. 2 & 3 if the length is less than 3.5 mm the fringe spacing will be greater than 100 nm, the bandwidth of a typical SLED spectrum, and therefore not suitable for observing the shift in the interference spectrum. Selecting a shorter length will also result in a reduced sensitivity but that can be improved by infiltrating the microholes with suitable hygroscopic materials. Based on our experimental observations and considering the above explained factors we suggest the best lengths for an efficient humidity sensing to be in the range from 3.5 mm to 100 mm.

The response of the PCFI to humidity variations is found to be reversible and repeatable with low hysteresis. Under laboratory conditions it is reusable, but humidity is a truly analytical measurement in which the sensor must be in direct contact with the process environment. This of course has implications of contamination and degradation of the sensor to varying degrees depending on the nature of the environment. Possible contamination agents are dust particles and chemical vapours. So a further study of the sensor head contamination in different process environments and the observation of the shift in its response in those conditions are required in order to get a better understanding of the long term stability of our sensor in field applications. In the case of a PCFI based sensor this limitation can be overcome by different ways; a recalibration of the sensor head after a certain period of time and a subsequent reuse of the sensor head during another time interval, or, since the fabrication of the PCFI based sensor head is simple and cost effective, replacing the sensor head or attaching some filters to the sensor head by which it can be protected from contamination or an ultrasonic cleaning and subsequent heating (which will remove the contaminants like dust particles without damaging the sensor head) is another method to make the sensor reusable after contamination.

A study of cross sensitivity to temperature reveals that the PCFI based humidity sensor is almost temperature independent. Conventional glass fibre relative humidity sensors require coatings and thus are always temperature dependent and, furthermore, since the majority of such sensors use polymer materials as coatings, they are not suitable for use in hightemperature applications. One significant advantage of the sensor explained here is that the sensor head is made of single material silica. This suggests that apart from low and room temperature applications the PCF interferometer based humidity sensor can also be used in harsh and high-temperature environments to monitor humidity.
