Specialty Fibers for Discrete and Distributed Sensing ...

Specialty Fibers for Discrete and Distributed Sensing Application Joohyun Koh, K. Bennett, R. Bennett, X. Chen, C. K. Chien, M.-J. Li, and D. Nolan Specialty Fiber Research & Development, Corning Incorporated, SP-TD-02-2, Corning, NY 14831 Tel: 607-974-3450, E-mail: [email protected] ABSTRACT Over the past decades, discrete or distributed Fiber Optic Sensing (FOS) applications have seen an increased acceptance in many areas. High level optical and mechanical reliability of optical fiber is necessary to guarantee reliable performance of FOS. In this paper, we review recent research and development activities on new specialty fibers. The main approaches to enhancing fiber attributes include new refractive index profile design and fiber coating modification. Keyword: Specialty optical fiber; Fiber optic sensor, POTDR, Hermetic coating, Polyimide coating

INTRODUCTION Over the last few decades, optical fibers have been widely employed in telecommunication application due to high transmission capacity, immunity to electromagnetic interference, etc. Optical fibers have also been used in point and/or distributed sensing applications [1-3]. For example, Fiber Bragg Grating (FBG) enables single or multi-point sensing for Structural Health Monitoring (SHM) [4]. Distributed temperature and/or pressure sensing is capable through different scattering mechanisms, including Rayleigh [5], Raman [6, 7], and Brillouin [8] scattering. Applications for measurement include (i) polarization change for security monitoring in critical infrastructure, (ii) strain in civil structures, and (iii) pressure and temperature in oil and gas wells, platforms, and pipelines. Standard acrylate coated Single Mode (SM) and Multi Mode (MM) fibers have been widely used for FOS application. However, there is a great need to improve optical and mechanical attributes of fibers in order to enhance sensitivity and optical/mechanical reliability. There are two primary components in optical fibers, the glass and the coating. Sensor manufacturers can benefit from modifications to each of these components individually or in combination. Modification of glass in optical fiber includes tuning of refractive index profile. For polarization-sensitive distributed sensing system, preserve linear state of polarization over long length is desirable in the area requiring precise polarization control in the sensor system. Modification of coating in optical fiber completely depends on operating environment. For most telecommunication application, acrylate polymers have been applied as coating materials. However, it is not suitable for sensing application because acrylate-coated fibers are not hermetic and limiting operating temperatures. For example, optical fiber experiences high temperature, high pressure, moisture, hydrogen and other harmful species (e.g., CO2, and H2S) in typical oil/gas down hole environment or concrete embedded system. Amorphous carbon-based hermetic coating provides a protective layer which prevents ingress of molecular water or hydrogen into silica glass of the fiber. The presence of hermetic coating enables one to not only improve mechanical integrity but also maintain low optical loss in harsh environments. Polyimide coating is suitable for use in applications requiring to operation at high temperatures or under high strain environments. Therefore, hermetic coating combined with polyimide coating provides improved mechanical and optical reliability against harsh environments [9]. At the 1st part of this paper, we propose new fiber which is designed for conducting polarization-sensitive distributed sensing of local perturbation along the deployed fibers. The fiber is specially designed to have longer beatlength (≡ low birefringence) and high uniformity of the linear birefringence. At the 2nd part in this paper, we present recent progress of a hermetic/polyimide coated fibers with enhanced hermeticity. We discuss fiber performance, in particular resistance to H2 ingression and mechanical properties. In addition, we propose lifetime model of polyimide material at high temperatures.

Smart Sensor Phenomena, Technology, Networks, and Systems 2009, edited by Norbert G. Meyendorf, Kara J. Peters, Wolfgang Ecke, Proc. of SPIE Vol. 7293, 729312 · © 2009 SPIE · CCC code: 0277-786X/09/$18 · doi: 10.1117/12.811701 Proc. of SPIE Vol. 7293 729312-1

GLASS MODIFICATION: INTRUSION SENSING SYSTEM INCORPORATED WITH INDEX PROFILE MODIFIED FIBER In infrastructure monitoring, distributed Fiber Optic Sensing (FOS) systems provide information on perturbations along the path, which can trigger signal in case there is a substantial and/or abnormal change of strain along the fibers. Thus, distributed FOS systems have important applications in ensuring the early detection of deterioration of critical infrastructures and timely repairing when necessary. FOS systems that utilize polarization effects require sensing a change of polarization state due to small perturbation in the optical fiber. Such changes can be produced, for example, by stress, fiber bends, or changes in pressure. The occurrence and location of a disturbance in the vicinity of a localized portion of the optical fiber can be determined based on the recognition that such a disturbance changes the polarization conditions of light backscattered from the disturbed portion of the optical fiber. The change of the state of polarization in the optical fiber can be detected by backscattered light utilizing a Polarization-Sensitive Optical Time Domain Reflectometer (POTDR) device. By inserting a polarization component such as a polarizer into the OTDR path, the polarization change can exhibit itself in the form of intensity change. Figure 1 illustrates schematically a POTDR based detection system. In this proposed system, OTDR launches the pulse light through a polarization controller, and directed through the polarizer, and then into the sensing fiber. Polarization controller is employed to align the state of polarization in the sensing fiber to be incident on the polarizer to maximize the amount of light launched into the sensing fiber. Polarizer utilized in the POTDR systems of Fig. 1 serves two main purposes. First, inline polarizer assures that light going into the sensing fiber is polarized, although this could also be achieved without the inline polarizer (e.g. by using an OTDR that already emits polarized light). More importantly, inline polarizer serve as analyzer for light that is backscattered from the sensing fiber toward OTDR, which light provides information on the state of polarization of the sensing fiber. Without the polarizer in place to analyze the backscattered light from the sensing fiber, the OTDR would only detect total intensity of the light pulse. However, by incorporating the polarizer, information related to the state of polarization along the sensing fiber can also be obtained. As a result, the state of polarization of a light pulse traveling in the sensing fiber evolves continuously in both forward and backward propagating direction when the light has been back reflected. Alternatively, one can use a pair of optical circulator to route outgoing light and backscattered light through separate paths. In this case, separate polarizers as polarization controller and analyzer can be used.

1

c:rr t::' F

2

4

3

(EX8)_

CD

Figure 1: Schematics of POTDR-based detection system. Numbering is as follows; 1: OTDR, 2: polarization controller, 3: polarizer, 4: sensing fiber

Figure 2 illustrates schematically a simplified POTDR trace taken using the setup shown in Fig. 1, with signal intensity I versus distance D along the sensing fiber. More specifically, plot A illustrates the intensity I1 of the backscattered optical signal at a time T1. Plot B illustrates the intensity I2 of the backscattered optical signal at a time T2= T1 +ΔT, where the fiber was perturbed (for example, by the intruder) at the time T2. Plot C is an intensity difference, I2 - I1. Plot C clearly shows that plots A and B were identical up to the position D0, but became different beyond distance D0 along the fiber, which is the site of the perturbation. Most fibers exhibit different levels of birefringence non-uniformity together with mode coupling that occurs along the fiber length. Consequently, a fiber having a varying beatlength associated with the birefringence nonuniformity may not exhibit a clear periodic POTDR trace as shown in Fig. 2. For example, the POTDR trace may be “noisy” because periodic polarization evolution of the light propagating through the fiber is frequently disrupted by random changes of the fiber’s birefringent axis and/or the levels of birefringence. It is also possible that in a distributed sensing system that uses standard SM fibers, the fiber beatlength of some of such fibers could fall below the resolution of the POTDR, in which case the variation of the intensity along the fiber will be averaged out and will not be

Proc. of SPIE Vol. 7293 729312-2

resolvable. This occurs because such variation of the intensity will be significantly reduced within one OTDR pulse width, resulting in low signal to noise ratio. It leads to the situation that some of commonly used SM fibers may not be adequate for sensing applications, or that many measurements have to be averaged and/or filtered to improve signal to noise ratio. Therefore, newly designed fiber having low birefringence, longer beatlength, and higher birefringence uniformity was proposed. It provides more sensitive and/or accurate measurement in POTDR based sensing system.

Distance along the sensing fiber (D)

Figure 2: Simplified POTDR trace without (Plot A) and with (plot B) disturbance on sensing fiber and signal difference (plot C).

Figure 3: Schematics of sensing fiber employed in POTDR-based detection system. a and b corresponds to minor and major axis of elliptical core.

A sensing fiber with low and uniform birefringence is fabricated using the Outside Vapor Deposition (OVD) process. First, a fiber preform was made with higher birefringence (for example a preform that can yield an unspun fiber with a uniform beatlength between 1.5 and 20 m, measured at 1550 nm) and then fiber an be produced by spinning the preform during the draw process to reduce the effective birefringence. It results in significant increase in effective beatlength, for example, between 10 m and 200 m. Figure 3 shows a schematic of sensing fiber where “a” and “b” represents minor and major axis of the elliptical core in the fiber. The elliptical core fiber is fabricated to introduce controlled amount of birefringence. Figure 4 shows birefringence of fiber (defined as B = 2π/LB, where LB = Beatlength) as a function of core ellipticity (e). It was found that the relationship between B and e is linear. We can effectively fabricate elliptical core fiber having desired level of birefringence with this plot. A sensing fiber is eventually made by bi-directional or sinusoidal spinning during fiber draw process in order to have low effective birefringence (LB’) with high level of beatlength uniformity [10]. It should be noted that birefringence of unspun fiber should be higher than non-uniform background birefringence, which will ultimately be reduced by spinning. As a result, effective beatlength (LB’) after spinning would be higher than beatlength (LB) of unspun fiber, at least LB’≥3LB. 200

Delta, % (arb. units)

Birefringence (1/m)

150

100

50

0 0

0.05

0.1

0.15

0.2

0.25

Radius (arb. units)

0.3

Ellipticity (1-a/b)

Figure 4: Birefringence as a function of ellipticity of core, defined as 1-a/b shown in Fig. 3.

Figure 5: Schematic of index profiles for proposed bend insensitive fiber


Optionally, the sensing fiber can be designed to have low bending loss. The sensing fiber can be locally bent to a tight diameter during installation, so it is preferable that the sensing fiber would be bending insensitive. This low bending loss is achieved, for example, by profile tuning which incorporate a low index trench. Figure 5 shows exemplary refractive index profile of such fiber. Such a low bend insensitive fiber can prevent interference of the fiber bending loss during determination of intensity signal change induced by the polarization signal. Also, low bending loss makes the sensing fiber suitable for deployment in harsh environment, which can broaden the applicability of the detection system. To illustrate how the proposed scheme works, we have made the fibers according to proposed design and conducted measurement on selected fibers. A preform with elevated fiber birefringence is made and the fibers were subsequently drawn. First, unspun fiber was drawn for the purpose to know the fiber beatlength without the involvement of fiber spinning. The beatlength of unspun fiber was measured to have a value of 3 m at 1550 nm. Then, actual sensing fiber was made by spinning during draw. In order to minimize deployment-dependence of measured beatlength, the fiber was prepared on large diameter spool with zero tension in wind. POTDR traces were taken before and after intentional perturbation on sensing fibers in 3 km span. Figures 6-8 shows how the location of a perturbation can be detected by utilizing autocorrelation. First, the POTDR traces before and after perturbations are obtained. Such traces are shown in the Figs. 6 and 7. As exhibited by the zoomed section in Figure 6, the effective beatlength is around 60 m, which is far longer than 3 m beatlength in the unspun fiber made from the same preform. This is what is expected by the proposed scheme of making highly uniform fibers with low birefringence. The traces shown in Figures 6 and 7 appear to be similar to one another. The differences are in the fine details which are not clearly shown in these figures as limited by resolution. We performed the following steps in order to identify the precise location of the perturbation. As shown in Fig. 6, polarization related information is generally embedded in the background that depicts fiber attenuation. It is noted that the center line of the POTDR trace has a negative slope. Since the polarization sensitive information carries the information related to the local perturbation, we subtracted the line with negative slope from the POTDR trace in the first step. We now have only the information related to the local polarization information in the “processed POTDR traces”. The original OTDR traces can be labeled as Pb(z) and Pa(z) respectively, for the trace before and after the perturbation. The removal of the linear slope results in processed POTDR trace which is described by the equation S i ( z ) = Pi ( z ) − (ci + d i z ) , where i= b or a, representing traces obtained ‘before’ and ‘after’ respectively, and ci and di are the two parameters that uniquely determine the straight line taken out from Pi(z), which are determined by linear regression resulting in a better fitting of the overall trace. 24 23

24

25

21 1.8

1.9

2.0

23 22 21 20

2.1

OTDR Signal (dB)

OTDR Signal (dB)

25

Before Perturbation 22

24 23 22 21 20

1.2 1.41.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 Position (km)

Figure 6: Exemplary POTDR trace obtained before the perturbation. A section is zoomed to see the quasi-periodic variation due to polarization evolution along the fiber.

After Perturbation

1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 Position (km)

Figure 7: Exemplary POTDR trace obtained after perturbation.


Finally, the location of the perturbation can be obtained by building the autocorrelation function of the two processed POTDR traces. We calculate the autocorrelation according to the following equation, z + 0.5 w

0.4

1 S b ( z ' ) S a ( z ' )dz ' w z −0∫.5 w

where, w is the width of the specific window used to calculate the autocorrelation. The selection of the window width w is not critical. The width w, for example, can be between 100 m and few thousand meters. Using a window width of 200 m, as an example, we calculate the autocorrelation function as shown in Figure 8. The position at which the autocorrelation curve first crosses the zero level is the position where perturbation takes place. In this case, the position of the perturbation is located at 2.32 km, which agrees with the position we set to incur the perturbation.

AutoCorrelation

τ ( z) =

0.3

Perturbation Site

0.2 0.1 0.0 -0.1

1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8

Position (km)

Figure 8: Exemplary autocorrelation function obtained from the processed POTDR traces obtained before and after the perturbation. The perturbation site is the first position that such function reaches zero.

COATING MODIFICATION: HERMETIC/POLYIMIDE COATED FIBERS The acrylate coating on a contemporary telecommunications fiber serves two main purposes [11]. The primary function of the coating is to provide protection for the glass against abrasions and environmental exposure. The second function which has evolved over time is to provide a cushioning effect to minimize the transmission of external forces to the glass thereby providing extra protection. Telecommunication fibers are typically exposed to temperatures ranging from -50 to 85oC. Operation in harsher environments requires modification of the design and materials of the coating system. Addition of a hermetic coating to a fiber with an acrylate coating has been widely applied in undersea telecom application [12]. Hermetic coating provides a protective layer which prevents ingress of molecular water or hydrogen into silica glass of the fiber. It also enables highly reliable deployment of the fiber under smaller coil diameters. The presence of hermetic coating enables one to not only improve mechanical integrity but also maintain low optical loss in harsh environments. Traditionally, coating materials for optical fiber used in adverse environments are silicones and fluoro polymers. However, the typical use temperature is around 170~180°C for silicone and 250°C for fluoro polymers. For applications at or above 300°C, polyimides are the preferred choice for their superior thermal stability at high temperature, excellent mechanical properties, and chemical inertness. Unlike most of coating materials for optical fibers, polyimide is thermally curable material. Thus, the curing process of polyimide coating involves drying the solvent and follow by curing the polyamic acid to polyimde through the imidization step. We present recent progress of fiber development efforts which employs hermetic combined with polyimide coatings. Thermal Stability of Polyimide Coating Two polyimide materials were initially evaluated in this study. The thermal stability is one of the most important attributes to the high temperature application and was evaluated in two ways; temperature ramping and isothermal test. Weight change is continuously monitored by TGA (ThermoGravimetry Analysis) under two different ramping rates, 1 oC/min and 10 oC/min in temperature ramping experiment. Figure 9 shows polyimide A (PA) has a higher onset temperature of weight loss than polyimide B (PB). At 20% weight loss, PA has a higher decomposition temperature than PB, 572 oC vs. 525 oC at 1°C/min and 634 oC vs.

They are labeled as PA and PB for convenience.

T onset @1°C/min T onset @10°C/min T 20% wt. loss @1°C/min T 20% wt. loss @10°C/min


PB 520 596 525 600

PA 573 624 572 634

Table 1: Thermal stability of polyimide materials

600°C/min, respectively. Table 1 summarizes measured results. In an accelerated isothermal degradation measurement at 500°C, PA lasts 246 minutes over PB’s 44 minutes at 20% weight loss. Measured result shows PA could have a lifetime 5 times longer than PB, as shown in Figure 10. As a result, PA was selected to incorporate fiber products as coating material. 120

100

80

TGA Weight Loss, %

TGA Weight Loss, %

100 o

PA, 10 C/min PB, 10oC/min

60

o

PA, 1 C/min PB, 1oC/min

40 20 0

80

PB 60 40 20 0

0

200

400

600

800

PA

0

100

o

200

300

400

500

600

Time (min)

Temperature ( C)

Figure 9: Thermal stability of polyimide materials, measured by TGA in temperature ramping mode

Figure 10: Comparison of thermal stability of PA and PB, measured by TGA in isothermal mode at 500°C

Optical Performance of Hermetic / Polyimide Fibers In fiber optic sensing system, especially for oil/gas application, the transmission characteristics of optical fibers should be stable at high temperature and high H2 pressure environment. H2 absorption is common to silica-based optical fibers. In particular, it escalates at high temperature and high H2 pressure environment. Investigation had been done to clarify the cause of transmission loss of optical fibers which is classified as follows; (1) Reversible loss increase due to the vibration of H2 molecules which is normally appeared around 1240 nm in wavelength [13, 14] (2) Irreversible loss increase due to the chemical reaction between H2 and silica, which is further classified into two mechanisms a. Loss increase due to OH formation which is normally appeared at 1380 nm (Si-OH) and 1410 nm (Ge-OH) [15] b. Loss increase in short wavelength which is related to lattice defects [16] Given above mentioned features of H2 incorporation into silica network, our analysis is focused on loss increase around 1240, 1380~1420 nm. For multi-mode fibers, 1000-1100 nm is of interest which falls into DTS (Distributed Temperature Sensing) operating window. Figure 11 shows typical hydrogen absorption spectra in both hermetic and non-hermetic polyimide coated silica fibers. Fibers under test are 50/125 multimode fibers. Both fibers were treated under 100 %, 1 atm H2 at 150oC for 168 hours. The loss peak at 1240 nm (H2 molecules) and broad peak around 1400 nm (Si-OH, Ge-OH) are seen in non-hermetic MM fibers. In addition, the loss in typical operating window for DTS system (1000-1100 nm) is affected by the hydrogen absorption. On the other hand, hermetic MM fiber does not show any sign of H2 absorption. Hence, the optical fiber with appropriate hermetic protection against H2 diffusion is needed in order to have extended lifetime in down-hole environment. Next, we investigated onset temperature of H2 absorption for hermetic/polyimide coated MM fibers. Fibers were tested under 100 %, 1 atm H2 for 48 hours at temperature ranging from 150oC to 200oC. Spectral attenuation was measured before and after each H2 aging steps. Optical response against H2 aging is shown in Fig. 12. By judging loss increase around 1240 nm and 1380 nm, onset temperature of H2 absorption is around 185oC, which seems more obvious at 200oC. In parallel, a few commercially available hermetic/polyimide coated fibers were tested at the same condition. It was found that fiber sample shown in Fig. 12 exhibits about 10~20oC higher onset temperature against H2 aging.


I'

- Untreated MM fiber

- Untreated - Treated at lUOtC

- -H2/Hiqfl temp treated tnt-flemetic MM fiber -

- - Treated at l7OtC

- - H2/Hiqfl temp treated Hermetic MM fiber

Treated at I 9OtC

---Treated at 200tC

tOO

1000

1200

1400

1600

900

1000

1400

1200

Wavelength (nm)

Wavelength (nm)

Figure 11: Hydrogen absorption spectra for both nonhermetic and hermetic polyimide coated multimode fibers. Test conditions include 100 %, 1 atm H2 at 150oC for 168 hours.

Figure 12: Spectral attenuation for the hermetic/polyimide coated MM fiber tested under 100 %, 1 atm H2 for 48 hours. The temperature varies from 150oC to 200oC.

Lastly, fiber sample was tested in longer time, extending to 696 hours. Figure 13 shows spectral attenuations for hermetic/polyimide coated MM fiber before and after being exposed to 100 % 1 atm H2 environment at 185oC for 696 hours. Growth of attenuation at 1240 nm and 1380 nm is appeared in spectra but nearly no change at 1000-1100 nm which is critical operating window for DTS application.

- Untreated ---Treated at l8SoC for 000 flrs

Ann

lnnn

l2flfl

l'nn

lAflfl

Figure 13: Spectral attenuation for hermetic/polyimide coated MM fiber tested under 100 %, 1 atm H2 at 185oC for 696 hours.

Mechanical Properties of Hermetic / Polyimide Fibers Dynamic fatigue test was performed for both fresh and treated fibers. It is a tensile testing which involves gripping the fiber at both ends and pulling it until failure occurs. The fiber must be gripped carefully so no damage occurs. Gripping is achieved by wrapping the fiber around two 76 mm Teflon coated capstans and then the fiber is pulled apart. Testing was done using a 50 cm gauge length with a 500 mm/min crosshead speed in a controlled environment of 50% RH and 22° C. The advantages to this type of testing are the longer gauge length in which you will encounter more isolated flaws as opposed to the shorter gauge length of the 2 point bend test. The test data was fitted and analyzed with Weibull distribution as shown in Figure 14. Fiber under test is hermetic/polyimide coated MM fiber which was treated at 180oC for 48 hours. Strength

Figure 14: Failure probability of hermetic/ polyimide coated MM fiber measured at 0.5 m gauge length test


distribution is slightly shifted after exposing to high temperature, but staying in the range of 400-500 kpsi which is desirable for cabling process.

SUMMARY Use of optical fibers for sensing application requires high reliability, and sometimes unique characteristics depending on system design. Such a requirement can be met by modification of the glass and coating in the fiber. We first propose new fiber design for conducting the polarization-sensitive distributed sensing application. The fiber is specially designed to have longer beatlength (low birefringence), high uniformity of the linear birefringence, and optionally low bending loss. The birefringence properties of the fiber are achieved by elevating the fiber intrinsic birefringence above the non-uniform background level during preform making and subsequently draw the fiber using bidirectional or sinusoidal spinning. We demonstrated distributed intrusion sensing system incorporated with POTDRbased system. It enables to map the location of the perturbation at a particular fiber position to a physical location. Secondly, we demonstrated that hermetic/polyimide coated fibers can operate reliably at temperatures up to 190oC for extended period of time even in the presence of H2. We understand that hermetic layer can no longer effectively protecting H2 diffusion into glass at or above 185~190oC. However, extended lifetime at least by factor of 2-3 would be benefit for sensor manufacturers. The challenges faced in the near future include developing sensor and fiber combinations that meet the overall reliability and financial needs of the market. Closer interaction between the fiber designers and the sensor designers/manufacturers should bear fruitful results.

REFERENCE 1. Van Steenkiste, R. J. And Springer, G. S., Strain and temperature measurement with fiber optic sensors, 1997 (Technomic, Lancaster, PA). 2. Mrad, N. Optical Fiber Sensing Technology: Introduction and evaluation and application. In Encyclopedia of smart materials, vol. 2, p715, 2002 (John Wiley & Sons, Inc., New York). 3. P. David and S. Peter, Distributed Temperature Sensing using Fiber-Optics, 2000 Electricity Engineers Association Annual Conference (Auckland, New Zealand, June 16-17, 2000). 4. A. Mendez, Fiber bragg grating sensors and SHM applications: a market overview, 6933-05, SPIE Smart Structure/NDE (San Diego, CA 2008). 5. D. K. Gifford, S. T. Kreger, A. K. Sang, M. E. Froggatt, R. G. Duncan, M. S. Wolfe, and B. J. Soller, Sweptwavelength interferometric interrogation of fiber Rayleigh scatter for distributed sensing applications, 6770-14, SPIE Optics East (Boston, MA 2007). 6. S. Johnson, Fiber optic distributed temperature measurement in wells, Pruett Industries, Inc. (Bakersfield, CA 1996). 7. G. Fernando, Fibre optic sensor systems for monitoring composite structures, Reinforced Plastics, Volume 49, Issue 11, 41-49, December 2005. 8. Z. Wu and H. Zhang, Brillouin scattering based structural sensing in high temperature environment, 6757-02, SPIE Optics East (Boston, MA 2007). 9. C. H. Wang, A. Soufiane, I. Majid, K. Wei, and G. Drenzek, High reliability hermetic optical fiber for oil and gas application, 17th Int. Conference on Optical Fiber Sensors, SPIE Vol. 5855, p563 (SPIE, Bellingham, WA, 2005) 10. X. Chen, T. Hunt, M. J. Li, and D. Nolan, Properties of polarization evolution in spun fibers, Optics Letters, 28, 2028-2030 (2003). 11. I Kouzmina, C.K. Chien, P. Bell, E. Fewkes, Corning® CPC® Protective Coating – An Overview, Corning Optical Fiber White Paper WP3703 (2003). 12. K.E. Lu, G.S. Glaesemann, M.T. Lee, D.R. Powers, J.S. Abbott, Mechanical and hydrogen characteristics of hermetically coated optical fibre, Opt and Quant Elec, 22, 227-237, (1990). 13. K. J. Beales, D. M. Cooper, J. D. Rush, M. Fox, K. W. Plessner, and S. J. Stannard-Powell, Increased attenuation of optical fibers caused by diffusion of hydrogen, 9th ECOC, post deadline pap. (Oct. 1983). 14. K. Mochizuki, Y. Namahira, M. Kuwazura, and Y. Iwamoto, Effects of hydrogen on infrared absorption characteristics in optical fibers, WB2, OFC 84 (1984). 15. K. Mochizuki, Y. Namihira, M. Kuwazuru, and Y. Iwamoto, Behavior of hudrogen molecules absorbed on silica in optical fibers, IEEE J. Quantum Electron, Vol. QE-20, p694 (1984). 16. N. J. Pitt, A. Marshall, J. Irven, and S. Day, Long term interactions of hydrogen with single mode optical fibers, 10th ECOC, 15A2 (1984).
