- Email: [email protected]
Detection of high humidity by optical fibre sensing at telecommunications wavelengths
15 January 1998
Optics Communications 146 Ž1998. 90–94
Detection of high humidity by optical fibre sensing at telecommunications wavelengths D.C. Bownass ) , J.S. Barton, J.D.C. Jones Physics Department, Heriot-Watt UniÕersity, Riccarton, Edinburgh, EH14 4AS, UK Received 9 July 1997; accepted 3 September 1997
Abstract We demonstrate the proof of principle for a novel fibre optic sensor for the detection of high humidity. The device uses single-mode fibre at telecommunications wavelengths, making it particularly suitable for application in passive optical networks. The sensing mechanism relies on a humidity-induced refractive index change in an organic film coated onto the cladding of a bent fibre. The sensor was shown to respond at relative humidities ) 62% and ) 80% using gelatine and polyethylene oxide films respectively, by the onset of a strong interaction between the LP01 mode and a whispering gallery mode. q 1998 Elsevier Science B.V. PACS: 42.81.P; 92.60.J; 84.40.U Keywords: Fibre-optic sensors; Humidity detection; Whispering gallery interaction; Network monitoring; Telecommunications
1. Introduction It is known that the lifetimes of components used in passive optical communications networks depend on environmental conditions, with failure hastened by exposure to high temperature and humidity w1,2x. Network monitoring may be desirable, preferably based on passive sensors compatible with passive networks. Fibre optic sensors have already been developed for variables such as temperature and strain w3–5x, with potential for network monitoring. This paper is concerned with the development of a passive optical sensor for humidity monitoring. Sensors have been reported for the detection of water in the liquid, rather than vapour, phase w6,7x. This may not be sufficient, as components such as optical splitters could be affected by high humidity, well before liquid water enters the device housing. If a humidity sensor were incorporated close to vulnerable components, then preventative mainte-
)
E-mail: [email protected].
nance prior to system failure would be feasible, leading to a reduction in costs. Several fibre optic humidity sensors have been described previously, but none are suitable for passive optical communications systems. For example, point sensors have been reported which use the end of a fibre coated with either a metal w8x or dielectric w9x film. These sensors are incompatible with a serial array and the use of optical time domain reflectometry ŽOTDR. for interrogating the array. Others have used special fibres with the cladding removed and replaced with either porous silica w10x or a special polymer w11x, but require multimode fibre, making them incompatible with the standard components used in a telecommunications application. Another technique employs the spectral absorption of CoCl 2 entrapped in a gelatine w12x or polymer w13x host which is then coated onto the fibre. These sensors require multimode fibre, and the use of visible wavelengths, again making them unsuitable for a telecommunications network. We have recently investigated an alternative type of fibre-optic humidity sensor w14x in which the cladding of
0030-4018r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 0 3 0 - 4 0 1 8 Ž 9 7 . 0 0 5 2 5 - 7
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
the fibre is polished away and an overlay is applied whose refractive index depends on relative humidity, thus influencing the insertion loss of the polished section of the fibre. In the present work, we develop the idea of using a humidity-sensitive overlay, but avoid the requirement for precise polishing of the fibre cladding. Instead, the fibre is bent in the sensing region, so that the overlay influences the propagation of whispering gallery modes in the fibre cladding. The sensor described here uses single-mode fibre and wavelengths in the telecommunications range. It is suitable for time division multiplexing in a serial array with interrogation by OTDR, and is small, simple and lightweight.
91
water content of the film will cause it to swell slightly, resulting in a reduction in the overall density and hence a corresponding drop in refractive index. When the film has a refractive index greater than that of the silica cladding, as occurs at low humidities in the experiments described here, there is only a very weak Fresnel reflection of the WG mode at the cladding-overlay interface, and the interference with the core guided mode is negligible. At high humidities, the film has a refractive index lower than that of the cladding and hence total internal reflection of the WG mode can occur, resulting in strong interference with the core mode, revealed by oscillations in the transmission of the bend as a function of wavelength w15,16x.
2. Theory The sensor is based upon the refractive index change with humidity of a film applied to the cladding of a bent single-mode fibre. The bend causes a whispering gallery ŽWG. mode w15x to propagate in the cladding, and interact with the interface between the fibre and the film overlay. With various possible overlay materials, the refractive index decreases as the humidity increases due to the increased water content in the film. Materials which exhibit this behaviour include gelatine, polyethylene oxide, polyvinyl alcohol as well as many other polymers. The change in optical properties with humidity can be seen from the following relation w11x for the refractive index n of a composite material Že.g., gelatine and water. n2 y 1 n2 q 2
sÝ i
ri R i
ž / Mi
,
where r i is density, Mi is molecular weight and R i is defined as the molecular refractivity of the ith component material. A change in any of the three parameters will change the refractive index of the film; an increase in
3. Experiment A short section of buffer was removed from the fibre Žsingle-mode with a cut-off wavelength of 1.216 mm., which was then bent 1808 around a 9-mm-radius mandrel. In initial experiments, gelatine was chosen as the overlay and a solution was applied over the fibrermandrel and allowed to dry. As the operation of the device only depends upon reflection at the cladding-film interface, no control of the film thickness was required provided that it was much greater than the penetration depth of the evanescent wave at this point, although the actual thickness was of the order of 100 mm. The sensor was sealed in a chamber in which the humidity could be controlled. An electronic temperature and humidity sensor was also placed inside the chamber. The experimental arrangement is shown in Fig. 1. The basis of the experiment is to detect interference between the core-guided and WG modes, whose onset indicates the condition of high relative humidity. The interference is revealed by periodic oscillations in the
Fig. 1. Experimental set-up to detect high humidity by the use of a polymer film cast over an unbuffered fibre bent around a mandrel.
92
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
Fig. 2. Wavelength scan showing sudden appearance of WG modes at humidities above ; 62% RH — gelatine overlay, 1808 bend, 9-mm radius.
insertion loss of the bend as a function of wavelength. The insertion loss was measured by using a tungsten halogen lamp as a ‘‘white light’’ source, coupled into the optical fibre via a grating monochromator to provide wavelength control. At the output of the fibre, an InGaAs photodiode detector was used with lock-in amplification referenced to a mechanical chopper in the input fibre. Fig. 2 shows the result of performing wavelength scans under different relative humidity ŽRH. conditions. The ‘‘dry’’ plot corresponds to a humidity of 40% RH and no oscillations due to WG interference are present. At 62% RH, the plot remains the same Žnot shown.; however, an increase to 69% RH Ždenoted as the ‘‘humid’’ plot. shows
the sudden appearance of oscillations due to WG interference. These oscillations persisted for all humidities above this level. It is important that the sensor response is reversible; this was confirmed by cycling the humidity. At low humidities, the WG oscillations disappeared, and returned again when the humidity was increased. It has been shown in separate experiments, that when the arc length of the bent fibre is greater than the WG interaction length, then the periodicity of the WG interaction can contain harmonics of the fundamental oscillation predicted by Harris and Castle w15x. For a standard telecommunications grade fibre, with the buffer stripped and bent around an 18-mm-diameter mandrel, the theoreti-
Fig. 3. Confirmation of refractive index change in gelatine with changing humidity by measurement of Fresnel reflection from a coated fibre end at a wavelength of 633 nm.
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
93
Fig. 4. Appearance of WG modes at humidities above ; 80% RH — PEO overlay, 1208 bend, 6-mm radius.
cal period for the fundamental oscillation is ; 500 nm at a wavelength of 1500 nm. The oscillations observed in this experiment have a period of ; 125 nm, corresponding to the fourth harmonic of the fundamental. To confirm the mechanism by which the sensor operates, the internal Fresnel reflection from the end of a fibre coated with gelatine was measured as a function of humidity, in order to determine the relative humidity at which the refractive indices of the gelatine and the fused silica fibre were equal. To obtain a strong signal, a HeNe laser at 633-nm wavelength and multimode fibre were used. Fig. 3 shows how the reflected signal fell as the refractive index of the gelatine approached that of the silica. Fig. 3 is plotted against an absolute scale of water vapour partial pressure to allow for temperature fluctuations during the experiment. The point at which the reflected signal abruptly drops corresponds to the same relative humidity at which WG modes suddenly appear in Fig. 2, i.e. 1800 Pa partial pressure of water vapour, equivalent to a range of 60–68% RH at the temperature of the sensor characterisation experiment Ž; 22–248C.. We have neglected the small effect of the relative differences in dispersion of the refractive indices of fused silica and gelatine. As a further demonstration of the principle, the experiments were repeated for a variety of bend radii and overlay materials. The choice of bend radius and arc length determine the spectral response, and the humidity Žvia the refractive index of the overlay. determines the strength of the WG interactions. In addition to gelatine, polyethylene oxide ŽPEO. and polyvinyl alcohol ŽPVA. solutions were cast onto the bent fibres. The PVA films tended to crack and peel off when dry, but the PEO films proved to be very durable and completely reversible in their humidity response. The change in insertion loss for a device with a 1208 bend around a 6-mm-radius mandrel and a PEO overlay is shown in Fig. 4. In this case, the threshold humidity Žat ; 228C. is 80% RH, corresponding to when
the WG interactions become significant, i.e. when the refractive indices of the PEO and optical fibre are equal.
4. Discussion The experimental work required a wavelength scan to be performed to verify the presence of WG modes. In a real application, only two probe wavelengths would be required to detect the onset of WG mode interference and provide normalisation against intensity fluctuations. If the sensor were interrogated at a wavelength corresponding to a maximum, strong attenuation would occur when humid. The bend geometry determines the position Ži.e., the phase. of the oscillations, and would be chosen such that suitable points in the transmission plot coincided with telecommunications wavelengths. By the use of OTDR, a number of sensors deployed in series could be interrogated simultaneously and the position of any humid sensors would be found. The insertion loss per sensor is approximately 1 dB for the dry state at 1300 nm, which is acceptable if the device were intended to be used on a dedicated sensor fibre. An OTDR dynamic range of 34 dB would allow approximately 10 sensors on the fibre after taking other losses into account, with sufficient dynamic range to detect the humidity change. This device acts as a high humidity detector, rather than measuring a continuous value. The point at which the oscillations appeared was approximately 62% and 80% RH Žat ; 22–248C. for gelatine and PEO, respectively, which are in a suitable humidity range for the current application. Gelatine was used initially for simplicity, as it is known to be reversibly affected by humidity, and is very easy to apply. However, it is susceptible to damage from bulk water and to mould growth. The polyethylene oxide is a better choice of material because it is not susceptible to mould, although it could be damaged by immersion in
94
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
liquid water. However, this can be overcome by chemically modifying the polymer to make it insoluble yet still hygroscopic w7x. Additionally, it can be seen that the use of different materials leads to a different switching point for the sensor, allowing them to be adapted to respond at different humidity levels. The key to the response of the device is the change in refractive index of the overlay material with humidity from being greater than that of silica to lower. We have made initial measurements of the refractive index of PEO by measuring the Fresnel reflection from the distal end of a fibre coated with PEO. The refractive index was found to change from around n f 1.5 to n f 1.4, at a wavelength of 1.31 mm, for relative humidities of 70% and 80%, respectively. There will also be some temperature cross-sensitivity due to changes in the refractive indices of the cladding and overlay materials. A change in the refractive index of the cladding will modify the optical path difference between WG and core modes and hence will phase shift the interference fringes; but the effect is estimated to be less than one-tenth of a fringe for a 308C temperature change w16x, and can be compensated by choosing probe wavelengths in phase quadrature. Temperature induced changes in the refractive index of the overlay could potentially change the humidity level when the device switches, i.e. the point when the cladding and overlay indices are equal, but this is also expected to be a relatively small effect. For example, we have measured d nrdT Žalso at l s 1.31 mm. for PEO as being in the order of y10y4 8Cy1, as opposed to the change in refractive index with humidity of ) 0.1 for an increase from 70% to 80% RH. A potential problem with this device is repeatability of manufacture of sensors with identical spectral characteristics. This is because of the nature of the WG interference, rather than the properties of the overlay material. Figs. 2 and 4 clearly demonstrate that different bend lengths and radii give different responses, although this is a rather extreme example. For this reason, we are currently investigating the refractive index as a function of humidity and temperature for a variety of overlay materials, and continuing our investigations of an alternative sensor design w14x.
5. Conclusions A fibre optic sensor has been proposed that is potentially suitable for detecting high humidity in a telecommunications network. A humidity-responsive film with a reversible refractive index change was overlaid on a bent
fibre. At high humidities, the film has a refractive index lower than that of the silica cladding, causing total internal reflection of the WG mode and strong oscillations in transmitted power as a function of wavelength. This change in refractive index of the film with humidity was also observed independently, confirming the WG model. The sensor is intended to be compatible with serial deployment and interrogation by OTDR means in a passive optical network.
Acknowledgements The authors thank Peter Wilson, Nigel Evans and David Stockton of BT Laboratories for many helpful discussions. This research was jointly funded by BT Laboratories and the Engineering and Physical Sciences Research Council, whom DCB thanks for the provision of a CASE studentship.
References w1x I.J. Wilkinson, Electron. Lett. 29 Ž1993. 1137–1139. w2x M. Gadonna, T. Draycott, S. Gundersen, in: 13th Annual Conference on European Fibre Optic Communications and Networks, 1995, pp. 65–69. w3x K. Shimizu, T. Horiguchi, Y. Koyamada, Optics Lett. 20 Ž1995. 507. w4x L. Wosiniski, B. Sahlgren, M. Breidne, R. Stubbe, J.-P. Betend-Bon, Interferometric Fibre Sensing SPIE 2341 Ž1994. ´ 15. w5x A.M. Vengsarkar, W.C. Michie, L. Jankovic, B. Culshaw, R.O. Claus, J. Lightwave Technol. 12 Ž1994. 170. w6x S. Tomita, T. Hiroyuki, N. Kasahara, J. Lightwave Technol. 8 Ž1990. 1829. w7x W.C. Michie, B. Culshaw, I. McKenzie, M. Konstantakis, N.B. Graham, C. Moran, F. Santos, E. Bergqvist, B. Carlstrom, Optics Lett. 20 Ž1995. 103. w8x A.D. Stuart, P.E. Grazier, Int. J. Optoelectron. 3 Ž1988. 177. w9x F. Mitschke, Optics Lett. 14 Ž1989. 967. w10x K. Ogawa, S. Tsuchiya, H. Kawakami, T. Tsutsui, Electron. Lett. 24 Ž1988. 42. w11x C. Ronot, H. Archenault, H. Gagnaire, J.P. Goure, N. Jaffrezic-Renault, T. Pichery, Sensors Actuators B 11 Ž1993. 375. w12x A.P. Russell, K.S. Fletcher, Anal. Chim. Acta 170 Ž1985. 209. w13x D.S. Ballantine, H. Wohltjen, Anal. Chem. 58 Ž1986. 2883. w14x D.C. Bownass, J.S. Barton, J.D.C. Jones, Optics Lett. 22 Ž1997. 346. w15x A.J. Harris, P.F. Castle, J. Lightwave Technol. 4 Ž1986. 34. w16x R. Morgan, J.D.C. Jones, P.G. Harper, J.S. Barton, Optics Comm. 85 Ž1991. 17.
Optics Communications 146 Ž1998. 90–94
Detection of high humidity by optical fibre sensing at telecommunications wavelengths D.C. Bownass ) , J.S. Barton, J.D.C. Jones Physics Department, Heriot-Watt UniÕersity, Riccarton, Edinburgh, EH14 4AS, UK Received 9 July 1997; accepted 3 September 1997
Abstract We demonstrate the proof of principle for a novel fibre optic sensor for the detection of high humidity. The device uses single-mode fibre at telecommunications wavelengths, making it particularly suitable for application in passive optical networks. The sensing mechanism relies on a humidity-induced refractive index change in an organic film coated onto the cladding of a bent fibre. The sensor was shown to respond at relative humidities ) 62% and ) 80% using gelatine and polyethylene oxide films respectively, by the onset of a strong interaction between the LP01 mode and a whispering gallery mode. q 1998 Elsevier Science B.V. PACS: 42.81.P; 92.60.J; 84.40.U Keywords: Fibre-optic sensors; Humidity detection; Whispering gallery interaction; Network monitoring; Telecommunications
1. Introduction It is known that the lifetimes of components used in passive optical communications networks depend on environmental conditions, with failure hastened by exposure to high temperature and humidity w1,2x. Network monitoring may be desirable, preferably based on passive sensors compatible with passive networks. Fibre optic sensors have already been developed for variables such as temperature and strain w3–5x, with potential for network monitoring. This paper is concerned with the development of a passive optical sensor for humidity monitoring. Sensors have been reported for the detection of water in the liquid, rather than vapour, phase w6,7x. This may not be sufficient, as components such as optical splitters could be affected by high humidity, well before liquid water enters the device housing. If a humidity sensor were incorporated close to vulnerable components, then preventative mainte-
)
E-mail: [email protected].
nance prior to system failure would be feasible, leading to a reduction in costs. Several fibre optic humidity sensors have been described previously, but none are suitable for passive optical communications systems. For example, point sensors have been reported which use the end of a fibre coated with either a metal w8x or dielectric w9x film. These sensors are incompatible with a serial array and the use of optical time domain reflectometry ŽOTDR. for interrogating the array. Others have used special fibres with the cladding removed and replaced with either porous silica w10x or a special polymer w11x, but require multimode fibre, making them incompatible with the standard components used in a telecommunications application. Another technique employs the spectral absorption of CoCl 2 entrapped in a gelatine w12x or polymer w13x host which is then coated onto the fibre. These sensors require multimode fibre, and the use of visible wavelengths, again making them unsuitable for a telecommunications network. We have recently investigated an alternative type of fibre-optic humidity sensor w14x in which the cladding of
0030-4018r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 0 3 0 - 4 0 1 8 Ž 9 7 . 0 0 5 2 5 - 7
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
the fibre is polished away and an overlay is applied whose refractive index depends on relative humidity, thus influencing the insertion loss of the polished section of the fibre. In the present work, we develop the idea of using a humidity-sensitive overlay, but avoid the requirement for precise polishing of the fibre cladding. Instead, the fibre is bent in the sensing region, so that the overlay influences the propagation of whispering gallery modes in the fibre cladding. The sensor described here uses single-mode fibre and wavelengths in the telecommunications range. It is suitable for time division multiplexing in a serial array with interrogation by OTDR, and is small, simple and lightweight.
91
water content of the film will cause it to swell slightly, resulting in a reduction in the overall density and hence a corresponding drop in refractive index. When the film has a refractive index greater than that of the silica cladding, as occurs at low humidities in the experiments described here, there is only a very weak Fresnel reflection of the WG mode at the cladding-overlay interface, and the interference with the core guided mode is negligible. At high humidities, the film has a refractive index lower than that of the cladding and hence total internal reflection of the WG mode can occur, resulting in strong interference with the core mode, revealed by oscillations in the transmission of the bend as a function of wavelength w15,16x.
2. Theory The sensor is based upon the refractive index change with humidity of a film applied to the cladding of a bent single-mode fibre. The bend causes a whispering gallery ŽWG. mode w15x to propagate in the cladding, and interact with the interface between the fibre and the film overlay. With various possible overlay materials, the refractive index decreases as the humidity increases due to the increased water content in the film. Materials which exhibit this behaviour include gelatine, polyethylene oxide, polyvinyl alcohol as well as many other polymers. The change in optical properties with humidity can be seen from the following relation w11x for the refractive index n of a composite material Že.g., gelatine and water. n2 y 1 n2 q 2
sÝ i
ri R i
ž / Mi
,
where r i is density, Mi is molecular weight and R i is defined as the molecular refractivity of the ith component material. A change in any of the three parameters will change the refractive index of the film; an increase in
3. Experiment A short section of buffer was removed from the fibre Žsingle-mode with a cut-off wavelength of 1.216 mm., which was then bent 1808 around a 9-mm-radius mandrel. In initial experiments, gelatine was chosen as the overlay and a solution was applied over the fibrermandrel and allowed to dry. As the operation of the device only depends upon reflection at the cladding-film interface, no control of the film thickness was required provided that it was much greater than the penetration depth of the evanescent wave at this point, although the actual thickness was of the order of 100 mm. The sensor was sealed in a chamber in which the humidity could be controlled. An electronic temperature and humidity sensor was also placed inside the chamber. The experimental arrangement is shown in Fig. 1. The basis of the experiment is to detect interference between the core-guided and WG modes, whose onset indicates the condition of high relative humidity. The interference is revealed by periodic oscillations in the
Fig. 1. Experimental set-up to detect high humidity by the use of a polymer film cast over an unbuffered fibre bent around a mandrel.
92
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
Fig. 2. Wavelength scan showing sudden appearance of WG modes at humidities above ; 62% RH — gelatine overlay, 1808 bend, 9-mm radius.
insertion loss of the bend as a function of wavelength. The insertion loss was measured by using a tungsten halogen lamp as a ‘‘white light’’ source, coupled into the optical fibre via a grating monochromator to provide wavelength control. At the output of the fibre, an InGaAs photodiode detector was used with lock-in amplification referenced to a mechanical chopper in the input fibre. Fig. 2 shows the result of performing wavelength scans under different relative humidity ŽRH. conditions. The ‘‘dry’’ plot corresponds to a humidity of 40% RH and no oscillations due to WG interference are present. At 62% RH, the plot remains the same Žnot shown.; however, an increase to 69% RH Ždenoted as the ‘‘humid’’ plot. shows
the sudden appearance of oscillations due to WG interference. These oscillations persisted for all humidities above this level. It is important that the sensor response is reversible; this was confirmed by cycling the humidity. At low humidities, the WG oscillations disappeared, and returned again when the humidity was increased. It has been shown in separate experiments, that when the arc length of the bent fibre is greater than the WG interaction length, then the periodicity of the WG interaction can contain harmonics of the fundamental oscillation predicted by Harris and Castle w15x. For a standard telecommunications grade fibre, with the buffer stripped and bent around an 18-mm-diameter mandrel, the theoreti-
Fig. 3. Confirmation of refractive index change in gelatine with changing humidity by measurement of Fresnel reflection from a coated fibre end at a wavelength of 633 nm.
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
93
Fig. 4. Appearance of WG modes at humidities above ; 80% RH — PEO overlay, 1208 bend, 6-mm radius.
cal period for the fundamental oscillation is ; 500 nm at a wavelength of 1500 nm. The oscillations observed in this experiment have a period of ; 125 nm, corresponding to the fourth harmonic of the fundamental. To confirm the mechanism by which the sensor operates, the internal Fresnel reflection from the end of a fibre coated with gelatine was measured as a function of humidity, in order to determine the relative humidity at which the refractive indices of the gelatine and the fused silica fibre were equal. To obtain a strong signal, a HeNe laser at 633-nm wavelength and multimode fibre were used. Fig. 3 shows how the reflected signal fell as the refractive index of the gelatine approached that of the silica. Fig. 3 is plotted against an absolute scale of water vapour partial pressure to allow for temperature fluctuations during the experiment. The point at which the reflected signal abruptly drops corresponds to the same relative humidity at which WG modes suddenly appear in Fig. 2, i.e. 1800 Pa partial pressure of water vapour, equivalent to a range of 60–68% RH at the temperature of the sensor characterisation experiment Ž; 22–248C.. We have neglected the small effect of the relative differences in dispersion of the refractive indices of fused silica and gelatine. As a further demonstration of the principle, the experiments were repeated for a variety of bend radii and overlay materials. The choice of bend radius and arc length determine the spectral response, and the humidity Žvia the refractive index of the overlay. determines the strength of the WG interactions. In addition to gelatine, polyethylene oxide ŽPEO. and polyvinyl alcohol ŽPVA. solutions were cast onto the bent fibres. The PVA films tended to crack and peel off when dry, but the PEO films proved to be very durable and completely reversible in their humidity response. The change in insertion loss for a device with a 1208 bend around a 6-mm-radius mandrel and a PEO overlay is shown in Fig. 4. In this case, the threshold humidity Žat ; 228C. is 80% RH, corresponding to when
the WG interactions become significant, i.e. when the refractive indices of the PEO and optical fibre are equal.
4. Discussion The experimental work required a wavelength scan to be performed to verify the presence of WG modes. In a real application, only two probe wavelengths would be required to detect the onset of WG mode interference and provide normalisation against intensity fluctuations. If the sensor were interrogated at a wavelength corresponding to a maximum, strong attenuation would occur when humid. The bend geometry determines the position Ži.e., the phase. of the oscillations, and would be chosen such that suitable points in the transmission plot coincided with telecommunications wavelengths. By the use of OTDR, a number of sensors deployed in series could be interrogated simultaneously and the position of any humid sensors would be found. The insertion loss per sensor is approximately 1 dB for the dry state at 1300 nm, which is acceptable if the device were intended to be used on a dedicated sensor fibre. An OTDR dynamic range of 34 dB would allow approximately 10 sensors on the fibre after taking other losses into account, with sufficient dynamic range to detect the humidity change. This device acts as a high humidity detector, rather than measuring a continuous value. The point at which the oscillations appeared was approximately 62% and 80% RH Žat ; 22–248C. for gelatine and PEO, respectively, which are in a suitable humidity range for the current application. Gelatine was used initially for simplicity, as it is known to be reversibly affected by humidity, and is very easy to apply. However, it is susceptible to damage from bulk water and to mould growth. The polyethylene oxide is a better choice of material because it is not susceptible to mould, although it could be damaged by immersion in
94
D.C. Bownass et al.r Optics Communications 146 (1998) 90–94
liquid water. However, this can be overcome by chemically modifying the polymer to make it insoluble yet still hygroscopic w7x. Additionally, it can be seen that the use of different materials leads to a different switching point for the sensor, allowing them to be adapted to respond at different humidity levels. The key to the response of the device is the change in refractive index of the overlay material with humidity from being greater than that of silica to lower. We have made initial measurements of the refractive index of PEO by measuring the Fresnel reflection from the distal end of a fibre coated with PEO. The refractive index was found to change from around n f 1.5 to n f 1.4, at a wavelength of 1.31 mm, for relative humidities of 70% and 80%, respectively. There will also be some temperature cross-sensitivity due to changes in the refractive indices of the cladding and overlay materials. A change in the refractive index of the cladding will modify the optical path difference between WG and core modes and hence will phase shift the interference fringes; but the effect is estimated to be less than one-tenth of a fringe for a 308C temperature change w16x, and can be compensated by choosing probe wavelengths in phase quadrature. Temperature induced changes in the refractive index of the overlay could potentially change the humidity level when the device switches, i.e. the point when the cladding and overlay indices are equal, but this is also expected to be a relatively small effect. For example, we have measured d nrdT Žalso at l s 1.31 mm. for PEO as being in the order of y10y4 8Cy1, as opposed to the change in refractive index with humidity of ) 0.1 for an increase from 70% to 80% RH. A potential problem with this device is repeatability of manufacture of sensors with identical spectral characteristics. This is because of the nature of the WG interference, rather than the properties of the overlay material. Figs. 2 and 4 clearly demonstrate that different bend lengths and radii give different responses, although this is a rather extreme example. For this reason, we are currently investigating the refractive index as a function of humidity and temperature for a variety of overlay materials, and continuing our investigations of an alternative sensor design w14x.
5. Conclusions A fibre optic sensor has been proposed that is potentially suitable for detecting high humidity in a telecommunications network. A humidity-responsive film with a reversible refractive index change was overlaid on a bent
fibre. At high humidities, the film has a refractive index lower than that of the silica cladding, causing total internal reflection of the WG mode and strong oscillations in transmitted power as a function of wavelength. This change in refractive index of the film with humidity was also observed independently, confirming the WG model. The sensor is intended to be compatible with serial deployment and interrogation by OTDR means in a passive optical network.
Acknowledgements The authors thank Peter Wilson, Nigel Evans and David Stockton of BT Laboratories for many helpful discussions. This research was jointly funded by BT Laboratories and the Engineering and Physical Sciences Research Council, whom DCB thanks for the provision of a CASE studentship.
References w1x I.J. Wilkinson, Electron. Lett. 29 Ž1993. 1137–1139. w2x M. Gadonna, T. Draycott, S. Gundersen, in: 13th Annual Conference on European Fibre Optic Communications and Networks, 1995, pp. 65–69. w3x K. Shimizu, T. Horiguchi, Y. Koyamada, Optics Lett. 20 Ž1995. 507. w4x L. Wosiniski, B. Sahlgren, M. Breidne, R. Stubbe, J.-P. Betend-Bon, Interferometric Fibre Sensing SPIE 2341 Ž1994. ´ 15. w5x A.M. Vengsarkar, W.C. Michie, L. Jankovic, B. Culshaw, R.O. Claus, J. Lightwave Technol. 12 Ž1994. 170. w6x S. Tomita, T. Hiroyuki, N. Kasahara, J. Lightwave Technol. 8 Ž1990. 1829. w7x W.C. Michie, B. Culshaw, I. McKenzie, M. Konstantakis, N.B. Graham, C. Moran, F. Santos, E. Bergqvist, B. Carlstrom, Optics Lett. 20 Ž1995. 103. w8x A.D. Stuart, P.E. Grazier, Int. J. Optoelectron. 3 Ž1988. 177. w9x F. Mitschke, Optics Lett. 14 Ž1989. 967. w10x K. Ogawa, S. Tsuchiya, H. Kawakami, T. Tsutsui, Electron. Lett. 24 Ž1988. 42. w11x C. Ronot, H. Archenault, H. Gagnaire, J.P. Goure, N. Jaffrezic-Renault, T. Pichery, Sensors Actuators B 11 Ž1993. 375. w12x A.P. Russell, K.S. Fletcher, Anal. Chim. Acta 170 Ž1985. 209. w13x D.S. Ballantine, H. Wohltjen, Anal. Chem. 58 Ž1986. 2883. w14x D.C. Bownass, J.S. Barton, J.D.C. Jones, Optics Lett. 22 Ž1997. 346. w15x A.J. Harris, P.F. Castle, J. Lightwave Technol. 4 Ž1986. 34. w16x R. Morgan, J.D.C. Jones, P.G. Harper, J.S. Barton, Optics Comm. 85 Ž1991. 17.