Analytica Chimica Acta 1106 (2020) 71e78

Contents lists available at ScienceDirect

Analytica Chimica Acta

journal homepage: www.elsevier.com/locate/aca

An online pH detection system based on a microfluidic chip

Yu Lu , Qianru Feng , Ruohu Zhang , Hui Lu , Jingting Su , Yiping Cui , Li Zhu *
Advanced Photonics Center, School of Electronic Science & Engineering, Southeast University, Nanjing, 210096, Jiangsu, China

h i g h l i g h t s g r a p h i c a l a b s t r a c t

An online pH detection system with

artificial neural network (ANN) to
read out pH value in real time.

Robust in extreme acid and alkaline
conditions with large dynamic ranges
of 5 M [Hþ] - pH 3.0 and pH 6.0 - 2 M
[OH].

High reversibility, repeatability, sta-
bility and long life-time.

a r t i c l e i n f o a b s t r a c t

Article history: An online pH detection system is very critical in monitoring the sudden change of pH, especially in
Received 19 September 2019 strongly acidic and alkaline conditions. We developed a pH sensing chip which works in the range of 5 M
Received in revised form [Hþ]-pH 3.0 and pH 6.0-2 M [OH] with response time of 90 s. The sensing chip was formed by coating a
14 January 2020
pH sensing membrane onto the wall of a microfluidic chamber. The pH sensing membrane was prepared
Accepted 28 January 2020
Available online 29 January 2020
by chemically immobilizing m-Cresol purple in polyvinyl alcohol (PVA). The pH detection system con-
sisted of light source, pH sensing chip and photodiode (PD). Once the pH of fluid flowing in the sensing
chip changed, the intensity of transmitted light changed. The intensity of transmitted light was converted
Keywords:
Online pH detection
to voltage, which was the function of pH value, by the PD. A feed-forward artificial neural network (ANN)
Microfluidic chip with error back-propagation training algorithm was employed to model the behavior of the pH sensor
High acidity and high alkalinity and read out pH values of unknown solutions. The pH detection system shows high stability with
Artificial neural network increasing the ionic strength. It also possesses properties of repeatability, reversibility and long life-time.
These advantages make the proposed pH detection system a promising solution for online detection of
pH values in harsh conditions.
© 2020 Elsevier B.V. All rights reserved.

1. Introduction the range of pH, but cannot give the precise value of pH. The
common quantitative measurement of pH includes electrochemical
pH measurement has received much attention in many fields, pH sensors [7e9] and optical sensors [10e14]. Fundamentally, the
such as medical diagnostics [1,2], agriculture [3], beverage in- electrochemical pH sensor, such as glass electrode pH meter, con-
dustries [4], process control [5] and engineering [6]. Using color sists of a pH-responsive electrode and a reference (unvarying)
change of indicators or pH test strips is the simplest way to show electrode. When the two electrodes are immersed in a solution, the
potential difference between electrodes is directly related to the pH
of the solution. Absorption- and luminescence-based pH sensor are
two main types of optical pH sensors. The absorbance or fluores-
* Corresponding author.
E-mail address: [email protected] (L. Zhu). cence of pH-responsive regents is quantificationally related to the

https://doi.org/10.1016/j.aca.2020.01.063
0003-2670/© 2020 Elsevier B.V. All rights reserved.
72 Y. Lu et al. / Analytica Chimica Acta 1106 (2020) 71e78

pH. However, these methods usually work in an offline way, thus Surface Profiler (Ambios, America). A LSP04-1A constant flow
cannot meet the requirements of some practical applications. For pump (Longer Precision Pump, China) equipped with pipes was
example, real-time pH monitoring is especially needed in chemical employed to inject solution of various pH into the pH sensing chip.
industry because pH variation affects chemical reaction efficiency A KW-4B Spin Coater (SETCAS Electronics, China) was used to spin
and production quality greatly [15]. coating pH sensing membrane onto glass slides. An Exemplar Pro
In addition to online measurement, pH sensors for high acidity optical fiber spectrophotometer (BWTEK, America) was used to
and alkalinity (pH < 3 or pH > 13) are particularly demanded in record transmitted spectra. FTIR analysis was carried out using a
industries such as photographic developers and wastewater treat- Nicolet iS 50 FT-IR Fourier Transform Infrared Spectrometer
ment facilities [16]. Conventional pH meters with glass electrodes (ThermoFisher Scientific, China). A Digilent Analog Discovery 2
suffer from errors and chemical deterioration in high acidity and (Digilent, America) transferred voltage data to the computer via the
alkalinity range [17]. USB interface and displayed results in WaveForms software.
The measurement range of the optical pH sensor can be
extended by integrating multiple indicators with different pKa 2.3. Preparation of the sensing membrane
values in a sensor [12e14]. It is also found that covalent binding
indicators to solid matrix can greatly widen the dynamic response 0.765 g m-Cresol purple (2.0 mmol), 4.0 ml 37e40% formalde-
range of optical sensors [18]. Many acid-base indicators such as hyde aqueous solution (correspond to formaldehyde 49.4 mmol),
phenol red [19], phenolphthalein [20], brilliant yellow [21] Congo 0.50 g NaOH and 20.0 ml distilled water were added into a three-
red [22], orange (II) [23] have been reported to be immobilized on neck flask, which was equipped with a magnetic stirrer, a N2 inlet
various solid substrates to broaden dynamic working ranges. These and a condenser. The mixture reacted at boiling temperature under
pH sensors can work in either high acidity or high basicity condi- nitrogen atmosphere for 8 h forming m-Cresol purple-
tions. However, they work in offline way and cannot meet the re- formaldehyde (abbreviated as CPF) reaction solution. 1.5 g PVA
quirements of real-time measurement. was dissolved in 20.0 ml deionized water at 100  C under N2 at-
In this work, we develop an online pH detection system working mosphere. Then, 0.6 ml CPF reaction solution was added into PVA
in both high acidity and high basicity conditions. m-Cresol purple is aqueous solution and reacted for 2 h. The viscous mixture was spin-
immobilized in the PVA membrane via hydroxymethyl groups coated onto a glass slide with a speed of 1200 rpm and then baked
produced in m-Cresol purple-formaldehyde reaction. The prepared for 1 h at 90  C. The processes were repeated 3 times. After that, the
pH sensing membrane is coated on the inner wall of a microfluidic glass slide was soaked in deionized water for 24 h at room tem-
chip to integrate optical and fluidic functionalities. Light Emitting perature to remove unbonded indicators and the pH sensing
Diode (LED) with 561 nm central wavelength works as light source membrane was eventually prepared. The SEM image of the pH
in acidic condition while 599 nm in alkaline condition. The trans- sensing membrane is shown in Fig. 1(a). The morphology of the
mitted light is detected by a PD and converted light intensity into membrane is homogenous and crack-free which attributes to the
voltage which is correlated to pH value. Error back-propagation reliable performance of the pH sensing chip. The thickness of the
ANN algorithm is applied to treat the nonlinear response curve of pH sensing membrane measured by surface profiler is 3.6 mm.
the pH sensor. The proposed pH detection system responds rapidly
to both extremes of pH scales. In addition, measuring pH value at 2.4. Fabrication of the pH sensing chip
microfluidic scale enables continuous monitoring with ultra-low
reagent consumption. Spectra characteristics, ionic strength ef- Fabrication steps of the pH sensing chip are shown in Fig. 1(b).
fect, response curve, repeatability, reversibility and long-term sta- The cover plate with microchannel is made of poly-
bility of the sensor are fully discussed in this paper. dimethylsiloxane (PDMS) using the replica molding technique.
Briefly, the mixture of pre-polymer of PDMS and cross-linker was
2. Materials and methods cast on the micro-channel mold which was fabricated using a 3D
printer. The PDMS replica with designed microchannel was peeled
2.1. Reagents off from the mold after been baked at 60  C for 12 h. After being
cleaned by oxygen plasma and punched the inlet and outlet, the
m-Cresol purple, formaldehyde aqueous solution (37e40% (w/ PDMS cover plate was bonded to the glass slide which was coated
w), Analytical Reagent), PVA (degrees of polymerization: 1750 ± 50, the sensing membrane. To achieve permanent bonding, the chip
degree of alcoholysis: 88%), sodium hydroxide (NaOH), hydro- was baked at 65  C for 3 h and the pH sensing chip was finally
chloric acid (HCl), acetic acid, boric acid and phosphoric acid were fabricated. The photograph of the pH sensing chip is shown in
purchased from China Medical Company. BrittoneRobinson buffers Fig. 1(c).
(pH 1.8e11.9) were obtained by titrating a mixed acid solution
(0.04 M acetic acid, 0.04 M boric acid and 0.04 M phosphoric acid) 3. Results and discussion
with 0.2 M NaOH to the required pH [24]. The higher acidic and
alkaline solutions were prepared by diluting HCl and NaOH, 3.1. Immobilization of m-Cresol purple in the PVA
respectively. The pH values were determined by a PHB-5 digital pH
meter (Weiye Instrument, China). Deionized distilled water was m-Cresol purple is a sulfophenolphthalein and its molecular
used throughout experiments. structure is shown in Fig. 2. Phenolic hydroxyl is the auxochrome of
m-Cresol purple and plays an important role in color change. In
2.2. Instrumentation phenol-formaldehyde reaction, the formaldehyde mainly attacks
the ortho and para position of the phenolic hydroxyls, rather than
UVevis absorption measurements were performed on a UV- the phenolic hydroxyls themselves [25]. Therefore, the phenolic
3600 UV-VIS-NIR spectrophotometer (SHIMADZU, Japan). The hydroxyls remain intact after phenol-formaldehyde reaction and
scanning electron microscopy (SEM) image of pH sensing mem- m-Cresol purple keeps its pH sensitivity after it is covalent bound to
brane was taken on a Zeiss Ultra Plus field emission scanning PVA by condensation reaction.
electron microscope with an accelerating voltage of 1 kV. The To verify if hydroxymethyl groups have been successfully pro-
thickness of the pH sensing membrane was measured with XP-2 duced in CPF reaction solution, FTIR spectra of CPF powder and m-
 Y. Lu et al. / Analytica Chimica Acta 1106 (2020) 71e78 73

Fig. 1. (a) SEM image of the pH sensing membrane. (b) Fabrication steps of the pH sensing chip. (c) Photograph of the pH sensing chip.

wavelengths redshifted from 432, 527 and 578 nm to 434, 561 and
599 nm. 561 nm and 599 nm were selected as the detection
wavelengths in low pH and high pH measurement, respectively. It is
particularly noteworthy that the dynamic response range of the pH
sensing membrane is extended from pH 1.2-pH 2.8e5 M [Hþ]-pH
3.0 and pH 7.4-pH 9.0 to pH 6.0-2 M [OH] compared with that of
m-Cresol purple molecule. These changes might be due to the
newly produced chemical bond between m-Cresol purple and PVA
[26e28]. FTIR spectra (Fig. 3(b)) showed that the ring stretch band
at 1567 cm1 appeared in the pH sensing membrane, which sug-
gested that m-Cresol purple had been covalently immobilized in
the PVA membrane.

3.3. Equilibrium response time

Fig. 2. Molecular structure of m-Cresol purple.

To measure the response time of the pH sensing chip, various

buffer solutions were injected into the chip by a flow pump suc-
Cresol purple were taken and shown in Fig. 3(a). The CPF powder cessively. The velocity of the buffer was about 18 mm s1. The po-
was prepared by freeze-drying the CPF reaction solution. There was sition of each optical element was carefully adjusted to make sure
a stronger and wider hydroxyl absorption peak at 3385 cm1 in CPF. that the light emitted from LED was perpendicularly incident onto
At the same time, the intensity of aromatic CeH stretching vibra- the pH sensing membrane. The transmitted spectra through the pH
tion band at 3048 cm1 diminished in CPF and the shoulder peak of sensing chip were collected by an optical fiber spectrophotometer.
aliphatic CeH stretching vibration at 2830 cm1 appeared. It sug- The integration time was set to 10 ms and transmitted spectra were
gested that formaldehyde should have reacted with m-Cresol acquired every 3 s by the upper computer of the optical fiber
purple and hydroxymethyl groups have been produced in CPF. spectrophotometer. The response curve of the pH sensing chip was
obtained by plotting the transmitted light intensity at 561 nm
3.2. Absorption characteristics of the pH sensing membrane (acidic condition) and 599 nm (alkaline condition) over time.
Fig. 6(a) shows the continuous response curves of the pH sensing
The absorption spectra of m-Cresol purple aqueous solutions chip in acidic solutions at 561 nm while Fig. 6(b) in alkaline solu-
and the pH sensing membrane at various pH values are shown in tions at 599 nm. It is found that the absorption of pH sensing chip
Fig. 4 and Fig. 5, respectively. From absorption spectra, we found reached 95% of its steady state in less than 90 s. The very thin
the pH sensing membrane kept chromogenicity of m-Cresol purple (3.6 mm) thickness of the sensing membrane could be one of the
in both acid and alkaline solutions, but maximal absorption reasons which contributes to the fast response time, because the

Fig. 3. (a) FTIR spectra of m-Cresol purple and CPF. (b) FTIR spectra of PVA membrane and pH sensing membrane. (For interpretation of the references to color in this figure legend,
the reader is referred to the Web version of this article.)
74 Y. Lu et al. / Analytica Chimica Acta 1106 (2020) 71e78

Fig. 4. Absorption spectra of m-Cresol purple in different buffer solutions: (a) acidic condition: pH 1.2, pH 1.4, pH 1.8, pH 2.0, pH 2.2, pH 2.4, pH 2.6, pH 2.8. (b) alkaline condition:
pH 7.4, pH 7.6, pH 7.8, pH 8.0, pH 8.2, pH 8.4, pH 8.6, pH 8.8, pH 9.0. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this
article.)

Fig. 5. Absorption spectra of pH sensing membrane in different buffer solutions: (a) acidic condition: pH 3.0, pH 2.7, pH 2.0, pH 1.0, pH 0.4, 1 M [Hþ], 2 M [Hþ], 3 M [Hþ], 4 M [Hþ],
5 M [Hþ]. (b) alkaline condition: pH 6.0, pH 6.3, pH 7.4, pH 9.0, pH 10.2, pH 11.1, pH 11.6, pH12.1, pH 12.5, pH 13.6, 1 M [OH], 2 M [OH].

Fig. 6. Response curve of the pH sensing chip. (a) at 561 nm for alternative change of pH from 2.8 to 0.2 and 3 M [Hþ]. (b) at 599 nm for alternative change of pH from 10.2 to 11.2,
12.8 and 1 M [OH].

thinner the membrane, the shorter the protons crossing time. wavelength. The relative standard deviations (RSDs) of given pH
Another possible reason is that the hydrophilicity of PVA mem- buffers were less than 0.6%.
brane could make the hydronium ions diffuse fast in the membrane The absorbance of the membrane at 433.5 nm was kept
and greatly shorten the response time [29]. recording while it was stored in deionized water at 25  C for over 60
days. As shown in Fig. 8, no significant variation in the absorbance
3.4. Repeatability, reversibility and stability was observed during this period. The decrease of the absorption
signal was less than 4.2% within 2 months, which means there is
In order to monitor the repeatability and reversibility of the almost no m-Cresol purple leaching from the membrane and the
sensing membrane, buffer solutions with different pH (10.2, 11.2, membrane keeps it sensitivity to pH change, shown in Fig. 9. These
12.8, 1 M [OH]) were injected into the sensor by a flow pump results indicate that the proposed pH sensing chip has the advan-
subsequently. The continuous flow response curve was recorded by tage of long-term stability. The comparison of performances be-
an optical fiber spectrophotometer for several cycles. Fig. 7 shows tween the proposed pH sensing chip and some typical pH sensors
the absorbance changes of the membrane versus time at 599 nm in literature reports is shown in Table 1.
 Y. Lu et al. / Analytica Chimica Acta 1106 (2020) 71e78 75

indicators [32]. Effects of ionic strength on absorption responses of

m-Cresol purple and the pH sensing membrane in both acid and
alkaline solutions were studied by gradual increasing the concen-
tration of NaCl and KCl to 0.5 M. Results are shown in Fig. 10. m-
Cresol purple was obviously sensitive to ionic strength and the
absorbance of m-Cresol purple increased more than 14% in acid
solution and 47% in basic solution compared with solution without
electrolyte. However, the variations of absorbance of the pH
sensing membrane were only 0.9% and 1.7% in acid and alkaline
condition, respectively. We consider the possible reason contrib-
uting to the negligible cross-sensitivity towards ionic strength
could be as follows: the PVA is uncharged and the surface potential
of pH sensing membrane approaches to zero when m-Cresol purple
is immobilized into the polymer, which could build a neutral
Fig. 7. Response repeatability and reversibility of the membrane in four different pH environment to resist the disturbance of ionic strength to ioniza-
buffers: pH 10.2, 11.3, 12.8 and 13.8 (from bottom to top). tion/protonation of indicator molecules [32]. Therefore, the sensi-
tivity of m-Cresol purple to ionic strength should be greatly
reduced when it was immobilized into PVA. The robust perfor-
mance makes the pH sensing membrane a good choice for pH
detection in complex environment.

3.6. pH detection system with ANN

The schematic diagram of the pH detection system is illustrated

in Fig. 11. LEDs (LED1: 599 nm; LED2: 561 nm) worked as the light
sources. After being collimated by lenses group, beam was
perpendicularly incident onto the pH sensing membrane. The in-
tensity of transmitted light which was the function of pH value was
converted to voltage by the PD. Digilent Analog Discovery 2 worked
as an A/D converter and transferred voltage data to the computer
via the USB interface.
As shown in Fig. 12, the absorbance of the pH sensing membrane
is not linear dependent on pH value, either in acid or alkaline
conditions. ANN has shown excellent results in modeling complex
Fig. 8. Absorbance of the pH sensing membrane recorded at 433.5 nm over 60 days.
nonlinear relationships. Therefore, a feed-forward ANN having a
single hidden neuron layer with back-propagation training algo-
rithm was employed to model the spectral behavior of the pro-
3.5. Effects of the ionic strength posed sensor.
Generally, ANN adapts itself to the nonlinear mechanism by
Ionic strength represents the concentration of ions in a solution. investigating the input-output relationship using a hyperspace
Common cationic, such as Naþ and Kþ at high concentrations often equation. Training is the process of inputting multiple sets of input-
influence the response of a pH sensor and cause errors in mea- output relationships from external environment. During training,
surements. Ionic strength affects the performance of indicator in the coefficients of the equation such as number of nodes in each
two ways: (1) electrolyte solutions may absorb light in the working hidden layer, learning rate and momentum are optimized to fit the
range of indicators; (2) ionic strength impacts the ionization/pro- training data satisfactorily. To optimize the pH detection system,
tonation of an indicator molecule and changes the ratio of conju- the liquid samples were divided into two subsets: training and test
gate base and conjugate acid, and finally alters absorbance of pH sets. Forty samples were selected within the pH range from 5 M

Fig. 9. Absorption spectra of pH sensing membrane in different buffer solutions after two months of storage: (a) acidic condition: pH 3.0, pH 2.3, pH 1.1, 1 M [Hþ], 3 M [Hþ], 5 M
[Hþ]. (b) alkaline condition: pH 6.0, pH 7.2, pH 8.3, pH 9.3, pH 10.2, pH 11.5, pH 12.4, pH 13.6, 2 M [OH].
76 Y. Lu et al. / Analytica Chimica Acta 1106 (2020) 71e78

Table 1
Comparison between the proposed pH sensing chip and reported pH sensors.

Year Indicator Dynamic pH range Response time

2015 [10] pyranylidene malononitrile derivative (DPM) 1.0-2.0 <70 s

2008 [12] m-cresol purple and bromocresol green 1.04-8.70 <120 s
2018 [22] Congo red 0e4.5 150 s
2016 [23] orange (II) 3e10 120 s
2013 [30] purple cabbage pigment 2e11 <120 s
2016 [31] anthocyanins 1e10 180 s
This work m-Cresol purple 0.7-3.0 and 6.0e14.3* 90 s

* 5 M [Hþ]-pH 3.0 and pH 6.0-2 M [OH] are corresponding to 0.7-3.0 and 6.0e14.3.

Fig. 10. Ionic strength effects on m-Cresol purple and pH sensing membrane. (a) m-Cresol purple; (b) pH sensing membrane. (For interpretation of the references to color in this
figure legend, the reader is referred to the Web version of this article.)

Fig. 11. Set-up of the pH detection system.

[Hþ]-pH 3.0 and from pH 6.0-2 M [OH] as training set and of m-Cresol purple. Therefore, the current pH detection system
seventeen samples with various pH values as test set. After training, could be expected to measure real samples in the physiological
the pH detection system was applied to predict the pH values of range of pH (6e8) if their absorptions do not overlap with m-Cresol
unknown buffer solutions and results are shown in Fig. 13. The purple’s.
squared correlation coefficient R2 is 0.9996 and mean square error
is 0.004 which show high reliability of the detection system. 4. Conclusion
Since the pH sensing membrane showed negligible cross-
sensitivity towards ionic strength, we thought it could be used to A pH detection system based on a microfluidic chip is developed
measure pH values of some biological samples around physiolog- for online detection of pH values in extreme acid and alkaline
ical condition (pH 6e8). We tested 4 protein solutions, which were conditions. It has large dynamic ranges of 5 M [Hþ]-pH 3.0 and pH
cytochrome c, PMP22-TM4 (WT) and PMP22-TM4 (G150D) in 0.1 M 6.0-2 M [OH]. It also shows negligible cross-sensitivity towards
KPi buffer at pH 7.2, and Lysozyme in 0.1 M KPi buffer at pH 7.0, in ionic strength, which makes it works properly in complex envi-
the system. ronments, such as biological samples. The pH sensing chip has a
Measured pH values were 9.73, 7.26, 7.27 and 6.95, respectively. rapid response time of 90 s and possesses properties of revers-
Except cytochrome c, our system worked well with the other pro- ibility, repeatability, high stability and long life-time. With the
teins. The measurement errors of the three proteins were 0.83%, application of ANN, the pH detection system can read out pH value
0.97% and 0.71%, respectively. The large error in the case of cyto- of unknown solutions precisely. These excellent performances
chrome c should be attributed to the large absorbance at 599 nm of make the proposed pH detection system a promising solution for
heme, which is partially overlapped with the basic absorption peak online pH detection in harsh conditions.
 Y. Lu et al. / Analytica Chimica Acta 1106 (2020) 71e78 77

References

[1] B.A. Webb, M. Chimenti, M.P. Jacobson, D.L. Barber, Dysregulated pH: a perfect
storm for cancer progression, Nat. Rev. Canc. 11 (2011) 671e677.
[2] M.T. Ghoneim, A. Nguyen, N. Dereje, J. Huang, G.C. Moore, P.J. Murzynowski,
C. Dagdeviren, Recent progress in electrochemical pH-sensing materials and
configurations for biomedical applications, Chem. Rev. 119 (2019)
5248e5297.
[3] M. Futagawa, T. Iwasaki, H. Murata, M. Ishida, K. Sawada, A miniature inte-
grated multimodal sensor for measuring pH, EC and temperature for precision
agriculture, Sensors 12 (2012) 8338e8354.
[4] C. Bohnke, H. Duroy, J.L. Fourquet, pH sensors with lithium lanthanum titanate
sensitive material: applications in food industry, Sensor. Actuator. B Chem. 89
(2003) 240e247.
 th, E. Pretsch, Spectroscopic in
[5] E. Lindner, T. Zwickl, E. Bakker, B.T.T. Lan, K. To
situ imaging of acid coextraction processes in solvent polymeric ion-selective
electrode and optode membranes, Anal. Chem. 70 (1998) 1176e1181.
[6] P.L. Fuhr, D.R. Huston, Corrosion detection in reinforced concrete roadways
and bridges via embedded fiber optic sensors, Smart Mater. Struct. 7 (1998)
217e228.
[7] R.G. Du, R.G. Hu, R.S. Huang, C.J. Lin, In situ measurement of Cl concentra-
Fig. 12. Absorbance of the pH sensing membrane at different pH values (black:
tions and pH at the reinforcing steel/concrete interface by combination sen-
561 nm; red: 599 nm). (For interpretation of the references to color in this figure sors, Anal. Chem. 78 (2006) 3179e3185.
legend, the reader is referred to the Web version of this article.) [8] S. Zaman, M.H. Asif, A. Zainelabdin, G. Amin, O. Nur, M. Willander, CuO
nanoflowers as an electrochemical pH sensor and the effect of pH on the
growth, J. Electroanal. Chem. 662 (2011) 421e425.
[9] S. Zhuiykov, E. Kats, K. Kalantar-zadeh, M. Breedon, N. Miura, Influence of
thickness of sub-micron Cu2O-doped RuO2 electrode on sensing performance
of planar electrochemical pH sensors, Mater. Lett. 75 (2012) 165e168.
[10] D.Y. Kim, H. Kim, H.J. Kim, A sol-gel-derived optical sensor applicable for
determination of extremely low pH values, Bull. Kor. Chem. Soc. 36 (2015)
473e478.
[11] A. Moutsiopoulou, A.K. Andreopoulou, G. Lainioti, G. Bokiasa, G. Voyiatzis,
J.K. Kallitsis, Quinoline-functionalized cross-linked poly(vinyl acetate) and
poly(vinyl alcohol) nanoparticles as potential pH-responsive luminescent
sensors, Sensor. Actuator. B Chem. 211 (2015) 235e244.
[12] M. Shamsipur, F. Abbasitabar, V. Zare-Shahabadi, Shahabadi, M. Akhond,
Broad-range optical pH sensor based on binary mixed-indicator doped sol-gel
film and application of artificial neural network, Anal. Lett. 41 (2008)
3113e3123.
[13] S. Dong, M. Luo, G. Peng, W. Cheng, Broad range pH sensor based on solegel
entrapped indicators on fibre optic, Sensor. Actuator. B Chem. 129 (2008)
94e98.
[14] B.D. Gupta, S. Sharma, long-range fiber optic pH sensor prepared by dye
doped sol-gel immobilization technique, Optic Commun. 154 (1998)
282e284.
[15] M. Hosseini, R. Heydari, M. Alimoradi, A novel pH optical sensor using methyl
orange based on triacetylcellulose membranes as support, Spectrochim. Acta
Fig. 13. Correlation between actual and predicted pH values. 128 (2014) 864e867.
[16] J. Lin, Recent development and applications of optical and fiber-optic pH
sensors, Trac. Trends Anal. Chem. 19 (2000) 541e552.
[17] A. Safavi, M. Bagheri, Novel optical pH sensor for high and low pH values,
Declaration of competing interest Sensor. Actuator. B Chem. 90 (2003) 143e150.
[18] S. Capel-Cuevas, M.P. Cue llar, I. de Orbe-Paya
, M.C. Pegalajar, L.F. Capitan-
Vallvey, Full-range optical pH sensor array based on neural networks,
The authors declare that they have no known competing Microchem. J. 97 (2011) 225e233.
financial interests or personal relationships that could have [19] Z. Liu, J. Liu, T. Chen, Phenol red immobilized PVA membrane for an optical pH
appeared to influence the work reported in this paper. sensor with two determination ranges and long-term stability, Sensor.
Actuator. B Chem. 107 (2005) 311e316.
[20] Z. Liu, F. Luo, T. Chen, Phenolphthalein immobilized membrane for an optical
pH sensor, Anal. Chim. Acta 510 (2004) 189e194.
CRediT authorship contribution statement [21] N. Raoufi, F. Surre, T. Sun, M. Rajarajan, K.T.V. Grattan, Wavelength dependent
pH optical sensor using the layer-by-layer technique, Sensor. Actuator. B
Yu Lu: Investigation, Software, Validation, Writing - original Chem. 169 (2012) 374e381.
[22] M.G.K. Salmani, G.H. Rounaghi, M. Chamsaz, An optical sensor for determi-
draft, Writing - review & editing. Qianru Feng: Software, Investi- nation of low pH values based on covalent immobilization of Congo red on
gation, Writing - original draft. Ruohu Zhang: Investigation, triacetyl cellulose films via epichlorohydrin, Sensor. Actuator. B Chem. 254
Funding acquisition. Hui Lu: Investigation. Jingting Su: Investiga- (2018) 177e181.
[23] R. Heydari, M. Hosseini, A. Amraei, A. Mohammadzadeh, Preparation of a
tion. Yiping Cui: Supervision, Funding acquisition. Li Zhu: novel pH optical sensor using orange (II) based on agarose membrane as
Conceptualization, Validation, Writing - review & editing, Project support, Mater. Sci. Eng. C 61 (2016) 333e337.
administration, Supervision, Funding acquisition. [24] H.T.S. Britton, R.A. Robinson, CXCVIII.-Universal buffer solutions and the
dissociation constant of veronal, J. Chem. Soc. (1931) 1456e1462.
[25] A. Knop, L.A. Pilato, Reaction mechanisns, in: Phenolic Resins Chemistry, in:
Applications and Performance Future Directions, Springer, Berlin, 1985,
Acknowledgements pp. 24e61.
[26] Z. Liu, F. Luo, T. Chen, Polymeric pH indicators immobilized PVA membranes
The authors gratefully acknowledge the financial supports from for optical sensors of high basicity based on a kinetic process, Anal. Chim. Acta
519 (2004) 147e153.
the National Natural Science Foundation of China (61378045,
[27] P. Hashemi, M.M. Abolghasemi, Preparation of a novel optical sensor for low
61601118), the Natural Science Foundation of Jiangsu Province pH values using agarose membranes as support, Sensor. Actuator. B Chem.
(BK20131297), and Scientific Research Foundation for the Returned 115 (2006) 49e53.
[28] G.J. Mohr, O.S. Wolfbeis, Optical sensors for a wide pH range based on azo
Overseas Chinese Scholars, State Education Ministry of the People’s
dyes immobilized on a novel support, Anal. Chim. Acta 292 (1994) 41e48.
Republic of China.
78 Y. Lu et al. / Analytica Chimica Acta 1106 (2020) 71e78

[29] E. Wang, K.F. Chow, V. Kwan, T. Chin, C. Wong, A. Bocarsly, Fast and long term pH sensor based on immobilization of red grape extract on bioorganic agarose
optical sensors for pH based on sol-gels, Anal. Chim. Acta 495 (2003) 45e50. membrane, Sensor. Actuator. B Chem. 224 (2016) 391e395.
[30] X. Chen, Z. Gu, Absorption-type optical pH sensitive film based on immobi- [32] J. Janata, Do optical sensors really measure pH? Anal. Chem. 59 (1987)
lized purple cabbage pigment, Sensor. Actuator. B Chem. 178 (2013) 207e211. 1351e1356.
[31] M.M. Abolghasemi, M. Sobhi, M. Piryaei, Preparation of a novel green optical