**Durability of Functionalized Carbon Structures with Optical Fiber Sensors in a Highly Alkaline Concrete Environment**

**Kort Bremer 1,\*, Lourdes S. M. Alwis 2, Yulong Zheng 1, Frank Weigand 3, Michael Kuhne 4, Reinhard Helbig <sup>3</sup> and Bernhard Roth 1,5**


Received: 4 May 2019; Accepted: 12 June 2019; Published: 18 June 2019

**Abstract:** The paper presents an investigation into the durability of functionalized carbon structures (FCS) in a highly alkaline concrete environment. First, the suitability of optical fibers with different coatings—i.e., acrylate, polyimide, or carbon—for the FCS was investigated by subjecting fibers with different coatings to micro/macro bending and a 5% sodium hydroxide (NaOH) (pH 14) solution. Then, the complete FCS was also subjected to a 5% NaOH solution. Finally, the effects of spatial variation of the fiber embedded in the FCS and the bonding strength between the fiber and FCS was evaluated using different configurations —i.e., fiber integrated into FCS in a straight line and/or with offsets. All three coatings passed the micro/macro bending tests and show degradation after alkaline exposure, with the carbon coating showing least degradation. The FCS showed relative stability after exposure to 5% NaOH. The optimum bonding length between the optical fiber and the carbon filament was found to be ≥150 mm for adequate sensitivity.

**Keywords:** structural health monitoring (SHM); functionalized carbon structure (FCS); carbon reinforced concrete (CRC); fiber optic sensor (FOS); optical glass fiber

#### **1. Introduction**

The replacement of conventional techniques used for structural health monitoring (SHM) based on electrical means (i.e., strain gauges for strain measurement) [1,2] with fiber optic sensors (FOSs) has seen increased popularity over the last decade due to the number of advantages they possess over conventional schemes. The glass (silica) construction of fiber optics renders them robust and capable of withstanding harsh and corrosive environments [3–7]. The low attenuation of optical glass fiber enables them to be interrogated over a long length (i.e., several 100 s of kilometers). The fact that a multitude of sensors can be multiplexed within a single strand of fiber together with the possibility of utilizing a wider bandwidth makes FOSs most suitable for applications that require extraction of data from a vast amount of sensing elements such as those required for SHM of large structures [8–12]. Moreover, the sensors, although multiplexed along one fiber, have the capability to monitor several parameters simultaneously and thus provide the opportunity to assess not only, for example, the strain levels but also other chemical parameters, such as humidity and pH. Thus, it is evident from the

current literature that the utilization of FOSs for SHM has not yet reached its full potential and requires further investigations paired with advances in chemical and materials engineering. In fact, the recent merging between textile and sensor engineering has seen rapid developments in wearable technology, especially in the biomedical engineering field [13–15], which provides much motivation for adopting advances in chemical/textile engineering in other areas of sensing, i.e., SHM.

Conventional FOSs used for SHM have been mainly grating-based and require adequate attachment of the sensor region to the structure. This had been mainly achieved using binding agents, such as epoxy [16,17]. The installation of such a sensing scheme not only requires specialist trained personnel in fiber optics to carry out the installation, but also, the accuracy of the sensing mechanism relies heavily on the efficiency of the intermediate transfer agent, i.e., epoxy. In addition, the hassle of having to somehow attach the fiber on the structure makes it unattractive to civil engineers, who are not used to dealing with cables that are of micro-meter-diameter-level dimension [18]. An ideal solution would be to embed the sensors on to the strengthening element of the structure itself, which saves time and costs, increasing the efficiency of the overall scheme.

Projects such as the Carbon Concrete Composite (C3) initiative in Germany aim to replace traditional steel reinforced concrete (SRC) as a building material with carbon reinforced concrete (CRC) [19]. The inherent characteristics of CRC possess a considerable number of advantages over conventional SRC, i.e., light weight, lifespan, thermal and electrical conductivity properties, flexibility of fabrication, efficiency and immunity to risk of corrosion. Further, CRC can be constructed in thin layers with high tensile strength, which makes it most suitable for intelligent building construction. The direct integration of FOSs into CRC forms an advanced field of research, which needs careful investigation into the durability of the said integration and thus its practicality in long-term use. To this end, it is of vital importance to evaluate whether the physical construction of the optical glass fiber (preferably single-mode fiber with acrylate, polyimide or carbon coating) is able to withstand its host environment, i.e., concrete.

The combination of the high alkaline environment in concrete and the concrete mechanical stress adversely affects not only the integrity of the optical fiber but the strength of the bonding between the carbon fiber reinforcement and the optical fiber. In a worst-case scenario, the optical fiber can be damaged to a level of non-operation. The sensing region must not detach from the structure/reinforcement as a result of mechanical stress or be decomposed due to the high alkaline surrounding. These considerations emphasize the level of attention needed to evaluate the mechanical properties of the optical fiber embedded into concrete.

Over the past few decades, much work has been focused on the survival of glass in high alkaline concrete environments [20], leading to embedding FOSs in concrete to evaluate its durability [21,22]. Research has shown that polymeric coatings increase the durability of optical fibers in concrete [21]. In particular, optical silica glass fibers with high alkaline resistance were obtained by coating them with polyetherimide [23], fluorine-polymer [24], and carbon coating [25].

Previous work by the authors presented a Functionalized Carbon Structure (FCS), where fiber optic sensors were "woven" into carbon fiber reinforcement polymer (CFRP) strands in order to evaluate the bonding strength between the sensor element and concrete [26]. This was focused on FOSs embedded in FCSs and textile net structures (TNSs) based on alkaline resistant glass. This work was further extended to embed FCSs in concrete blocks to evaluate the performance of the FOSs and the detailed analysis on the performance and the results can be found elsewhere [27]. However, when FCSs are embedded into concrete structures, the highly alkaline environment would degrade the FCS, thus limiting its life-span and operation. The strength of resistance to these highly alkaline environments would depend on the type of fiber used and the material of the protective-coating surrounding the fiber, i.e., acrylate, polyimide, or carbon. To this end, the paper presented here consists of a detailed analysis on the durability of FCS embedded with optical glass fiber sensors in a highly alkaline concrete environment. In addition, the spatial variation of the optical fiber inside the FCS was also analyzed to evaluate its effects on the bonding between the optical glass fiber and FCS.

#### **2. Materials and Methods**

#### *2.1. Fabrication of one Dimensional (1D) and Two Dimensional (2D)-FCS*

Figure 1 depicts the experimental concept of embedding the FCS in concrete for its SHM, as evaluated in the work presented herewith.

**Figure 1.** Schematic of the application of functionalized carbon structures (FCS) for the reinforcement and structural health monitoring (SHM) of concrete structures.

The fabrication of the FCS has already been reported [26,27]. The fabrication technique was developed at the Saxon Textile Research Institute (STFI) in Chemnitz, Germany and is based on embroidering carbon fibers and optical glass fibers simultaneously in a grid-like format on a polyvinyl alcohol (PVA) nonwoven substrate. PVA was chosen for this purpose since it can be easily dissolved in hot water as well as it protects the FCS. The advantage of this fabrication technique is that depending on the application, tailored FCS can be produced with various forms of lattice structures, multiple layers of carbon filaments as well as with different configuration of the optical glass fibers inside the FCS. The latter is, in particular, interesting to optimize the bond between the textile carbon structure and the optical glass fiber and thus to optimize the sensor response.

In order to explore the sensor response and the resistance to highly alkaline concrete environment of the FCS, one dimensional (1D) and two dimensional (2D) FCSs have been fabricated. The 1D-FCSs have been fabricated, in particular, to investigate the sensitivity of the FCS to the highly alkaline concrete environment. To do this, a single strand of carbon filaments of 400, 800 or 1600 tex have been simultaneously embroidered with SM fiber Corning ClearCurve (CC) on the PVA nonwoven substrate. After dissolving the PVA, the length of each 1D-FCS was measured to be 300 mm.

The 2D-FCSs have been fabricated in order to evaluate whether the shape of the optical glass fiber inside the FCS has an impact on the sensor response. To explore this, FCSs have been fabricated with different configurations of the optical glass fiber, as shown in Figure 2. In total, the fiber is integrated into the FCS in three different configurations. In the first case, (a) the fiber is straight; in the second case, (b) the fiber is placed with a slight offset; and in the third case, (c) the fiber meanders along the material. The purpose of the second and third configurations is to explore spatial variation of the optical fiber inside the FCS and whether this results in a stronger bond between the optical glass fiber and the FCS and thus facilitates an optimized sensor response. The FCS fabricated for the investigation had a dimension of 500 <sup>×</sup> 110 mm2 and a grid size of 10 mm <sup>×</sup> 10 mm. These were fabricated using 1600 tex carbon filaments and optical glass fibers of type Corning CC with acrylate coating. For the second configuration, the offset of the optical glass fiber was introduced at half length of the FCS with an offset of one grid element. In case of the third configuration, the size of each meander was 2 × 3 grid elements. In case of the offset and meander configuration, no significant attenuation of the light propagating the optical fiber of the FCS could be measured.

**Figure 2.** 2D-FCSs with different optical fiber sensor configurations. Straight (**a**), with an offset (**b**) and meander (**c**) fiber configurations were applied to investigate the bond between optical fiber and the carbon filament.

#### *2.2. Optical Glass Fibers*

Since FCSs are designed for reinforced concrete structures, the optical fibers inside FCSs have to withstand the highly alkaline concrete environment and still be fully capable of expected operations in terms of sensing and light guidance. Usually a fiber coating is applied in order to protect the optical silica glass fiber against impacts and thus to avoid mechanical degradation. However, when, for instance, the fiber is subjected to harsh environments, such as highly alkaline concrete environments, the fiber coating might be destroyed and thus the optical glass fiber might lose its mechanical stability. Therefore, commercially available optical fibers with different fiber coatings have been investigated in terms of their suitability to be applied for FCS applications. For the investigation, the Corning CC with acrylate coating (ClearCurve ZBL), the OFS Fitel Clearlite with polyimide (PI) (F21976) coating as well as the OFS GEOSIL with carbon/polyimide (C/PI) (BF06159) coating were chosen.

#### *2.3. Evaluating Sensitivity to Micro- and Macro-Bending of Optical Fibers*

In addition, an advantage of optical fibers applied for FCSs, discussed in Section 2.2, is that they can be processed during the embroidery fabrication process, i.e., any breaks while embroidering on the PVA nonwoven substrate can be avoided as well as the introduced light attenuation inside the optical glass fiber due to micro- and macro-bending can be neglected. Hence, in order to investigate their sensitivity to micro- and macro-bending when embroidered on the substrate, an experimental set-up was designed that consisted of a 3D printed mold and a power meter (dB-meter from FiboTec). The 3D printed mold enables the periodic and reproducible bending of the glass fiber with a bending radius of five millimeters and the power meter is used to monitor the introduced light attenuation inside the optical glass fiber. The minimum bending radius of 5 mm was determined from the fabrication process of the FCS. In Figure 3, a picture of the experimental setup consisting of the 3D printed mold and the power meter is illustrated.

**Figure 3.** Experimental setup to evaluate the sensitivity to micro- and macro-bending effects on the applied optical glass fibers.

#### *2.4. Simulated Highly Alkaline Concrete Environment*

Moreover, the impact of the highly alkaline concrete on the optical fiber was simulated using a 5% NaOH solution (pH 14) [24]. The failure stress of the optical fibers as well as the sensor response to the FCSs was determined using the tensile testing machine (MFC Sensortechnik T3000) after being exposed to the 5% NaOH solution.

The 5% NaOH solution was used to simulate the influence of concrete pore water on the glass fiber and on the fiber coating, respectively. In case of determining the failure stress of the optical glass fibers for each exposure time, the mean value from five fiber samples were calculated. Moreover, the length of the fiber samples under test was 385 mm for the Corning CC and 280 mm for the OFS Clearlite and OFS GEOSIL. The length of the acrylate coated fiber was different due to the fixation of this fiber to the tensile testing machine. However, the length of all fibers-under-test was kept equal. Prior to the tensile tests, all fiber samples were strained up to 1 N and to additionally visualize the fiber surface after exposure to the highly alkaline 5% NaOH solution, images of the fibers were taken using a scanning electron microscope (SEM) at the Saxon textile research institute (STFI). To determine the sensor response of the FCSs after being exposed to simulated concrete pore water, the 300 mm long 1D-FCS was immersed in the 5% NaOH solution for three months. After the exposure, the 1D-FCSs was dried for one day and then subsequently mounted on the tensile testing machine. The sensor response was measured using the experimental setup described in Section 2.5.

#### *2.5. Evaluating Force Transfer of FCS*

In order to investigate the sensor response of the FCS a fiber optic Mach-Zehnder (MZ) interferometer was applied [25]. The previously developed fiber optic MZ interferometer enables the investigation of the force transfer between the FCS and the optical glass fiber (fiber optic sensors) and thus the characterization of the bond between these two elements of the FCS. As illustrated in Figure 4, the applied fiber optic MZ interferometer consists of a broadband light source (BBS) (Opto-Link C-Band ASE), two fiber optic 3 dB couplers, two fiber arms, and a spectrometer (OSA) (Ando AQ6317B). The fiber optic MZ interferometer was set-up with only SM fiber components and all fiber components were spliced together. Moreover, the FCS was spliced to one arm of the developed interferometer in order to determine the strain transfer between the optical fiber and the FCS. When force is applied to the FCS using the tensile testing machine, the related length change of the optical fiber of the FCS causes a phase difference (Δϕ) between the light (of wavelength λ) travelling in both fiber arms. This in turn results in a change of the interference pattern (displayed in the subset of Figure 4) from which the length change Δ*L* of the optical fiber inside the FCS can be determined as follows:

$$L = \frac{\lambda\_c^2}{n\_{core} \cdot \Delta\lambda} \tag{1}$$

In Equation (1), *ncore*, λ*c*, and Δλ are the refractive index of the fiber core, the central wavelength of the light source, and spectral difference between two adjacent maximums of the measured transmission spectrum of the Mach-Zehnder interferometer, respectively. The refractive index value for the fiber core was taken as 1.45.

**Figure 4.** Fiber optic Mach-Zehnder (MZ) interferometer to characterize the sensor response from the FCS under test.

#### **3. Results**

#### *3.1. Response of Optical Fibers to Macro- and Micro-Bending*

The results of the micro- and macro-bending tests of the optical glass fibers from Section 2.3 are summarized in Table 1. All three optical fibers show relatively less sensitivity to the applied bending. However, the best results were obtained for the two optical glass fibers from OFS (NA = 0.17). Therefore, due to less sensitivity to bending and thus the relatively low light attenuation, all three fiber types are suitable for the application in FCS.

**Table 1.** Results of the micro- and macro-bending test of the optical glass fibers with different protective coatings. CC: ClearCurve; PI: Polyimide; C/PI: Carbon/Polyimide.


#### *3.2. Response of Optical Fibers to Simulated Alkaline Pore Water*

The impact of the highly alkaline environment of the concrete on the optical glass fibers is summarized in Table 2 and Figure 5. The failure strain tests of the polyimide coated fiber had to be stopped after 14 days. This is due to the detachment of the polyimide coating from the glass fiber (Figure 5b). In addition, for the test on the polyimide coated fiber after been exposed to alkaline 5% NaOH solution, only three fiber samples could be measured, since all other fiber samples were already broken during mounting to the tensile testing machine. In case of the optical glass fiber with carbon coating, two different fabrication batches were evaluated. Fiber samples of the carbon coated optical glass fiber from the first fabrication batch were already consumed after 28 days of exposure to alkaline pore water. Therefore, a second batch of this fiber was ordered to at least conduct the resistant experiments to alkaline pore water over a period of one year. As can be seen in Table 2, the carbon coated optical glass fiber from the first fabrication batch did not show a clear trend towards a lower failure strain for increasing alkaline attack. However, in the case of the optical fibers with carbon coating from the second fabrication batch, a continuous degradation of the failure strain for increasing alkaline attack was observed. It is assumed that the defects in the coating from the second fabrication batch of the carbon coated optical glass fiber had allowed simulated pore water to attack the silica glass matrix of the fiber cladding. It can also be seen in Table 2 that the optical glass fiber with an acrylate coating shows continuous mechanical degradation with increasing alkaline attack.

**Table 2.** Change of the mean failure strain of the optical glass fibers with different fiber coating for different exposure times to highly alkaline 5% NaOH solution. N specifies the number of samples per test.


Figure 5 shows different fiber coatings after being subjected to simulated alkaline attack. As can be clearly seen in Figure 5b, the polyimide coating is detached from the fiber after only 14 days of alkaline attack. No defects on the fiber coating could be detected for the optical glass fibers with acrylate (Figure 5a) and carbon (Figure 5c) coatings. The results obtained are consistent with [24,25], where acrylate and carbon coated fibers showed resistance against the highly alkaline environment.

**Figure 5.** Scanning electron microscope (SEM) images of the optical glass fiber with acrylate (**a**), polyimide (**b**), and carbon/polyimide (**c**) coatings after 14- and 28-days of alkaline attack, respectively.

#### *3.3. Response of 1D-FCS to Simulated Alkaline Pore Water*

The response of the FCS (equipped with Corning CC) to alkaline pore water was investigated using 400, 800, and 1600 tex 1D-FCSs. In Table 3, the sensitivity of 300 mm long FCSs to applied force are illustrated before and after exposure to the simulated alkaline pore water solution over a period of three months. For each tex number, three 1D-FCS samples were measured for increasing and decreasing force before and after the alkaline attack. In addition, before the measurements commenced, the 1D-FCS samples were strained prior to the measurement up to 5 N and the maximum applied strain was measured to be 6.7 mE during the tests. From Table 3, it follows that in general the FCSs are relatively stable against alkaline pore water attack. One reason for this might be the PVA that is used to stabilize the FCS and known to be relatively inert against chemicals. Furthermore, FCSs with higher tex numbers show less sensitivity to applied force (due to the increased cross-section) but in turn also less variation of the sensor response before and after the alkaline attack.

**Table 3.** Change of the sensor response from 1D-FCS due to exposure times to highly alkaline 5% NaOH solution. σ specifies the standard deviation of the measurement.


#### *3.4. Sensor Response of FCS with Di*ff*erent Optical Fibre Configurations*

In Figure 6, the sensor response of the FCSs with different optical fiber configurations are illustrated (for all three different fiber configurations the Corning CC was applied). Before the measurements were performed, the FCS samples were strained to a force of 40 N and the maximum applied strain was 2.4 mE during the tests. In case of configuration one (straight arrangement, Figure 6a) and two (offset arrangement, Figure 6b), a linear response to an applied force of 6.4 <sup>×</sup> 10−<sup>4</sup> mm/N (R2 = 0.96) and 6.7 <sup>×</sup> <sup>10</sup>−<sup>4</sup> mm/<sup>N</sup> (R<sup>2</sup> <sup>=</sup> 0.99), respectively, with a relatively low hysteresis of 4.4 <sup>×</sup> <sup>10</sup>−<sup>5</sup> mm/N and 1.9 <sup>×</sup> <sup>10</sup>−<sup>5</sup> mm/N, respectively, were obtained. In case of the third configuration (meander arrangement, Figure 6c), no correlation between the applied force and the length change of the optical fiber of the FCS could be observed (R2 = 0.25). The obtained result for configuration three suggests that due to the meander shape and thus due to the periodic spatial variation of the optical fiber inside the FCS, the bond between the optical fiber and the carbon filament is less strong (less bonding length) and thus the FCS is less sensitive to applied force. Therefore, the obtained results (from the three different fiber configurations) suggest that for sufficient sensor sensitivity, the bonding length between the optical fiber and the carbon filament to be at least ≥150 mm long (compare configuration two).

**Figure 6.** Sensor response for the 2D-FCSs with straight (bonding length 300 mm) (**a**), offset (bonding length 150 mm) (**b**), and meander (bonding length 60 mm) (**c**) optical fiber configurations.

#### **4. Conclusions**

The results verified that all three different optical glass fibers can be applied for FCS applications. Following this, the resistance of the optical glass fibers to highly alkaline environment was evaluated where all three fiber coating materials showed degradation and thus not provide full protection against highly alkaline attack. However, optical glass fibers with carbon coating showed the most promising results. Carbon coated optical glass fibers of two different fabrication batches were evaluated. One fabrication batch showed almost no degradation due to alkaline attack while the failure strain of fibers of the second fabrication batch decreased with increasing exposure time to highly alkaline solution. Therefore, it is assumed that micro defects of the fiber coating of the second fabrication batch caused the degradation of the failure strain. The evaluation of the sensor response of the 1D-FCS before and after exposure to the NaOH solution indicated that the FCSs are relatively stable against alkaline pore water attack. One reason for this might be that the PVA that is used to stabilize the FCS is known to be relatively inert against chemicals. In addition, the response of the fiber optic sensors inside the FCS were investigated for three different fiber configurations. The purpose of the second and third configurations was to explore the spatial variation of the optical fiber inside the FCS and whether this results in a stronger bond between the optical glass fiber and FCS. The obtained results suggest that the bonding length between the optical fiber and the carbon filament should be at least ≥150 mm long in order to provide sufficient sensor sensitivity. Currently, work is ongoing to develop a theoretical model that will be used to simulate the work presented herewith, paving way to evaluate the optimum conditions for improved durability.

**Author Contributions:** Conceptualization, K.B., F.W., and B.R.; software, K.B. and M.K.; validation, K.B. and Y.Z., F.W. and M.K.; formal analysis, K.B., M.K., and F.W.; investigation, K.B., Y.Z., F.W., L.S.M.A., and M.K.; writing—original draft preparation, K.B. and L.S.M.A.; writing—review and editing, B.R.; supervision, B.R. and R.H.; project administration, B.R. and R.H.

**Funding:** This research was funded by the Bundesministerium fuer Bildung und Forschung (BMBF) within Grant Number 03ZZ0345 (Carbon Concrete Composite (C3)). B.R. acknowledges support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy within the Cluster of Excellence PhoenixD (EXC 2122). The publication of this article was funded by the Open Access fund of Leibniz Universität Hannover.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Intensity-Modulated PM-PCF Sagnac Loop in a DWDM Setup for Strain Measurement**

**Mateusz M ˛adry 1,\*, Lourdes Alwis <sup>2</sup> and El ˙zbieta Bere´s-Pawlik <sup>1</sup>**


Received: 9 April 2019; Accepted: 1 June 2019; Published: 11 June 2019

#### **Featured Application: Potential application of the presented setup for strain measurement in a DWDM configuration.**

**Abstract:** A novel intensity-modulated Sagnac loop sensor based on polarization-maintaining photonic crystal fiber (PM-PCF) in a setup with a dense wavelength division multiplexer (DWDM) for strain measurement is presented. The sensor head is made of PM-PCF spliced to single-mode fibers. The interferometer spectrum shifts in response to the longitudinal strain experienced by the PM-PCF. After passing the Sagnac loop, light is transmitted by a selected DWDM channel, resulting in a change in the output optical power due to the elongation of PM-PCF. Hence, appropriate adjustment of spectral characteristics of the DWDM channel and the PM-PCF Sagnac interferometer is required. However, the proposed setup utilizes an optical power measurement scheme, simultaneously omitting expensive and complex optical spectrum analyzers. An additional feature is the possibility of multiplexing of the PM-PCF Sagnac loop in order to create a fiber optic sensor network.

**Keywords:** fiber optic sensor; Sagnac loop; intensity-modulated; DWDM; strain sensor

#### **1. Introduction**

Optical fiber sensors have been widely explored due to their potential advantages, i.e., immunity to electromagnetic interferences, compact size, lightweight and high sensitivity [1]. They could be applied as sensors for temperature [2,3], refractive index [4,5], humidity [6,7] or strain [8–14]. In fact, strain determination is one of the most important factors in structural health monitoring (SHM) [8]. So far, different fiber strain sensors have been presented, for example Mach–Zehnder interferometers [9,10], Fabry–Pérot interferometers [11], fiber Bragg gratings [12] or long period gratings [13,14]. In addition, Sagnac interferometers with highly birefringent fibers are also used for strain measurement. For instance, polarization-maintaining fibers (PMF) were proposed to be sensor heads for strain detection [15]. However, conventional PMFs, i.e., PANDA or bow-tie, are susceptible not only to strain, but also to ambient temperature. To overcome the temperature cross-sensitivity of PMF, Dong et al. [16] and Han [17] demonstrated the use of polarization-maintaining photonic crystal fiber (PM-PCF) in order to achieve strain sensing, which is inherently not sensitive to ambient temperature. Fu et al. presented a pressure sensor by applying PM-PCF within a Sagnac loop achieving a sensitivity of 3.42 nm/MPa [18]. The PM-PCF Sagnac loop was also incorporated within a fiber ring laser (FRL) to evaluate its environmental stability [19]. However, current setups based on spectral analysis in order to convert shifts in the transmission spectrum of the interferometer in response to elongation or pressure are not convenient and portable. Therefore, intensity-modulated

optical fiber sensor setups are of interest in order to avoid expensive and advanced spectral analysis instruments [10,20]. Hence a novel, cost-effective, highly sensitive PM-PCF Sagnac loop strain sensor connected to a DWDM (Dense Wavelength Division Multiplexer) based on optical power measurement is herewith proposed. The incident light is modulated by the interferometer and output optical power is measured after passing through the DWDM. The elongation of PM-PCF causes a shift in the interferometer spectrum, resulting in different output spectra. Dong et al. showed a possibility of ~30 mε elongation of PM-PCF, which indicates good sensitivity, durability and efficient application in the field of strain sensing [16]. In the meantime, PM-PCF has low temperature sensitivity due to its structure, which was previously shown to be ~0.3 pm/1 ◦C [16,21].

The idea of the presented setup relies on a shift in the interferometer spectrum, which has an influence on the transmitted light modulated by DWDM. Therefore, investigation of axial strain is performed only by optical power meters. Approximately 11 dB of output optical power change was experimentally measured for elongation in a range of 0–2000 με, which proves good sensing capabilities. The highest achieved sensitivity is approximately 0.01 dB/με with maximal resolution of 1 με. By applying reference measurement, incident light fluctuation could be eliminated. Additionally, the proposed setup may be easily multiplexed to create a sensor network consisting of PM-PCF Sagnac interferometers.

The work presented here considers one possible sensor configuration utilizing PM-PCF. It sheds light on a new approach with the use of PM-PCF Sagnac loops and optical power meters in order to determine the elongation of the fiber itself. One great advantage of the proposed system is that it is cost-effective in comparison to the current literature that utilizes spectral analysis measurement, where an interrogation unit is also quite expensive [16,17,22]. Currently, the PM-PCF fabrication technology is well developed as evidenced by the wide commercial availability of these fibers. Only a small length of PM-PCF is needed to complete a single sensor unit. Another feature of the proposed setup is its fairly high resolution compared to other wavelength-based setups employing this type of photonic crystal fiber [16,17,22]. A disadvantage of the demonstrated setup is the need for careful adjustment of PM-PCF Sagnac interferometer spectrum with the DWDM channel. However, preparation of proper PM-PCF length to correlate the spectrum should not to be a challenging problem. Another issue, which needs to be taken into consideration regarding practical implementation, is the protection of single-mode fibers in order to avoid any bending and elongation of the non-sensing length of the fiber.

This paper, for the first time to the best of the authors' knowledge, presents a PM-PCF Sagnac loop sensor setup connected to a DWDM for sensing applications. The paper presents the theoretical background followed by the proposal for a sensor network as well as experimental results.

#### **2. Theory**

The Sagnac interferometer relies on the phase difference between two counter propagating light beams. Introducing highly birefringent fibers inside a loop provide different light paths, which results in a specific interferometer pattern at the output. The phase difference can be formulated as follows [16]:

$$
\varphi = \frac{2\pi \text{BL}}{\lambda} \tag{1}
$$

where B is the birefringence of the PM-PCF known as the difference between effective refractive indices of the fast and slow axis, respectively, L is the length of the fiber and λ refers to the light wavelength. Due to the determination of the phase difference (ϕ), the transmission spectrum of the interferometer could be presented according to the following equation [16]:

$$\mathbf{T} = 1 - \frac{\cos \varphi}{2} \tag{2}$$

Indeed, the transmission spectrum is a period function depending on the phase difference (ϕ). The wavelength spacing between two adjacent interferometer fringes could be approximated by the following function [16]:

$$
\Delta\lambda = \frac{\lambda^2}{\text{BL}}\tag{3}
$$

According to Equation (3), the wavelength spacing of interferometer fringes directly depends on parameters of the fiber, i.e., birefringence and length. Elongation of the fiber leads to a change of phase difference (Δϕ) between counter propagating light waves, which could be expressed by the following formula [16]:

$$
\Delta\varphi = \frac{2\pi}{\lambda} (\Delta \text{LB} + \text{L}\Delta \text{B}) \tag{4}
$$

As a consequence, this change of phase difference causes the interferometer spectrum to red-shift due to an increase in longitudinal strain. The temperature effect is negligible because of the PCF structure. In the proposed setup, the output power could be estimated as an integral over the common spectrum of the Sagnac interferometer (TINT) and filter function of a given DWDM channel (TDWDM) with respect to incident broadband light emission (TBLS):

$$\mathbf{P}\_{\rm out} \approx \int \mathbf{T}\_{\rm BLS}(\lambda) \mathbf{T}\_{\rm DWDM}(\lambda) \mathbf{T}\_{\rm INT}(\lambda) \, d\lambda \tag{5}$$

The strain induces change in phase difference, which results in a shift in the interferometer spectrum. Thus, the change of optical power could be approximated as follows:

$$
\Delta \mathbf{P} \approx \int \mathbf{T}\_{\text{BLS}}(\lambda) \mathbf{T}\_{\text{DWDM}}(\lambda) \left( \frac{1 - \cos \Delta \varphi}{2} \right) d\lambda \tag{6}
$$

Change of phase difference caused by elongation has an influence on the transmitted power. Both the sensitivity and the measurement range are related to the edge slope of the interferometer and the spectral characteristics of DWDM. According to Equation (3), the wavelength spacing of interferometer fringes depends on both incident light wavelength and parameters of the fiber, i.e., birefringence and length. Thus, by the adjustment of the length of the PM-PCF, the measurement range could be modified.

#### **3. Proposed Sensor Network**

The sensor network consists of a broadband light source, which is split into *N* number of Sensor Units. A small fraction of light is coupled out to the optical power reference measurement in order to eliminate power fluctuations. One sensor unit refers simply to the PM-PCF Sagnac loop. The main part of light propagates through the fiber coupler, demultplexer (DEMUX) and multiplexer (MUX). At the end, each DWDM channel is assigned to a given detector, which corresponds to the PM-PCF Sagnac loop. The whole scheme of the sensor network proposed is shown in the Figure 1.

In Figure 1 the OPM (REF) refers to the optical power meter used for reference measurement. OPM corresponds to the optical power meter used for measurement of optical power at the output of the n-th sensor unit, MUX is a multiplexer and DEMUX is a demultiplexer. The experimental setup was performed with one PM-PCF Sagnac loop (sensor unit) in order to prove the concept of the sensor network. Multiplexing of sensor units could lead to the obtaining of the proposed sensor network by selecting different wavelengths as it operates on a wavelength division multiplexing scheme. The presented sensor network proposal could then be implemented practically. The operating wavelength range of DWDM is consistent with the International Telecommunication Union (ITU) recommendations.

**Figure 1.** The proposed sensor network based on DWDM (Dense Wavelength Division Multiplexer) PM-PCF (Polarization-Maintaining Photonic Crystal Fiber) Sagnac loops.

#### **4. Experimental Setup**

The experimental sensor setup is presented in Figure 2 and consists of a broadband light source (superluminescent light-emitting diode, λpeak = 1544.4 nm, FWHM = 45.5 nm, Thorlabs), two fiber couplers 1 × 2 (split ratio—95:5), a 3-port circulator, a fiber coupler 1 × 2 (split ratio—50:50), a polarization controller, polarization-maintaining photonic crystal fiber (PM-PCF), a dense wavelength division multiplexer (DWDM, 100 GHz, 8 channels, Fiberon) and two hand-held optical power meters (OPM, Detector: InGaAs, Resolution 0.01 dB, Grandway).

**Figure 2.** The scheme for the intensity-modulated PM-PCF based Sagnac loop strain sensor. Inset: The image of PM-PCF cross-section.

The used PM-PCF is commercially available and its properties are herein presented: core diameter 6.3/4.4 ± 0.5 μm, outer cladding 125 ± 5 μm, coating diameter 240 ± 10 μm, length—40 cm, birefringence: <sup>≥</sup><sup>4</sup> <sup>×</sup> <sup>10</sup><sup>−</sup>4, attenuation <sup>≤</sup>3 dB/km at 1550 nm. The image of the cross-section of fiber is fully presented as an inset in Figure 2.

The first fiber coupler 1 × 2 (95:5) was used to eliminate fluctuations from the light source. The fluctuation of any light source output power could be observed in a function of time. In order

to prevent fluctuations, the fiber coupler is proposed to be implemented in the setup. Otherwise, the sensor output power value could be disturbed by light source power variation, thereby affecting strain determination accuracy. A circulator was placed within the sensor setup so as to provide the light propagation in the proper direction and to prevent any light reflections from affecting the superluminescent diode. In order to prepare the Sagnac interferometer, the splicing process of PM-PCF to SMF (single-mode fiber) was taken into consideration to enhance repeatability and minimize losses. Following the parameters presented by Xiao et al. [23], the PM-PCF was spliced to SMF using a commercial fusion splicer (FSU975, Ericsson). Additionally, appropriate adjustment of the polarization state within the Sagnac loop was made to provide adequate interferometer fringe visibility. A DWDM was incorporated to modulate the intensity of light by adjusting its spectrum with the edge slope of the PM-PCF Sagnac interferometer. Both spectrum of the DWDM channel and the Sagnac loop interferometer are shown in Figure 3.

**Figure 3.** The spectrum of selected DWDM channel and PM-PCF Sagnac loop interferometer.

Spectral analysis from Figure 3 reveals that the interferometer fringe spacing is approximately 6.5 nm (near λ = 1550 nm) with a fringe visibility of ~20 dB. The selected DWDM channel spectrum overlaps the slope of the interferometer spectrum, which ensures adequate operation of the proposed sensor. A shift in the interferometer spectrum provides different values of output optical power. The slope of the interferometer spectrum is also crucial for sensor setup capabilities as it is directly related to the fringe spacing and the measurement range of the sensor. Hence, the length of the PM-PCF needs to be controlled in order to meet specific requirements.

In summary, Table 1 presents all components used within the experimental sensor setup with specified parameters.


**Table 1.** The parameters of used components.

#### **5. Results and Discussion**

#### *5.1. Strain Response of the PM-PCF Sagnac Interferometer*

Firstly, the strain response of the PM-PCF Sagnac interferometer was investigated over the range of 0–1500 με in steps of 250 με through the use of translation stages in order to prove the sensing idea. The spectra of the interferometers are presented in Figure 4 as well as the wavelength shift as a function of elongation.

**Figure 4.** (**a**) The interferometer spectrum shift due to strain, (**b**) the strain response of a selected wavelength dip from the interferometer.

An interferometer fringe at 1546.3 nm was selected to analyze shifts due to elongation of the PM-PCF. A linear response to strain is observed, which is in agreement with the literature. The sensitivity of the PM-PCF Sagnac loop is determined to be approximately 0.98 pm/με.

#### *5.2. Strain Response of the Intensity-Modulated DWDM PM-PCF Sagnac Loop Sensor*

The proposed sensor setup as depicted in Figure 2 was investigated by monitoring the output light after passing the Sagnac loop and the DWDM due to elongation of PM-PCF. The fiber was stretched over a range of 0–2000 με in steps of 250 με. The output transmission spectra are shown in Figure 5.

It could be observed from Figure 5 that the intensity of the output light is different due to the applied strain, i.e., the elongation of PM-PCF influences the spectrum of PM-PCF, which shifts towards longer wavelengths and thus the light coupling out from the Sagnac loop is accordingly modulated by the DWDM. The integral over the output spectrum corresponds to measured optical power, which determines the elongation of the fiber. An increase in strain causes an increase in output light. The intensity levels of the output spectrum are different due to the influence of edge slope spectrum of the interferometer. Thus, an investigation into the optical power change was conducted using the

reference (*Pref*) and output power (*Pout*), which eliminates the fluctuation of incident light. Multiple measurements were performed in order to examine the proposed experimental setup. The relationship between the change of optical power and the axial strain is presented in Figure 6.

**Figure 6.** The optical power change as a function of axial strain.

The analysis of measurement data shows that the change in optical power exceeds 11 dB (~11.2 dB) within 2000 με with a negligible deviation between the performed measurements, maximally ± 0.03 dB. A nonlinear response to strain is observed, which could be a result of the spectral correlation between the interferometer and DWDM. The quadratic function was selected as a fitting function (*R*<sup>2</sup> = 0.999) determined by the following equation:

$$
\Delta P = -1.925 \cdot 10^{-6} \cdot \text{S}^2 + 0.009454 \cdot \text{S} + 0.03112 \tag{7}
$$

In the equation above *S* refers to the strain value applied on PM-PCF (με) and Δ*P* corresponds to the change in output power (dB). To determine sensitivity at given strain values, the first order derivative of the fitting function was calculated as follows:

$$\frac{\partial(\Delta \mathbf{P})}{\partial \mathbf{S}} = -3.85 \cdot 10^{-6} \cdot \mathbf{S} + 0.009454 \tag{8}$$

Thus, the sensitivity is approximately 0.01 dB/με at the initial value (0 με) and 0.002 dB/με at 2 mε, respectively. Assuming the resolution of the standard optical power meter used in the experiment

(0.01 dB), the resolution of the proposed setup changes within the range of 1 με to 5 με was the result of the measurement.

Due to the use of an optical spectrum analyzer with a resolution of 0.01 nm and assuming the use of the experimental PM-PCF Sagnac interferometer (sensitivity of 0.98 pm/με according to the Figure 4), the possible resolution could be ~10 με, which is lower than expected from the proposed intensity-modulated sensor setup.

The setup presented by Dong et al. had a sensitivity of 0.23 pm/με and a resolution of ~43 με [16]. A similar setup demonstrated by Han achieved a sensitivity of ∼1.3 pm/με [17]. Another comparable sensitivity was achieved by Orlando et al. [22], i.e., 1.21 pm/με or 1.11 pm/με depending on whether PM-PCF is uncoated or coated (acrylate). Thus, compared to wavelength-based sensor setups employing this type of PCF [16,17], the presented sensor exhibits a higher resolution (~1 to 5 με) and reduces system cost through the replacement of OSA (Optical Spectrum Analyzer) by optical power meters. Moreover, an easy multiplication of PM-PCF Sagnac loops is possible in order to create a sensor network as its operation relies on DWDM.

It is also necessary to account for possible error resulting from temperature effects. This constraint had already been thoroughly investigated in the literature regarding this type of fiber, [16,21,22] where almost inherent insensitivity to ambient temperature had been found in PM-PCF (~0.3 pm/ ◦C). Assuming a temperature variation of 50 ◦C (0–50 ◦C), the induced error could be approximately 15 με. It evidently shows that thermal effect could be negligible in the proposed setup. Nevertheless, the advantage of this setup is the utilization of cost-effective optical power measurement instead of spectral analysis. The measurement range is related to physical parameters of PM-PCF. An increase of PM-PCF length causes relatively smaller spacing of interferometer fringes, which reduces the measurement range of the sensor setup.

#### **6. Conclusions**

An intensity-modulated PM-PCF Sagnac loop for strain measurement has been presented and experimentally verified. The sensing part contains a Sagnac interferometer using a PM-PCF, which was subjected to axial strain. Due to the elongation of the PM-PCF, the interferometer spectrum shifts. By adjusting the DWDM, elongation of PM-PCF could be determined by direct output optical power measurement without a need for an OSA. Experimental results pointed out an increase in the output optical power as a function of longitudinal strain experienced by the PM-PCF. The setup has a maximal sensitivity of 0.01 dB/με and resolution of 1 με when measured using standard optical power meters. Additionally, this setup could be multiplexed in order to build a fiber sensor network by exploiting different DWDM wavelengths. It greatly reduces cost due to replacement of expensive and advanced OSA with optical power meters.

**Author Contributions:** Conceptualization, M.M. and E.B.-P.; investigation, M.M.; methodology, M.M.; supervision, E.B.-P.; validation, L.A.; writing—original draft, M.M.; writing—review & editing, L.A. and E.B.-P.

**Acknowledgments:** This research was co-financed by the Designated Subsidy for Young Scientists, no. 0402/0159/18 and statutory funds of the Telecommunications and Teleinformatics Department, Wrocław University of Science and Technology, no. 0401/0023/18.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
