JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 13, NO.
7, JULY 1995
A Novel Technique for Optical Fiber pH
Sensing Based on Methylene Blue Adsorption
Bruno J.-C. Deboux, Elfed Lewis, Patricia J. Scully, and Robert Edwards
Abstrucf-A novel fiber optic pH sensor is reported. The
sensor is based upon surfacic adsorption of methylene blue dye,
producing absorption in the evanescent field surrounding the
sensing fiber. The sensor is based upon a 200 pm diameter Plastic
Clad Silica (PCS) fiber which has its cladding removed in order
to expose its core. The linear range of operation is between
pH 3 and pH 9 and its response time is considerably shorter
than indicator based probes which measure spectral changes of
pH sensitive chemicals. Effects of measurand ionic strength are
shown to be negligible which is a further important advantage
of this sensor over other optical fiber pH sensors. The sensor
design is sufficiently flexible to allow it to be configured in probe
(reflective) or in-line (transmissive) form. The latter allows the
application of time domain reflectometry technique which may
be used to achieve multipoint (distributed) sensing.
I. INTRODUCTION
A CCURATE measurement of pH is required in diverse
and challenging fields provided by industry, medicine
and the environment. On-line monitoring of pH is needed
for process control in extreme conditions such as those posed
by nuclear reactor environment [l] or waste water treatment
plant [ 2 ] .Clinicians require measurement of blood pH [3] and
other physiological fluids during surgery. Monitoring of pH
underground water involves harsh environments, and extreme
ionic concentrations.
Optical fiber sensors meet these challenges because they
have many advantages compared with conventional (potentio-
metric) pH sensors. For example, immunity to electromagnetic
interference, electric and intrinsic safety, possibility of minia-
turization, and bio-compatibility, make an optical fiber pH
sensor ideal for in vivo blood monitoring in electromagnet-
ically noisy intensive care environments. With the sensor
configuration reported, it is possible to extend the method
to quasi-distributed (multipoint) sensing. Such a method is
particularly suitable for applications involving long lengths of
pipework. The novel sensing mechanism overcomes problems
of accurate pH measurement under conditions of extreme ionic
concentration, for example, underground water monitoring
(e.g., slow response time).
Research to date has concentrated on changes in the optical
properties of colorimetric or fluorimetric pH indicators. Optical
fiber sensors based on such methods measure by reflection or
through interaction with the evanescent field changes in the
absorption due to changes in color of colorimetric indicators
Manuscript received December 2, 1994; revised February 20, 1995.
The authors ace with the Optical Fiber Sensors Research Group, School
of Electrical and Electronic Engineering, Liverpool John Moores University,
Liverpool L3 3AF, England, UK.
IEEE Log Number 9412214.
0733-8724/95$04.00 0 1995 IEEE
~ .-
1407
[4]. The same optical configurations have been used to monitor
the evolution of a fluorimetric (i.e., fluorescence intensity [ 5 ] ,
decay time [6]) pH indicators. These sensors have inherent
limitations. The range over which the output is linear is very
limited (typically 2 pH units) and signal processing corrections
suffer from the low sensitivity offered by the indicators outside
their linear range. More significantly, pH color indicators do
not measure the activity of OH- ions and are consequently
sensitive to the ionic strength of the monitored medium
(corrections taking into account the actual ionic strength of
the medium are possible but this parameter is very difficult
to evaluate accurately). The cost of these techniques could
also be very high due to the price of UV optics needed for
fluorimetric work as well as the complication of immobilizing
the indicator in a polymer [7].
The sensor described in this paper uses a novel sensing
technique. Interactions between the silica of the stripped
core of an optical fiber and hydroxyl ions of a solution,
induce modification of the surface charge of the silica. This
electrostatic attraction changes the concentration of oppositely
charged absorbing chromophores of methylene blue, within
the evanescent field region surrounding the fiber core. The
resultant changes in transmitted light intensity are a linear
function of pH over a measured range of 3-9-pH units. The
surfacic concentration of absorbing polar molecules appears
to be a function of the activity of OH- ions. The sensor
therefore measures the actual value of the pH and is immune to
changes in the ionic strength of the monitored medium. Other
advantages include a high sensitivity of 0.75 dB/pH and a fast
response time of less than one minute in a solution of normal
ionic strength, limited only by the semi-permeable membrane
used to prevent the methylene blue liquid dye modulator from
contaminating the measurand solution.
In Section 11, the optical and chemical principles of op-
eration of this new pH sensor are explained in more detail.
The theoretical hypothesis is tested (Section 111) by practical
experiments. In Section IV, the sensor is characterized, with
specific attention to the effect of ionic strength. Section V
discusses methods of sensor optimization tailored to specific
applications. Section VI contains a summary of the sensor
characteristics, and a discussion of future work.
11. THEORETICAL HYPOTHESIS
A. Optical Principle
If the core and cladding of an optical fiber are assumed to
have locally constant refractive indices, each propagating ray
of light can be treated as an individual plane wave. When
,
 1408 - JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 13, NO. 7, JULY 1995

0.08/

0.04
O -O6I

L 0.021

Fig. 1. Schematic of an optical fiber waveguide. 0 is the incident angle, 01

is the angle of reflection. n,l and nco are the refractive indices of the core 0.05 0.1 0.15 0.2 0.25
and cladding. p is the radius of the fiber core. L is the length of the exposed Incident Angle &(rad)
fiber core. n m b is the refractive index of the aqueous solution. Fig. 2. Waveguide geometry coefficient as a function of the angle of
incidence.
such a ray is reflected at the boundary between a denser
and a rarer optical medium, the fields associated with the
absorption coefficient (7) which was calculated [8], [9]. This
plane wave extend beyond the interface. These fields have
has been further developed to obtain an expression for A ( & )
an amplitude which decreases exponentially with increasing
in (4).
distance from the boundary and are consequently referred to
as evanescent fields. If the evanescent electric field interacts A = -Y= - Y (3 1
(Y C E
with an absorbing rarer medium, attenuation of the power
of the propagating ray will result. Under most experimental
x . nmb. COS e . cot e
A(ea) = (4)
conditions, this power absorption can be deduced from the 2rp, . n:, . cos2 dcmb . dsin2 e - sin2dcmb
Beer-Lambert law (1) o = c0s-l (sin & / n c o ) . (5)
pa = p, . e-CEAL. (1) Variables are defined in Fig. 1.
Where Po is the power of the ray before absorption, C, With Ocmb = sin-l(n,b/n,,), the critical angle of the
the concentration of the absorbing substance, E , the molar waveguide formed by the fiber core and the methylene blue
absorption coefficient and L, the optical path length. A is a solution.
coefficientderived from the geometry of the optical waveguide pe is the effective diameter of the fiber core. As the angle of
(waveguide geometry coefficient). incidence of the propagating ray approaches the critical angle
1 The absorbance is expressed as (2) Bc, the actual diameter of the fiber becomes [lo]:

AbsdB = - (x)
In 10
. EALC. Pe = P dp. + (6)
where dp is the penetration depth of the evanescent field within
I Equatjon (2) indicates that absorbance (Abs)is linearly propor- the cladding and is given by [ l l ]
tional to concentration (C), and this relationship is exploited
by the sensor described in this paper. As explained (Section x
dp = (7)
11-B),the concentration of an absorbing substance (e.g. methy- 2.r.nC0 Jsin2e-(n,ln,,2'
lene blue) at a silica surface is linearly related to the pH.
Therefore the absorbance of a silica sensing element can be Equation (4) gives the value of A as a function of the angle
shown to be a linear function of the pH. of incidence (ea) and is plotted in Fig. 2. This shows that
The optical fiber pH sensor described uses the bare core only modes close to the critical angle of the fiber (e,) are
of a chemically stripped Plastic Clad Silica (PCS) fiber as a strongly absorbed, which is significant when designing the
dense optical medium (n,, = 1.46) and an aqueous solution optical sensor. These can be described as the Poynting vector
of methylene blue as a rare absorbing medium (n,b = 1.33) of guided electromagnetic modes.
(Fig. 1). To determine the waveguide geometry dependencies of
In order to develop a theoretical model for the absorbance the power distribution, the power in each mode (Pi(&))
of this sensing element, the value of the waveguide geometry must be multiplied by the waveguide geometry coefficient
coefficient (A) must be determined. A is only valid for for this mode (A(&)). The total waveguide coefficient is
meridional rays which restricts the quantitative validity of the therefore given by the ratio between the sum of all geometric
model. In practice, skew rays and tunnelling rays contribute dependencies of the power in each mode (Pi(&)). A ( & ) and
highly to the absorbance of the sensor. A more accurate the sum of the incident .power in each mode (Pi(&))(8).
model has not been developed yet due to the difficulty to
determine exactly the mode distribution within multimode l e c p a ( e .iA) ( e i ) .dei
fibers. However, a qualitative model allows identification of A= (8)
optimization parameters (launch angles, mode filtering) as well ~ " ' P i ( d i.>
ddi
as relative effects of different waveguide geometries (core
diameter) or parasitic measurand variations (temperature). An where Oc = sin-'[nc0 . cos (sin-'(ncl/nc0))] is the critical
expression for A (3) can be derived from the evanescent angle of the optical fiber.
DEBOUX et al.: A NOVEL TECHNIQUE FOR OF'TICAL FIBER pH SENSING BASED ON METHYLENE BLUE ADSORPTION 1409

0 1 -~ ~ ~
" - ~- -
o 20 U) m m im ix) 140 Is0 ~en m 1 i* I 1 5 I 3, 6 65 I 75

Concentration(mmoUL) PH
Fig. 3. Theoretical plot of absorbance as a function of concentration of a Fig. 4 Companson between theoretically predicted and expenmentally ob-
methylene blue for a 30 mm sensing element. tained results Methylene blue bulk concentration Cb = 25 mgll

A collimated beam focused by a lens onto the fiber has the
power distribution given in (9), [12]
A has been computed for the following parameters:
p = 100,um, n,, = 1.457,
n,b = 1.33, E = 1.4 x lo5,
All modes are launched into the fiber which gives a value for
A of 0.125 x
Fig. 3 shows a plot of the estimated absorbance versus the
concentration.
It is possible to obtain a direct comparison between practical
results and theoretical evaluation by equating pH changes to
absorbing substance concentration changes within the evanes-
cent field region. We assume that neutral pH values (pH 7)
give a surfacic concentration equal to the bulk concentration
and low pH values (pH 3) give no absorption. It is therefore
possible to deduce the following equations:
Cb Cb
c, = -
4
x pH - - x .
4
C, is the surfacic concentration, c b is the bulk concentration
and pH is the actual pH value.
For a bulk concentration of 35 mfl, sensitivity values are
0.05 dB/pH for the theoretical model and 0.75 dB/pH for the
sensor. The difference was expected and due to the failure of
the meridional rays model. However, when taking into account
discrepancies between the theoretical and experimental absorp-
tion coefficient (y) reported by other researchers [ 111 one can
derive a correction coefficient Kc. from their results (12)
- ytheo
cf - -x 200
Texp
the response of the sensor. This correction coefficient, although
not yet justified theoretically, is still useful for the engineering
and optimization of the sensing element.
(9)
B. Chemical Principle
In all crystalline silicates, the central silicon atom is sur-
ncl = 1.43 rounded by four oxygen molecules in a tetrahedral arrange-
L = 30 mm. ment. Each tetrahedron can be linked to either one, two, three
or four other tetrahedra by Si-0 bonds at the comers, giving
a three-dimensional framework. The surface of the silica is
neutral, and has no net surface charge or potential. When silica
is immersed in aqueous solution, H+ and OH- ions and H2O
react with the surface to form an amphoteric hydroxylated
layer [ 131. The charge on the surface of the silica will therefore
depend directly on the composition of the solution. At low
pH, where there is a high concentration of H+ ions, a positive
surface charge will be induced on the silica. As the pH of the
solution is increased the concentration of H+ ions is reduced
until at high pH (alkaline) the solution contains an excess of
OH- ions, inducing a net negative surface charge on the silica.
These ions interact reversibly with the surface of the silica, and
the adsorption of protons or hydroxyl ions can be represented
by the (13)
-SOH$ -SiOHoz SiO- + H20. (13)
From the equation, it can be seen that at high pH, the surface
of the silica will have an induced negative surface charge,
which will subsequently decrease as the pH of the solution is
made more acid.
Methylene blue has a positively charged chromophore which
will be attracted to the negatively charged surface of the
silica at high pH. As the pH of the solution is reduced, the
electrostatic attraction between silica and methylene blue will
also be reduced. Consequently, the concentration of methylene
(12) blue at the surface of the silica will be lower.

When the waveguide coefficient ( A ) is multiplied by K,., the

graph in Fig. 4 is obtained.
Theoretical and experimental plots show good agreement.
Interpretation of this result highlights a constancy in error
factors due to the inaccuracy of the meridional ray model.
This model combined with the correction coefficient K,f can
nonetheless be used to evaluate within an order of magnitude
111. EXPERIMENTAL
VALIDATION
Experimental work was separated into three distinct areas.
First, the optical mechanism was simulated in a simplified
configuration, in order to test the theoretical hypothesis. Sec-
ondly, changes in the transmission spectrum of the optical fiber
sensor as a function of pH were measured when methylene
 1410 - JOURNAL OF LlGHTWAVE TECHNOLOGY, VOL. 13. NO. 7, JULY 1995

a7

am
--.
I
i HeNe1a.w I---- - -am

s.-

--
I
e aw t
I

i Hdk
,
Ham!
i l

Fig. 5. Measurement of transmission of microscope slide immersed in

methylene blue solution, as a function of pH of the solution. The concentration
of methylene blue is constant. ::!- -
3 4 5 I r 8
~

9
-_-
to
-
11
.
IP

0.0, -
PH
Fig. 7. Experimentally derived absorbance of alcian blue solution immersed
microscope slides as a function of pH.

Bare core

Fig. 6. Experimentally derived absorbance of methylene blue solution im-

mersed microscope slides as a function of pH.
Methylene Blue

blue was in surface contact with the stripped core of the Fig. 8. (a) Measurement of transmission spectrum of sensor. (b) Transverse
fiber. It was shown that spectral transmission of methylene and longitudinal cross sections through sensing element.
blue solution alone, is unaffected by pH. Finally, the spatial
intensity distribution of modes at the output fiber indicated
that the attenuation taking place was due to evanescent field
modulation.

A. Silica/Methylene Blue Interaction

If the theoretical explanation described in the previous sec-
tion is valid, it must be possible to experimentally determine
the methylene blue/silica interaction as follows.
A glass microscope slide was immersed in four separate
solutions of methylene blue. These solutions were prepared
with the same concentration (80 mmol/l) but with different
! pH values. The transmittance of the glass slide was measured
using the optical set-up described in Fig. 5. Fig. 6 shows the
plot of the transmittance as a function of pH. Fig. 6 confirms
the experimental observations, showing the absorption to be
linear with pH. These results must be treated as qualitative
since it is not possible to evaluate the thickness of the
absorbing layers of adsorbed methylene blue molecules.
The experiment was reproduced, but with alcian blue. As
was the case with methylene blue, this dye showed a pH depen-
dent adsorption to the surface of a microscope slide (Fig. 7).
Although it displays a pH dependence, measurements with the
alcian blue were unstable and were highly nonreproducible
(15% of full scale).
B. Spectral Characteristics of the Measurand
The absorption spectrum of the sensor at three different pH
values was measured using the experimental set-up described
in Fig. 8. These results are illustrated in Fig. 9, and show
that the changes in the transmission of the sensor with pH
are easily detected. Higher pH values cause greater negative
charging of the silica core of the fiber and thus attract a larger
Wavelength(nm)
Fig. 9. Absorption spectrum of the sensor for different pH values.
number of polar chromophores within the evanescent field
region, resulting in the increased absorption.
The absorption spectrum of a solution of methylene blue
contained in a 10-mm pathlength cuvette was shown to be
constant despite changes in pH. Typical spectrum is shown in
Fig. 10, which was obtained with the same optical configura-
tion as in Fig. 8.
The real part of the refractive index of a thin liquid film of
methylene blue solutions of different pH has also been shown
to be independent of pH (to the precision of 50.0005 refractive
index units achieved using an Abbe refractometer).
It was confirmed that the absorption changes were not due to
modifications of the optical properties of the methylene blue
but to an interaction between the dye and the silica of the
fiber core.
Fig. 11 shows that the peak wavelength of the absorption
spectrum peak, as measured by the sensor, is pH sensitive,
shifting by -2 nm/pH. Further investigation of this effect is
required, as it may form a wavelength dependent method of
pH monitoring.
DEBOUX et al.: A NOVEL TECHNIQUE FOR OPTICAL.FIBER pH SENSING BASED ON METHYLENE BLUE ADSORPTlON 1411

-.!
ao

~ - 5 w 5 S o m E n m m B o m ~ 12. The spatial intensity distribution of emitted light from sensing fiber
Fig.
Wavelength(nm) was recorded using CCD camera and image processing software..

Fig. 10. Absorption spectrum of a bulk solution of methylene blue measured

with an optical spectrum analyzer.

0.57

m 7m m Bo Ea Bo

Wavelength(nm)
Fig. 11. Peak absorption wavelength shift. The value of this shift is about
-2 nm/pH.

C. Mode Selective Absorption

If the attenuation of light intensity with pH is shown to be
mode dependent by observing the spatial intensity distribution
of the output of the sensing fiber, it can be concluded that
these variations are due to changes in the complex refractive
index at the surface of the fiber which affect the evanescent
field. If the changes are not due to evanescent field absorption,
but to a modification of the optical properties of the silica, the
beam profile at the output of the fiber will have a uniform
spatial intensity distribution.
Fig. 12 shows how the intensity distribution of the emittance
cone forming the numerical aperture of the output fiber was
captured with a CCD camera, and image processed. The pH of
the methylene blue was varied and the corresponding sensor
output beam profiles were recorded and normalized. A logic
AND was performed between corresponding pixels of the
captured beam intensity profiles as shown in Fig. 13. At low
pH [Fig. 13(b)] minimal attenuation occurred and the beam
profile was similar to that obtained using only distilled water in
contact with the stripped core fiber. However, Fig. 13(a) shows
a ring of highly attenuated modes under high pH conditions
corresponding to strong molecular absorption at the methylene
(b)
bluekladding interface. It can be seen that only the modes
Fig. 13. Logic AND between output beam intensity profile at different pH
close to the cut-off frequency of the fiber are highly attenuated. values. (a) Distilled water and methylene blue solution at pH 10. (b) Distilled
water and methylene blue solution at pH 3.
IV. SENSOR
CHARACTERISTICS
All experiments reported were performed using the sensor objective to obtain the required angular magnification. The
configuration described in Fig. 14. The source is a Helium- sensing fiber is a 200 pm PCS (Plastic Clad Silica) fiber with
Neon laser and launched into a high NA (0.65) microscope the core stripped using methyl chloride over a length of 30
 1412 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 13, NO 7. JULY 1995

ference in absorbance between two consecutive measurements

r p , , Meter at the same pH, and was found to be better than 10%. The
Sensor hysteresis over a pH range from 3-10 pH units, was less than
5%. Precision was less than 0.1 pH units. Use of a stabilized
HeNe Power
Laser Meter HeNe laser or laser diode light source could improve these
parameters by an order of magnitude.
Plastic Cup
Another important parameter for a sensor is the response
Fig. 14. Output power of pH sensor is measured as a function of pH of the time. At normal ionic strength (distilled water) the chemical
methylene blue solution. pH is measured with a glass electrode.
reaction is instantaneous, therefore the response time is only
limited by the characteristics of the semi-permeable membrane
and the physical dimension of the capsule (Section Vj. With
the configuration reported here, the response time is between
10 s (low pH) and 1 min (high pH). Low ionic strength
solutions could increase the response time by slowing down
the chemical reaction and thus need to be investigated.

Fig. 15. Longitudinal cross section of the sensing element. The methylene A . Effect of Ionic Strength
blue modulator is enclosed inside a capsule of semi-permeable material.
The sensor configuration of Fig. 14 was used with a tem-
5 perature stabilized laser diode and without the semi-permeable
45 membrane. Four solutions of identical concentration and pH
(pH 6.9) were prepared. Sodium chloride was added to three
-5 ,
4

35
solutions in concentrations of 5 , 10, 15 g/l. The absorbance of
each solution was measured and differences of no more than
9 25

c 1% were found, showing that the sensor is not sensitive to

4
n ' high ionic concentration.
a ' 5

1 Further experiments involving lower ionic strength, standard

05 ionic strength solutions, and testing over the whole pH range
0
3 5 5 7 B 9
need to be performed, but these early results clearly show that
P" this new sensing technique is significantly less sensitive to
ionic strength than other optical pH sensing techniques [ 141.
Fig 16. Plot of the absorbance of the \ensor a tunction of pH tor
methylene blue concentration of 35 mg/l

V. OPTIMIZATION
mm. The detector is a stabilized silicon detector and power
meter (AND0 AQ-2 105).
The capsule enclosing the sensing fiber and methylene A. Optical Characteristics
blue measurand is detailed in Fig. 15. The body is a win- The sensor described is easy to tailor to specific applications.
dowed stainless steel tube (0 4 mm) sealed by a glued For example, for a point sensor, signal attenuation is not
semi-permeable membrane (Alpha 400 dialiser membrane). really a problem and high sensitivity is a requirement. Also,
The semi-permeable membrane allows the entrainment of reflectometry techniques require a low parasitic (insertion)
hydroxyl ions while confining the dye molecules. loss thus allowing a large number of sensing points to be
A typical response of the sensor is shown in Fig. 16. The interrogated. In order to optimize the sensor with regards to
absorbance in dB is a linear function of the pH over the these specificities several parameters may be tuned.
measured range (linear to better than 0.2%). The pH range Dye concentration is one of these parameters. It determines
1 1
extends from pH 3-9 but should in theory extend to pH 10 the absorbance of the sensing element and therefore influ-
(Section 11) retaining a linear dependency on the absorbance. ences the insertion loss. A plot of the absorbance versus
The sensitivity is also very good at 0.25 dB/pH. the concentration of the solution is given in Fig. 17. The
The insertion loss is defined as the ratio of the transmitted absorbance can be fitted to the square root of the concentration
power of the sensing fiber with its cladding intact and stripped as described elsewhere [ 1 I]. This nonlinear dependency of the
in aqueous medium. For the configuration of Fig. 14, insertion absorbance is explained by the electrostatic shielding caused
loss was shown to be around 1.3 dB. Compared to the output by the adsorbed methylene blue molecules which reduce the
variation of intensity of the sensor with pH of 3 dB, the absorption expected from high concentrations.
I
insertion loss is sufficiently small to enable this sensor to be Dye concentration also determines the absorption range of
i
I
used in a distributed configuration, interrogated using optical the sensor (number of dB/full pH range). Fig. I8 shows that
time domain reflectometry techniques. range is also a function of the square root of the concentration.
The sensor characteristics are as follows, but their accuracy This could also be attributed to the molecular shielding which
is limited by the stability of the Helium-Neon laser light reduces the proportion of dye molecules able to interact with
source. The repeatability of the sensor is defined as the dif- the fiber at higher pH values.
DEBOUX et al.: A NOVEL TECHNIQUE FOR OPTICAL FIBER pH SENSING BASED ON METHYLENE BLUE ADSORPI?ON 1413

Launch Optics _PCS

_ _ kibre
Sensor Detector
,I= -~
~

-c
7\ _____
l ' ,
e-
, \
Bare Core
-1 Methylene Blue

1 b .

o t
0
~ - -
10 W m U1 eu m m Fig. 20. Sensor sensitivity was optimized by mode filtering at launch optics
Coaecatntioa(mgL) using band-pass spatial filter mask. Only modes near the critical angle of the
sensing element are allowed through the mask.
Fig. 17. Absorbance of the sensor as a function of the bulk concentration of
the methylene blue modulator. Experimental results are fitted to a function of
the square root of the concentration.
as illustrated in Fig. 2. As a consequence, it is possible to
improve the sensitivity of the sensor by suppressing the modes
not contributing to the sensing process. An experiment using
the optical set-up is illustrated in Fig. 20. The launch optics
included a 0.4 NA microscope objective to collimate the beam
from the laser diode (wavelength: 635 nm) and a 0.65 NA
objective to provide the appropriate angular magnification.
A circular mask was inserted in the collimated beam to
produce the required incident angles. An increase in sensitivity
0 c
- ,
0 10 m 30 YI 50 so rn was measured, but the low power contained in the allowed
Concentntioa(mgL) modes combined with the difficulty of maintaining the mode
Fig. 18. Plot of the ratio of the absorbance for a full scale variation of pH distribution over any length of fiber make this technique
over the pH range as a function of the bulk methylene blue concentration. incompatible with distributed sensing. For the point sensor,
Experimental results are fitted to a function of the square root of the the increase in sensitivity does not, in most cases, justify the
concentration.
increased complexity of the optical system, as the value of
0.75 dB/pH is sufficiently high.
I
Z I ./ B. Response Erne
At normal ionic strength (distilled water), the semiperme-
able membrane is the only factor limiting the response time
of the sensor. It is possible to reduce this time by choosing
a thinner membrane (faster diffusion time) and by optimizing
the ratio between measurand volume and membrane surface.
Ultimately, tubes of semi-permeable material may be used.
,--o- ~, -__ ~

Another factor influencing the choice of membrane is

0 5 i 0 1 5 m n m ~ ~ ~

Sensing Length(mm) its susceptibility to contamination. Contaminants such as

Fig. 19. Absorbance of the sensor as a function of the length of the stripped
fiber core.
Using these two graphs and by maximizing the derivative
of one of the two functions it is relatively easy to find the
optimum dye concentration for each application. For a point
sensor as used in the laboratory experiments, the optimal value
was 30 m a .
The rel&ionship between absorbance and sensing length is
shown in Fig. 19. Increase in the dye concentration results
in an increase in the number of molecules available for
interaction, but an increase in sensing length leads to a linear
increase in evanescent field volume. Thus, the good correlation
with the Beer-Lambert law, in that absorbance is proportional
to pathlength, is explained.
From results shown in Section 111-C, the sensing is highly
mode dependent and only modes near the cut-off wavelength
of the waveguide optical fibedaqueous medium are absorbed
limescale or biofilms (polymer enclosed bacterial colonies
[15]) can obstruct some of the pores thus increasing the
diffusion time. A more thorough investigation into suitability
of different membranes to different applications would involve
the long term testing of such membranes in the presence of
various contaminants.
VI. CONCLUSION
Preliminary results of tests on a novel optical pH sensing
technique are reported. The technique not only has many
advantages over conventional potentiometric sensors (small
size, electrical safety) but as a superior performance to other
optical fiber pH sensors. Most of the problems associated with
optical sensors based on absorption or fluorescence changes
of pH indicators are overcome. These sensors offer a limited
linear range of typically 2 pH units as opposed but the reported
sensor has a range of 6 pH units. More importantly, ionic
strength of the medium under test strongly affects the spectral
 1414 - JOURNAL OF LIGHTWAVE “ N O u X i Y , VOL. 13, NO. 7, JULY 1995
characteristics of pH indicators, but the reported sensor is
based on OH- ions activity, and t l y s still measures the real
pH value at high ionic CoacenrratiOns.
Optimization of the sensor enables it to be tailored to
specific applications by consideration of sensitivity, insertion
loss and response time of the sensor.
It is necessary to confirm these preliminary results with
extensive tests. Long km stability of the silica core must be
evaluated with regards to continuous immersion and degra-
dation of the surface by the methylene blue dye. Another
factor affecting long term stability is the contamination of
the semi-permeable membrane by biofilms or limescale and
different materials must be tested in iddustrial environment.
As reported here, high ionic strength ddes not adversely affect
the sensor, but low ionic strength such as found in underground
water might affect the response time of the sensor making it
impractical for environmental sensing.
The design of the sensor can be improved by modifying
the geometry of thecapsule to reduce its volume. A smaller
capsule will also improve the response time by reducing
the ratio (dye volume)/(membrane sdrface). Multiwavelength
monitoring by providing on-line calibration of the optical sys-
tem and compensation for common mode noise will strengthen
the possibility of many industrial applications.
In conclusion, the reported sensing techhique is highly en-
couraging, but further tests rem& to be performed particularly
with regards to its potential for industrial applications.
REFERENCES
[l] S. Motellier, M. H. Michels, B. hueauk, and P. Toulhoat,“Fiber-optic
pH sensor foin situ applications,” Sensors and Actuators B. vol. 11, pp.
467473, 1993.
[2] A. Holobar and B. H Weigl et d.,“Experimental results on an optical
pH measurcmt system for bioreactors,” Sensors and Actuators E, vol.
11: pp. 425430, 1993.
[3] J. I. Peterson, S. R. Goldstein, R. V. Fitzgerald, and D.K.Buckhold,
“Fiber optic pH probe for physiological use,” AMI. Chem., vol. 52, p.
864, 1980.
[4] C. Dafu, C. Qiang, H. Jinghong, C. Jinge, L. Yating, and Z. Zemin.
“Optical-fiber sensor,” Sensors and Actuators E, vol. B12, pp. 29-32,
Mar. 1993.
[ 5 ] B. D. M d r a i t h , “ E n h a n d ehnescent wave sensor based on sol-gel
derived porous glass coatings,” Sensors and Actuators E, vol. 11, pp.
29-34, 1993.
[6] S. Draxler and M. E. Lippitsch, “Optical pH sensors using fluorescence
decay time,” Sensors rmd Actmtors B. vol. 11, pp. 421-424, 1993.
[7] L. L. Blyler, R A. Lieberman, L. G. Cohen, J. A. Ferrara, and J.
B. MacChesney, “Optical fiber chemical sensors utilizing dye-doped
silicone polymer claddings,” holymer Engineering and Science, vol. 29,
mid-Sept. 1989.
[8] V. Ruddy, “An effective attenuation coefficient for evanescent wave
spectroscopy using multimode fibers,” Fiber Integr. Opt., vol. 9, pp.
142-150, 1990.
[9] B. D. Gupta, “Fiber optic evanescent field absorption sensor: Effect of
launching condition and the geometry of the sensing region,” Opt. Eng.,
vol. 33, pp. 1864-1868, June 1994.
[lo] G. Stewart and B. Culshaw, “Optical waveguide modelling and design
for evanescent field chemical sensors,” Opt. and Qunntwn Electron;
vol. 26, pp. 249-259. 1994.
[ l l ] V. Ruddy, B. Ma~Graith,and J. A. Murphy, “Spectroscopy of fluids
using evanescent wave absorption on multimode fiber,” SPIE vol. 1 I72
Chemical, Biochemical, & Environmental Sensors,, pp. 83-92, 1989.
[12] A. W.Snyder and J. D. Love, Optical Waveguide Theory. Chapman
and Hall, 1983.
[13] R. L. Snyder, R.A. Condrate. and P. F. Johnson, Advances in Materials
Characterization n,Materials Science Research Vol 19. New York:
Plenum, 1985, pp. 189-201.
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.
[14] G. F. Kirkbright, R. Narayaaaswami, and N. A. Welti, “Fiberqtic
pH p b e based on the use of an immobilized colorimetric indicator,”
Analyst. vol. 109, Aug. 1984.
[15] W. T. Wimpenny, ‘The spatial organization of biofilm,” Inaugauml
meeting of the British Biofilm Club, circulated preprints, Mar. 31-Apr.
2, 1993, pp. 1-3.
Bruno J.-C. Deboux was born in 1970 and received
the University Diploma @UT), in Electrical and
Electronic Engineering in 1991 h m the University
Paris XI (Cnchan), Paris, France and B.Eng.degree
with Honours in Electrical and Electronic Engineer-
ing in 1992 fnrm Liverpool Polytechoic, Liverpool,
UK.’
In 19p2 he joined the Optical Fiber Sensor Re-
search Group, Liverpool John Moores University,
and ~ since been engaged in research on opti-
cal fiber sensors. His research work includes the
development of patented sensors for pH measurements and biofilm detection.
Etfed Lewis was born in 1959 and graduated from
the University of Liverpool in 1981 with the B.Eng.
degree (Hons.) in Electronic Engineering. In 1987
he redeived the Ph.D. degree from the University
of ’Liverpool for optical investigations during the
cutrent zero phase of switchgear arcs.
Following two years research and development
in optical fibers sensors with BICC Telecom Cables
Lfd. (1985-1987). he joined the optical fiber sen-
sors nsearch group at the University of Liverpool
(1967-1989) as a Senior Research Assistant. In
1989 he joined the Liverpool John Moores University and helped establish
the present Optical Fiber Sensor,Researchgroup (1991 approx.). His present
research interests include optical fiber sensors for biological and environmental
monitoring.
hqid.J. S d y was born in 1964 and graduated
in 1985 from the Uaiversity of Manchester with
the 3.Sc. &gee (Hons) in Physics. She received
the M.Sc. degree in Instnunentation and Analytical
SEience from W S T , worked for GEC Industrial
Controls PLC, and then received the Ph.D. degree
from the University of Liverpool, in optical sensors
for physiological monitoring in magdetic resonance
scanners.
In 1990 she was appointed as Lecturer in Applied
physics at Liverpool John Moores University and
joined the newly establis,bed Optical Fiber Sensors Research Group. Resent
research interests include optital fiber sensors for environmental and biological
monitoring.
Rokrt Edwards was born in 1%7. He studied
Chemistry at University College Cardiff, UK, and
gained a Fist Class Honours degree in 1988. He
stayed at Cardiff University to study the Chem-
istry of Archaeological Lead and Tin Artefacts, and
was awarded the Ph.D. degree in 1991. Before his
present post, he was a Postdoctoral Reseaher at
Hull University, UK, investigating a novel method
for trace enzyme determination using chemilumi-
nesce and flow injection analysis techniques.
He was appointed as a Lecturer in Inorganic and
Environmental Chemistry at Liverpool John Moores University in August
1992. His research interests include investigating the geochemistry of heavy
(toxic) metals, and methods for their analysis; novel techniques for the in situ
reclamation of contaminated land and the use of fiber otics for remote and
trace element analysis.