Sensors 2013, 13, 4067-4087; doi:10.

3390/s130404067
OPEN ACCESS

sensors
ISSN 1424-8220
www.mdpi.com/journal/sensors
Article

Fabrication Quality Analysis of a Fiber Optic Refractive Index

Sensor Created by CO2 Laser Machining
Chien-Hsing Chen 1, Bo-Kuan Yeh 2, Jaw-Luen Tang 1 and Wei-Te Wu 2,*
1
Department of Physics, National Chung Cheng University, Chiayi 621, Taiwan;
E-Mails: [email protected] (C.-H.C.); [email protected] (J.-L.T.)
2
Department of Biomechatronics Engineering, National Pingtung University of Science and Technology,
Pingtung 912, Taiwan; E-Mail: [email protected]

* Author to whom correspondence should be addressed; E-Mail: [email protected];

Tel.: +886-8-770-3202 (ext. 7599); Fax: +886-8-774-0420.

Received: 28 December 2012; in revised form: 14 March 2013 / Accepted: 22 March 2013 /
Published: 26 March 2013

Abstract: This study investigates the CO2 laser-stripped partial cladding of silica-based
RSWLF ILEHUV ZLWK D FRUH GLDPHWHU RI  ȝP ZKLFK HQDEOHs them to sense the refractive
index of the surrounding environment. However, inappropriate treatments during the
machining process can generate a number of defects in the optic fiber sensors. Therefore,
the quality of optic fiber sensors fabricated using CO2 laser machining must be analyzed.
The results show that analysis of the fiber core size after machining can provide preliminary
defect detection, and qualitative analysis of the optical transmission defects can be used to
identify imperfections that are difficult to observe through size analysis. To more precisely
and quantitatively detect fabrication defects, we included a tensile test and numerical
aperture measurements in this study. After a series of quality inspections, we proposed
improvements to the existing CO2 laser machining parameters, namely, a vertical scanning
pathway, 4 W of power, and a feed rate of 9.45 cm/s. Using these improved parameters, we
FUHDWHG RSWLFDO ILEHU VHQVRUV ZLWK D FRUH GLDPHWHU RI DSSUR[LPDWHO\  ȝP QR REYLRXV
optical transmission defects, a numerical aperture of 0.52 ± 0.019, a 0.886 Weibull
modulus, and a 1.186 Weibull-shaped parameter. Finally, we used the optical fiber sensor
fabricated using the improved parameters to measure the refractive indices of various
solutions. The results show that a refractive-index resolution of 1.8 × 10í4 RIU (linear
fitting R2 = 0.954) was achieved for sucrose solutions with refractive indices ranging
between 1.333 and 1.383. We also adopted the particle plasmon resonance sensing scheme
Sensors 2013, 13 4068

using the fabricated optical fibers. The results provided additional information, specifically,
a superior sensor resolution of 5.73 × 10í5 RIU, and greater linearity at R2 = 0.999.

Keywords: CO2 laser machining; optical fiber sensor; refractive index sensing

1. Introduction

Various biosensors, such as the electrochemical sensors developed by Clark and Lyons [1], use a
galvanometer to measure the glucose concentration of a solution and achieve the measurement objectives.
Additionally, semiconductor ion-sensitive biosensors adopt a semiconducting structure that comprises
metal-insulating field effect transistors (MISFET) [2]. Another example are the optical fiber sensors
that exploit the optical fiber transmission characteristics to achieve sensing objectives, such as
evanescent wave and surface plasmon resonance technologies [3].
Of the various biosensor types, optical fiber biosensors offer the unique characteristic of no
electromagnetic interference. Small, lightweight, and with the potential for miniaturization, optic fibers
can be used not only to transmit light signals, but also as the primary sensing element. Optical fibers
are widely employed for engineering and environmental control and in mechanical and biological
developments [4].
Optical fibers have a three-layer structure that comprises a silica-based fiber core, a polymer
cladding, and a coating of harder polymer as the outermost layer that protects the fiber. Various
methods and structures to provide optical fibers with sensing capabilities have been developed,
including fiber Bragg grating [5], fiber-optic interferometers [6], and window-type optical fiber
sensors [7]. Among them, window-type optical fiber sensors, as shown in Figure 1, have the simplest
structure; only partial removal of the coating material is required to expose the fiber core beneath.
Once exposed, the window-type optical fiber structure allows sensors in a test environment to conduct
ambient refractive index sensing using the attenuated total reflection (ATR).
The current methods for stripping part of the optical fiber material can be broadly divided into
mechanical and chemical methods. The most common of the many mechanical fiber optic stripping
methods involve polishing the stripper or fiber [8]. However, the fact that the fiber optic stripper can
potentially damage the fiber core presents a significant disadvantage. The fiber polishing method
typically requires more expensive equipment, although it does offer high machining accuracy. The
chemical method involves the use of various solutions such as sulfuric acid, which was employed by
Matthewson [9]. The optical fiber was soaked in sulfuric acid before heating it to between 180 and
200 °C to soften and strip the outer coating material. Nonetheless, etching quality is also difficult to
control because a slight error can generate unexpected processing phenomena that affect the sensing
quality. Researchers have also employed the flame vaporization technique by exploiting the melting
point characteristics of various layers of the optical fiber cable. This technique is used to vaporize the
outer cladding material, exposing the glass fiber core. Although easy to process, the processing scope
and duration of this method is difficult to control, rendering it unsuitable for extended research [10].
Other research teams have employed lasers and precision lenses on laser processing platforms with a
moving mechanism. This method of removal is less time-consuming compared to the other two
Sensors 2013, 13 4069

methods and the parameters are easier to control. Regarding laser processing, the precision lens on the
laser processing platform tends to age, which may affect the accuracy of the moving platform and lead
to cause experimental errors. This can also lead to inferior processing results, problems such as an
inability to correctly remove materials, and/or changes in the material properties and costs of heating
the area because of excessive laser energy [11,12]. However, regardless of which method is employed,
they all provide the same disadvantage, that is, a lack of comprehensive post-processing quality control
procedures.
Based on the above discussion, and to further understand the basic characteristics of window-type
optical fiber sensors, we used CO2 laser processing as the fiber optic sensor processing method in this
study because the laser-processing parameters are convenient to configure and provide a wider range
of basic characteristics. Studies of common optical fiber sensors typically investigate only fiber optic
sensor fabrication methods or the resolution and sensitivity of back-end sensing applications; the
processing quality of the sensing area is rarely examined [13–15]. Poor-quality processing, such as
over-processing resulting in excessive removal or modification of material, can reduce the sensor
resolution and sensitivity, cause light scattering in the sensing area, insufficient sensing power, or
functional surface coverage during subsequent surface plasmon resonance (SPR) or particle plasmon
resonance (PPR) detection [16,17]. Therefore, the purpose of this study was to eliminate defects or
residue from the sensing area of window-type optical fiber sensors. We examined the CO2 laser
processing results for the sensing area and established a CO2 laser processing quality inspection
method. Finally, a window-type optical fiber sensor was developed according to the optimal
processing conditions identified in this study, and the sensor performance was subsequently verified.

Figure 1. Schematic of the fiber sensor: (a) crude fiber; and (b) fiber sensors (window type).

2. Experimental Section

The adjustable parameters of the CO2 laser machine (Model Mercury-II M-12, LaserPro Inc.,
New Taipei, Taiwan) employed in this study included the processing power (1 W to 10 W), processing
speed (0.63 cm/s to 63 cm/s), laser-sourced cooling nozzle pressure (0.1 MPa at less than 60 psi), and
focusing position (adjusted by altering the Z-axis in the machine’s three-axis displacement platform).
The non-adjustable parameters were the laser pulse width (± ȝV DQG SXOVH IUHTXHQF\
(5 KHz). This study primarily analyzed the sensing area quality of laser-processed optical fiber sensors.
To conduct various quality analyses more accurately, we set the fixed length of the sensing region to
FPWKHIRFXVLQJSRVLWLRQRQWKHILEHUFRUHWRPPWKHODVHUSXOVHZLGWKWR“ȝVWKHODVHU
pulse frequency to 5 KHz, and the air nozzle gauge pressure to 0.3 MPa for air processing. In this
study, we considered the laser processing power and speed, in addition to self-developed fixtures, to
explore the laser processing path. Figure 2(a) shows the processing fixture used to attach the processed
optical fibers; this fixture is capable of attaching five optical fibers simultaneously. Figure 2(b) shows
Sensors 2013, 13 4070

a rotating fixture with a central hole packed tightly with optical fibers ready for processing in lockstep
rotation. Using the preset structure highlighted at every 60°, the operator can process optical fibers
every 60° a total of six times.

Figure 2. Schematic of the laser machining fixtures: (a) the processing fixture; and (b) the
rotating and fixed optical fiber fixture.

The yellow portion of the structure shown in Figure 3 is the optical fiber to be processed, the blue
dotted line denotes the established processing direction of the optical fiber, and the red arrows and
circular patterns represent the moving path and processing area of the laser source. When the laser
source and the blue dotted reference line move horizontally during processing, the structure adopts the
parallel machining condition; otherwise, the vertical machining condition is employed.

Figure 3. Schematic of the laser processing path: (a) parallel machining; and
(b) vertical machining.

(a) (b)
Sensors 2013, 13 4071

The research goals of this study were to propose a comprehensive method for assessing the quality
of window-type optical fibers processed by lasers, the convenience of light coupling in subsequent
sensing measurements, and the mechanical strength of manufactured window optical fibers. Therefore,
we employed a multi-mode glass optical fiber with a 400-ȝPILEHUFRUH, manufactured by Newport®
under the model number F-MBC, as the optical fiber. Regarding the size and structure, the optical fiber
comprised a 400-ȝPILEHUFRUH-ȝPcladding, and 730-ȝPFRDWLQJDVVKRZQLQ)LJXUH 4. The fiber
core was made of silica material, the cladding was made of hard and brittle polymer, and the coating
was made of Tefzel material. These materials were used to provide the optical fiber with superior
mechanical protection.

Figure 4. Schematic of the optical fiber.

By adjusting various methods for deploying the laser power, processing speed, and processing path,
we defined the following processing defects and circumstances:
1. The fiber core and coating material is either completely or partially removed, as shown in
Figure 5.
(a) The fiber core and coating material are partially removed.
(b) The fiber core and coating material are insufficiently removed.
(c) The fiber core and coating material are completely (maybe even excessively) removed.
2. The surface of the fiber core is altered, as shown in Figure 6.
(a) The surface changes are not uniform.
(b) The surface changes are uniform.
3. The fiber core shape changes, as shown in Figure 7.
(a) Wavy pattern
(b) Excessive removal
Sensors 2013, 13 4072

Figure 5. Schematic of an optical fiber cable: (a) partially removed; (b) insufficiently
removed; and (c) completely removed.

Figure 6. Schematic of optical fiber defects: (a) non-uniform changes in property; and
(b) uniform changes in property.

Figure 7. Schematic of an optical fiber: (a) wavy pattern; and (b) excessive removal.

Therefore, to avoid the defects caused by laser processing, which can affect the sensing capability of
optical fiber sensors, negatively influence the sensor performance, or modify other functionalized
surfaces, we established a quality analysis method for identifying these defects. This quality analysis
method includes the following four items:
(1) Size measurement
This element can identify lacking, incomplete, or excessive removals and distinguish between the
two defect types, namely, wavy pattern and excessive removal, during fiber core modifications. The
measuring instrument is shown in Figure 8. The optical fiber sensor location point measurements
are shown in Figure 9. First, data from Points 1 and 6 were removed because the cumulative
thermal effect at these two points was relatively lower than the processing level in other areas. To
prevent the values measured at these two points from affecting the values measured at normal
processing areas, we use them to verify the total processing length at the sensing area. Then, the
additional diameter of the processing core was used as the average value of the four remaining data
groups (the remainder of the core, DcR).
Sensors 2013, 13 4073

Figure 8. Schematic of the measuring instrument: (a) fiber connecter; (b) measuring
fixture; (c) processed fiber; and (d) micro-stage.

Figure 9. Schematic of the measurement position.

(2) Light transmission defect detection method

The light transmission defect detection method was primarily based on internal optical fiber core
propagation by optically coupling the laser to the process. For this study, we used the structure shown
in Figure 10 to guide the laser light inside the optical fiber sensor as it propagates. First, an optical
collimator was used to focus the laser source (wavelength: 532 nm; power: 10 mW) inside the optical
fiber transmission line (NA = 0.27) during light propagation. At this time, light from the source was
transmitted to the end with an NA value of 0.27 and then guided through the optical fiber adapter as it
propagated within the optical fiber core (NA = 0.37). The Optical fiber core was then examined to
confirm that the confined light source was guided to the core of the optical fiber when propagating. A
diagram of the measurement is shown in Figure 11. When the surface is irregular or contains debris
from processing, according to the optical transmission principle, irregular surfaces in the path of light
transmissions cause the light to diffuse. Observation of the light diffusion phenomenon facilitates the
achievement of quality analysis objectives.
Sensors 2013, 13 4074

Figure 10. Diagram of the components of an optical light transmission defect: (a) laser
source; (b) collimator; (c) fiber cable (NA = 0.27); (d) fiber adapter; (e) fiber connecter;
and (f) fiber sensor (NA = 0.37).

Figure 11. Schematic of the tool used for qualitative analysis of the optical transmission
defects: (a) laser source; (b) collimator; (c) fiber cable; (d) fiber adapter; (e) fiber
connecter; (f) processed fiber; and (g) micro-stage.

(3) Numerical aperture (NA)

The two quality analysis methods previous mentioned were used to perform preliminary quality
analysis of the laser-processed optical fiber sensors. However, to ensure that no defects remain
undetected, additional in-depth investigations of the internal sensing area may be necessary. Therefore,
we established a numerical aperture measurement platform to measure the post-processing numerical
aperture of optical fiber sensors. The architecture in Figure 12 shows that parallel optical modulation
was first conducted. Subsequently, the focusing lens and optical fiber coupler were used to focus and
couple the beam onto the core inside the processed optical fiber sensor for light propagation. To
measure the NA value of the optical fiber sensor area, we partially cut the processed optical fiber
sensor, as shown in Figure 12(a) and 12(b), and employed the grinding method to smooth the sectioned
surface for further measurement. After the light coupled to the core layer was propagated, it scattered
Sensors 2013, 13 4075

from the back-end of the truncated plane to the rear, forming a circular light spot with a diameter of D.
This was then attached to the optical sensor using the moving platform to measure the size of the
circular light spot with a diameter of D at various lengths L. According to Equation (1) [18], the L and
D values measured can be employed to determine the numerical aperture value. If the sensing area of
the processed optical fiber sensor shows homogeneous surface modification, the NA value of the
sensing area decreases. This is because the width of the possible transmission light path is reduced by
surface modifications. In this study, we used this method as a basis for identifying defects.
D
NA sin T (1)
2 2
D  4L

Figure 12. Schematic of the numerical aperture measurement instrument.

a)
b)

(4) Weibull tensile test [19,20]

The Weibull tensile test is suitable for analyzing the tensile properties of hard and brittle
materials. The partial exposure of the optical fiber’s core silica material during the laser-stripping
procedure either alters the material properties or results in internally generated processing defects. If
window-type optical fiber sensors contain defects not detected by the previously described quality
analysis method, a Weibull tensile test can be employed to exploit the changing trends in tensile
failure characteristics. Figure 13 shows a tensile test diagram. Data obtained from the tensile test are
analyzed using the Weibull distribution. The mathematical formula for the Weibull distribution is
shown as Equation (2), where P is the cumulative probability of failure, ı is the applied tension
force (kgf), Ș is the scale parameter, ı0 is the minimum breaking tension force (kgf), and Ȧ is the
Weibull modulus:
V V0 Z
P 1  exp[ ( ) ] (2)
K
Sensors 2013, 13 4076

Figure 13. Schematic of the tensile test (Model FS-1002, Lutron Inc., New Taipei,
Taiwan): (a) force gauge; (b) measuring fixture; (c) translation stage; (d) gauge value
display; and (e) processed fiber.

(5) Sensing experiment

This study investigated whether the processing parameters selected based on the quality analysis
results offer sensing capabilities. The setup of the sensing experiment employed for this study is shown
in Figure 14. First, a function generator (Model 33220A, Agilent Inc., Santa Clara, CA, USA) was
used to generate a direct current pulse with square waves of a 1-KHz frequency and 1-V voltage to
drive the green light-emitting diode (LED) light source (Model EHP-AX08LS-HA/SUG01-P01,
Everlight Inc., New Taipei, Taiwan). Next, the total reflection characteristic of optical fiber was used
to direct the light into the sensing area for ATR sensing. Subsequently, the optical signal was further
directed through the optical fiber to the back-end signal capture device for photoelectric signal
conversion and processing. The microfluidic chip was made of poly(methyl methacrylate): PMMA
(chip size: 50 u 20 u 8 mm; micro-channel size: 50 u 1 u 0.9 mm) and fabricated using a computer
numerical control engraving machine by our group [21]. The sensing fiber was packaged inside the
microfluidic chip. The input and output ports of the microfluidic chip were used to infuse the solution
flowing through the sensing fiber. The sensing environment involved the injection of various sucrose
solution concentrations into a sensing microfluidic chip using a syringe for further sensing testing. For
the experiment, we first prepared deionized (DI) water. The relationship between the sucrose
concentration and refractive index of 1.333 to 1.383 RIU is shown in Table 1 [7]. After the sensing
experiment, to compare our results with those reported in previous literature [13], we calculated the
resolution of the sensor. Then, we modified the gold nanoparticles in the sensing area to conduct
PPR-sensed environmental refractive index measurements [13]. The gold nanoparticles modified in the
optical fiber sensing area were prepared by the study researchers [7].
Sensors 2013, 13 4077

Figure 14. Schematic of the experimental setup for creating sensing measurements:
(a) function generator (Agilent Inc. Model: 33220A); (b) LED light source (Model:
EHP-AX08LS-HA/SUG01-P01, Everlight Inc.); (c) sensing chip; (d) photo diode (Model
PD-ET2040, EOT Inc., Traverse, MI, USA); (e) lock-in amplifier (Model 7225, Signal
Recovery Inc., Oak Ridge, TN, USA); and (f) computer.

Table 1. Refractive index of various sucrose solution concentrations [7].

RIU Wt %
DI water 1.333 0
No.1 1.343 6.8
No.2 1.353 13.25
No.3 1.363 19.45
No.4 1.373 25.4
No.5 1.383 31.05

3. Results and Discussion

3.1. Size Analysis

This study used the parameter scanning method to identify superior parameters. First, the result of a
large-range parameter measurement was employed to obtain a near residual diameter (DcRRIȝP
at a processing speed of 9.45 to 25.2 cm/s and to arrange a parameter scanning experiment in this
processing speed range. Figure 15 is a diagram showing the relationship between the power of various
parallel spindle paths and DcR; the results indicate that as the processing power increases, the residual
diameter gradually decreases under various processing speeds. From the same perspective of
processing energy, as the processing speed increases, the total energy loss per unit area per unit of time
causes DcR to gradually increase. Figure 16 is a diagram of the vertical spindle processing relationship,
which exhibits the same conditions. The line graph results in Figures 15 and 16 show that the sloping
trend of parallel spindle processing exceeds that of vertical spindle processing. This suggests that the
Sensors 2013, 13 4078

material removal processing change rate under horizontal conditions is comparatively greater. In
addition, the standard deviations of the data points from the two processing paths show that parallel
spindle path processing is greater than vertical spindle path processing. The cause of this phenomenon
may be the greater horizontal processing path displacement, because cumulative structure errors can
lead to inconsistent processing quality.

Figure 15. Schematic of the size analysis results (Scanning path: parallel).

9.45 cm/s
550 12.6 cm/s
Near residual diameter-DRC (Pm)

15.75 cm/s
18.9 cm/s
22.05 cm/s
500 25.2 cm/s

450

400

350

4 5 6 7 8 9 10
Power (W)

Figure 16. Schematic of the size analysis results (Scanning path: vertical).

9.45 cm/s
12.6 cm/s
550 15.75 cm/s
Near residual diameter-DRC (Pm)

18.9 cm/s
22.05 cm/s
25.2 cm/s
500

450

400

350
4 5 6 7 8 9 10
Power (W)

To ensure a residual diameter of nearly 400 µm using laser processing while avoiding the
non-processed removal of material from the fiber core layer, we established an acceptable range for the
residual diameter:
395 µm < (average value +|standard deviation|) < 400 µm
Sensors 2013, 13 4079

This range was used to select the following parameters for the second quality analysis process, as
shown in Table 2.

Table 2. Size analysis access parameters.

Scanning Path Power (W) Velocity (cm/s) Avg. DCR ȝP Std. DCR ȝP
4 9.45 392.9 3.48
4 12.6 395.8 0.88
Parallel 5 12.6 393.1 2.18
5 15.75 393.6 2.51
6 15.75 395.3 3.75
4 9.45 393.5 1.77
5 12.6 397.1 2.77
Vertical
8 18.9 393.8 3.36
10 22.05 395.6 1.12

3.2. Qualitative Analysis of Optical Transmission Defects

The nine sets of data obtained through size measurements satisfy the criteria for superior parameters
(within the expected range). The optical defect transmission method was used to qualitatively analyze
the quality of the processing area. The results in Table 3 show that under parallel processing machining
conditions, with a laser processing power of 4 W and a processing speed of 9.45 cm/s, excessively low
processing speeds result in exorbitant removal of material, generating wavy patterns on the surface of
the processing area, as shown in Figure 17. Additionally, regarding the vertical processing machining
condition, for optical fiber sensors processed with a power of 5 to 10 W and a speed of 12.6 to
22.05 cm/s, some of the material could not be removed correctly, resulting in the optical leakage
phenomenon, as shown in Figure 18. Figure 19 is a diagram of the schematic without defects.

Table 3. Results of the qualitative analysis of optical transmission defects.

Machining parameter Result
Scanning Path Power (W) Velocity (cm/s) Result Reason
4 9.45 × Wavy pattern
4 12.6 O Without light leakage
Parallel 5 12.6 O Without light leakage
5 15.75 O Without light leakage
6 15.75 O Without light leakage
4 9.45 O Without light leakage
5 12.6 × Removal incomplete
Vertical
8 18.9 × Removal incomplete
10 22.05 × Removal incomplete
Sensors 2013, 13 4080

Figure 17. Schematic of the wavy pattern defect.

Figure 18. Schematic of the incomplete removal defect.

Figure 19. Schematic without any obvious optical transmission defect.

3.3. Numerical Aperture Measurement

Preliminary analysis of the quality of laser-processed optical fiber sensors was conducted using two
quality analysis methods. We then established an NA measurement platform to measure the NA of
the processed optical fiber sensors. During the experiment, we measured the NA of preprocessed
optical fibers, obtaining an average value of 0.385 ± 0.01. Compared to the factory specification of
0.37 ± 0.02 with Į = 0.01, the average value obtained using the two quality analysis methods did not
differ significantly. Therefore, the measurement platform developed in this study is feasible.
After conducting the above experiment, we used the quality analysis results to perform separate NA
measurement experiments. The results are shown in Table 4 and are relatively small for several NA
groups. Because the refractive index between the fiber coating and core layers was modified,
according to Equation (3), where ncladding decreases and ncore remains the same, the NA value of the
material increases [18]. The F-test in an analysis of variance (ANOVA) was used to analyze each
group at a VLJQLILFDQFH OHYHO RI Į   8VLQJ WKH ULJKW-tailed test method with an F significance
value as the determination basis, we found that a significant relationship existed between Groups 1, 2,
and 3. The value of Group 4 was relatively small, whereas that for Group 5 was large in comparison to
the other groups. This result was included in the next quality analysis process to identify defects:

NA { sin T c ncore  ncladding

2 2
(3)
Sensors 2013, 13 4081

Table 4. Results of numerical aperture measurements.

Scanning Power Velocity Number of NA NA
Path (W) (cm/s) samples Avg. Std.
Crude fiber 0.385 0.01
No. 1 4 12.6 0.482 0.023
No. 2 5 12.6 0.497 0.021
Parallel 10
No. 3 5 15.75 0.491 0.023
No. 4 6 15.75 0.449 0.032
No. 5 Vertical 4 9.45 0.520 0.019

3.4. Weibull Tensile Test

By converting the diagram of stress and cumulative damage probability created using the Weibull
tensile test, we can plot a Weibull graph. Figure 20 shows that the information in Table 5 can be
obtained after linearly fitting the five parameter groups. The larger the Weibull PRGXOXVHTXDWLRQȦLQ
the Weibull GLVWULEXWLRQWKHJUHDWHUWKHUHOLDELOLW\$QRWKHULPSRUWDQWSDUDPHWHULVȘwhich belongs to
the scale parameter. The larger this value, the greater the damage to the material. Additionally, the
wider the fracture stress distribution, the less homogeneous the material. By contrast, the smaller the
value, the more homogeneous the material. Therefore, the resulting table shows that Group 4 has a
superior Weibull modulus. However, regarding the results of NA measurements, this group had the
smallest value. Thus, to prevent defects, Group 4 was ignored. Statistical analysis of the NA
measurement results indicated that Groups 1, 2, and 3 had no significant differences. However, the
Weibull tensile analysis results showed that various parameters changed excessively. To avoid the
occurrence of non-detected defects, these groups were also eliminated. Therefore, the parameters of
Group 5 were selected as the optimum processing parameters for this study.

Figure 20. Weibull distribution.

Sensors 2013, 13 4082

Table 5. Weibull tensile test results.

Machining parameter Number ımin ımax
Scanning Power Velocity of Ȧ Ș R2
samples (kgf) (kgf)
Path (W) (cm/s)
No. 1 4 12.6 0.591 0.878 0.14 3.95 0.913
No. 2 5 12.6 0.922 1.476 0.25 4.05 0.833
Parallel
No. 3 5 15.75 30 0.753 0.942 0.1 4.1 0.969
No. 4 6 15.75 1.008 1.368 0.1 5.9 0.963
No. 5 Vertical 4 9.45 0.886 1.186 0.25 4.15 0.974

3.5. Refractive Index Sensing Measurements

The quality inspection method previously described was used to select processing parameters for
the vertical spindle processing path. A superior set of parameters, including a processing power of 4 W
and a processing speed of 9.45 cm/s, was used to create optical fiber sensors for the refractive index
sensing experiments. In Figure 21, the X-axis represents the time and the Y-axis represents the signal.
Figure 21 shows that the experimental intensity decreased as the injection concentration increased
(refractive index increment). When the signal measurement was complete, the sensor resolution was
calculated. In Figure 22, the X-axis represents the refractive index value and the Y-axis represents the
signal level of each concentration after the average signal was normalized. The relationship diagram in
Table 6 shows the sensor slope (m) used to calculate the sensor resolution. After repeating the
experiment three times, we obtained an average sensor resolution value of 1.8 × 10í4 RIU.

Figure 21. Plot of the ATR fiber sensors’ temporal responses to injections of increasing
sucrose solution concentrations and refractive indices.

0.204 DI water
DI water
0.202 1.343
0.200 1.353
Intensity (V)

0.198 1.363
0.196
1.373
0.194
0.192
1.383
0.190
-200 0 200 400 600 800 1000 1200 1400 1600 1800
Time (s)
Sensors 2013, 13 4083

Figure 22. Plot of the ATR fiber sensor response versus the refractive index of the
sucrose solution.

Table 6. The sensing experiment results for ATR fiber sensor.

Coefficient of variation RI resolution 3 ıP
No. Sensing slope (m)
(ı = Standard of deviation/m) (RIU)
1 1.106 7.65 × 10í5 2.08 × 10í4
2 1.217 4.23 × 10í5 2.23 × 10í4
3 1.058 1.07 × 10í5 1.09 × 10í4
Average sensor resolution values 1.8 × 10í4

This study also used the superior processing parameters obtained through quality analysis to create
window-type optical fiber sensors by modifying gold nanoparticles [13]. The PPR sensing method was
used to conduct three iterations of the refractive index sensing experiment. A diagram of the
relationship between the experiment time and signal is shown in Figure 23. The relationship between
the average intensity of the refractive index of a measured signal and the refractive index is shown in
Figure 24. The results of the sensing resolution after linear regression are shown in Table 7. The
average sensor resolution was approximately 5.73 × 10í5 RIU. Compared to unmodified ATR sensors,
the resolution of the sensor was an order of magnitude smaller as that reported in previous literature,
although the sensor length was similar [13]. The results of the linearity comparison modified sensing
method are shown in Figure 25. The quality analyzed optical fiber sensors have a superior degree of
linearity compared to the non-quality analyzed optical fiber sensors reported in previous studies. This
may be because the removal of optical fiber material could not be confirmed in related literature. Thus,
quality-analyzed window-type optical fiber sensors possess basic ATR sensing abilities, with the
sensor resolution reaching a 10í4 RIU level. Once the modified gold nanoparticles sensed using the
PPR method were excited, they created a regional plasma resonance response that effectively increased
the sensor resolution and linearity.
Sensors 2013, 13 4084

Figure 23. Plot of the PPR fiber sensors’ temporal response to injections of increasing
sucrose solution concentrations and refractive indices.
0.245
DI water DI water
0.240
0.235 1.343
0.230
1.353
Intensity (V)

0.225
0.220 1.363
0.215
1.373
0.210
0.205 1.383

-500 0 500 1000 1500 2000 2500 3000 3500 4000

Time (s)

Figure 24. Plot of the PPR fiber sensor response versus the refractive index of the
sucrose solution.

1st
1.00 2nd
3rd
0.98
0.96
0.94
I/I0

0.92
Equation y = a + b*x
0.90 Adj. R-Square 0.99956 0.99977 0.99986
Value Standard Error

0.88 B
B
Intercept
Slope
4.04769
-2.28539
0.03265
0.02395
D Intercept 5.02932 0.03131

0.86 D
F
Slope
Intercept
-3.0213
3.72781
0.02297
0.01626
F Slope -2.04206 0.01193
0.84
1.34 1.35 1.36 1.37 1.38 1.39
RIU

Table 7. Results of the sensing experiment for PPR fiber sensor.

No. Sensing slope (m) ı value RI resolution ıP (RIU)
1 2.285 1.07 × 10í5 1.41 × 10í5
2 3.021 6.15 × 10í5 6.1 × 10í5
3 2.042 6.6 × 10í5 9.69 × 10í5
Average sensor resolution values 5.73 × 10í5
Sensors 2013, 13 4085

Figure 25. Plot of the sensor response versus the refractive index of the sucrose solution.

1.00 Chau et al.[4] (R2=0.9653)

present study (R2=0.9997)
0.95

0.90
I/I0

0.85

0.80

0.75
1.34 1.35 1.36 1.37 1.38 1.39
RIU

4. Conclusions

The original optical fiber structure of window-type optical fiber sensors has changed according to
current sensing needs. The current structure exposes the fiber core to facilitate contact with the sensing
environment, and is adopted to measure the environmental refractive index according to the principle
of gradually reducing the total reflective sensing. The partial removal method selected for the optical
fibers used in this study was the laser thermal removal method. Although adjusting the parameters
using this method is extremely easy, because of the thermal effects of removal, if the parameters are
not appropriately controlled, defects are likely to result, such as insufficient removal, excessive
removal, or property changes. The purpose of removing material is to ensure the core layer of the
optical fiber contacts the environment directly. However, if defects exist and the refractive index does
not match as expected, the sensitivity of optical fiber sensors may be reduced. The sensing principle
employed in this study was the ATR sensing method because the penetration depth of the evanescent
wave is extremely limited. In addition, the evanescent wave decays exponentially into the surrounding
media because of its distance from the fiber core. However, if other defects exist, the sensing ability is
inevitably reduced. In this study, we proposed four methods for analyzing the laser processing quality,
that is, size measurements, light transmission defect detection, NA measurement, and a Weibull tensile
test. They were employed to verify the quality of the sensing area of optical fiber sensors after laser
processing. These methods can also facilitate coating the functional surface, thereby increasing the
surface coating rate and sensor sensitivity.
After quality analysis, the remaining processed sensor region measured ȝP and did not possess
light transmission defects. The NA value of the sensing area was 0.52 ± 0.019, with superior
processing parameters; that is, a Weibull WHQVLOH WHVWHG PRGXOH QXPEHU Ȧ   VFDOH SDUDPHWHU
Ș PLQLPXPGHVWUXFWLYHSXOOLQJIRUFHRINJIDQGPD[LPXPGHVWUXFWLYHSXOOLQJIRUFHRI
4.15 kgf. The processing parameters also include a vertical processing path with a processing power of
4 W and a processing speed of 9.45 cm/s. The parameter-processed optical fiber sensors have ATR
Sensors 2013, 13 4086

refractive index sensing capabilities. The average sensor resolution measured was approximately
1.8 × 10í4 RIU (R2 = 0.954), with the PPR sensing resolution reaching 5.73 × 10í5 RIU (R2 = 0.999).
The above discussion indicates that in this study, we successfully established a quality analysis
method for laser-processed optical fiber sensors. The detection method provides superior quality
verification of fiber optic sensors before application by including a quantitative analysis of size, NA,
and material properties to eliminate any concerns regarding reduce ATR sensing capability or
insufficient functional surface coverage.

Acknowledgments

The authors thank Tai-Huei Wei and Lai-Kwan Chau for their laboratory assistance. The partial
support of the National Science Council (NSC) of Taiwan under Contract Nos. NSC-99-2120-M-194-
005-CC1; NSC101-2120-M-194-001-CC2; NSC-99-2112-M-194-004-MY3; NSC-101-2221-E-020-
010-MY3 is gratefully acknowledged.

References

1. Brooks, S.L.; Ashby, R.E.; Turner, A.P.F.; Calder, M.R.; Clarke, D.J. Development of an on-line
glucose sensor for fermentation monitoring. Biosensors 1987, 3, 45–56.
2. Lundstrom, I.; Spetz, A.; Winquist, F.; Albery, W.J.; Thomas, J.D.R. Semiconductor biosensors
[and Discussion]. Philos. Trans. Roy. Soc. Lond. B Biol. Sci. 1987, 316, 47–60.
3. Janata, J. Principles of Chemical Sensors; Plenum Press: New York, NY, USA, 1989.
4. Cheng, S.-F.; Chau, L.-K. Colloidal gold-modified optical fiber for chemical and biochemical
sensing. Anal. Chem. 2002, 75, 16–21.
5. Wang, J.-N.; Tang, J.-L. Feasibility of fiber bragg grating and long-period fiber grating sensors
under different environmental conditions. Sensors 2010, 10, 10105–10127.
6. Wei, T.; Han, Y.; Tsai, H.-L.; Xiao, H. Miniaturized fiber inline Fabry-Perot interferometer
fabricated with a femtosecond laser. Opt. Lett. 2008, 33, 536–538.
7. Chen, C.H.; Chao, T.C.; Li, W.Y.; Shen, W.C.; Cheng, C.W.; Tang, J.L.; Chau, L.K.; Wu, W.T.
Novel D-type fiber optic localized plasmon resonance sensor realized by femtosecond laser.
J. Laser Micro Nano Eng. 2010, 5, 1–5.
8. Chiu, M.H.; Chiu, P.C.; Liu, Y.H.; Zheng, W.D. Single-mode D-type optical fiber sensor in
spectra method at a specific incident angle of 89°. Sens. Transd. 2009, 5, 41–46.
9. Yoon, H.-J.; Kim, C.-G. The mechanical strength of fiber Bragg gratings under controlled UV
laser conditions. Smart Mater. Struct. 2007, 16, 1315–1319.
10. Chau, L.-K.; Lin, Y.-F.; Cheng, S.-F.; Lin, T.-J. Fiber-optic chemical and biochemical probes
based on localized surface plasmon resonance. Sens. Actuators B Chem. 2006, 113, 100–105.
11. Vengsarkar, A.M.; Lemaire, P.J.; Judkins, J.B.; Bhatia, V.; Erdogan, T.; Sipe, J. E. Long-period
fiber gratings as band-rejection filters. J. Lightwave Technol. 1996, 14, 58–65.
12. Malki, A.; Humbert, G. Investigation of the writing mechanism of electric-arc-induced
long-period fiber gratings. Opt. Lett. 2003, 42, 3776–3779.
Sensors 2013, 13 4087

13. Chiang, C.-Y.; Hsieh, M.-L.; Huang, K.-W.; Chau, L.-K.; Chang, C.-M.; Lyu, S.-R. Fiber-optic
particle plasmon resonance sensor for detection of interleukin-ȕ LQ V\QRYLDO IOXLGV.
Biosens. Bioelectr. 2010, 26, 1036–1042.
14. Zhao, L.; Jiang, L.; Wang, S.; Xiao, H.; Lu, Y.; Tsai, H.-L. A High-Quality Mach-Zehnder
Interferometer Fiber Sensor by Femtosecond Laser One-Step Processing. Sensors 2010, 11,
54–61.
15. 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.
16. Homola, J.; Yee, S.S.; Gauglitz, G. Surface plasmon resonance sensors: review. Sens. Actuators B
Chem. 1999, 54, 3–15.
17. Salamon, Z.; Macleod, H.A; Tollin, G. Surface plasmon resonance spectroscopy as a tool for
investigating the biochemical and biophysical properties of membrane protein systems, II:
Applications to biological systems. BBA Rev. Biomembranes 1997, 1331, 131–152.
18. Michael, B.; Eric, W.V.S. Fiber Devices and Systems for Optical Communications; McGraw-Hill:
Orlando, FL, USA, 2002.
19. Alexey, F.K.; Sergei, L.S.; Alexandr, N.D.; Evgeny, M.D. Mechanical strength and fatigue of
microstructured optical fibers. In Proceedings of Optical Fiber Communication Conference
(OFC), Anaheim, CA, USA, 25–29 March 2007.
20. Liu, X.Y.; Dai, G.C. Surface modification and micromechanical properties of jute fiber mat
reinforced polypropylene composites. eXPRESS Polym. Lett. 2007, 1, 299–307.
21. Wang, J.-N. A Microfluidic long-period fiber grating sensor platform for chloride ion
concentration measurement. Sensors 2011, 11, 8550–8568.

© 2013 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 license
(http://creativecommons.org/licenses/by/3.0/).