Analytica Chimica Acta 1091 (2019) 50e58

Contents lists available at ScienceDirect

Analytica Chimica Acta

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

Automated analytical microsystem for the spectrophotometric

and red wines
monitoring of titratable acidity in white, rose
Natalia Sa

ndez a, Antonio Calvo-Lo

pez a, Susana S.M.P. Vidigal b, Anto

nio O.S.S. Rangel b,

n Alonso-Chamarro a, *
Julia
a
Group of Sensors and Biosensors (GSB), Departament de Química, Edifici Cn, Universitat Auto noma de Barcelona, Bellaterra, 08193, Barcelona, Spain
b
Universidade Cato
lica Portuguesa, CBQF - Centro de Biotecnologia e Química Fina - Laborato

rio Associado, Escola Superior de Biotecnologia, Rua Diogo
Botelho 1327, 4169-005, Porto, Portugal

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

Low-cost polymeric analytical

microsystem.

Titratable acidity determination us-
ing optical detection.

Complete automation of the
microsystem.

Working range covering the common
titratable acidity content of musts
and wines.

Real wine samples successfully
analysed.

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

Article history: The design, construction and evaluation of a low-cost cyclic olefin copolymer (COC)-based continuous
Received 31 July 2019 flow microanalyzer with optical detection to determine the titratable acidity content of wine is here
Received in revised form presented. The analysis method is based on the monitoring of the blue coloration decrease of a buffered
9 September 2019
bromothymol blue (BTB) solution in the presence of the acidic compounds of wine. The microanalyzer
Accepted 18 September 2019
Available online 20 September 2019
monolithically integrates the required microfluidic motifs as well as an optical flow cell where the
measurements are performed by using a miniaturized and versatile photometric detection system.
Fluid management is totally automated by the use of computer-controlled microvalves, permitting the
Keywords:
Lab on a chip
automatic calibration of the system as well as the automatic sampling, including in-line dilution and
Miniaturization analysis. The reduced size of the whole system along with its high simplicity and automation make it
Automation suitable for its application to the on-line monitoring of titratable acidity during wine-making processes.
Cyclic olefin copolymer With the optimal conditions, a linear range up to 0.50 g L1 tartaric acid, a quantification limit (LOQ) of
Titratable acidity 0.01 g L1 and a detection limit (LOD) of 0.004 g L1 were obtained, covering the most common acidity
Wine acidity content of musts and wines. A sampling rate up to 26 h1 could be achieved, consuming less than 3 mL of
inexpensive reagents per analysis and requiring no pretreatment of the sample. The microsystem has
been successfully applied to the quantification of the titratable acidity content of several wine samples,
being the results in excellent agreement with the ones obtained by the reference method.
© 2019 Elsevier B.V. All rights reserved.

* Corresponding author.
E-mail address: Julian.Alonso@uab.es (J. Alonso-Chamarro).

https://doi.org/10.1016/j.aca.2019.09.052
0003-2670/© 2019 Elsevier B.V. All rights reserved.

ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa
1. Introduction
Wine acidity has an important and complex influence on several
characteristics of wine, being directly related to its microbiological
and chemical stability and having a strong impact on the organo-
leptic attributes of the final product [1]. Acidity of musts and wines
can be assessed by the determination of several parameters. From
these, titratable acidity is considered one of the most important
ones owing to its strong relationship, along with pH, with the
previously mentioned properties [1e4]. Its measurement and
adjustment (when deemed necessary) is actually performed at
several steps of the wine-making process in order to guarantee a
final high-quality product.
The official methods for determining titratable acidity consist in
the titration of the sample with a strong base to a specified
endpoint, pH 7.0 [5] or pH 8.2 [6], by potentiometric or colorimetric
detection using phenolphthalein [5] or bromothymol blue (BTB) [6]
as pH indicators. These classical acid-base titrations are laborious
and time-consuming, require large sample and titrant volumes and
usually involve skilled personnel.
Due to the need for more simple, rapid and cost-effective
methods displaying a higher degree of automation and that could
be easily implemented in routine analysis, several alternative
methods have been proposed. These include, for example, flow-
based systems for the automated titration of samples with poten-
tiometric [7] and colorimetric [8e13] endpoint determinations and
flow-based systems that do not involve the titration of the samples
but the construction of a calibration curve by potentiometric [14] or
colorimetric detection [15e19]. The latter are based on the moni-
toring of the absorbance of a solution containing an acid-base in-
dicator when it is mixed with a standard/sample, being the change
in the absorbance proportional to the acidity of the solution.
The aforementioned methods presented several advantages in
comparison to the classical titration ones, including higher auto-
mation, lower reagent and sample consumption and higher
throughput. However, some of them required complex manifolds
such as dialysis systems [18] or active mixers [7,9e11,13], used
expensive instrumentation [12,19], involved some sort of sample
pretreatment [8,14,17] or displayed narrow working ranges [15,16].
Several strategies can be used for solving some of the previously
mentioned drawbacks, such as the use of automated fluid man-
agement peripherals, miniaturized analytical platforms or smaller
dedicated detection systems.
In order to increase the miniaturization and automation of flow-
based systems, the use of miniaturized computer-controlled com-
mutators for fluid management has been frequently employed.
Systems including this type of actuators, such as multipumping
[20,21] and multicommuted [22e24] flow systems, offer several
advantages in comparison to the classical flow-based ones,
including a higher miniaturization of the setup (due to the reduced
dimensions of the commutators), the reduction of sample and re-
agents consumption (due to the precise insertion of volumes), the
increase in the reproducibility (due to the complete automation of
the processes and the minimal operator intervention) and a
reduction in the overall costs (owing to the previously mentioned
advantages) [22,24]. Furthermore, these systems are flexible and
versatile, as several hydrodynamic and chemical parameters can be
modulated by changing the programmed actuation of the com-
mutators instead of physically modifying the setup.
One of the most important factors for the development of a
functional miniaturized analytical system is the appropriate se-
lection of the substrate material. During the last decade, the use of
cyclic olefin polymers (COPs) and copolymers (COCs) as substrate
materials for Lab on a Chip (LOC) applications has become more and
more extensive due to their promising properties which include
51
high chemical resistance, low water absorption and high biocom-
patibility, among others [25,26]. One of the most relevant charac-
teristics of COPs/COCs is the availability of a wide range of materials
with different glass transition temperatures (Tg), enabling the use
of a rapid and simple multilayer approach for the construction of
the devices [27,28]. Furthermore, their high transparency in the
visible and near ultraviolet regions of the spectrum makes them
suitable for methods requiring optical detection, as they allow to
perform through-layer optical measurements [25,29].
On the other hand, the advances in optoelectronics have
permitted the development of new smaller and low-cost light
sources (such as light emitting diodes, LEDs) and detectors (such as
photodiodes), thus permitting the miniaturization and cost
reduction of the optical detection systems and, therefore, of the
whole experimental setups [30e34].
The purpose of the work here described was the development of
a miniaturized, versatile and simple COC-based continuous flow
microsystem for the determination of titratable acidity in musts
and wines requiring no sample pretreatment and displaying a wide
working range. The system was developed in order to be easily
applied to the on-line monitoring of this parameter in must and
wine during wine-making processes, using computer-controlled
commutators for increasing its automation and miniaturization.
The determination method of the proposed system was based on
the monitoring of the blue coloration decrease of a buffered BTB
solution in the presence of acidic compounds, using an inexpensive,
miniaturized and versatile optical detection system for performing
the absorbance measurements. In order to demonstrate the appli-
cability of the microsystem, several wine samples were analyzed.
2. Experimental
2.1. Reagents and materials
The microanalyzer was constructed by using COC layers from
TOPAS Advanced Polymers (Florence, KY, USA) of diverse thick-
nesses and grades: two 1 mm and one 400 mm Topas® 5013 COC
layers and two 25 mm Topas® 8007 COC layers.
All used chemicals were of analytical reagent grade. All solutions
were prepared in MilliQ water and were degassed prior to use. BTB
(Sigma-Aldrich, Saint Louis, MO, USA) indicator solutions were
prepared by accurate weighing and dissolution in phosphate buffer
aqueous solutions. Potassium phosphate monobasic (KH2PO4,
Sigma-Aldrich) and potassium phosphate dibasic anhydrous
(K2HPO4, Fluka, Buchs, Switzerland) were used for preparing the
buffer solutions. Standard solutions of L-(þ)-tartaric acid (Sigma-
Aldrich) were prepared by dilution of a stock standard solution of
5 g L1. All tested samples were commercially available. No pre-
treatment, either degassing or dilution, was required prior to
analysis. For the reference method, a 0.1 M sodium hydroxide
(NaOH, Fisher, Pittsburgh, PA, USA) aqueous solution was used as
titrant. The solution was previously standardized by using potas-
sium hydrogen phthalate (KHP, Sigma-Aldrich).
2.2. Fabrication of the microanalyzer
The description of the general multilayer microfabrication pro-
cess used for the construction of the COC-based microanalyzer can
be found elsewhere [35]. The process is based on the bonding of
several micromachined COC layers of different glass transition
temperatures (Tg) that, when bonded, create the final internal 3D
structures. For the device here described, Topas® 5013 COC plaques
(Tg ¼ 130  C) were used as structural layers and contained the
fluidic motifs, whereas Topas® 8007 COC foils (Tg ¼ 75  C) were
used as bonding layers. The fabrication process can be divided into
52
ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa
three main steps: the design of the device, the machining of the
layers and the bonding of the microstructured layers (Fig. 1). The
design was performed by using CAD software and consisted of
three different layers (Fig. 2A): the central layer, which contained
the main fluidic structures, the top layer, which contained inlet and
outlet holes, and the bottom layer, which did not contain any fluidic
feature. The fluidic motifs were machined onto the substrate by
using a computer numerically controlled (CNC) micromilling ma-
chine (Protomat C100/HF, LPKF Laser & Electronics, Germany). The
bonding of the layers was performed in a uniaxial hydraulic press
(Francisco Camp, Granollers, Spain) at 102  C and 6 bar. An
aluminum support with four alignment pins ensured the precise
alignment of the layers and therefore the accurate structuring of
the internal fluidic motifs. The integration of the fluidic connections
was performed once the construction of the microanalyzer was
finished (Fig. 2B). The final dimensions of the device were
30 mm  50 mm x 2.5 mm, and its final total weight, without the
fluidic connections, was of approximately 3 g; the total dead vol-
ume of the device was 326 mL.
The microanalyzer was designed to have three different chan-
nels for the introduction of fluids. Liquids entering the two first
channels merged at a Y-shaped confluence point and were then
mixed in a first short serpentine micromixer. A second confluence
point connected the third channel with the previous micromixer.
This time, a longer serpentine channel ensured the complete
mixture of the fluids. The length of the micromixers was based in
previous works [36]; simple verification experiments were per-
formed here. All the channels for liquid introduction as well as the
two serpentine micromixers were 1.0 mm in height and 0.8 mm in
width.
The second micromixer ended into an optical detection chamber
of 4.5 mm in diameter and 1.0 mm in height (i.e. 1.0 mm optical
path length). The chamber was designed in order to minimize the
most common problems related to the presence and the circulation
of gas bubbles while monitoring absorbance, such as an increase in
the noise and the appearance of transitory signals or baseline drifts
[36]. The smooth contours of the cell prevented both the formation
and accumulation of bubbles, which are usually favored by abrupt
changes in the geometry of the fluidic features [37]. Besides, an
increment in the flow rate at the exit of the chamber was used for
favoring bubble evacuation in case any was produced. The change
in the flow rate was achieved by a drastic decrease in the cross-
section dimensions of the outlet channel in comparison to the
detection chamber. Liquid flowed outside the device through a
single outlet channel of 1.0 mm in height and 0.4 mm in width. The
dimensions of this channel were smaller than those from the inlet
channels and the mixing channels in order to produce the afore-
mentioned change in the flow rate at the exit of the detection
chamber.
The external shape of the device was designed to perfectly fit
the insertion port of the detection system (see 2.3 Experimental
Fig. 1. Main steps of the fabrication process.
setup), thus guaranteeing an accurate and reproducible alignment
of the optical flow cell with respect to the LED and the photodiode
[38].
2.3. Experimental setup
The complete system setup (Fig. 3) can be divided into three
main parts: the fluid management peripherals, the microanalyzer
and the optical detection system. The fluid management periph-
erals consisted of a four-channel peristaltic pump (Gilson Minipuls
3, Gilson, WI, USA) equipped with 0.64 mm internal diameter
Tygon tubing (Ismatec, Wertheim, Germany) and four three-way
solenoid valves (161T031, NResearch, West Caldwell, NJ, USA).
Teflon tubing of 0.80 mm internal diameter (Scharlab SL, Setmenat,
Spain) was used for the fluidic connections between the periph-
erals. An automated controller (FlowTest, BioTray, Villeurbanne,
France) was used for the actuation of the solenoid valves. Pro-
gramming of the actuation of the valves was executed in a personal
computer (PC) by using the dedicated CosDesigner software.
The optical detection system used was previously developed by
our research group [39] and subsequently adapted to this work. The
system consisted of a LED with an emission peak centered at
621 nm (HLMP-EH1A, Avago, Digi-Key Electronics, Thief River Falls,
MN, USA) and a Si photodiode with an effective area of 33 mm2
(S1337-66BR, Hamamatsu Photonics, Hamamatsu, Japan), both
mounted into a compact poly(methyl methacrylate) (PMMA)
structure where the microanalyzer was inserted into. The insertion
structure was based in a “Lock-and-Key” concept [38] for allowing a
reproducible positioning of the device with respect to the LED and
the photodiode (Fig. 4). The LED and the photodiode were con-
nected to a printed circuit board (PCB) that, in turn, was connected
to a data acquisition card (DAQ) (NI USB-6211, National In-
struments, Austin, TX, USA). The DAQ was responsible for the
modulation of the LED and the acquisition and transference of the
detected signal to a PC. A digital lock-in amplification was used for
processing the raw data, increasing the signal-to-noise ratio and
permitting the operation of the system in ambient light conditions
without requiring any physical amplifier.
2.4. Flow manifold
The operating procedure of the analysis consisted in the
continuous pumping of a buffered BTB indicator solution into the
microanalyzer through channel A (see Fig. 3), which was in-line
diluted by the continuous pumping of H2O through channel B.
Channel C was used for the introduction of the stock standard so-
lution and the samples. Fixed volumes of these solutions were
injected and in-line diluted through the actuation of the three-way
solenoid microvalves. When the acidic solutions were mixed with
the diluted indicator solution inside the microanalyzer, a change in
the color of the latter was produced. The mixture was directed

ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa 53

Fig. 2. A: CAD design of the structural layers (Topas® 5013 COC) of the microanalyzer. Layers a and b are of 1 mm, whereas layer c is of 400 mm. Topas® 8007 COC foils of 25 mm are
used as bonding layers and are placed in between the structural plaques. B: Photograph of the final device: a) fluidic connections; b) microfluidics; c) optical detection chamber.

Fig. 3. Schematic illustration of the experimental setup. PP: peristaltic pump; W: waste; Vx: three-way solenoid microvalve.

towards the detection chamber, where the optical detection system
measured and monitored the absorbance changes. In order to avoid
interference from the sample species absorbing at the same
wavelength, a blank was measured for each sample. The mea-
surement of the blank was performed by the automated substitu-
tion of the buffered BTB solution by a phosphate buffer solution
(blank solution) using a three-way solenoid microvalve located at
the reagent solution stream.
3. Results and discussion
3.1. Design and optimization of the analytical microsystem
The main goal of the work here presented was the development
of a dedicated analytical microsystem for the determination of
titratable acidity in wine requiring no pretreatment of the samples
and that could be easily applied to the on-line monitoring of this
54
ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa
Fig. 4. Illustration of the “Lock-and-Key” insertion structure of the optical detection
system. A: Exploded view, showing the photodiode (a), the LED (b), the several PMMA
layers that constitute the polymeric structure (c) and the four bolts that secure it (d). B:
Reversible positioning of the device, perfectly fitting the insertion port and thus
permitting an accurate and reproducible alignment of the optical detection chamber
with respect to the LED and the photodiode.
parameter during wine-making processes. In order to achieve such
purposes, a wide working range along with a high degree of
automation and robustness of the system were necessary. The
different parameters that could influence the system, including the
flow rate, the injection volume, the indicator solution composition
and the microvalves actuation, among others, were studied and
optimized using a univariate optimization procedure.
3.1.1. Optimization of the chemical and hydrodynamic variables
Several acid-base indicators have been reported for the deter-
mination of titratable acidity in wine, dedicating special attention
to phenolphthalein [8,9,12,18,19] and bromothymol blue [15e17],
which are the two indicators used in the Organisation Internationale
de la Vigne et du Vin (OIV) [5] and the Association of Official
Analytical Chemists (AOAC) [6] official methods, respectively. Due
to the intrinsic coloration of wine, the use of BTB as indicator was
here preferred, thus being the analysis based on the monitoring of
the decrease in the blue coloration of the indicator when mixed
with the acidic standard/sample solutions. The experimental
determination of the linear range for the fading of the blue color-
ation was carried out and it was established for pH values ranging
from 7.8 to 6.9. The pH of the indicator solution was therefore
adjusted to 7.8 during the whole work.
Tartaric acid was selected as the model acid for preparing the
standard solutions since it is the main contributor to titratable
acidity in musts and wines [1,2]. The targeted minimum working
range was established from 0.05 g L1 to 0.50 g L1, which corre-
sponded to a titratable acidity from 1.25 g L1 to 12.5 g L1
expressed as tartaric acid for a 25-fold dilution of the samples. This
range would cover the most common titratable acidity content of
musts and wines, being from 6 g L1 to 9 g L1 and from 4 g L1 to
6.5 g L1 tartaric acid, respectively [40].
Initially, the indicator solution used consisted in an 8 mg L1 BTB
aqueous solution adjusted to pH 7.8. With these conditions, mini-
mal injection volumes of highly diluted tartaric acid solutions were
capable of producing the total color change of the indicator from
blue to yellow. Therefore, in order to increase the resistance of the
indicator solution to change its pH, BTB solutions were prepared in
phosphate buffer solutions [41]. Predictably, an increment in the
buffer capacity of the indicator solution led to a reduction in the
sensibility but a noticeable increase in the linear range. The BTB
concentration was also studied, obtaining an improvement in both
sensibility and linear range by increasing the BTB concentration.
From the several compositions evaluated, the minimal reagent
concentrations which covered the targeted linear range were
500 mg L1 BTB and 35 mM phosphate buffer.
The increase in the buffer capacity of the indicator solution was
not only useful for broadening the linear range but also for
increasing the stability of the indicator solution in terms of pH. A
highly stable pH of the indicator solution would actually be
mandatory for the application of the microsystem to the on-line
monitoring of titratable acidity during wine-making processes.
The stability of several BTB solutions adjusted to pH 7.8 and
differing on their phosphate buffer content was studied. The pH of
these solutions was measured during 10 days, showing a slow
progressive acidification due to the dissolution of CO2 from ambient
air. Despite the pH variation was modest, a minor loss in pH could
have a relevant effect on the working range of the determination,
especially when considering the reduced working range of the in-
dicator (from pH 7.8 to pH 6.9). The obtained results demonstrated
that BTB indicator solutions with a phosphate buffer concentration
of minimum 25 mM would be required for less than a 5% loss in the
working range during 10 days. The indicator solution selected as
optimal was therefore expected to be stable for at least 10 days,
since its composition involved a phosphate buffer concentration
higher than 25 mM.
The evaluation of the indicator solution composition was per-
formed by using the previously optimized flow rate and injection
volume, both selected by considering the requirements for a proper
system functioning. Limitations due to the use of a multi-
commutation scheme for the dilution of standards and samples
were taken into consideration (see 3.1.2 Optimization of the fluid
management peripherals). 434 mL min1 per channel was chosen as
the optimum flow rate, demonstrating the best compromise be-
tween the baseline noise (due to the inherent pulsating flow of the
peristaltic pump) and the required value for a feasible and suitable
actuation of the microvalves. The selection of the optimal injection
volume, 289 mL, was mainly based on the required injection time
for the proper automated 25-fold in-line dilution of the sample at
the optimal flow rate. The injection time which corresponded to an
injection volume of 289 mL at a flow rate of 434 mL min1 was 40.0 s.
The initial conditions, the evaluated ranges and the selected
values for the studied parameters are summarized in Table 1.
3.1.2. Optimization of the fluid management peripherals actuation
As mentioned before, several three-way solenoid microvalves
Table 1
Optimization of the chemical and hydrodynamic variables.
Variable Initial value Tested range Optimum value
Injection volume (mL) 7.5 7.5e486 289
Channel flow rate (mL min1) 750 434e1500 434
Stock standard solution:
[tartaric acid] (g L1) 0.5 e 0.5
Indicator solution:
pH 7.8 e 7.8
[BTB] (mg L1) 8 8e500 500
[Phosphate buffer] (mM) 0 0e35 35
Blank solution:
pH 7.8 e 7.8
[Phosphate buffer] (mM) 35 e 35

ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa
were implemented to the microsystem in order to increase its
automation and miniaturization. The valves were programmed for
the automatic calibration of the system as well as for the automatic
sample analysis, including its in-line dilution, by using a multi-
commutation approach. The selected controller permitted the
programming of the actuation of the microvalves by means of very
simple and intuitive PC software and, once the actuation program
was transferred to the apparatus, it could be executed with no
further need of the controller to be connected to the PC.
An example of a complete operation sequence of the valves for
both calibration and sample analysis can be found in Fig. 5 (see also
Fig. 3 for better understanding). After automatically filling the
system with the required solutions (not included in Fig. 5), the
autocalibration was carried out. The process consisted in the
analysis of several standard solutions produced by the in-line
dilution of a single stock standard solution by the actuation of V1.
Steps a to c exemplify the injection of the stock standard solution
being diluted to 1:10 (a), diluted to 1:2 (b) and without dilution (c).
The analysis of a sample required from several steps, starting by the
opening of V3 while the sample was being introduced to the system
up to V2 (d). After cleaning the system by the introduction of H2O
up to V3 (e), the latter was closed (f). Once the baseline was
recovered, the sample was injected and in-line diluted to 1:25 by
the actuation of V2 (g). In order to perform the measurement of the
blank, the buffered BTB solution was automatically substituted by a
phosphate buffer solution by the actuation of V4 (h). Once the new
baseline was reached, the sample was injected and in-line diluted
to 1:25 anew (i).
Several valves were tested for the automatic dilution of stan-
dards and samples using different dilution profiles; the obtained
results were compared to the ones obtained by the injection of
manually prepared standard and sample solutions. In order to set
the appropriate dilution sequence for the autocalibration, the
limitations of the microvalves were considered. The precision of the
dilution process was dependent on the accuracy of the commuta-
tion of the valves as well as on the stability of the flow rate, being
0.2 s the lowest commutation speed used which gave an acceptable
reproducibility of the dilutions (relative standard deviation,
RSD < 5%). The number of commutation cycles per injection was set
to 20 for the standard dilutions in order to homogenize the possible
error coming from the commutation of the valves. Since the dilu-
tion required for the sample (1:25) was significantly higher than
Fig. 5. Operation sequence of the valves for both calibration and sample analysis. V1 and V2 were in charge of the dilution of the stock standard solution and the sample,
respectively; V3 was used for discarding waste solutions before entering the microanalyzer; and V4 permitted the selection between the BTB indicator solution and the phosphate
buffer solution. The times here represented are for guidance purposes only.
55
the one for the standard solutions (maximum 1:10), the use of the
same number of cycles was not affordable. Thus, a parallel experi-
ment for assessing the optimum actuation for the dilution of the
sample was performed. Despite no statistically significant differ-
ence for a 95% confidence level (n ¼ 5) was found between the
average peak height of two different replicates for minimum
commutation speeds ranging from 0.1 s to 0.4 s, the repeatability
was considerably higher for cycles with minimum commutation
speeds of 0.2 s and 0.3 s. On the other hand, when analyzing the
absorbance peak profiles, an incomplete mixture of the sample-
H2O plugs could be noticed for cycles of 0.3 s “on” - 7.2 s “off”,
becoming more evident for cycles with higher minimum commu-
tation times. Consequently, 0.2 s “on” - 4.8 s “off” was selected as
the optimum commutation cycle for the sample dilution. The
optimized actuation of the valves for the dilution of the stock
standard solution and the sample can be found in Table 2.
The actuation of the microvalves was evaluated by comparing a
calibration curve obtained by the automated injection and dilution
of the stock standard solution with a calibration curve where the
diluted standard solutions were manually prepared. A synthetic
sample of 5 g L1 tartaric acid which was in-line diluted to 1:25 (i.e.
to 0.2 g L1) and a manually diluted sample of 0.2 g L1 tartaric acid
were analyzed. No statistically significant differences were found
between the two calibration curves (n ¼ 18, 95% confidence level)
nor between the obtained results for the in-line and the manually
diluted sample (n ¼ 5, 95% confidence level), thus demonstrating
an excellent performance of the microvalves. Repeatability of the
actuation of the valves was also assessed by performing successive
automated injections (n ¼ 10) of different tartaric acid standard
solutions, all of them showing RSD values below 5%: 2.3% for the
0.05 g L1, 0.5% for the 0.20 g L1 and 0.8% for the 0.50 g L1. A
synthetic sample of 5 g L1 tartaric acid was also injected and in-
line diluted to 1:25, obtaining an RSD of 1.2% (n ¼ 10).
Dispersion of the injected plugs in flow-based systems increases
with the increase of the distance from the injection point to the
detection point [42]. In order to minimize the difference between
the dispersion of the standard plugs and the sample plugs, the
distance between V1 and V2 (see Fig. 3) was kept to the minimum
affordable one. Two different calibration experiments were per-
formed: the first consisted in using V1 for the dilution of the stock
standard solution and V2 for the dilution of the sample; the second
was performed by inverting the roles of the valves V1 and V2. No
56
ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa

Table 2
Optimized actuation program for obtaining the diluted standard solutions from the stock standard solution and for the sample injection and in-line dilution.

Injection time (s) ton (s) toff (s) Number of cycles Dilution factor Stock standard solution (g L1) Diluted solution (g L1)

Standard solutions
40.0 0.2 1.8 20 1:10 0.50 0.05
0.4 1.6 1:5 0.10
0.8 1.2 2:5 0.20
1.2 0.8 3:5 0.30
1.6 0.4 4:5 0.40
40.0 e e e 0.50
Sample
40.0 0.2 4.8 8 1:25 e e

statistically significant differences were found between the cali- sample, 500 mg of BTB and 35 mmol of phosphate and producing
bration curves (n ¼ 18, 95% confidence level) nor the sample anal- 3 mL of waste per assay.
ysis results (n ¼ 5, 95% confidence level), thus demonstrating no
significant effect of the distance between the two valves.
3.3. Real sample analysis

3.2. Analytical performance
The repeatability of the method was evaluated by performing
three different calibration analyses in three different working days
under identical hydrodynamic and chemical conditions and using
newly prepared solutions. Standards ranging from 0.05 g L1 to
0.50 g L1 were automatically analyzed in triplicate by using the
aforementioned optimal conditions (see Tables 1 and 2). Fig. 6
displays the recorded signal and the calibration curve obtained
for one of the experiments. No statistically significant differences
(n ¼ 18, 95% confidence level) were found between the three cali-
bration replicates, obtaining a mean slope value of 0.472 with a RSD
of 1.2% and therefore demonstrating an excellent between-day
repeatability. Detection and quantification limits were calculated
as 3 times and 10 times the standard deviation of the baseline,
respectively, obtaining mean values of 0.004 g L1 tartaric acid
(LOD) and 0.01 g L1 tartaric acid (LOQ). With the optimal condi-
tions, a sampling rate of 26 h1 was achieved, consuming 289 mL of
The proper performance of the analytical microsystem was
assessed by determining the titratable acidity of twelve different
samples: five white wines, two rose
wines and five red wines. The
results were compared to those obtained by analyzing the samples
using the AOAC reference method. For the AOAC method, 5 mL of
degassed wine were added to 100 mL MilliQ water and were then
titrated to pH 8.2 against a 0.1 M NaOH aqueous solution. The
titrant solution had been previously standardized using KHP and
phenolphthalein as pH indicator.
For the titratable acidity determination using the developed
microsystem, the samples were injected in quintuplicate and in-line
diluted to 1:25. In order to eliminate the interference from the
colored species, a blank was performed for each sample by following
the same process but automatically substituting the buffered BTB
solution (indicator solution) by a phosphate buffer solution (blank
solution). The absorbance of the blank for red wine samples was
considerably high, whereas rose
wine samples showed a relatively
low absorbance; no signal was observed for white wine samples.

Fig. 6. Recorded signal and calibration curve for one of the calibration experiments using the optimal conditions. A multicommutation dilution program was used for the in-line
preparation of standard solutions of (a) 0.05 g L1, (b) 0.10 g L1, (c) 0.20 g L1, (d) 0.30 g L1, (e) 0.40 g L1 and (f) 0.50 g L1 tartaric acid.

ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa 57

Table 3

and red wine samples by
Titratable acidity (TA) of the wine samples determined by the reference method and the developed method. The titratable acidity content of rose
applying the color correction is also included.

Sample Wine type Reference method Developed method

Color correction No color correction

1 1
TA (g L tartaric acid) TA (g L tartaric acid) % Error TA (g L1 tartaric acid) % Error

1 Red 5.7±0.1 5.6±0.2 1.8 5.4±0.2 5.3

2 Red 5.1±0.1 5.1±0.1 0.0 4.8±0.1 5.9
3 Red 4.9±0.2 4.9±0.2 0.0 4.5±0.2 8.2
4 Red 5.0±0.2 4.87±0.06 2.6 4.66±0.06 6.8
5 Red 5.5±0.1 5.3±0.1 3.6 5.0±0.1 9.1
6

Rose 4.4±0.1 4.2±0.2 4.5 4.2±0.2 4.5
7

Rose 6.0±0.1 5.9±0.1 1.7 5.8±0.1 3.3
8 White 4.5±0.1 4.62±0.09 2.7
9 White 5.5±0.1 5.44±0.07 1.1
10 White 5.9±0.2 5.7±0.2 3.4
11 White 6.2±0.2 6.21±0.08 0.2
12 White 6.6±0.1 6.85±0.07 3.8

The obtained results can be found in Table 3. A low relative error
(below 5%) was obtained for all the analyzed samples, thus
demonstrating the excellent performance of the microsystem.
Whilst rose
samples did not show a noticeable improvement in the
relative error by applying the color correction, the results obtained
for red wine samples were substantially improved by measuring
the blank. Nonetheless, relative error values below 10% were ob-
tained for all the red wine samples without applying the color
correction. A 10% relative error could be acceptable for rapid
routine analysis and it would avoid the need of one of the micro-
valves (V4, see Fig. 3), thus simplifying the system and reducing its
size and cost, apart from reducing the analysis time and the reagent
and sample consumption.
The results obtained by the two methods were compared by
applying a paired t-test. No statistically significant difference was
found between the reference method and the developed method:
tcalc (1.2852) < ttab (2.2010) for n ¼ 12 and 95% confidence level. A
linear relationship between the two methods could be established,
being TA Developed method ¼ (1.1 ± 0.1) $ TA Reference
2 NORTE-01-0145-FEDER-000030.
method þ (0.3 ± 0.8), R ¼ 0.9668, for n ¼ 12 and 95% confidence
level.
The developed microsystem demonstrated an excellent perfor-
mance when compared to other similar systems which can be
found in the literature, such as permitting the analysis of highly
colored wines [10] or avoiding any sample pretreatment [8,14,17].
The advantages of the system here described in comparison to
other previously reported systems include a wider working range
[15,16] as well as the use of a simpler manifold [7,9e11,13,18] and
cheaper instrumentation [12,19], among others.
4. Conclusions
A miniaturized analytical microsystem for the spectrophoto-
metric monitoring of titratable acidity in wine has been developed
and successfully applied to the analysis of several wine samples.
The system has demonstrated an excellent throughput as well as a
wider working range, a reduced sample and reagent consumption
and a higher miniaturization and automation in comparison to
other previously reported systems. The simple, miniaturized and
low-cost system here presented involves no sample pretreatment
and produces minimal waste. Besides, the complete automation of
the analytical process, including calibration and sampling, has been
achieved, therefore eliminating any human intervention. Its per-
formance has been compared to the AOAC official method, proving
an adequate accuracy and precision of the measurements. The high
robustness, miniaturization and automation of the system would
permit its application to the on-line continuous monitoring of
titratable acidity in wineries, thus contributing to a better control of
the wine-making process and a higher quality of the final product.
Declaration of competing interest
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
Acknowledgements
lia S

Nata andez acknowledges the financial support from the
MINECO-FEDER through the project CTQ2017-85011-R and from
the Generalitat de Catalunya through the project 2017 SGR 220.
Susana Vidigal acknowledges the financial support of Fundo
Europeu de Desenvolvimento Regional (FEDER), under Programa
Operacional Regional do Norte - Norte2020 through the project
References
[1] R.S. Jackson, Wine Science: Principles and Applications, third ed., Academic
Press, Elsevier, Burlington, MA, USA, 2008 https://doi.org/10.1016/B978-0-12-
381468-5.00006-3.
[2] P. Ribe
reau-Gayon, Y. Glories, A. Maujean, D. Dubourdieu, Handbook of
Enology Volume 2: the Chemistry of Wine Stabilization and Treatments,
second ed., John Wiley & Sons, Ltd, Chichester, UK, 2006 https://doi.org/
10.1002/0470010398.
[3] R.A. Plane, L.R. Mattick, L.D. Weirs, An acidity index for the taste of wines, Am.
J. Enol. Vitic. 31 (1980) 265e268.
[4] M. Amerine, E.B. Roessler, C.S. Ough, Acids and the acid taste. I. The effect of
pH and titratable Acidity, Am. J. Enol. Vitic. 16 (1965) 29e37.
[5] Total acidity (method OIV-MA-AS313-01), in: Compend. Int. Methods Wine
Must Anal., International Organisation of Vine and Wine, Paris, France, 2018.
[6] W. Horwitz, G.W. Latimer (Eds.), Official Methods of Analysis of AOAC IN-
TERNATIONAL, eighteenth ed., AOAC INTERNATIONAL, Gaithersburg, MD,
USA, 2010.
[7] T.J. Cardwell, R.W. Cattrall, G.R. O’Connell, J.D. Petty, G.R. Scollary, Potentio-
metric detectors for the determination of acidity by discontinuous flow
analysis, Electroanalysis 4 (1992) 805e810, https://doi.org/10.1002/
elan.1140040811.
[8] M.J.A. Lima, B.F. Reis, Fully automated photometric titration procedure
employing a multicommuted flow analysis setup for acidity determination in
fruit juice, vinegar, and wine, Microchem. J. 135 (2017) 207e212, https://
doi.org/10.1016/j.microc.2017.09.016.
[9] A.J.C. Garcia, B.F. Reis, Instrumentation and automated photometric titration
procedure for total acidity determination in red wine employing a multi-
commuted flow system, J. Autom. Methods Manag. Chem. (2006) 1e8, https://
doi.org/10.1155/JAMMC/2006/83247 (2006).
[10] R.S. Honorato, M.C.U. Araújo, R.A.C. Lima, E.A.G. Zagatto, R.A.S. Lapa, J.L.F. Costa
Lima, A flow-batch titrator exploiting a one-dimensional optimisation algo-
rithm for end point search, Anal. Chim. Acta 396 (1999) 91e97, https://
58
ndez et al. / Analytica Chimica Acta 1091 (2019) 50e58
N. Sa
doi.org/10.1016/S0003-2670(99)00366-9.
[11] E.N. Gai~ ao, R.S. Honorato, S.R.B. Santos, M.C.U. Araújo, An automated flow-
injection titrator for spectrophotometric determinations of total acidity in
wines, using a single standard solution and gradient calibration, Analyst 124 [28]
(1999) 1727e1730, https://doi.org/10.1039/a905188k.
[12] J. Marcos, A.
Ios, M. Valc
arcel, Automatic titrations in unsegmented flow
systems based on variable flow-rate patterns, Anal. Chim. Acta 261 (1992)
489e494, https://doi.org/10.1016/0003-2670(92)80230-5. [29]
[13] T.J. Cardwell, R.W. Cattrall, G.J. Cross, G.R. O’Connell, J.D. Petty, G.R. Scollary,
Determination of titratable acidity of wines and total acidity of vinegars by
discontinuous flow analysis using photometric end-point detection, Analyst [30]
116 (1991) 1051e1054, https://doi.org/10.1039/an9911601051.
[14] K. Vahl, H. Kahlert, L. von Mühlen, A. Albrecht, G. Meyer, J. Behnert, Deter-
mination of the titratable acidity and the pH of wine based on potentiometric [31]
flow injection analysis, Talanta 111 (2013) 134e139, https://doi.org/10.1016/
j.talanta.2013.02.057.
[15] S.S.M.P. Vidigal, A.O.S.S. Rangel, A flow-based platform for measuring the
acidity parameters in wine, Talanta 168 (2017) 313e319, https://doi.org/ [32]
10.1016/j.talanta.2017.03.029.
[16] S.S.M.P. Vidigal, A.T.C. Ramos, A.O.S.S. Rangel, Flow-based system for the
determination of titratable Acidity in wines, Food Anal. Methods 9 (2016) [33]
2241e2245, https://doi.org/10.1007/s12161-016-0408-7.
[17] E. Mataix, M. Luque de Castro, Sequential determination of total and volatile
acidity in wines based on a flow injection-pervaporation approach, Anal.
Chim. Acta 381 (1999) 23e28, https://doi.org/10.1016/S0003-2670(98)00677- [34]
1.
[18] A.O.S.S. Rangel, I.V. To
th, Sequential determination of titratable acidity and
tartaric acid in wines by flow injection spectrophotometry, Analyst 123
(1998) 661e664, https://doi.org/10.1039/a707477h.
[19] M. Peris-Tortajada, A. Maquieira, R. Puchades, Automated determination of
total acidity in wines by flow injection analysis, Am. J. Enol. Vitic. 44 (1993) [35]
118e120.
[20] J.L.F.C. Lima, J.L.M. Santos, A.C.B. Dias, M.F.T. Ribeiro, E.A.G. Zagatto, Multi-
pumping flow systems: an automation tool, Talanta 64 (2004) 1091e1098, [36]
https://doi.org/10.1016/j.talanta.2004.04.034.
[21] J.L.M. Santos, M.F.T. Ribeiro, A.C.B. Dias, J.L.F.C. Lima, E.E.A. Zagatto, Multi-
pumping flow systems: the potential of simplicity, Anal. Chim. Acta 600 [37]
(2007) 21e28, https://doi.org/10.1016/j.aca.2007.01.029.
[22] F.R. Rocha, B.F. Reis, E.A. Zagatto, J.L.F. Lima, R.A. Lapa, J.L. Santos, Multi-
commutation in flow analysis: concepts, applications and trends, Anal. Chim. [38]
Acta 468 (2002) 119e131, https://doi.org/10.1016/S0003-2670(02)00628-1.
[23] M.A. Feres, P.R. Fortes, E.A.G. Zagatto, J.L.M. Santos, J.L.F.C. Lima, Multi-com-
mutation in flow analysis: recent developments and applications, Anal. Chim.
Acta 618 (2008) 1e17, https://doi.org/10.1016/j.aca.2008.04.039. [39]
[24] M. Catala
Icardo, J.V. García Mateo, J. Martínez Calatayud, Multicommutation
as a powerful new analytical tool, TrAC Trends Anal. Chem. (Reference Ed.) 21
(2002) 366e378, https://doi.org/10.1016/S0165-9936(02)00505-8.
[25] P.S. Nunes, P.D. Ohlsson, O. Ordeig, J.P. Kutter, Cyclic olefin polymers: [40]
emerging materials for lab-on-a-chip applications, Microfluid. Nanofluidics 9
(2010) 145e161, https://doi.org/10.1007/s10404-010-0605-4. [41]
[26] F. Bundgaard, G. Perozziello, O. Geschke, Rapid prototyping tools and methods
for all-Topas® cyclic olefin copolymer fluidic microsystems, Proc. Inst. Mech.
Eng. C J. Mech. Eng. Sci. 220 (2006) 1625e1632, https://doi.org/10.1243/
09544062JMES295. [42]
[27] J. Steigert, S. Haeberle, T. Brenner, C. Müller, C.P. Steinert, P. Koltay,
N. Gottschlich, H. Reinecke, J. Rühe, R. Zengerle, J. Ducre
e, Rapid prototyping
of microfluidic chips in COC, J. Micromech. Microeng. 17 (2007) 333e341,
https://doi.org/10.1088/0960-1317/17/2/020.
A. Calvo-Lo
pez, M. Puyol, J.M. Casalta, J. Alonso-Chamarro, Multi-parametric
polymer-based potentiometric analytical microsystem for future manned
space missions, Anal. Chim. Acta 995 (2017) 77e84, https://doi.org/10.1016/
j.aca.2017.08.043.
O. Ymbern, N. Sa
ndez, A. Calvo-Lo
pez, M. Puyol, J. Alonso-Chamarro, Gas
diffusion as a new fluidic unit operation for centrifugal microfluidic platforms,
Lab Chip 14 (2014) 1014e1022, https://doi.org/10.1039/c3lc51114f.
P.K. Dasgupta, I.-Y. Eom, K.J. Morris, J. Li, Light emitting diode-based detectors,
Anal. Chim. Acta 500 (2003) 337e364, https://doi.org/10.1016/S0003-
2670(03)00575-0.
F.-B. Yang, J.-Z. Pan, T. Zhang, Q. Fang, A low-cost light-emitting diode induced
fluorescence detector for capillary electrophoresis based on an orthogonal
optical arrangement, Talanta 78 (2009) 1155e1158, https://doi.org/10.1016/
j.talanta.2009.01.033.
K.M.G. de Lima, A portable photometer based on LED for the determination of
aromatic hydrocarbons in water, Microchem. J. 103 (2012) 62e67, https://
doi.org/10.1016/j.microc.2012.01.003.
M. O’Toole, K.T. Lau, R. Shepherd, C. Slater, D. Diamond, Determination of
phosphate using a highly sensitive paired emitteredetector diode photo-
metric flow detector, Anal. Chim. Acta 597 (2007) 290e294, https://doi.org/
10.1016/j.aca.2007.06.048.
F. Pujol-Vila, P. Gime
nez-Go
mez, N. Santamaria, B. Antúnez, N. Vigue
s,
M. Díaz-Gonz
alez, C. Jime
nez-Jorquera, J. Mas, J. Sacrist

an, X. Mun~ oz-Berbel,
Portable and miniaturized optofluidic analysis system with ambient light
correction for fast in situ determination of environmental pollution, Sens.
Actuators B Chem. 222 (2016) 55e62, https://doi.org/10.1016/
j.snb.2015.07.095.
O. Ymbern Llorens, Development of Centrifugal Microfluidic Platforms Based
on Polymer Microfabrication Technology, Universitat Auto  noma de Barcelona,
2015.
A. Calvo Lo
pez, Disen ~ o, construccio

n y evaluacio

n de analizadores miniatur-
izados para su aplicacio
n aeroespacial, medioambiental, alimentaria, bio-
me
dica e industrial, Universitat Auto noma de Barcelona, 2017.
S. Burger, M. Schulz, F. von Stetten, R. Zengerle, N. Paust, Rigorous buoyancy
driven bubble mixing for centrifugal microfluidics, Lab Chip 16 (2016)
261e268, https://doi.org/10.1039/C5LC01280E.
O. Ymbern, M. Berenguel-Alonso, A. Calvo-Lo
pez, S. Go
mez-de Pedro,
D. Izquierdo, J. Alonso-Chamarro, Versatile lock and key assembly for optical
measurements with microfluidic platforms and cartridges, Anal. Chem. 87
(2015) 1503e1508, https://doi.org/10.1021/ac504255t.
S.G. Pedro, M. Puyol, D. Izquierdo, I. Salinas, J.M. de la Fuente, J. Alonso-Cha-
marro, A ceramic microreactor for the synthesis of water soluble CdS and CdS/
ZnS nanocrystals with on-line optical characterization, Nanoscale 4 (2012)
1328, https://doi.org/10.1039/c2nr11525e.
C.S. Ough, M.A. Amerine, Methods for Analysis of Musts and Wines, second
ed., John Wiley & Sons, Ltd, New York, NY, USA, 1988.
A. Calvo-Lo
pez, O. Ymbern, D. Izquierdo, J. Alonso-Chamarro, Low cost and
compact analytical microsystem for carbon dioxide determination in pro-
duction processes of wine and beer, Anal. Chim. Acta 931 (2016) 64e69,
https://doi.org/10.1016/j.aca.2016.05.010.
V. Cerda, L. Ferrer, J. Avivar, A. Cerd
a, Flow Analysis: A Practical Guide, first ed.,
Elsevier, Amsterdam, Netherlands, 2014.