This is the accepted version of the journal article: Calvo-Lopez, Antonio; Ymbern, Oriol; Izquierdo, David; [et al.]. «Low cost and compact analytical microsystem for carbon dioxide determination in production processes of wine and beer». Analytica chimica acta, Vol. 931 (August 2016), p. 64-69. DOI 10.1016/j.aca.2016.05.010 This version is available at https://ddd.uab.cat/record/288013 under the terms of the license Accepted Manuscript Low cost and compact analytical microsystem for carbon dioxide determination in production processes of wine and beer Antonio Calvo-López, Oriol Ymbern, David Izquierdo, Julián Alonso-Chamarro PII: S0003-2670(16)30605-5 DOI: 10.1016/j.aca.2016.05.010 Reference: ACA 234576 To appear in: Analytica Chimica Acta Received Date: 20 April 2016 Accepted Date: 6 May 2016 Please cite this article as: A. Calvo-López, O. Ymbern, D. Izquierdo, J. Alonso-Chamarro, Low cost and compact analytical microsystem for carbon dioxide determination in production processes of wine and beer, Analytica Chimica Acta (2016), doi: 10.1016/j.aca.2016.05.010. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT PT RI U SC Color change of pH indicator Wine and beer CO2 monitoring CO2 (g) AN In- (aq) HIn (aq) 0.00000 500 CO2 diffusion 1000 Absorbance (AU) -0.01000 M 2500 -0.02000 0 100 CO2 detection D 5000 -0.03000 mg L-1 CO2 TE 7500 -0.04000 10000 -0.05000 EP 0 500 1000 1500 2000 2500 3000 Microanalyzer Time (s) C AC ACCEPTED MANUSCRIPT 1 Low cost and compact analytical microsystem for carbon dioxide 2 determination in production processes of wine and beer 3 Antonio Calvo-Lópeza, Oriol Ymberna, David Izquierdob,c and Julián Alonso- 4 Chamarroa,* a PT 5 Group of Sensors and Biosensors, Department of Chemistry, Autonomous University of 6 Barcelona, Edifici Cn, 08193 Barcelona, Spain RI b 7 Centro Universitario de la Defensa (CUD), Academia General Militar, Carretera de Huesca s/n, 8 Zaragoza, 50090, Spain SC c 9 Grupo de Tecnologías Fotónicas (GTF), Aragon Institute of Engineering Research (I3A), 10 Universidad de Zaragoza, Mariano Esquillor ed. I+D+i, Zaragoza, 50018, Spain 11 U *Corresponding author. E-mail address: Julian.Alonso@uab.es; Tel: +34935812149 AN 12 Abstract M 13 The design, construction and evaluation of a low cost, cyclic olefin copolymer (COC)-based D 14 continuous flow microanalyzer, with optical detection, to monitor carbon dioxide in bottled TE 15 wines and beers as well as in fermentation processes, is presented. The microsystem, 16 constructed by computer numerically controlled (CNC) micromilling and using a multilayer EP 17 approach, integrates microfluidics, gas-diffusion module and an optical flow-cell in a single 18 polymeric substrate. Its size is slightly bigger than a credit card, exactly 45 x 60 x 4 mm in the C 19 microfluidic and diffusion module zone and 22.5 x 40 x 3 mm in the flow-cell zone. The gas- AC 20 diffusion module is based on a hydrophobic polyvinylidene fluoride (PVDF) membrane, which 21 allows the transfer of the carbon dioxide present in the sample to a bromothymol blue (BTB) 22 pH-sensitive acceptor solution, where the color change is measured optically. The detection 23 system consisted of a LED with an emission peak at 607 nm and a photodiode integrated in a 24 printed circuit board (PCB). The obtained analytical features after the optimization of the 25 microfluidic platform and hydrodynamic variables are a linear range from 255 to 10,000 mg L-1 26 of CO2 and a detection limit of 83 mg L-1 with a sampling rate of 30 samples h-1. 1 ACCEPTED MANUSCRIPT 27 Keywords: Lab on a chip, Cyclic Olefin Co-polymer, Miniaturization, Gas-diffusion, Carbon 28 dioxide, Optical detection. 29 1. Introduction 30 The amount of carbon dioxide (CO2) is an important parameter in the monitoring of PT 31 fermentation process in wines and beers [1] and it is also a factor that affects their organoleptic 32 properties [2,3]. For these reasons, CO2 is routinely controlled in the production processes of RI 33 these beverages in wineries and breweries. SC 34 There were traditionally three different official methods to determine CO2 in these samples: the 35 manometric, the enzymatic and the titrimetric, being the last one selected as first option due to U 36 its advantages over the others methods [4,5]. This procedure consisted in a titration with sulfuric AN 37 acid between pH 8.6 and 4.0 of the alkalinized CO2, using a pH electrode. Despite of its 38 advantages, the titrimetric method shows some difficulties. In order to overcome them, direct M 39 measurements of dissolved carbon dioxide, especially those based on electrochemical methods, 40 have also been proposed [6]. However, those methods based on Severinghaus principle show a D 41 long response time and sometimes low reproducibility and, in general, they are often expensive, TE 42 not miniaturized and not easily automatable. As another alternative, automated systems based 43 on continuous flow methodologies with spectrophotometric detection and with gas-diffusion EP 44 separation step, have been used [7-9]. In this way, different quality parameters such as the 45 analysis time, the sample and reagent consumption and the reproducibility of the measures C 46 without the need of skilled laboratory personal, were improved. In these systems, the process to AC 47 determine CO2 usually starts when the sample is injected into a carrier solution which is mixed 48 with an acidic donor solution. At that moment, all the inorganic carbon species are converted to 49 CO2, which diffuses through a gas-diffusion membrane. Then, a pH-sensitive acceptor stream 50 collects the CO2 causing a change in the pH which is detected by a change in the absorbance of 51 the acid-base indicator. Nevertheless, these systems present a narrow working range that 52 difficult the determination of CO2 in beers (about 5000 mg L-1) and wines (from about 1000 mg 2 ACCEPTED MANUSCRIPT 53 L-1 in slightly sparkling up to 12000 mg L-1 in very highly sparkling wines) [1,8]. In addition, 54 sample and reagent consumption are still too high and the size of the experimental set-up makes 55 them not portable, limiting their use in both discrete in-situ sample analysis and on-line 56 monitoring of fermentation process. 57 To overcome these drawbacks, the miniaturization of analytical systems can be applied to obtain PT 58 the so-called micro Total Analysis Systems (µTAS) [10,11]. Among all the different materials 59 that have been used to fabricate these miniaturized analyzers, such as glass, silicon or ceramics RI 60 [12], polymers offer great advantages over the others. They allow a fast and easy fabrication SC 61 using inexpensive facilities providing low cost, small and compact microanalyzers. In addition, 62 depending on the polymer employed, they have a good transparency in the UV-Vis range, good U 63 mechanical resistance, high chemical inertia against most acids and alkalis and high AN 64 biocompatibility [13-16]. Furthermore, polymers technology allows the fabrication of 65 microsystems using a multilayer approach in order to obtain tridimensional structures through M 66 the use of adhesives between layers in the bonding process. However, when adhesives are used, 67 the obstruction of the microchannels can occur and there may be leaking of liquids or D 68 delamination between layers. To overcome these drawbacks, cyclic olefin co-polimer (COC) TE 69 arises to provide a thermolamination process using the same material substrate as a sealing 70 agent. Thus, taking advantage from the different glass transition temperatures in which COC EP 71 can be formulated, a hermetic bonding of tridimensional structures can be achieved through the 72 interlink of polymer chains on the surfaces of the layers when they are put together under a C 73 certain pressure and temperature [17, 18]. This technology allows also the monolithic AC 74 integration of other elements such as hydrophobic/hydrophilic polymeric membranes, 75 conductive paths or electrodes into the multilayer COC devices, which permits to integrate 76 different steps of the analytical procedure such as sample dosage, sample pretreatment and 77 separation, mixing, reaction, detection, among others [13,19, 20]. 78 Finally, the advances in optoelectronics, which have let the appearance of new small light 79 sources such as light-emitting diodes (LEDs) [21-25] and small detectors such as photodiodes 3 ACCEPTED MANUSCRIPT 80 [21-24], allow the possibility to reduce the size of the experimental set-up for optical 81 measurements compared to the conventional equipments, making them easily miniaturized 82 [18]. 83 Herein, we propose a low-cost, COC-based microsystem with a gas-diffusion step integrated for 84 the spectrophotometric detection of CO2 using bromothymol blue (BTB) as pH-sensitive PT 85 acceptor solution. The microanalyzer integrates microfluidics, gas-diffusion separation step and 86 an optical detection cell. These features, joined to the simplicity of the detection setup, make the RI 87 entire system easily portable, with a reduced consumption of sample and reagents and a wider SC 88 linear working range. To demonstrate its applicability, different wine and beer samples have 89 been analyzed. 90 2. Experimental U AN 91 2.1. Reagents and materials M 92 The microanalyzer was fabricated with plaques and foils of COC purchased from Topas 93 Advanced Polymers (Florence, KY, USA) in different grades and thicknesses: Topas 5013 D 94 plaques of 500 µm and 1 mm thickness, and Topas 8007 foils of 25 µm and 50 µm thickness. TE 95 A hydrophobic membrane made of polyvinylidene fluoride (PVDF) with 150 µm thickness and 96 0.45µm pore diameter (Millipore) was used to separate the analyte from its sample matrix by EP 97 gas diffusion. 98 All reagents employed in this work were of analytical grade. All solutions were prepared in C 99 Milli-Q water. Carbon dioxide standard solutions were prepared daily by successive dilutions of AC 100 a 50,000 mg L-1 stock of Na2CO3 (Panreac) and degassed. As a donor solution, a 0.2M H2SO4 101 acidic solution (Sigma-Aldrich), containing 60 mg L-1 of H2O2 (Panreac) in order to avoid the 102 SO2 interference by oxidizing it to sulfate [7], was used to convert all inorganic carbon species 103 into carbon dioxide. As an acceptor solution, a 12 mM NaH2PO4/Na2HPO4 (Panreac) buffer 104 solution (pH 7.6), containing 0.06 mM of BTB (Sigma Aldrich), was used. 105 2.2. Fabrication of the microanalyzer 4 ACCEPTED MANUSCRIPT 106 The fabrication process of COC-based microsystems used, which is based in the lamination of 107 different COC layers with different glass transition temperatures (Tg), is described in detail 108 elsewhere [19,20]. Topas 6013 plaques with Tg=130 ºC were used as fluidic structural layers for 109 the mechanization of all motifs of the design (microchannels and flow cell) and Topas 8007 110 foils with Tg=75 ºC were used between them as sealing layers. The fabrication process consists PT 111 of four main steps: prototype design, fluidic motifs machining, integration of additional non- 112 COC elements (such as electrodes and gas-diffusion membrane) and final lamination. The RI 113 prototype design was performed with a CAD software (Figure 1) taking into account previous 114 experimental results obtained by our research group and described elsewhere [20]. The SC 115 microsystem was constructed using five structural layers that, once overlapped, provide the 116 three-dimensional structure required for this application (Figure 1A). The dimensions of the 117 U microanalyzer were 45 x 60 x 4 mm in the microfluidic and diffusion module zone and 22.5 x AN 118 40 x 3 mm in the flow-cell zone which is inserted into the detection system based on a lock-and- 119 key concept [18] described in the 2.3 Experimental setup section. The microfluidics inside the M 120 microsystem included three liquid inlets (Figure 3). Two of them converge in a T-shape D 121 confluence point. At this point the carrier solution, where the sample is injected, and the H2SO4 TE 122 solution are mixed to obtain carbon dioxide from all inorganic carbon species present in the 123 sample following the next chemical reactions: CO32-(aq) + H+ ⇆ HCO3-(aq) , HCO3-(aq) + H+ ⇆ EP 124 H2CO3(aq), H2CO3(aq) ⇆ CO2(aq) + H2O, CO2(aq) ⇆ CO2(g). The mixed stream is carried to the 125 diffusion membrane unit where, along a meander microchannel, the carbon dioxide diffuses C 126 through the PVDF membrane towards the acceptor solution. The BTB pH-indicator acceptor AC 127 stream sets to pH 7.6 with 12 mM phosphate buffer collects the carbon dioxide diffused 128 changing its absorbance (from blue color to yellow color at wavelength of 607 nm) due to the 129 pH modification induced. Finally, both the donor and the acceptor solutions are carried to the 130 corresponding waste outlets. 131 All fluidic motifs (holes and channels) were machined onto the polymeric plaques by means of 132 a computer numerically controlled (CNC) micromilling machine (Protomat C100/HF, LPKF, 5 ACCEPTED MANUSCRIPT 133 Spain). The dimensions of the microchannels throughout the microsystem were 0.4 mm wide 134 and 0.3 mm height except in the zone where the diffusion process takes place. In order to 135 maximize the contact area between the donor and acceptor solutions with the diffusion 136 membrane, the dimensions of the channels in both cases were 1 mm wide and 0.1 mm height. 137 The diameter of the detection flow-cell was 4.5 mm with an optical path-length of 1 mm. The PT 138 total microsystem dead volume was 200 µL. In order to increase the lifetime of the gas- 139 diffusion membrane and to prevent its breakage, a non-coincident inlet streams configuration, in RI 140 the gas diffusion unit, was used [20]. Thus, the stress on the membrane caused by the incidence 141 of the two streams at the same point was avoided. SC 142 The integration of the PVDF membrane was made as follows. It was cut, washed with deionized U 143 water, dried and placed properly with its more hydrophobic face in contact with the donor AN 144 solution [20]. As it can be seen (Figure 1B), some holes were made on the external contour of 145 the PVDF membrane to favor the penetration of the sealing foil (Topas 8007) during the M 146 lamination process in order to prevent liquid leakage between layers. In order to avoid the 147 flattening of the membrane and the layers deformation during the thermo-compression process D 148 as a result of the incorporation of the membrane, an extra 50 µm thick Topas 8007 sealing foil TE 149 with a machined hole with the dimension of the membrane area is placed between the two 150 machined COC layers [19, 20]. EP 151 COC layers alignment and final lamination was performed in a thermo-compression press 152 (Francisco Camps, Granollers, Spain) at 102 ºC and 4 atm using an aluminum support with 4 C 153 fiducial alignment pins. Therefore, the whole microsystem acts as a monolithic substrate with AC 154 the different layers perfectly bonded between them. Finally, fluidic connectors were fixed onto 155 the polymer inlet/outlet ports with a holder and screws (Figure 1B). 156 2.3. Experimental setup 157 The detection system developed previously by our research group has been described elsewhere 158 [18,24]. It consists in a compact and robust optical reader for microfluidic platforms which is 6 ACCEPTED MANUSCRIPT 159 composed by a structure with a Printed Circuit Board (PCB) for the optical detection electronics 160 and an insertion port, designed with a lock and key concept, to allow a perfect alignment 161 between the light source, the measurement flow-cell and the detector (Figure 2). The PCB 162 structure integrates as a light source a Light Emitting Diode (LED) with an emission peak 163 centered at 607 nm (Kingbright, Taipei, Taiwan) and as a detector, a PIN Hamamatsu S1337- PT 164 66BR with 33 mm2 of active area photodiode. The signal generated was acquired by means of a 165 Data Acquisition Card (DAQ) NI USB-6211 from National Instruments (Austin, Texas, US), RI 166 which transfers the signal from the PCB, via USB interface, to a personal computer, where it is 167 processed using a digital lock-in amplifier. The data collection rate was 10 s-1. The use of lock- SC 168 in amplification allowed to increase the signal-to noise ratio and to work in ambient light 169 conditions avoiding thus the effect of stray light on the measures [18,24]. 170 U On the other hand, the continuous flow system setup is shown in Figure 3. It consists of an AN 171 external peristaltic pump (Minipuls 3, Gilson, Wisconsin, US) using Tygon® tubing (Ismatec, 172 Wertheim, Germany) with 1.14 mm internal diameter and a six-way injection valve (Hamilton M 173 MVP, Reno, US). Teflon tubing (Scharlab, S. L., Cambridge, England) of 0.8 mm internal D 174 diameter was used to connect the external elements to the microsystem. TE 175 3. Results and discussion 176 3.1. Design and optimization of the analytical microsystem EP 177 The main goal of this work is the development of a low cost, simple, selective and robust 178 microanalyzer for the determination of carbon dioxide in bottled wine and beer and in their C 179 fermentation processes at breweries or wineries, which monolithically integrates the AC 180 microfluidic platform, the gas diffusion module and the detection flow-cell. This microsystem 181 has to meet all the requirements for the proposed application, such as easy automation, small 182 size, limited consumption of reagents and a wide working range. 183 In a previous work [20], the design of the diffusion module was optimized in order to enhance 184 the robustness of the separation process and to increase the lifetime of the PVDF diffusion 185 membrane keeping suitable analytical features such as sensitivity, baseline stability and 7 ACCEPTED MANUSCRIPT 186 detection limit. In this work, variables such as the flow mode (concurrent or countercurrent) and 187 the flow rate ratio between the acceptor and donor solutions inside the diffusion module were 188 tested, concluding that the same configuration previously optimized (countercurrent flow mode 189 and 2:1 donor: acceptor ratio) is also the better choice for the present work. Additionally, the 190 microfluidic structures and the hydrodynamic variables has been designed and optimized PT 191 respectively to minimize the sample and regents consumption. 192 The shape of the flow-cell has also been redesigned and optimized introducing smooth-contour RI 193 geometries, without any sharp corners, and taking into account the Venturi effect and SC 194 Bernoulli’s principle [26]. Thus, the formation and retention of bubbles was minimized inside 195 the flow-cell. Despite of this, if eventually bubbles appear, their evacuation is favored due to the U 196 pressure-velocity difference of the fluid between the flow-cell and the outlet stream (Figure AN 197 1Bd). For this reason, noisy and transient signals and baseline drifts related with the presence of 198 bubbles in the flow-cell were avoided, increasing the overall robustness of the microsystem. M 199 In addition, the shape of the microsystem was designed so that it can be inserted into a detection 200 system based on the lock and key concept (Figure 2), allowing a perfect and reversible D 201 alignment between the LED, the detection flow-cell and the photodiode and thus, obtaining a TE 202 robust and compact system, as demonstrated in a previous work [18]. 203 The influence of chemical variables (buffered acceptor solution, acidic donor solution and EP 204 potential interfering compounds) and hydrodynamic parameters (flow rate and sample injection C 205 volume) on the analytical features has also been evaluated using an optimization of each variable individually in order to achieve a compromise between the sensitivity of the analytical AC 206 207 measurements, the baseline signal stability, the linear working range, reagent and sample 208 consumption and the sample throughput. 209 According to the bibliography, bromocresol purple [9], phenol red [27] and bromothymol blue 210 (BTB) [7,19,27-29] have been used as pH-sensitive indicators dissolved in an acceptor stream to 211 determine carbon dioxide in real samples. In this work, BTB dissolved in a phosphate buffer 212 solution was selected because it showed a higher working range [29]. The BTB acceptor 8 ACCEPTED MANUSCRIPT 213 solution was tested at concentrations ranging from 0.03 to 0.08 mM, the phosphate buffer 214 strength was evaluated from 0.1 to 12 mM and the pH of the acceptor stream was evaluated 215 from 7 to 8. The optimal results were obtained using an acceptor solution of BTB 0.06 mM at 216 pH 7.6 with phosphate buffer strength of 12 mM. Using this buffered acceptor solution, the 217 peak height caused by a given CO2 concentration diminishes compared to that obtained with a PT 218 non-buffered acceptor solution. This fact means that the higher the buffer strength, the lower the 219 peak height for each CO2 concentration and the lower the sensitivity, but at the same time it RI 220 allows enlarging the operative working range [19,28,29]. Taking into account that the aim of the 221 present work is to be able to determine a range of relatively high carbon dioxide concentrations SC 222 in wines and beers, a very low limit of detection and high sensitivity are not required. In 223 addition, using the buffered acceptor solution, the changes in the pH caused by the absorption of 224 U atmospheric carbon dioxide are negligible. Nevertheless, in other applications when very low AN 225 concentrations of CO2 have to be determined, non-buffered acceptor solution should be used. In 226 these cases, the absorption of atmospheric carbon dioxide would be a serious problem which M 227 can be minimized introducing cartridges of soda lime at the air inlet of the bottles of reagents to D 228 completely remove the atmospheric CO2 [30]. TE 229 On the other hand, hydrochloric and sulfuric acids were evaluated as donor acidic solutions 230 without obtaining any significant differences between them in terms of peak height. However, EP 231 sulfuric acid was selected due to that hydrochloric acid can diffuse through the PVDF 232 membrane, since it has a higher vapor pressure, and can change the pH of the acceptor solution. C 233 The concentration of H2SO4 donor stream was varied from 0.2 to 0.6 M. The optimal results AC 234 were obtained with a donor solution of H2SO4 0.2 M. 235 Taking into account potential interfering compounds that can be found in these samples, sulfite 236 ion is the only one that has to be taken into account because it is the only chemical compound 237 present in wine samples able to diffuse in an acidic medium, as SO2, through the PVDF 238 membrane [7]. Its concentration in wine can be up to 300 mg L-1 SO2. For this reason H2O2 is 239 added to the donor stream in order to assure the total oxidation of SO2 to SO42- which cannot 9 ACCEPTED MANUSCRIPT 240 diffuse through the membrane. The concentration of hydrogen peroxide used was optimized 241 elsewhere [7], being 60 mg L-1. Therefore, the final composition of the donor stream was 0.2 M 242 of H2SO4 with 60 mg L-1 of H2O2. Despite of this, due to SO2 usually is in less proportion than 243 CO2 in wine, H2O2 could be suppressed without obtaining significant differences in the results. 244 Regarding the hydrodynamic parameters, the flow rate of the carrier, the donor and the acceptor PT 245 streams were varied from 100 to 500 µl min-1 and the sample injection volume was varied from 246 25 to 580 µL. A flow rate of 400 µl min-1 for carrier, donor and acceptor streams and a sample RI 247 injection volume of 25 µL were considered the optimal parameters. SC 248 3.2. Analytical performance U 249 Analytical characterization of the microsystem was carried out by successive calibrations AN 250 obtained using different standard solutions of Na2CO3 from 100 to 10,000 mg L-1. Figure 4 251 shows the recorded signal for one calibration. The obtained equation (n = 8 and 95% M 252 confidence) was Abs=-4.45·10-6 (±2·10-8) [CO2] + 0 (±1·10-4) with r2=0.9999. The detection 253 limit, calculated as three times the standard deviation of the blank, was 83 mg L-1. The linear D 254 working range corresponds from 255 to 10,000 mg L-1. Repeatability studies were performed by TE 255 successive injections of 2,500 and 5,000 mg L-1 Na2CO3 standard solution. Relative standard 256 deviation (n = 6, 95% confidence) of the signals were lower than 2% in both cases. EP 257 Reproducibility of the results obtained in different days using different reagent solutions was 258 also determined. Mean slope of 4.38·10-6 with RSD value lower than 2% was achieved, thus C 259 demonstrating the reproducible inter-day validation of the microsystem. These results showed AC 260 the robustness and the reliability of the whole experimental setup. A sampling rate of 30 261 samples h-1 was obtained with the optimized experimental conditions. 262 3.3. Real samples analysis 263 Real samples of wines and beers from different origin were analyzed to validate the proposed 264 microsystem. In order to cover a wide range of carbon dioxide concentrations, slightly sparkling 10 ACCEPTED MANUSCRIPT 265 wines, sparkling wines and beers were analyzed. With the aim of establish an accurate 266 comparison between the results obtained by both methods, before the analysis of the samples, 267 and following the procedure of the reference method [4,5], the bottles were kept at 4 ºC, they 268 were opened and their content was immediately treated with 30 mL of 10 M NaOH per liter of 269 sample in order to fix the concentration of CO2 and avoid its loss. Then the mixtures were PT 270 allowed to reach room temperature in closed bottles without air before their determination. It is 271 noteworthy that the developed microsystem can analyze real sample without any pretreatment. RI 272 As it has been stated previously, NaOH was added to allow an accurate comparison between 273 both methods. In the case that NaOH is not used, the longer the bottle is open, the more loss of SC 274 carbon dioxide there is, so this is an issue to take into account. In the same way, the 275 microsystem is able to determine CO2 online in fermentation processes in breweries and 276 U wineries. Due to the logistic difficulties, these kinds of samples were not determined. AN 277 The results obtained with the developed optical microsystem were validated by comparison with M 278 the ones obtained with the titrimetric reference method. This procedure, as have already 279 mentioned, consisted in a titration of alkalized CO2 with sulfuric acid, between pH 8.6 and 4.0, D 280 using a pH-meter. The obtained results are shown in Table 1. TE 281 As it can be seen, the results of the developed microsystem are not significantly different from 282 the ones obtained with the reference method using the paired t-test (tcalc = 1.595; ttab= 2.365; EP 283 tcalc