Paper

Ammonia detection in water with balloon-like plastic optical fiber sensor

, , , , , and

Published 29 June 2023 © 2023 IOP Publishing Ltd
, , Citation Nurfatihah Che Abd Rashid et al 2023 Eng. Res. Express 5 025080 DOI 10.1088/2631-8695/acdfbb

2631-8695/5/2/025080

Abstract

This work presents the demonstration of a plastic optical fiber (POF)-based ammonia sensor. The sensor head is formed by bending an unclad fiber optic sensor into a balloon-like structure. The bending radius of the balloon-like bent fiber optic sensor is varied from 1.0 to 2.5 cm. The performance of each sensor is tested using ammonia solution with concentration ranging from 0 to 15 mg l−1. Results show that the optimized performances of the proposed sensor occur when the bending radius is fixed at 1.5 cm. At this bending radius, the sensor illustrates the sensitivity of − 0.0024 (mg/l)−1 with linearity of 0.97 and resolution of − 4.17 mg l−1. For comparison, the bent sensor is compared to the straight sensor and performances of the former is found to be more superior. In addition, the balloon-like bent sensor is further tested with real water samples. The sensor sensitivity is found to be − 0.0022 (mg/l)−1 with linearity of 0.95 and resolution of − 4.54 mg l−1. The result shows that the sensor has comparable performance in the ammonia detection for both pure ammonia and real water samples. In essence, this balloon-like bent sensor functions without additional coating on the sensor head, making it favorable in terms of the simplicity of the design.

Export citation and abstract BibTeX RIS

1. Introduction

Fiber optic sensors have garnered considerable attention from researchers around the world due to their attractive features of easy installation, light weight, low cost and immunity from electromagnetic interference [1]. Optical fiber is originally used as an optical waveguide to transmits light with low loss through total internal reflection. As an optical waveguide, the light is contained within the cladding and core of the optical fiber. This circumstance prevents the propagating light from interacting with the surrounding medium. In order to facilitate the propagating light interact with the surrounding medium, approaches involving the modification to the fiber optic geometry such as removing a portion of the cladding is implemented. By removing a portion of the cladding, light propagating through the unclad region is partially exposed to the surrounding medium, thus allowing reactions like light absorption [2]. This method is commonly known as intrinsic fiber optic sensor.

In particular, intrinsic fiber optic sensor can be divided into three categories; Structured sensor [315], fiber grating sensor [16, 17] and interferometric sensor [1820]. The most popular method that has been reported in the literature is structured-type sensor, which includes straight unclad fiber [4, 7], tapered fiber [3], side polished fiber [12], reflective (tip) fiber [13, 14] and bending fiber [911, 15]. These sensors have been developed to detect properties such as temperature [7], liquid level [11] and chemical properties such as glycerin [12] and ammonia concentration [3, 4, 9, 13, 14].

Ammonia has been widely used in agricultural fertilizers, plastics, explosives and many other chemical and biological industries [20]. Despite the many applications of ammonia, excess amount of it, however, can be toxic for ecosystems and humans. Therefore, monitoring the level of ammonia for the purpose of environmental pollution detection and personal safety is highly desirable. For the detection of ammonia in water, structured-type fiber optic sensor such as straight unclad [5], tapered [3] and bent [6, 8, 9] fiber is widely used. Among these methods, the bent structure, especially u-shape [6, 8, 9], has been investigated by many researchers to determine the concentration of ammonia in water. This is because bent fiber has the potential to increase sensor sensitivity while also being easy to fabricate and suitable for point sensing [2]. Additionally, additional coating material such as, tin oxide [4], dye [14], sol-gel silica [6, 21] zinc oxide [8, 22] and oxazine 170 perchlorate [9, 13] were applied on the modified cladding to enhance the sensitivity and selectivity of the sensor in sensing ammonia concentration. However, applying additional coating on the sensor head will result in enhanced complexity on the design. Therefore, a balloon-like bent fiber optic sensor without additional coating on the sensor head is demonstrated in this work. The performances of the balloon-like bent sensor without coating are found to be comparable with previous developments with that of additional coating. In other words, this proposed work has the edge in terms of complexity of the design.

2. Experiment

2.1. Experimental setup

A clad modified fiber optic sensor is used to detect ammonia concentration in water. The clad-modified region of the sensor is prepared by using etching method in order to remove the cladding of the optical fiber. A plastic optical fiber (POF) from Avago with polymethyl methacrylate (PMMA) core and 50 cm long fiber is used as the sensing fiber throughout the experiment. The POF used has a core reflective index of 1.492 with the diameter of 980 μm and the cladding reflective index of 1.417 with the diameter of 20 μm. In this work, the sensor head is firstly prepared by removing the middle portion of the POF jacket using a stripper. The cladding of the sensing region of the fiber is then removed by dipping the sensing part in acetone for 10 s. Afterwards, the fiber is immediately immersed in DI water in order to neutralize interaction between the fiber and the acetone. Then, the surface of the sensing region is polished using sandpaper before being dipped again into DI water for cleaning. The FIS Fiber Optic Continuity Tester is used to test the sensing region of the fiber and the uniformity of the light is monitored as shown in figure 1. In ths work, no additional coating is applied on the surface of the modified cladding.

Figure 1.

Figure 1. Light uniformity observation using FIS Fiber Optic Continuity Tester. (a) Uniform light distribution on fiber core surface (b) Non-uniformity of light distribution on fiber core surfaces due to flaw in the cladding removal procedure.

Standard image High-resolution image

Figure 2 shows the setup used for this work consisting a sensing fiber is connected to an Ocean Optics HR4000 high-resolution spectrometer and an Ocean Optics DH-mini (UV–vis-NIR) light source . In this research, a straight and a bending fiber optic sensor are tested and compared. The setups for those sensors are illustrated in figures 2(a) and (b), respectively. In this research, a balloon-like bent sensor is introduced for bending effect to detect the ammonia concentration in water sample. At the beginning, an experiment to find out the optimal unclad length is conducted where the structure of straight fiber optic sensor is utilzed. The unclad length of the sensor is varied from 1.0 cm to 4.0 cm and the sensor performance of each unclad length is analysed. The best sensor with high sensitivity, linearity and resolution is further tested using balloon-like bent fiber optic sensor as shown in figure 2(b) for comparison. The bending radius of the balloon-like bent sensor head is varied from 1.0 to 2.5 cm. In this work, the bending structure is formed using a 3D printed fibre holder to fix the diameter of the bent structure. In addition, a shrink tube is employed at the glueing point to ensure that the curvature is maintained. This bending formation of the balloon-like bent structure is shown in figure 2(b). The sensor is tested in different ammonia concentrations and the performance is observed and analysed.

Figure 2.

Figure 2. Experimental setup for (a) straight and (b) balloon-like bend fiber optic sensor.

Standard image High-resolution image

In this intensity-based sensor, bending of optical fiber will induce loss [23]. This is because refractive index (RI) in the bent region is not uniform, unlike straight fiber. As the fiber is bent, the RI at the bending region varies. This condition leads to deviation of the light passing through the bending region from its center of curvature, resulting in light loss at the outer surface of the bending region to the surrounding medium. The the numerical aperture (NA) will change depending on RI variation which is stated as [2]:

Equation (1)

where ${{n}}_{{core}}$ and ${n}_{{cladding}}$ are the refractive index, RI of the fiber core and cladding, respectively, while $R$ stands for the radius of curvature, $r$ for the radius of the fiber core, and $a$ for the position of ray in the core, where ($-r\leqslant a\leqslant r$).

2.2. Sample preparation

In this work, the ammonia samples are prepared from the dilution of ammonia solution of HmbG chemical with deionized (DI) water. As illustrated in figure 3, the ammonia concentrations tested varied from 0 to 15 mg l−1 in 3 mg l−1 steps. Using the OceanView software, the light intensity of each sample is observed and recorded for the measurement. Three tests are conducted on each ammonia sample. The average value of the light intensity recorded for the three tests conducted for each concentration is analysed.

Figure 3.

Figure 3. Water samples with different ammonia concentration for pure ammonia sample.

Standard image High-resolution image

3. Result and discussion

In the first stage, the performance of the straight fiber optic sensor in detecting different ammonia concentration is determined. In this work, the output intensity, sensitivity, and linearity performance for an unclad length of 1 cm, 2 cm, 3 cm and 4 cm is analyzed and compared. The unclad length that produces high sensor sensitivity, linearity and resolution is chosen as the optimal unclad length of the sensor. Figure 4 illustrates the output intensity at 0 mg l−1 ammonia concentration for each unclad length. Based on figure 4, it can be observed that the light intensity decreases as the length of the unclad region increases. Moreover, the intensity spectrum shown in figure 4 shows that the sensor has the highest peak at 690 nm wavelength. Thus, the analysis on the sensor sensitivity and linearity is carried out on wavelength of 690 nm. Figure 5 depicts the relationship between normalized output intensity and ammonia concentration ranging from 0 to 30 m g /L . Based on figure 5, it can be seen that the normalized output intensity and ammonia concentration have a linear relationship. Therefore, the sensitivity of the sensor can be extracted out from the slope of the linear fit equation [24]. Furthermore, based on figure 5, it can be seen that the unclad length with the highest sensitivity and linearity is 2 cm. From the graph shown in figure 5, fiber optic sensor with unclad region of 2 cm gives sensitivity of 0.0011 mg l−1 and linearity of 0.7774. Thus, the 2 cm unclad length is chosen as the optimal unclad length in detecting variation of ammonia concentration.

Figure 4.

Figure 4. Output intensity as function of wavelength for different length of the unclad region.

Standard image High-resolution image
Figure 5.

Figure 5. Normalized output intensity as function of ammonia concentration for different length of the unclad region at operating wavelength of 690 nm.

Standard image High-resolution image

For further investigation, the sensor resolution is calculated using equation (2) [22].

Equation (2)

Equation (2) shows that higher sensor sensitivity will result in smaller sensor resolution. Small resolution value shows high sensor resolution performance which indicates the sensor reliability in detecting the smallest measurement. Based on equation (2), the calculated resolution for straight fiber optic sensor is 9.09 mg l−1.

In this work, the LOD is determined based on the the slope (sensitivity) of the calibration curve and the standard deviation of the linear graph with confidence factor of 3.3. The relationship of LOD is shown in the equation (3) [25]. Based on equation (3), the LOD obtained for the straight fiber optic sensor is 3.3780 mg l−1.

Equation (3)

In the second stage, the performance of the balloon-like bent sensor is analyzed. With bending radius variation from 1.0 to 2.5 cm and a step of 0.5 cm, the sensor is tested using different ammonia concentration. Based on figure 6, it can be observed that the output intensity is reduced significantly with the reduction of bending radius. This condition shows that bending of fiber causes losses to the fiber sensor. In theory, bending induced stress on the fiber surface. Stress on the fiber causes RI profile on the fiber surfaces to change, thus varying NA in the sensor head. Therefore, smaller bending radius will cause the RI profile changes more significantly, thus leading to more light loss to the surrounding medium [2].

Figure 6.

Figure 6. Output intensity as function of wavelength with varying bending radius (c = 15 mg l−1).

Standard image High-resolution image

Based on previous studies, reducing the bending radius will increase the sensor sensitivity. Nonetheless, for different sensing medium, there is an optimal bending radius that yeilds the highest sensitivity[2]. In this work, the optimum bending radius of the balloon-like bent sensor is analyzed by varying the bend radius from 1.0 cm to 2.5 cm. Figure 7 shows the sensor sensitivity and linearity for bend radius of 1.0 cm, 1.5 cm, 2.0 cm and 2.5 cm. Based on figure 7, the balloon-like bent sensor with 1.5 cm bending radius produces the highest sensitivity of −0.0024 m mg l−1 with linearity of 0.9710. Meanwhile, the calculated resolution and LOD is −4.17 mg l−1 and −2.3318 mg l−1, respectively. Moreover, it is clear from figure 7 that the output intensity decreases as the ammonia concentration increases. This circumstance contributes to the negative slope that represents the sensitivity of the sensor. As aforementioned, bending of fiber caused RI changes at the sensing region, thus increasing the NA at the bending region. Increasing of NA causes more light loss to the surrounding medium. This condition creates evanescence field around the sensing region within the surrounding medium where analyte in the surrounding medium absorbs the light used for the transmission. Therefore, as the amount of ammonia increases, more light is absorbed and it causes decreasing of light detected (negative slope) [23, 26]. On the other hand, the output intensity for straight fiber optic sensor increases linearly with the ammonia concentration as shown in figure 5. This is because straight sensor has uniform RI at the sensing area, resulting in the refractive index on the core remains constant. In theory, as ammonia concentration increases, the refractive index of the sample also increases. Due to this circumstance, when the ammonia concentration increases, the difference between the refractive index of sample and the refractive index of fibre core will decrease. This condition contributes to lower light loss, causing an increase in the output light intensity [27, 28].

Figure 7.

Figure 7. Normalized output intensity as function of ammonia concentration for different bending radius at operating wavelength of 690 nm.

Standard image High-resolution image

Figure 8 shows the output intensity for ammonia concentration of 0 mg l−1, 3 mg l−1, 6 m mg l−1, 9 mg l−1, 12 mg l−1, and 15 mg l−1. Based on figure 8, it can be seen that the intensity reduces as the ammonia concentration increases. This happens as a result of evanescence field that forms around the sensing region when bending is introduced. Evanescence field causes the analyte in the ammonia sample to absorb the light intensity that escapes from the sensing region [2, 26].

Figure 8.

Figure 8. Output intensity as function of ammonia concentration.

Standard image High-resolution image

For further analysis, the stability, stability precision and RSD of both sensors are further investigated. In this work, the response within a 180 s time frame is recorded for the investigation of the stability of the sensor. Figure 9 shows the time response for (a) straight and (b) balloon-like bent sensor in a 180 s time frame for ammonia concentration of 15 mg l−1. The time response for both sensors is recorded at the operating wavelength of 690 nm. Based on figure 9, it is obvious that the intensity for balloon-like bent sensor is highly stable throughout this time frame. On the other hand, the intensity for straight sensor has slight fluctuation within 120 to 150 s time frame.

Figure 9.

Figure 9. Sensor stability as a function of time in ammonia sample (c = 15 mg l−1).

Standard image High-resolution image

For further study, precision of the sensor is calculated based on the stability graph shown in figure 9. The mathematical formulation for the stability precision is as follows [25]:

Equation (4)

where ${I}_{{average}}$ is an average value of the measured data and ${I}_{n}$ is the normalized intensity value of the nth measurement. From the equation, the stability precision for the balloon-like bent fiber optic sensor is 99.90%, while the straight fiber optic sensor is 99.88% for 15 mg l−1 ammonia concentration.

In this work, the measurement is repeated three times. The recorded output intensity for straight fiber optic sensor with the length of unclad region of 2 cm and balloon-like bent fiber optic sensor with 1.5 cm bending radius is shown in figures 10(a) and (b), respectively. The measured data obtained from the first, second, and third experiments are shown in measurement 1, measurement 2 and measurement 3, respectively. The standard deviation, σ, gain from all three measurement is recorded and the RSD of the system is calculated.

Figure 10.

Figure 10. Output intensity as a function of ammonia concentration for (a) straight and (b) bend radius operating at 690 nm wavelength.

Standard image High-resolution image

Relative standard deviation (RSD) is a technique used to quantify the deviation, $\sigma $ of a group of data distributed around the mean, $\mu .$ Figure 11 shows the calculated RSD value for ammonia concentration ranging from 0 to 15 mg l−1. Based on figure 11, it can be observed that the minimum RSD For balloon-like bent sensor is 0.0039% at ammonia concentration of 9 mg l−1, while the maximum RSD is 0.1320% at ammonia concentration of 15 mg l−1. Meanwhile, for straight fiber optic sensor, the minimum RSD is 0.0154% at ammonia concentration of 15 mg l−1, while the maximum RSD is 0.2235% at ammonia concentration of 6 mg l−1.

Equation (5)

Figure 11.

Figure 11. Calculated RSD for straight and bent sensor operating at 690 nm.

Standard image High-resolution image

Table 1 shows the performance comparison between straight fiber optic sensor and balloon-like bent fiber optic sensor. The performance of each sensor is compared into six performance parameters; Sensitivity, linearity, resolution, LOD, precision and RSD. Based on table 1, it can be observed that the sensor with the balloon-like bent structure shows better performance in terms of sensor sensitivity, linearity, resolution, LOD, precision and RSD. This proves that balloon-like bent sensor improves the performance of sensor in detecting different ammonia concentration. Meanwhile, in comparison to previous developments as tabulated in table 2, the performances of this balloon-like bent sensor are comparable. Despite the absence of additional coating on the sensor head, the performances in terms of sensitivity and linearity are competitive with respect to others. This results from the utilization of the balloon-like fiber optics sensor where it introduces bending effect which enhances the exposure of evanescent waves for the interaction with ammonia in water [29, 30]. In other words, this sensor has the edge of design simplicity resulting from the avoidance of the additional coating.

Table 1. Performance comparison between straight and bend structure.

StructureSensitivity ((mg/L)−1)R2 Resolution (mg/L)LOD (mg/L)Precision (%)RSD (Avg.) (%)
Straight0.00110.77749.093.378099.880.10
Bend−0.00240.9710−4.17−2.331899.900.06

Table 2. Performance comparison with previous developments.

StructureSensitivity ((mg/L)−1)Linearity (R2)Coating
U-shape [9]−0.00200.9099Oxazine with PDMS
U-shape [13]0.00360.9677Oxazine with PDMS
Balloon-like bent [This work]−0.00240.9710No coating

For further investigation, the balloon-like bent sensor with unclad length of 2 cm and bending radius of 1.5 cm is tested using real water samples gathered in the vicinity of Batu Pahat, Johor, Malaysia. The water samples are first tested using Ammonia High Range Portable Photometer (HI96733) purchased from Hanna Instrument and the concentration obtained are 1.1 mg l−1, 2 mg l−1, 3.1 mg l−1, 7.9 mg l−1. 8.9 mg l−1 and 15.4 mg l−1. As the reference, the ammonia concentration of DI water is also measured with the commercial photometer which gives the reading of 0 mg l−1. For comparison, the real water samples are then tested using the proposed fiber optic sensor and the result obtained is shown in figure 12. Based on figure 12, the proposed balloon-like bent fiber optic sensor shows sensitivity of −0.0022 mg l−1 and linearity of 0.9545. Meanwhile, the calculated resolution, LOD, stability precision and RSD are found to be −4.54 mg l−1, −1.5220 mg l−1, 99 .91% and 0.05% respectively. The obtained result shows that the sensor has comparable performance in the ammonia detection for both pure ammonia and real water samples.

Figure 12.

Figure 12. Normalized output intensity as function of ammonia concentration in real water sample.

Standard image High-resolution image

4. Conclusion

In conclusion, this work demonstrates experimentally ammonia sensing utilizing an unclad balloon-like bent fiber optic sensor. For the sensor design, an uncladded fiber optic sensor is bent into a balloon-like shape in which it uses a 3D printed fiber holder to fix the bending diameter of the structure and a shrink tube to maintain the curvature. The sensor functions without additional coating on the sensor head. The sensor is tested using ammonia concentration varied from 0 mg l−1 to 15 mg l−1. Experimental results reveal that at the operating wavelength of 690 nm, the sensor shows linear relationship with ammonia concentration. It is found that the sensor shows the optimal performance at the unclad length of 2 cm. After variation of bending radius, the optimal bending radius is found at 1.5 cm with sensitivity of −0.0024 mg l−1, linearity of 0.9710, resolution of 4.17 mg l−1, stability precision of 99.9%, LOD of −2.3318 mg l−1 and RSD of 0.06%. The sensor is not only tested using pure ammonia solution, but also in real water samples. It is found that the sensor shows comparable performance in detecting ammonia in both pure ammonia and real water samples. In essence, despite the absence of additional coating on the sensor head, the performance of the sensor in terms of sensitivity and linearity are competitive with respect to other previous developments.

Acknowledgments

The authors thank Universiti Tun Hussein Onn Malaysia (UTHM) for supporting this research work under Post-Graduate Research Grant (GPPS) Code H620.

Data availability statement

All data that support the findings of this study are included within the article (and any supplementary files).

Please wait… references are loading.
10.1088/2631-8695/acdfbb