Refractive Index Sensing with D-Shaped Plastic Optical Fibers ... - MDPI

Download PDF
4 downloads 526 Views 3MB Size Report
Comment
Dec 13, 2016 - Fiber Marketing & Development Group, Tokyo, Japan) were selected ..... smo o th sig n a l, k avg. (a .u.) Refractive index, next (RIU). (a). 1.41.
sensors Article

Refractive Index Sensing with D-Shaped Plastic Optical Fibers for Chemical and Biochemical Applications Filipa Sequeira 1,2, *, Daniel Duarte 1 , Lúcia Bilro 1,3 , Alisa Rudnitskaya 2,4 , Maria Pesavento 5 , Luigi Zeni 6 and Nunzio Cennamo 6 1 2 3 4 5 6

*

Instituto de Telecomunicações, 3810-193 Aveiro, Portugal; [email protected] (D.D.); [email protected] (L.B.) CESAM, University of Aveiro, 3810-193 Aveiro, Portugal; [email protected] I3N/FSCOSD, Department of Physics, University of Aveiro, 3810-193 Aveiro, Portugal Department of Chemistry, University of Aveiro, 3810-193 Aveiro, Portugal Department of Chemistry, University of Pavia, 27100 Pavia, Italy; [email protected] Department of Industrial and Information Engineering, Second University of Naples, 81031 Aversa, Italy; [email protected] (L.Z.); [email protected] (N.C.) Correspondence: [email protected]; Tel.: +351-234-377-900

Academic Editor: W. Rudolf Seitz Received: 17 October 2016; Accepted: 7 December 2016; Published: 13 December 2016

Abstract: We report the optimization of the length of a D-shaped plastic optical fiber (POF) sensor for refractive index (RI) sensing from a numerical and experimental point of view. The sensing principle is based on total internal reflection (TIR). POFs with 1 mm in diameter were embedded in grooves, realized in planar supports with different lengths, and polished to remove the cladding and part of the core. All D-shaped POF sensors were tested using aqueous medium with different refractive indices (from 1.332 to 1.471) through intensity-based configuration. Results showed two different responses. Considering the refractive index (RI) range (1.33–1.39), the sensitivity and the resolution of the sensor were strongly dependent on the sensing region length. The highest sensitivity (resolution of 6.48 × 10−3 refractive index units, RIU) was obtained with 6 cm sensing length. In the RI range (1.41–1.47), the length of the sensing region was not a critical aspect to obtain the best resolution. These results enable the application of this optical platform for chemical and biochemical evanescent field sensing. The sensor production procedure is very simple, fast, and low-cost. Keywords: plastic optical fiber (POF); refractive index sensors; optical fiber sensors; remote sensing; chemical and biochemical sensing

1. Introduction In recent years, plastic optical fibers (POFs) have been well known for their successful use as optical fiber sensors (OFS) [1–3]. These fibers share some attributes with glass optical fibers (GOF), such as immunity to electromagnetic fields, small size and weight, but also allow the production of low-cost sensing systems. Compared to GOFs, POFs show higher resistance to harsh environments, flexibility, simpler manufacturing and handling procedures, great numerical aperture, and allow easy connectorization due to the large diameter of the fibers. Bilro et al. presented a review of POF sensor technology with a special focus on intensity variation schemes and low-cost solutions [4]. Jin and Granville reviewed the recent progress in POF sensors, focusing on intrinsic detection schemes [5]. OFSs can be classified as intrinsic and extrinsic, depending on whether the fiber is interacting with an analyzed medium or if it is used only as a waveguide that allows the propagation of the light to the sensing region, respectively. Further, the detection scheme can be based on reflexion, where the light source and detector are placed on the same side of the Sensors 2016, 16, 2119; doi:10.3390/s16122119

www.mdpi.com/journal/sensors

Sensors 2016, 16, 2119

2 of 11

fiber; or on transmission, if they are placed on opposite sides. In both detection schemes, there are several transduction mechanisms that can be employed, including sensors based on variations of the evanescent field or in spectroscopic methods (absorption, fluorescence, and refractive index, RI). The light is propagated in an optical fiber by total internal reflection (TIR), and in a standard optical fiber, the interaction of the evanescent field with the external medium is negligible (radiation penetrates into the fiber cladding, with penetration depth around hundreds of nanometers, and whose energy decays exponentially with the distance from the core–cladding interface). For an optical fiber to be used as a sensor, to detect changes in the external medium—such as variations of the external refractive index or changes occurring in a selective layer deposited on the fiber—the light that propagates in the fiber core should interact directly or be able to sense the variations occurring in the external medium or in the selective layer. One simple procedure is to partially or totally remove the cladding of the fiber where the core has to be exposed. This can be done by applying mechanical tapering or chemical etching, where the cladding is removed along the thickness of the fiber. The cladding can also be removed by side-polishing on one side of the fiber, and a D-shape can be obtained. Depending on the aim of the application, one can choose to remove only the cladding (partially or totally) or also part of the core. In a fiber with an exposed core, the external medium acts as a “substitute” cladding, and the light that propagates in the fiber will interact with this external medium, and changes of RI can be monitored. When the aim is chemical sensing by the use of selective layers, this layer will be the substitute cladding, and the changes that occur in this layer can also be detected and monitored; for example, by the use of molecularly imprinted polymers (MIPs), where the binding of the template molecule with the MIP causes a variation in the polymer matrix. Several studies related to RI-POF sensors can be found in the literature. Bilro et al. reported theoretical modelling of D-shaped POFs at different macrobending conditions and external RI, which was validated by experimental results [6]; the experimental study was performed with different conditions of macrobending and external RI for sensing regions with lengths of 1.3 cm, 1.5 cm, and 2.1 cm with a total depth of 550 µm, 640 µm, and 550 µm, respectively (total depth is the total thickness of the remaining core). Feng et al. studied evanescent field sensing with the refractive index of a tapered POF at the wavelengths 532 nm, 633 nm, and 780 nm with different tapered waist diameters. This work reported that the reduction of the diameter of a tapered POF and the increase of the number of tapered regions improves the sensitivity and linearity of the sensor response. The best performance of the tapered POF based RI sensors was achieved at 633 nm for RI ranging from 1.33 to 1.41 [7]. An optimization of depth and curvature radius of a D-shaped POF with 1 cm length, aiming to increase linearity range and sensitivity to RI (1.333–1.455) at 652 nm, was also reported by Feng et al., with the best results being obtained for a depth of 500 µm and a curvature radius of 5 cm [8]. Nevertheless, a further optimization scheme was reported by the same authors—Liu and Feng, obtaining the best results using a D-shape with 2 cm length and an excurvation structure, with the same conditions of bending radius and depth; however, sensitivity and resolution of the sensor were not specified [9]. A resolution of 10−3 –10−4 RIU (refractive index units) for RI range (1.333–1.403) was reported for a multi-D-shaped sensor by Chen et al.; the D-shaped regions (from three to seven, spaced 1 mm apart) were written by femtosecond laser pulses in a communication-grade multimode silica optical fiber (Corning 62.5/125 µm) with 100 µm depth, 250 µm width, and 1 mm length [10]. The transmitted power increased linearly when the sensor is exposed to sucrose solutions of increasing refractive index, although a complex procedure and expensive instrumentation are required for the sensor production. Cennamo et al. have reported several biological and chemical sensors based on SPR (surface plasmon resonance) in a D-shaped POF platform [11–13] with 1 cm length sensing region and with typical optical resolution of around 6 × 10−4 RIU [14]. In all these reported works, the length of the sensing region was not taken into account in the optimization of a sensor’s sensitivity to RI. The novelty of this work is the study of the sensitivity to RI and resolution of D-shaped POF sensors with the length of the sensing region using a low-cost procedure and an intensity-based detection scheme. Numerical and experimental results will be

Sensors 2016, 16, 2119

3 of 11

presented Sensors 2016, towards 16, 2119 the optimization of the length of the D-shaped sensor for RI ranging from 1.332 3 of to 11 1.471. Finally, this work reports the development of low-cost intensity-based POF sensors with resolution suitable for chemical and biochemical applications [15,16]. presented towards the optimization of the length of the D-shaped sensor for RI ranging3 from Sensors 2016, 16, 2119 of 11 1.332 to 1.471. Finally, this work reports the development of low-cost intensity-based POF sensors with 2. Materials and Methods presented towards the optimization of the length of the D-shaped sensor for RI ranging from 1.332 to resolution suitable for chemical and biochemical applications [15,16]. 1.471. Finally,available this workplastic reports optical the development of low-cost intensity-based POF sensors withOptical Commercially fibers (POFs) from Asahi Kasei (DB-1000, Plastic resolution suitable for chemical and biochemical applications [15,16]. Fiber Marketing & Development Group, Tokyo, Japan) were selected with the following 2. Materials and Methods characteristics: a polymethyl methacrylate (PMMA) core of 980 μm, a fluorinated polymer cladding 2. Materials and Methodsplastic optical fibers (POFs) from Asahi Kasei (DB-1000, Plastic Optical available of 10 Commercially μm, a numerical aperture of 0.5. The refractive index in the visible range of interest is about 1.49 Commercially available Group, plastic optical fibers (POFs) from Asahi Kasei Plasticcharacteristics: Optical Fiber Marketing & Development Tokyo, Japan) were selected with(DB-1000, the following for PMMA and 1.40 for the fluorinated polymer. Fiber Marketing & Development Group, Tokyo, Japan) were selected with cladding the following a polymethyl methacrylate (PMMA) core of 980 µm, a fluorinated polymer of 10using µm, Glycerin was purchased from Carlo Erba Reagenti, and the solutions were prepared characteristics: a polymethyl methacrylate (PMMA) core of 980 μm, a fluorinated polymer cladding a numerical aperture refractive of 0.5. The refractive index solutions in the visible range of interest is about 1.49 for deionized water. index ofThe therefractive tested measured in all experiments of 10 μm, a The numerical aperture of 0.5. index in thewas visible range of interest is about 1.49 using PMMA and 1.40 for the fluorinated polymer. an Abbefor Refractometer, RMI, frompolymer. Exacta and Optech Labcenter. PMMA and 1.40Model for the fluorinated Glycerin was purchased from Carlo and solutions prepared Glycerin was purchased from CarloErba Erba Reagenti, Reagenti, and thethe solutions werewere prepared using using deionized water. The refractive index of the tested solutions measuredinin experiments using an deionized water. The refractive index of the tested solutions was was measured allall experiments using 2.1. Sensors’ Development an Abbe Refractometer, Modelfrom RMI, Exacta from Exacta Optech Labcenter. Abbe Refractometer, Model RMI, andand Optech Labcenter. The fibers were cut to about 20 cm length using a fiber optic cutter from Rennsteig, and 2.1. Sensors’ Development embedded inDevelopment grooves on planar supports (with different lengths from 1 cm to 6 cm), as shown in 2.1. Sensors’ Figure 1. The fibers were cut to about 20 cm length using a fiber optic cutter from Rennsteig, and The fibers were cut to about 20 cm length using a fiber optic cutter from Rennsteig, and embedded embedded in grooves on planar supports (with different lengths from 1 cm to 6 cm), as shown in in grooves on 1.planar supports (with different lengths from 1 cm to 6 cm), as shown in Figure 1. Figure

(a) (a)

(b) (b)

Figure 1. D-shaped sensors: (a) lengths of the sensing region ranging from 1 cm (Sensor 1) to 6 cm Figure 1. D-shaped Figure 1. D-shaped sensors: sensors: (a) (a) lengths lengths of of the the sensing sensing region region ranging ranging from from 11 cm cm (Sensor (Sensor 1) 1) to to 66 cm cm (Sensor 6); (b) Close-up of Sensor 1, with sensing region of 1 cm length. (Sensor 6); (b) Close-up of Sensor 1, with sensing region of 1 cm length. (Sensor 6); (b) Close-up of Sensor 1, with sensing region of 1 cm length.

The planar supports used in this study were made with ABS plus™ production-grade ® 3D plus™ Thethermoplastic, planar supports used in this study were made with ABS production-grade with an instrumental error of 100were μm, produced by Mojo Printer (provided by The planar supports used in this study made with ABS plus™ production-grade ® ® ® Stratasyswith FDMan , Eden Prairie, MN,error USA).ofThe dimensions of the grooves were 13D mm width, 700 μm thermoplastic, instrumental 100 μm, produced by Mojo Printer (provided by ® thermoplastic, with an instrumental error of 100 µm, produced by Mojo 3D Printer (provided ® FDM ®, lengths depth, and from 1–6MN, cm (with 1 cm step). Stratasys Eden Prairie, USA). The dimensions of the grooves were 1 mm width, 700 μm ® ® by StratasysTheFDM , Eden Prairie, MN, USA). The dimensions of the grooves weresupport 1 mm width, D-shaped obtained by polishing the fibers embedded in the planar depth, and lengths from sensors 1–6 cmwere (with 1 cm step). 700 µm with depth, and lengths from 1–6 cm (with 1 cmand step). sandpaper of 5 μm (LFG5P), 3 μm (LFG3P), 1 μm (LFG1P), with a “figure eight” pattern. The D-shaped sensors were obtained by polishing the fibers embedded in the planar support the surface of thewere fibersobtained was polished a D-shape, cladding and partinofthe the planar core wassupport TheWhen D-shaped sensors by into polishing the the fibers embedded with sandpaper of 5Figure μm (LFG5P), 3 μm (LFG3P), and 1 μm(h)(LFG1P), with a “figure eight” removed (see 2). The depth of the produced sensors can be easily calculated through thepattern. with sandpaper of 5 µm (LFG5P), 3 µm (LFG3P), and 1 µm (LFG1P), with a “figure eight” pattern. When the surface of d,the was polished into aregion, D-shape, fiber diameter, andfibers the half width of the sensing w. the cladding and part of the core was When the surface of the fibers was polished into a D-shape, the cladding and part of the core was removed (see Figure 2). The depth of the produced sensors (h) can be easily calculated through the removed (see Figure 2). The depth of the produced sensors (h) can be easily calculated through the fiber diameter, d, and the half width of the sensing region, w. fiber diameter, d, and the half width of the sensing region, w.

Figure 2. Image of a D-shaped fiber, where d is the fiber diameter, r the radius, h the depth, and w the half width of the sensing region.

Figure 2. Image of a D-shaped fiber, where d is the fiber diameter, r the radius, h the depth, and w the Figure 2. Image of a D-shaped fiber, where d is the fiber diameter, r the radius, h the depth, and w the half width of the sensing region. half width of the sensing region.

Sensors 2016, 16, 2119 Sensors 2016, 16, 2119

4 of 11 4 of 11

The The depth depth of of the the fiber, fiber,h,h,isistherefore thereforecalculated calculatedby bysimple simpletrigonometric trigonometricequations: equations: h  i r. cossin−1 w ,, h =ℎ r.=cos r where: where: w , and sin θ = = , and r h cos θ = = ℎ r

(1) (1)

(2) (2) (3) (3)

The The D-shaped D-shaped POF POF region region was was evaluated evaluated by by scanning scanning electron electron microscopy microscopy (SEM, (SEM, model model Zeiss Zeiss SUPRA35, Berlin, Germany). Figure 3 shows the analysis of the sensing area for three of the six sensors. SUPRA35, Berlin, Germany). Figure 3 shows the analysis of the sensing area for three of the six The width of width the sensing obtained, 2w, is around µm, equivalent to a depth, of 325 h, µm sensors. The of theregion sensing region obtained, 2w, is760 around 760 μm, equivalent to ah,depth, of (Equation (1)). 325 μm (Equation (1)).

(a)

(b)

Figure 3. SEM images of the D-shaped sensors: (a) Sensor 2; (b) Sensor 5. Figure 3. SEM images of the D-shaped sensors: (a) Sensor 2; (b) Sensor 5.

2.2. Optical Sensing Configuration 2.2. Optical Sensing Configuration Sensing in an intensity-based configuration allows for the measurement of the transmitted light Sensing in an intensity-based configuration allows for the measurement of the transmitted light that passes through the sensing area. The experimental setup (as shown in Figure 4a) comprised a that passes through the sensing area. The experimental setup (as shown in Figure 4a) comprised a stabilized power supply, LED (Avago SFH757V, Broadcom Limited, San Jose, CA, USA), stabilized power supply, LED (Avago SFH757V, Broadcom Limited, San Jose, CA, USA), wavelength wavelength centered at 650 nm, see Figure 4b), an optical coupler (50:50), two photodetectors centered at 650 nm, see Figure 4b), an optical coupler (50:50), two photodetectors (Avago SFH250V, (Avago SFH250V, Broadcom Limited, San Jose, CA, USA), and an oscilloscope (Picoscope). Output Broadcom Limited, San Jose, CA, USA), and an oscilloscope (Picoscope). Output data, time, and data, time, and voltage of the reference and sensor signals—in mV (Vreference and Vsensor, voltage of the reference and sensor signals—in mV (Vreference and Vsensor , respectively)—were logged respectively)—were logged into a PC by means of Picoscope’s software. The self-referenced transmitted into a PC by means of Picoscope’s software. The self-referenced transmitted signal (k) was used to signal (k) was used to correct source fluctuations and variations due to external conditions, as shown correct source fluctuations and variations due to external conditions, as shown in Equation (4): in Equation (4): Vsensor k == Vre f erence

(4)(4)

After adding adding the the test test solution solution to to the the D-shaped D-shaped sensor, sensor, the the signals signals were were recorded recorded for for 55 min, min, and and After the mean mean value value of of the the referenced referenced signal signal (k) (k) with with respective respective mean mean absolute absolute relative relative error error (MARE) (MARE) was was the calculated with with MATLAB MATLABsoftware. software. calculated

Sensors 2016, 16,16, 2119 Sensors 2016, 2119

5 of 1111 5 of

50,000

40,000

Counts (a.u.)

30,000

20,000

10,000

0 100

200

300

400

500

600

700

800

900

Wavelength (nm)

(a)

(b)

Figure 4. (a) Outline of the experimental setup; (b) LED spectrum. Figure 4. (a) Outline of the experimental setup; (b) LED spectrum.

2.3. Refractive Index Measurements 2.3. Refractive Index Measurements All the D-shaped POF sensors were tested using glycerin solutions with increasing refractive All the D-shaped POF sensors were tested using glycerin solutions with increasing refractive index varying from 1.332 to 1.471. In each performed test, the refractive index of the tested solutions index varying from 1.332 to 1.471. In each performed test, the refractive index of the tested solutions was measured with an Abbe Refractometer. was measured with an Abbe Refractometer. Each test started by adding deionized water to the D-shaped sensor. Measurements in water Each test started by adding deionized water to the D-shaped sensor. Measurements in water (which had the lowest refractive index) were used for signal normalization, according to Equation (5): (which had the lowest refractive index) were used for signal normalization, according to Equation (5): (5) k= k norm = solution (5) k water The D-shaped sensor was washed twice using the next test solution, with higher refractive Theto D-shaped was washed twice using the next testprevious solution,solution. with higher index, index, clean thesensor surface and eliminate any residues of the The refractive new test solution towas cleanthen the surface any residues(for of the previous The new test for solution then added, and eliminate the signals recorded 5 min). Thissolution. procedure was used all thewas tests and added, and the signals recorded (for 5 min). This procedure was used for all the tests and all sensors. all sensors. Three replicated measurements of the sensor response to refractive index (next )(nwere performed to ext) were performed Three replicated measurements of the sensor response to refractive index validate the obtained results.results. For each sensor, meanthe value of the threeoftests ) and the respective to validate the obtained For each the sensor, mean value the (k three tests (k avg ) and the avg standard deviation (δkavg ) were obtained. The obtained. sensitivityThe (S) of the D-shaped is definedsensor by: is sensitivity (S) ofsensor the D-shaped respective standard deviation (δkavg) were defined by: ∂k avg S = (6) =∂next (6) The resolution (∆n) is is defined asas the minimum amount ofof change inin refractive index that can bebe The resolution (Δn) defined the minimum amount change refractive index that can detectable, and can bebe defined as:as: detectable, and can defined 1 (7) ∆n = .δk avgmax S 1 (7) = . where δkavgmax is the maximum value of standard deviation obtained at the refractive index range ofwhere interest. δkavgmax is the maximum value of standard deviation obtained at the refractive index range of interest. 2.4. Sensor Modeling

2.4.ASensor Modeling simulation model of the sensor based on the model developed in reference [6] was implemented and compared to the model experimental thison model, generaldeveloped considerations must be noted: A simulation of the results. sensor For based the model in reference [6] was radiation losses in multimode POFs are usually calculated by a geometric approach; light is treatedmust as implemented and compared to the experimental results. For this model, general considerations individual and inlosses this case, losses occurPOFs by the interactions of light rays the interface layerslight of be noted:rays, radiation in multimode are usually calculated by awith geometric approach; the fiber core and external medium through refraction and reflection; uniform mode distribution (UMD) is treated as individual rays, and in this case, losses occur by the interactions of light rays with the was assumed; onlyofmeridional raysand were taken into account due torefraction their significant contribution to interface layers the fiber core external medium through and reflection; uniform radiation losses; the LED lightwas source was approximated to a Lambertian emitter, is I(θ) due = I0 .cos(θ), mode distribution (UMD) assumed; only meridional rays were taken intothat account to their where θ is the angle with respect to the fiber axis and I is the intensity for normal incidence; 0 significant contribution to radiation losses; the LED light source was approximated to a Lambertian emitter, that is I(θ) = I0.cos(θ), where θ is the angle with respect to the fiber axis and I0 is the intensity

Sensors 2016, 16, 2119

6 of 11

the polished surface will be considered as a macrobending, and its effects will also be modeled. The polished surface can be represented by an arc of a circle (concave surface) with length L and radius Rb0 obtained by: L2 + 4(r − h )2 Rb0 = − . (8) 8(r − h ) This bent surface will influence the power losses, being the bend angle (θ b0 ) given by: s 1−

θb0 = θc

 2r , Rb0 θc2

(9)

where θ c is the complementary angle of the critical angle for the core/cladding interface. The light rays will encounter the side-polished interface, and a fraction of energy will be lost by transmission. The loss reflection coefficient Rn can be calculated by the Fresnel’s equations for unpolarized light, through the amplitude reflection coefficients of the perpendicular and parallel polarization: R2⊥ + R2k Rn (θ ) = (10) 2 respectively, r r  2  2 ncl 1 − ncr sin θ − next cos θ ncr cos θ − next 1 − ncr sin θ next next and R (θ ) = r (11) r R⊥ (θ ) = k  2  2 ncr ncr ncl 1 − next sin θ + next cos θ ncr cos θ + next 1 − next sin θ with ncr and ncl as the fiber core and cladding refractive indices, respectively. Multiple reflections of a ray will occur over the whole sensitive section. The total number of internal reflections N(θ) that the light ray undergoes is the near integer obtained by N (θ ) =

L tan θ . 2(r + h )

(12)

To simulate the roughness of the sensing region, a dependent external refractive index power loss roughness coefficient (Rs ) was considered. This coefficient has a linear decrease profile until reaching the fiber cladding refractive index, and it is given by: Rs (next ) =

s n2H2O + n2cl

!

(n2ext + n2cl ),

(13)

where s stands for a roughness constant that characterizes the type of polish used, and is constant to all presented sensors in this work. The normalized transmitted output of the fiber will then be calculated by: Rθ N (θ ) Rs (next ) 0 b0 Rnext cos θsin θ dθ Pnext = . (14) ηRI = R θ N (θ ) Pn H2O Rs (n cos θsin θ dθ ) cR H2O

0

n H2O

3. Results and Discussion 3.1. Analysis of Sensitivity and Resolution vs. Length of D-Shaped POF Region Figure 5 shows the normalized smooth signal (kavg ) versus the refractive index (next ) as measured for all the investigated sensors. The symbols represent the experimental results, and the error bars represent the standard deviation. As shown in Figure 5, the response of the D-shaped sensors to refractive index (RI) is dependent on the length of the sensing region.

Sensors 2016, 2119 Sensors 2016, 16,16, 2119 Sensors 2016, 16, 2119

of 11 77 7ofof1111

Normalized Normalizedsmooth smoothsignal, signal,kavg kavg(a.u.) (a.u.)

1.3 1.3 1.2 1.2 1.1 1.1 1.0 1.0 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.51.32 1.32

Sensor 1 - 1 cm Sensor 1 - 1 cm Sensor 2 - 2 cm Sensor 2 - 2 cm Sensor 3 - 3 cm Sensor 3 - 3 cm Sensor 4 - 4 cm Sensor 4 - 4 cm Sensor 5 - 5 cm Sensor 5 - 5 cm Sensor 6 - 6 cm Sensor 6 - 6 cm 1.34 1.34

1.36 1.36

1.38 1.38

1.40 1.40

1.42 1.42

1.44 1.44

1.46 1.46

1.48 1.48

Refractive index, next (RIU) Refractive index, next (RIU)

Figure5.5.Normalized Normalized smoothsignal signal vs. refractive refractive index(RI) (RI) forall all D-shapedsensors, sensors, RI range Figure Figure 5. Normalizedsmooth smooth signalvs. vs. refractive index index (RI)for for allD-shaped D-shaped sensors,RIRIrange range 1.33–1.47. 1.33–1.47. 1.33–1.47.

1.20 1.20 1.18 1.18 1.16 1.16 1.14 1.14 1.12 1.12 1.10 1.10 1.08 1.08 1.06 1.06 1.04 1.04 1.02 1.02 1.00 1.00 0.98 0.98 1.33 1.33

1.2 1.2

Sensor 1 - 1 cm Sensor 21 -- 21 cm cm Sensor Sensor Sensor 23 -- 23 cm cm Sensor 43 -- 43 cm cm Sensor Sensor 54 -- 54 cm cm Sensor Sensor 5 5 Sensor 6 - 6 cm cm Sensor Fit 6 - 6 cm Linear Linear Fit

1.34 1.34

1.35 1.35

Normalized Normalized smooth smooth signal, signal, kavg kavg (a.u.) (a.u.)

Normalized Normalized smooth smooth signal, signal, kavg kavg (a.u.) (a.u.)

Two different responses were observed, and are depicted Figure For the D-shaped sensor Twodifferent differentresponses responseswere wereobserved, observed,and andare aredepicted depictedinin inFigure Figure6.6. 6.For Forthe theD-shaped D-shapedsensor sensor Two with 1 cm length (Sensor 1), the normalized smooth signal had a slight decrease in RI range between with1 1cm cmlength length(Sensor (Sensor1),1),the thenormalized normalizedsmooth smoothsignal signalhad hada aslight slightdecrease decreaseininRIRIrange rangebetween between with (1.33–1.37) RIU, and an exponential decrease the refractive index range (1.37–1.47) RIU. For all (1.33–1.37)RIU, RIU,and andan anexponential exponentialdecrease decreaseinin inthe therefractive refractiveindex indexrange rangeofof of(1.37–1.47) (1.37–1.47)RIU. RIU.For Forall all (1.33–1.37) other D-shaped sensors, the normalized smooth signal increased linearly with increasing the otherD-shaped D-shapedsensors, sensors,the thenormalized normalizedsmooth smoothsignal signalincreased increasedlinearly linearlywith withincreasing increasingRIRI RIinin inthe the other range (1.33–1.39) RIU, and an decrease of smooth signal with rangeof of of (1.33–1.39) andexponential an exponential exponential decrease of the the normalized normalized smooth with range (1.33–1.39) RIU,RIU, and an decrease of the normalized smooth signal withsignal increasing increasing RI was observed between (1.41–1.47) RIU. RI was observed betweenRIU. (1.41–1.47) RIU. RIincreasing was observed between (1.41–1.47)

1.36 1.36

1.37 1.37

1.38 1.38

Refractive index, next (RIU) Refractive index, next (RIU)

(a) (a)

1.39 1.39

1.40 1.40

1.1 1.1 1.0 1.0 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.51.41 1.41

Sensor 1 - 1 cm Sensor 21 -- 21 cm cm Sensor Sensor Sensor 23 -- 23 cm cm Sensor 3 3 Sensor 4 - 4 cm cm Sensor 54 -- 54 cm cm Sensor Sensor Sensor 56 -- 56 cm cm Sensor 6 - 6 cm Exponential Fit Exponential Fit

1.42 1.42

1.43 1.43

1.44 1.44

1.45 1.45

1.46 1.46

1.47 1.47

Refractive index, next (RIU) Refractive index, next (RIU)

(b) (b)

Figure 6. Normalized smooth signal vs. refractive index: (a) RI: 1.33–1.39, linear fitting; (b) RI: Figure6. 6. Normalized smooth signal vs. refractive (a) RI: 1.33–1.39, linear (b) RI: Figure Normalized smooth signal vs. refractive index: index: (a) RI: 1.33–1.39, linear fitting; (b)fitting; RI: 1.41–1.47, 1.41–1.47, exponential fitting. exponential fitting. 1.41–1.47, exponential fitting.

The sensitivity the resolution of the D-shaped sensors were calculated by (6) (7). Thesensitivity sensitivity and and the resolution ofof thethe D-shaped sensors werewere calculated by Equations Equations (6) and and(6) (7). The and the resolution D-shaped sensors calculated by Equations The obtained results are shown in Table 1 and Figure 7 for refractive index ranging from 1.33 to 1.39 The(7). obtained results results are shown in Table 1Table and Figure 7 for refractive index index ranging from 1.33 1.39 and The obtained are shown inabsolute 1 and Figure 7 for refractive ranging fromto1.33 RIU. In this range, the sensitivity is the value of the slope of the linear fitting obtained for RIU. In this range, the sensitivity is the absolute value of the slope of the linear fitting obtained for toeach 1.39sensor. RIU. In this range, the sensitivity is the absolute value of the slope of the linear fitting obtained each sensor. for each sensor. For Sensors linear increase the normalized smooth signal with the increase For Sensors2 22toto to6,6, 6,a aalinear linearincrease increaseofof ofthe thenormalized normalizedsmooth smoothsignal signalwith withthe theincrease increaseofof of For Sensors external refractive index was observed from 1.33 to 1.39 RIU. In this refractive index range, the external refractive index was observed from 1.33 to 1.39 RIU. In this refractive index range, the external refractive index was observed fromsensors 1.33 to were 1.39 RIU. In this refractive index range,ofthe sensitivity and resolution of the D-shaped strongly dependent on the length the sensitivity and resolution of the D-shaped sensors were strongly dependent on the length ofthe the sensitivity and resolution of the D-shaped sensors were strongly dependent on the length of sensing region. The sensitivity increased and the resolution decreased with increasing the length. sensing region. The sensitivity increased and the resolution decreased with increasing the length. sensing region. The sensitivity increased and −1the resolution decreased with the length. −3 increasing obtained The The best best results—sensitivity results—sensitivity of of 2.8271 2.8271 au.RIU au.RIU−1 and and resolution resolution of of 6.48 6.48 ×× 10 10−3 RIU—were RIU—were obtained for for Sensor 6 with 6 cm sensing length (see Table 1). Sensor 6 with 6 cm sensing length (see Table 1).

Sensors 2016, 16, 2119

8 of 11

The best results—sensitivity of 2.8271 au.RIU−1 and resolution of 6.48 × 10−3 RIU—were obtained for Sensor 6 with 6 cm sensing length (see Table 1). Table 1. RI: 1.33–1.39—Sensitivity and resolution of the D-shaped sensors, given by Equations (6) and (7). Sensor

Length/cm

Sensor 1 Sensor 2 Sensor 3 Sensor 4 Sensor 5 SensorsSensor 2016, 16, 6 2119

Sensitivity

1 2 3 4 5 6

Resolution

δkavgmax

10−1

0.0987 0.3699 0.8392 1.6467 2.2292 2.8271

1.86 × 4.95 × 10−2 2.18 × 10−2 1.11 × 10−2 8.21 × 10−3 6.48 × 10−3

0.0183

0.7787 0.9978 0.9982 0.9902 0.9986 0.9917

1.33–1.37

1.33–1.39 8 of 11

0.18

2.8

-3

Y = -0.939 + 0.630.x 2.4

Y= 8.12x10 + 0.731.e

0.16

(-1.417.x)

2

R = 0.999

2

0.14

R = 0.993 2.0

Resolution (RIU)

-1

RI Range

0.20

3.2

Sensitivity (au.RIU )

R2 (Linear Fit)

1.6 1.2 0.8

0.12 0.10 0.08 0.06 0.04

0.4

0.02

0.0

0.00 1

2

3

4

5

6

1

2

3

4

Length (cm)

Length (cm)

(a)

(b)

5

6

Figure 7. RI: 1.33–1.39: (a) Sensitivity and (b) resolution of the D-shaped sensors vs. length of the Figure 7. RI: 1.33–1.39: (a) Sensitivity and (b) resolution of the D-shaped sensors vs. length of the sensing region. For Forthe theD-shaped D-shapedsensor sensorwith with1 1cm cm length (Sensor resolution obtained is the for sensing region. length (Sensor 1), 1), thethe resolution obtained is for the refractive index range 1.33–1.37 refractive index range 1.33–1.37 RIU.RIU. Table 1. RI: 1.33–1.39—Sensitivity and resolution of the D-shaped sensors, given by Equations (6)

Sensitivity and resolution in the refractive index range from 1.41 to 1.47 RIU—where an and (7). exponential decrease of the normalized signal was observed—are shown in Table 2 and Figure 8. R2 In thisSensor range, theLength/cm sensitivity is Sensitivity given by the following Resolution RI Range δkavgmax expression: (Linear Fit) −1 R0×.n10 Sensor 1 1 0.0987 1.86 0.7787 1.33–1.37 (15) ext S( RI:1.41−1.47) = | R0 .A.e |. Sensor 2 2 0.3699 4.95 × 10−2 0.9978 0.9982 Sensor 3 3 0.8392 2.18 × 10−2 Table parameters 0.0183 given by Equations that allow the1.33–1.39 resolution −2 and (15), Sensor2.4RI: 1.41–1.47—Obtained 4 1.6467 1.11 × 10(7) 0.9902 and sensitivity of the D-shaped sensors to be calculated, respectively. 0.9986 Sensor 5 5 2.2292 8.21 × 10−3 0.9917 Sensor 6 6 2.8271 6.48 × 10−3 Sensor Length/cm R0 A δkavgmax RI Range R2 (Exponential Fit)

Sensor 1 25.25in the 0.9980 1.37–1.47 an −3.19 × 10−17 index range from 1.41 Sensitivity and1 resolution refractive to 1.47 RIU—where −14 Sensor 2 2 21.31 0.9968 −1.36 × 10was exponential decrease of the normalized signal observed—are shown in Table 2 and Figure 8. In 3 19.52 0.9917 2.38 × 10−13 this Sensor range, 3the sensitivity is given by − the following expression: 0.0286 Sensor 4 4 30.30 0.9995 1.41–1.47 −2.76 × 10−20 −30 . Sensor 5 5 45.41 0.9882 −( 4.91 × 10 |. | = . . (15) : . . ) Sensor 6 6 40.91 0.9916 −4.52 × 10−27

Considering the refractive index range from 1.41 to 1.47 RIU, the sensitivity and resolution of the D-shaped sensors were dependent on the refractive index. In this case, the length of the sensing area was not critical to obtain the best resolution. Resolution of about 10−3 RIU was obtained for refractive index of 1.44 or higher for all sensors. For example, for Sensor 3 with 3 cm sensing length, a resolution of 6.86 × 10−3 RIU (for 1.41 RIU) and 2.13 × 10−3 RIU (for 1.47 RIU) was obtained, with a sensitivity of 4.1709 au.RIU−1 and 13.4548 au.RIU−1, respectively—much higher than the values obtained in the RI range from 1.33 to 1.39 RIU. Table 2. RI: 1.41–1.47—Obtained parameters given by Equations (7) and (15), that allow the

Sensors 2016, 16, 2119 Sensors 2016, 16, 2119

9 of 11 9 of 11

26

0.022

24

Sensor 1 - 1 cm Sensor 2 - 2 cm Sensor 3 - 3 cm Sensor 4 - 4 cm Sensor 5 - 5 cm Sensor 6 - 6 cm

-1

Sensitivity (au.RIU )

20 18 16

0.018 0.016

14 12

Sensors 2016, 16, 2119 10 8 626

222

-1

Sensitivity (au.RIU )

020 1.41 18 16

0.014 0.012 0.010

9 of 11

0.008 0.006 0.022 0.004

Sensor 1 - 1 cm Sensor 2 - 2 cm Sensor 3 - 3 cm - 4 cm 1.42 Sensor 41.43 1.44 1.45 Sensor 5 - 5 cm Refractive index (RIU) Sensor 6 - 6 cm

14 12

0.020 0.002

1.46

1.47

0.018 0.000 1.41 0.016

Resolution (RIU)

424

Sensor 1 - 1 cm Sensor 2 - 2 cm Sensor 3 - 3 cm Sensor 4 - 4 cm Sensor 5 - 5 cm Sensor 6 - 6 cm

0.020

Resolution (RIU)

22

(a)

10

1.42

1.43

1.44

1.45

Refractive index (RIU)

0.014 0.012

Sensor 1 - 1 cm Sensor 2 - 2 cm Sensor 3 - 3 cm 1.46 Sensor 4 -1.47 4 cm Sensor 5 - 5 cm Sensor 6 - 6 cm

(b)

0.010

Figure 8. RI: 1.41–1.47: (a) resolution0.008 of the the D-shaped D-shaped sensors sensors vs. vs. length the Figure 8. RI: 1.41–1.47: (a) Sensitivity Sensitivity and and (b) (b) resolution of length of of the 8 0.006 sensing region. 6 sensing region. 0.004

4

0.002

3.2. Comparison between Simulation and Experimental Results 0.000 RIU, the sensitivity and resolution of the Considering the refractive index range from 1.41 to 1.47 0 1.41 1.42 1.43 1.44 1.45 1.46 1.47 1.41 1.42 1.43 1.44 1.45 1.46 1.47 The model described in Section applied to RIInfrom 1.39, a range interestarea for D-shaped sensors were dependent on2.4 thewas refractive index. this 1.33 case,tothe length of(RIU) theofsensing Refractive index Refractive index (RIU) − 3 chemical and biochemical applications. The results obtained with the simulations (solid lines) are was not critical to obtain the best resolution. Resolution of about 10 RIU was obtained for refractive (b) length, a resolution depicted in Figure 9 and compared with experimental results (symbols). index of 1.44 or higher for(a) all sensors. Forthe example, for Sensor 3 with 3 cm sensing 2

Normalized smooth signal, kavg (a.u.) Normalized smooth signal, kavg (a.u.)

of 6.86Figure × 10−8.3 RIU (for 1.41 RIU) and 2.13 × 10−3 RIU (for 1.47 RIU) was obtained, with a sensitivity RI: 1.41–1.47: (a) Sensitivity and (b) resolution of the D-shaped sensors vs. length of the −1 of 4.1709 au.RIU au.RIU−1 , respectively—much higher than the values obtained in the 1.3 sensing region.and 13.4548 Sensor 1 - 1cm RI range from 1.33 to 1.39 RIU. Sensor 2 - 2cm Sensor 3 - 3cm 3.2. Comparison between Simulation and Experimental Results Sensor 4 - 4cm 3.2. Comparison between Simulation and Experimental Results 1.2 Sensor 5 - 5cm Sensor The model described in Section 2.46 - 6cm was applied to RI from 1.33 to 1.39, a range of interest for The model described in SectionSimulation 2.4 was applied to RI from 1.33 to 1.39, a range of interest for chemical and biochemical applications. The results obtained with the simulations (solid lines) are chemical and biochemical applications. The results obtained with the simulations (solid lines) are 1.1 depicted in Figure 9 and compared with the experimental results (symbols). depicted in Figure 9 and compared with the experimental results (symbols).

1.01.3

1.2 1.33

Sensor 1 - 1cm Sensor 2 - 2cm Sensor 3 - 3cm Sensor 4 - 4cm Sensor 5 - 5cm 1.34 1.35 1.36 1.37 1.38 Sensor 6 - 6cm Simulation Refractive Index, n (RIU)

1.39

ext

1.1

Figure 9. Experimental and simulation results: normalized smooth signal vs. refractive index for all D-shaped sensors in the RI range (1.33–1.39). 1.0

In the RI range (1.33–1.39), a good correlation was observed between experimental results and the simulations performed for the D-shaped sensors with length equal to or greater than 4 cm. For 1.33 length, 1.34 1.35normalized 1.36 1.37 1.38 1.39 the sensors with shorter D-shape the transmitted signal predicted by the model Refractive Index,between next (RIU) the simulated and the experimental was higher than the observed one. This discrepancy data is possibly due to small variations in the morphology of the D-shape region (roughness and results: normalized refractive index Figure9.resulting 9.Experimental Experimental andsimulation simulation results: normalized smoothsignal signalvs. refractive indexfor forall all total Figure depth) fromand the manual process used for thesmooth preparation ofvs.the sensors. D-shaped D-shapedsensors sensorsininthe theRIRIrange range(1.33–1.39). (1.33–1.39).

4. Conclusions theRIRIrange range (1.33–1.39), a good correlation observed between experimental results and InInthe (1.33–1.39), a good correlation waswas observed between experimental results and the D-shaped POF refractive index sensors with different sensing region lengths have been the simulations performed forD-shaped the D-shaped sensors length or greater cm.the For simulations performed for the sensors with with length equalequal to or to greater than than 4 cm.4For characterized. Numerical and experimental results have shown that the resolution of this optical the sensors shorter D-shape length, the normalized transmitted predicted the model sensors with with shorter D-shape length, the normalized transmitted signalsignal predicted by theby model was platform is sufficiently low to enable further developments towards chemical and biochemical was higher than the observed one. This discrepancy between the simulated and the experimental sensing, while also allowing low volume sampling if equipped with an appropriate flow cell. The data is possibly due to small variations in the morphology of the D-shape region (roughness and total depth) resulting from the manual process used for the preparation of the sensors. 4. Conclusions

Sensors 2016, 16, 2119

10 of 11

higher than the observed one. This discrepancy between the simulated and the experimental data is possibly due to small variations in the morphology of the D-shape region (roughness and total depth) resulting from the manual process used for the preparation of the sensors. 4. Conclusions D-shaped POF refractive index sensors with different sensing region lengths have been characterized. Numerical and experimental results have shown that the resolution of this optical platform is sufficiently low to enable further developments towards chemical and biochemical sensing, while also allowing low volume sampling if equipped with an appropriate flow cell. The developed D-shaped sensors are easy to produce and allow a fast and low-cost sensing. The working refractive index range is of extreme importance in order to choose the best platform with the best performance in terms of sensitivity and resolution. The refractive index range from 1.33 to 1.39 RIU is of particular interest for chemical and biochemical applications. In this range, the sensitivity and the resolution strongly depend on the length of the sensing region, with the highest sensitivity (and lowest resolution of 6.48 × 10−3 RIU) being obtained for the sensor with 6 cm sensing length. In the RI range from 1.41 to 1.47 RIU, the sensitivity and resolution of the sensors depend on the refractive index of interest. In this region, the selected length of the sensing region in order to obtain the best performance should take the working refractive index into account. The biggest innovation achieved was the development of a POF sensor for RI sensing with 6.48 × 10−3 RIU resolution with simple and low-cost methods, and verification of the length-dependence of the sensing region in a D-shaped POF platform, for refractive index range 1.33–1.39. A linear response was obtained for all sensors with sensing length equal to 2 cm or higher. Although the obtained resolution is lower than the one obtained for a D-shaped POF-SPR platform (6 × 10−4 RIU [14]), it is the same order of magnitude as the one obtained for the sensor developed in reference [10] (10−3 –10−4 RIU), but without the need of expensive and complex instrumentation. Acknowledgments: This work is funded by FCT/MEC through national funds and when applicable co-funded by FEDER-PT2020 partnership agreement under the project UID/EEA/50008/2013 (project sWAT and Daniel Duarte research grant), Ph.D. fellowship (Filipa Sequeira: SFRH/BD/88899/2012) and investigator grant (Lúcia Bilro: IF/01664/2014; project INITIATE). Alisa Rudnitskaya wishes to acknowledge financial support from CESAM (UID/AMB/50017), FCT/MEC through national funds and the co-funding by the FEDER, within the PT2020 Partnership Agreement and Compete 2020 and through fellowship SFRH/BPD/104265/2014. The work was partly supported by Italian Ministry of University and Research (MIUR) PON 03PE_00155_1 OPTOFER. Authors wish to thank Pasquale Cirillo and Andrea Cirillo, from the Department of Industrial and Information Engineering, Second University of Naples, Aversa, Italy, for the production of the planar supports with grooves by Mojo® 3D Printer. Author Contributions: N.C., L.Z., F.S. and L.B. have conceived and designed the experiments; D.D. and L.B. have developed the model for the simulations; N.C. and F.S. have performed the images by SEM; all the authors have analyzed the results and have contributed to the writing of paper. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2.

3. 4. 5.

Gowri, A.; Sai, V.V.R. Development of LSPR based U-Bent Plastic Optical Fiber Sensors. Sens. Actuators B Chem. 2016, 230, 536–543. [CrossRef] Al-Qazwini, Y.; Noor, A.S.M.; Al-Qazwini, Z.; Yaacob, M.H.; Harun, S.W.; Mahdi, M.A. Refractive index sensor based on SPR in symmetrically etched plastic optical fibers. Sens. Actuators A Phys. 2016, 246, 163–169. [CrossRef] Park, J.; Park, Y.J.; Shin, J.-D. Plastic optical fiber sensor based on in-fiber microholes for level measurement. Jpn. J. Appl. Phys. 2015, 54, 028002. [CrossRef] Bilro, L.; Alberto, N.; Pinto, J.L.; Nogueira, R. Optical sensors based on plastic fibers. Sensors 2012, 12, 12184–12207. [CrossRef] [PubMed] Jin, Y.; Granville, A.M. Polymer Fiber Optic Sensors—A Mini Review of their Synthesis and Applications. J. Biosens. Bioelectron. 2016, 7, 194. [CrossRef]

Sensors 2016, 16, 2119

6. 7. 8. 9. 10. 11.

12.

13.

14. 15. 16.

11 of 11

Bilro, L.; Alberto, N.J.; Sá, L.M.; Pinto, J.L.; Nogueira, R. Analytical Analysis of Side-Polished Plastic Optical Fiber as Curvature and Refractive Index Sensor. J. Light. Technol. 2011, 29, 864–870. [CrossRef] Feng, D.; Liu, G.; Liu, X.-L.; Jiang, M.-S.; Sui, Q.-M. Refractive index sensor based on plastic optical fiber with tapered structure. Appl. Opt. 2014, 53, 2007–2011. Feng, D.; Zhang, M.; Liu, G.; Liu, X.-L.; Jia, D.-F. D-Shaped Plastic Optical Fiber Sensor for Testing Refractive Index. IEEE Sens. J. 2014, 14, 1673–1676. Liu, G.; Feng, D. Evanescent wave analysis and experimental realization of refractive index sensor based on D-shaped plastic optical fiber. Opt.-Int. J. Light Electron Opt. 2016, 127, 690–693. [CrossRef] Chen, C.H.; Tsao, T.C.; Tang, J.L.; Wu, W.T. A multi-D-shaped optical fiber for refractive index sensing. Sensors 2010, 10, 4794–4804. [CrossRef] [PubMed] Cennamo, N.; Varriale, A.; Pennacchio, A.; Staiano, M.; Massarotti, D.; Zeni, L.; D’Auria, S. An innovative plastic optical fiber-based biosensor for new bio/applications. The case of celiac disease. Sens. Actuators B Chem. 2013, 176, 1008–1014. [CrossRef] Cennamo, N.; D’Agostino, G.; Pesavento, M.; Zeni, L. High selectivity and sensitivity sensor based on MIP and SPR in tapered plastic optical fibers for the detection of L-nicotine. Sens. Actuators B Chem. 2014, 191, 529–536. [CrossRef] Cennamo, N.; Pesavento, M.; Lunelli, L.; Vanzetti, L.; Pederzolli, C.; Zeni, L.; Pasquardini, L. An easy way to realize SPR aptasensor: A multimode plastic optical fiber platform for cancer biomarkers detection. Talanta 2015, 140, 88–95. [CrossRef] [PubMed] Cennamo, N.; Massarotti, D.; Conte, L.; Zeni, L. Low cost sensors based on SPR in a plastic optical fiber for biosensor implementation. Sensors 2011, 11, 11752–11760. [CrossRef] [PubMed] Wang, X.D.; Wolfbeis, O.S. Fiber-Optic Chemical Sensors and Biosensors (2008–2012). Anal. Chem. 2013, 85, 487–508. [CrossRef] [PubMed] Wang, X.D.; Wolfbeis, O.S. Fiber-Optic Chemical Sensors and Biosensors (2013–2015). Anal. Chem. 2016, 88, 203–227. [CrossRef] [PubMed] © 2016 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/).