A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters

G Model SNB-23282; No. of Pages 6

ARTICLE IN PRESS Sensors and Actuators B xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb

A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters Thomas Steinsberger a , Patrick Kathriner a , Philipp Meier a , Alexander Mistretta a , Peter C. Hauser b , Beat Müller a,∗ a b

Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH-6047 Kastanienbaum, Switzerland The University of Basel, Department of Chemistry, Spitalstrasse 51, CH-4056 Basel, Switzerland

a r t i c l e

i n f o

Article history: Received 18 May 2017 Received in revised form 11 September 2017 Accepted 26 September 2017 Available online xxx Keywords: Alkalinity Micro-titrator Porewater Lakes Sediment

a b s t r a c t Alkalinity is an important parameter in oceans, lakes, groundwaters and sediment porewaters as a link to the global carbon cycle. It is determined by classic titration with acid where sufficient sample volume is available. However, application to the limited amounts of sediment porewater requires a different approach. A portable low cost coulometric micro-titrator based on a RuO2 pH-sensitive electrode and a Ag/AgCl reference electrode requiring 50 ␮l of total sample volume is presented. By using a distinct sandwich cell design, a well-defined titration volume could be achieved. The micro-titrator performed well within the targeted range of 1–10 mmol L−1 and a reproducibility within 3.5%. It was successfully applied to lake water and sediment porewater alkalinity measurements of Lake Lucerne and bears the potential for automation and in-situ applications. © 2017 Published by Elsevier B.V.

1. Introduction Alkalinity (Alk) is the capacity of a solution to neutralize acids and hence the sum of bases that it contains with pK values higher than a reference pH, minus acids present with a pK lower than the reference pH. Generally, alkalinity of waters from carbonaceous catchments are dominated by dissolved carbonates: Alk = [HCO3 − ] + 2[CO3 2− ] + [OH− ] − [H+ ]

(1)

whereas in seawater borate, B(OH)4 − is considered as an additional base [1]. Alkalinity in combination with either pH or dissolved inorganic carbon (DIC) allows to calculate the concentrations of the species H2 CO3 , HCO3 − , CO3 2− and pCO2 , the partial pressure of CO2 . Together with the concentration of Ca2+ , it gives access to the saturation state of calcite, which is essential to estimate dissolution and/or biogenic precipitation [2]. Calcite precipitation is triggered by supersaturation in the epilimnion caused by blooms of algae and cyanobacteria after the onset of photosynthesis in spring and early summer. In the top sediment layers, the pH decrease caused by the production of CO2 from mineralization of organic matter

∗ Corresponding author at: Seestrasse 79, CH-6047 Kastanienbaum, Switzerland. E-mail address: [email protected] (B. Müller).

leads to the dissolution of CaCO3 and release of Ca2+ and HCO3 − to the deep water of the lake. Alkalinity further quantifies the buffering capacity which shields lakes from acidification and therefore is crucial in areas with large acid loading (e.g. industrial areas or mining districts) [3]. Many reduction reactions (denitrification, sulfate reduction, reduction of Fe(III) and Mn(IV)) create alkalinity while oxidations (nitrification, mineralization of organic matter, oxidation of Fe(II) and Mn(II)) consume alkalinity and thus assist in characterizing and quantifying processes in sediments and the dynamics of the aquatic carbon cycle [4]. Alkalinity is determined by acidimetric titration. It is a conservative property and is thus not affected by gas exchange or pressure and temperature changes, therefore sample transport and storage is not critical. pH can easily be measured in situ with temperature-compensated potentiometric electrodes often integrated in submersible probes. While sufficient sample volumes are available for measurements in the water column, it is critical for sediment porewaters where only a few hundred microliters are available for analyses. Hence a titration method requiring minimal sample volumes is very desirable. For the titration of small sample volumes (<500 ␮l), several methods have been developed for in situ measurements and small sample size e.g. a burette driven system using fused silica capillaries and spectrophotometric end-point detection to reduce the sampling volume down to 120 ␮l [5]. A similar spectroscopic method

https://doi.org/10.1016/j.snb.2017.09.191 0925-4005/© 2017 Published by Elsevier B.V.

Please cite this article in press as: T. Steinsberger, et al., A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters, Sens. Actuators B: Chem. (2017), https://doi.org/10.1016/j.snb.2017.09.191

G Model SNB-23282; No. of Pages 6

ARTICLE IN PRESS

2

T. Steinsberger et al. / Sensors and Actuators B xxx (2017) xxx–xxx

was introduced by optically monitoring the indicator color change of acid-base end point titrations [6]. A direct alkalinity sensor was introduced by using an ion-selective proton pump membrane in combination with a pH probe separated by a thin layer of water sample [7,8]. Coulometric titration, the electrochemical generation of protons and/or hydroxide ions were introduced by using various electrodes e.g. electrodes made of Antimony/Antimony to produce hydroxide ions [9], gold electrodes to generate H+ and OH− by water hydrolysis in combination with a reference pH-sensitive ion-selective field effect transistor (ISFET) [10] or silver and/or platinum electrodes to offer various titration methods [11]. A solid state pH-sensor consisting of a pH-sensitive RuO2 electrode and a Ag/AgCl reference electrode was successfully developed by Colombo, Kappes and Hauser [12]. In all these devices, only molecular diffusion is responsible for the travel of OH− and H+ between the actuator electrode and the pH-sensitive electrode as no mixing or stirring is involved. Van der Schoot [16] further reported on the advantages of automated coulometric titration systems such as the high reproducibility and accuracy of diffusion driven micro titrators e.g the Orion FLASH titrator. For potentiometric sensors the relative pH change is monitored to accurately determine the end-point of titration reactions. We report on a new cell design, RuO2 plating, modified electronics and optimized software and apply the micro-titrator for the determination of alkalinity in lake waters and sediment porewaters. Performance and limitations of the micro-titrator are presented at measuring a vertical alkalinity concentration profile from the water column extending into the sediment porewater of Lake Lucerne, with simultaneous analyses using conventional titration for comparison. The presented micro-titrator is independent of a power source other than the supply from the laptop’s USB port and small enough to make it ideally suited for field applications and on-site measurements. It bears the potential of easy automation, underwater application and thus combination with probes used for in situ water column profiling.

Fig. 1. a) Photo of the alkalinity micro-titrator with match for scaling. b) Schematic side view of the micro-titration cell consisting of the carrier base plate, the etched ceramic wiring board, and the cover part with the flow channel and the in- and outlet made from acrylic glass. The whole micro-titrator cell has dimensions of 50 mm × 30 mm × 25 mm, and the flow channel 22 mm × 4 mm × 0.5 mm. Figure b is not to scale.

2. Methods 2.1. Cell design The micro-titration cell is comprised of three parts: the base, the etched wiring board, and the cover with the flow cell as shown in Fig. 1, analogous to the sandwich design of Colombo, Kappes and Hauser [12]. The flow channel was cut in the cover of acrylic glass framed by a channel for a sealing ring. This design allowed precise dimensioning of the channel, tight sealing and easy handling for maintenance. The flow channel was 22 mm long, 4 mm wide, and 0.5 mm high with a total volume of 44 ␮l. The sealing channel was 1.7 mm wide, 1 mm deep, and was filled with a non-conductive silicon sealing ring (Part. No. 15169417, Maagtechnic, Dübendorf, Switzerland). Four screws in the corners were applied to tightly seal the bottom and top parts. The effective volume of titration (the space between the actuator, the pH-electrode and the sealing ring) was ∼0.27 ␮l. 2.2. Titration wiring board The etched titration wiring board shown in Fig. 2 was a standard hydrocarbon/ceramic laminate (RO4350, Polytrona AG) custom designed and manufactured by Polytrona AG (Switzerland). The electrode tracks were produced on 35 ␮m thick copper lines by deposition of 15–17 ␮m bond gold (Ni/Au). The wiring board was covered with a solder stop mask to protect and isolate the electric tracks, except for the 5 mm long active ends of the electrode tracks

Fig. 2. Top down view of the titration wiring board with the gold counter electrode (CE), gold actuator electrode (AE), Ag/AgCl reference electrode (Ref), and RuO2 pHelectrode (pH). The gray area indicates the location of the flow channel.

that are exposed to the sample liquid in the titration cell (gray bar in Fig. 2). The pH-sensitive ruthenium dioxide (RuO2 ) layer (300 ␮m width) was placed between the two gold actuator electrodes (AE, 300 ␮m width) at a distance of 200 ␮m. The pH electrode potential was measured against a Ag/AgCl electrode (Ref, 500 ␮m width) positioned at a distance of 1200 ␮m. The 500 ␮m wide gold counter electrode (CE) was placed 700 ␮m away from the nearest actuator electrode to minimize interferences from OH− generated at the counter electrode. 2.3. Fabrication of the electrodes The reference electrode was electroplated with silver from a solution of 42 g L−1 K[Ag(CN)2 ], 43 g L−1 KCN, and 7.5 g L−1 K2 CO3 with a current density of 0.2 mA mm−2 for 30 min [13]. The Ag electrode was electro-polished as described in Colombo, Kappes and Hauser [12] and anodically oxidized in a solution of 50 g L−1

Please cite this article in press as: T. Steinsberger, et al., A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters, Sens. Actuators B: Chem. (2017), https://doi.org/10.1016/j.snb.2017.09.191

G Model SNB-23282; No. of Pages 6

ARTICLE IN PRESS T. Steinsberger et al. / Sensors and Actuators B xxx (2017) xxx–xxx

KCl for 8 min with a current density of 40 ␮A mm−2 to produce the AgCl-layer. The pH-sensitive RuO2 layer was produced by adapting and optimizing the procedures of Lowenheim [14]. Ruthenium was electrochemically deposited from a solution of 5 g L−1 potassium diazo-octachloro-nitrido-ruthenate (CAS-27316-90-1, Chempure, Germany), and 10 g L−1 ammonium formate (Part. Nr. 156264, Sigma-Aldrich), adjusted to pH 1.3 with hydrochloric acid at 70 ◦ C for 4 h with a current density of 20 ␮A mm−2 . An additional increase of RuO2 by oxidation e.g. by using HClO4 , was omitted. 2.4. Galvanostat circuitry and data acquisition The set-up of the coulometric titrator is illustrated in Fig. 3. The electrochemical cell is controlled with an Arduino Nano (www. arduino.cc), a programmable microcontroller board. For data acquisition and process control, the board is connected to a personal computer through the universal serial bus (USB). The USB port also acts as a power supply with a voltage of 5 V for the entire circuitry, allowing off-grid use of the instrument. The galvanostat circuitry is based on an operational amplifier (TL071, Texas Instruments, Austin, TX) with the actuator and counter electrodes in its negative feedback loop. Coulometric current control of the galvanostat is achieved by turning one of the digital output ports of the Arduino board to a logic high, connecting the corresponding input resistor to the 5 V power supply. The feed current setting and thus the rate of titration can be chosen by selecting the appropriate resistor. The potential difference between the pH-electrode and the reference electrode is measured with the help of an instrumentation amplifier (INA116, Texas Instruments) and transformed to a digital signal by a high precision analog-to-digital converter (LTC2400, Linear Technology, Milpitas, CA). The LTC2400 only accepts voltages between 0 V and +5 V. An offset 3.75 V was set without amplification of the voltage signal coming from INA116 to prevent negative voltage at the input of the LTC2400. Its output is transmitted to the Arduino board via the serial peripheral interface (SPI) bus. The entire operation is controlled by a software package consisting of a graphical user interface and data acquisition library running on the personal computer and a microcontroller firmware, for process control and data transmission, running on the Arduino board. The software package implements a light-weight data transmission and configuration protocol. This protocol provides a set of configuration commands for the microcontroller board, specifying the state of digital output ports and the source where input data is read from, and a standard way of transmitting measurement data. The graphical user interface and the data acquisition library are written in Python. It allows the user to set all necessary measurement parameters before a measurement is performed using the setup screen. The corresponding settings are then sent to the Arduino board. Upon successful completion of the setup phase the titration process is started and the measured voltages are transmitted from the Arduino to the personal computer, where it is plotted and saved to a file for later processing. The microcontroller firmware is written in C and only contains the code necessary to read data via the SPI bus, to set digital output pins and to transmit data as specified by the protocol. This reduction to the bare minimum allows for high measurement frequencies of around 50 Hz. The end points of the titrations were obtained from the first derivatives of the titration curves (signal vs. time) (see Fig. 4c) and could be reproduced to within a standard deviation of 3.5%.

3

samples were directly injected in the micro-titrator without prior filtration. Simultaneously, the same samples were analyzed with the standard laboratory procedure with an automatic titrator (862 Compact Titrosampler, Methrom AG, Switzerland). On the same day, a sediment core was retrieved in the vicinity of the water sampling location from Lake Lucerne (47◦ 1.057 N/8◦ 25.617 E) and porewater sampled immediately after return to the lab. Sediment porewater samples were collected using MicroRhizon filter tubes (1 mm diameter, 0.20 ␮m pore size, Rhizosphere Research Products, Wageningen, Netherlands) with high spatial resolution (5 mm resolution for the first 4.5 cm of the sediment) down to 10 cm sediment depths through pre-drilled holes (procedure described in detail in Torres et al., 2013). At least 700 ␮l of sediment porewater were extracted at each sampling depth to allow at least two micro-titrator measurements and a Gran-plot titration [1,15,16]; For Gran-plot titrations, 300 ␮l of porewater were used. Increments of 10 ␮l of 0.01 M HCl were added stepwise, and the corresponding pH measured with a combined ® pH micro-electrode (InLab Micro, Mettler-Toledo, Germany). The porewater retrieval time was between 10 to 60 s depending on sediment depth, i.e. sediment porosity. The porewater was directly injected into the micro-titrator channel via the syringe inlet. After each measurement, the electrodes were rinsed several times with Nanopure water. Calibration solutions were prepared from a 0.1 mol L−1 NaHCO3 stock solution (analytical grade, Part. No. S6014, Sigma-Aldrich) in Nanopure water (18.2 M, PureLab Ultra, ELGA). 200 ␮mol L−1 CaCl2 and 1.6 mgL−1 anthraquinone2,6-disulfonate (AQDS) was used as a background electrolyte for calibrations for sediment porewater measurements. 2.6. Diffusion processes As the effective titration volume was very low (∼0.27 ␮l) in a planar cell design, quasi two dimensional diffusion is seen as the main way of sample mixing [17]. The run-time of the titration is defined by the diffusion of H+ from the actuator electrode to the pH electrode and by the time required for the production of the amount of H+ needed to titrate a given amount of alkalinity. The size of the flow channel allowed a quasi-two-dimensional diffusion of H+ [11,12]. Based on this assumption, the diffusion time of H+ (tdiff ) is calculated according to Eq. (2) with t (s), x (distance between the electrodes in mm) and the diffusion coefficient of H+ (D = 9.3 × 10−3 mm2 s−1 ): tdiff = x2 (2D)−1

(2)

Using the size and placement of the electrodes, the minimum diffusion time of H+ to reach the pH electrode was calculated to be 13.4 s. It was experimentally determined by the time lapse from the application of a current at the electrodes in a solution with zero alkalinity until a change in the potential of the pH electrode. t was not related to the current applied. The production time of H+ ions was calculated with the deposition formula based on the second law of Faraday: n = I ∗ tdepo ∗ (z ∗ F)−1

(3)

where n is the amount of protons in moles, I the current in A, tdepo the time of deposition in s, z the charge, and F the Faraday constant (96 485.4 As mol−1 ). 3. Results and discussion

2.5. Lake water sampling and measurements 18 water samples were collected from the water column of the oligotrophic Lake Lucerne (47◦ 0.795 N/8◦ 21.557 E) between 0 and 150 m water depths on the 25th of November 2016. The water

In total 20 RuO2 -Ag/AgCl micro-titrator wiring boards were produced and tested. The sensitivity of the pH sensing system consisting of the pH-sensitive RuO2 electrode and a Ag/AgCl reference electrode was characterized with standard pH buffer solutions

Please cite this article in press as: T. Steinsberger, et al., A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters, Sens. Actuators B: Chem. (2017), https://doi.org/10.1016/j.snb.2017.09.191

G Model SNB-23282; No. of Pages 6

ARTICLE IN PRESS

4

T. Steinsberger et al. / Sensors and Actuators B xxx (2017) xxx–xxx

Fig. 3. Simplified circuit diagram of the coulometric titrator.

Fig. 4. a): Titration curves of a single HCO3 − standard solution (8 mmol L−1 ) using currents of 4.2 ␮A, 7.4 ␮A and 12.9 ␮A. b): Alkalinity titration curves of 1 mmol L−1 (blue), 2 mmol L−1 (orange), 4 mmol L−1 (red) and 8 mmol L−1 (green) standard solutions. The recorded potential is the response of the RuO2 pH sensor. Decreasing sensitivity is attributed to the aging of the pH-sensor. c): Alkalinity measurement (blue) and the corresponding 1st derivative (red) of the polynomial fitted measurement which is used to calculate the titration point. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(pH 4, 7 and 9) and yielded an average of 29.4 mV pH−1 . The subNernstian slope was attributed to the irregular deposition of the pH sensitive RuO2 as no oxidation step was included to increase the amount of RuO2 . During the aging of the pH-electrodes a decrease in sensitivity was observed. By adding an oxidation step, e.g. by using perchloric acid Colombo, Kappes and Hauser [12] increased the sensitivity to 40 mV pH−1 , yet still below the theoretical 59 mV pH−1 However, the sub-Nernstian pH sensitivity does not pose a problem for end point titration systems, as only relative pH changes are relevant and an inflection point could be determined accurately and with high reproducibility in the titration curve independent of the absolute sensitivity of the pH-electrode [12,17]. 3.1. Performance The titration of HCO3 − solutions were carried out by generating H+ ions at both actuator electrodes by hydrolysis of H2 O. Several series of 10 measurements of HCO3 − with various titration circuit boards yielded average standard deviations of <5%. The response of the RuO2 electrode to a change in pH (pH 4–pH 7–pH 9) were within several seconds similar to values observed by Colombo, Kappes and Hauser [12]. Linear five to six point HCO3 − calibrations between 1 mmol L−1 and 10 mmol L−1 resulted in average correlations of R2 = 0.99 (see Fig. 5). All calibrations showed an offset of ∼14.2 s attributed to the finite diffusion time between the actuator electrode and pH-electrode, which is close to the calculated value of 13.4 s based on the distance between actuator electrodes and the pH-sensitive electrode (see Eq. (2)). Hence, the theoretical detection limit is defined as the concentration where the line of a calibration curve intercepts with the value of 13.4 s. In practice, measurements with very low alkalinity concentrations (<1 mmol L−1 ) showed titration times close to the theoretical H+ diffusion time making a distinction tenuous, e.g. solutions of 0.5 mmol L−1 HCO3 − at a current of 10.5 ␮A showed titration times

Fig. 5. Linear alkalinity calibrations for 12.9 ␮A (388 k, blue), 10.5 ␮A (474 k, orange), 7.4 ␮A (676 k, green) and 4.2 ␮A (1200 k, purple). Slopes and intercepts were close to theoretical values given in brackets calculated with Eqs. (2) and (3): 388 k slope 3.85 (3.59)/14.2 (13.4), 474 k slope 4.49 (4.39)/16.8 (13.4), 676 k slope 5.22 (6.32)/14.2 (13.4) and 1200 k slope 10.5 (11.1)/14.2 (13.4). R2 values were all >0.99. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

of 14.4 s. Titration of low concentrations of alkalinity (1 mmol L−1 to 4 mmol L−1 ) were performed using low current (4.2 ␮A) with higher resistors (see Figs. 4 b and 5). This reduced the amount of H+ produced and thus lengthened the duration of the titration process. Conversely, high currents (12.9 ␮A) were applied for solutions with high alkalinity concentrations >4 mmol L−1 (see Fig. 4a). A complete measurement including rinsing of the electrodes was achieved in less than four minutes. Rinsing the titrator cell two times with nanopure water after each measurement was sufficient

Please cite this article in press as: T. Steinsberger, et al., A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters, Sens. Actuators B: Chem. (2017), https://doi.org/10.1016/j.snb.2017.09.191

G Model SNB-23282; No. of Pages 6

ARTICLE IN PRESS T. Steinsberger et al. / Sensors and Actuators B xxx (2017) xxx–xxx

5

to prevent memory effects. O2 and H2 gas formation was observed on the electrodes during alkalinity measurements using a binocular microscope. However, this gas formation did not show a detectable influence on the measurements. No bias towards other ions was observed although an influence of Cl− on the Ag/AgCl reference electrode could not be ruled out. The average life span of a micro-titrator plate was limited to about 400 measurements of laboratory standards or lake water samples, which was sufficient for a broad range of calibrations and measurements. The limiting factor was the pH-sensitive electrode where the RuO2 layer showed thinning and patches of disapperance of RuO2 probably due to abbrasion of the surface by repeated flushing of the channel, causing a rapid decline of the sensitivity of the pH sensor. Fewer measurements were possible when porewaters from anoxic sediments were analyzed. We attributed this behaviour to ongoing electrode poisoning by reduced Fe(II) naturally occuring in anoxic sediment porewaters, a phenomenon known to irreversibly damage the ion exchange reactions of the RuO2 surface [18]. In the upper sediments of oligotrophic Lake Lucerne, however, very low concentrations of Fe(II) allowed the completion of measurements before electrode poisoning. 3.2. Water column and sediment porewater alkalinity Samples collected from the water column of Lake Lucerne were analyzed with the micro-titrator and simultaneously titrated with the standard acidimetric method with an automatic titrator to the inflection point for comparison. Results from both methods were in close agreement showing a maximum relative deviation of 4.8% and an average deviation of 1.7% (see Fig. 6). The shape of the alkalinity profile in lakes in a calcareous catchment (Fig. 6a) originates from the biogenic precipitation of calcite in the productive surface layer, thus removing alkalinity – in this case dissolved calcite – from this zone. The precipitated particles sink through the water column where they slightly dissolve, and settle on the sediment surface, together with organic detritus from algae and zooplankton. Mineralization of this organic material with oxygen in the top sediment layers produces CO2 , causing the pH in the porewater to decrease. This process is most intense in the top few mm to cm of the sediment. Due to the decrease in pH settled calcite dissolves again in the porewater of the sediment leading to a corresponding increase in alkalinity. The results in Fig. 6 show that firstly, the water column was not homogeneously mixed during winter overturn at this location of Lake Lucerne but still kept the pattern of the abovementioned processes with 1.75 mmol L−1 Alkalinity at the top of the water column and 2.41 mmol L−1 at 150 m water depth, and secondly, that Lake Lucerne was a source of CO2 to the atmosphere during lake overturn as water in equilibrium with calcite and atmospheric pCO2 would result in an equilibrium alkalinity of only 1.24 mmol L−1 [2]. As sediment porewater measurements are limited by the amount of available sample volume, the use of the micro-titrator allowed an unprecedented high resolution alkalinity profile of the sediment of Lake Lucerne. At least 700 ␮l of sediment porewater were extracted from each sampling location between 0 to 10 cm sediment depth to retrieve enough sample volume for both, microtitrator and Gran-plot titration. By only using the micro-titrator, the porewater sample volume could be considerably reduced to 100–200 ␮l and thereby increase the sampling resolution down to 0.25 cm. Using calibration standards with pure carbonate salts for the sediment porewater analysis resulted in alkalinity concentrations ∼20% lower than the concentrations obtained with Gran-plot titrations. The application of matrix calibration for the sediment porewater improved the measurements to an average deviation of 3.6% (see Fig. 6b). In addition to HCO3 − , 1.6 mg L−1 anthraquinone2,6-disulfonate (AQDS), a proxy for dissolved organic carbon, and

Fig. 6. a) Alkalinity concentration profiles from the water column of Lake Lucerne (note that vertical depth is in meters); b) Sediment porewater alkalinity profile of Lake Lucerne (vertical depth is in millimeters). Orange full circles show mean values and standard deviation of measurements performed by the micro-titrator at each sampling point, while blue open circles depict standard acidimetric measurements in (a) and a single Gran-plot titration in (b). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

200 ␮mol L−1 of CaCl2 were added. We observed that the addition of AQDS and the background electrolyte (CaCl2 ) close to approximated values of the sediment porewater concentrations was essential to improve the measurements. Further, using higher concentrations of both, AQDS or CaCl2 , did not considerably improve the measurements. The resulting vertical porewater concentration profile traced the dissolution of calcite in the sediment and thus the increase of alkalinity appropriately and compared well with the much more laborious Gran-plot titration which also requires larger sample volumes. 4. Conclusion The presented alkalinity micro-titrator can easily be produced in the laboratory once the procedure is established. Its small size, low cost, low power consumption and off-grid application make it ideally suited for measurements in small sample volumes, for on-

Please cite this article in press as: T. Steinsberger, et al., A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters, Sens. Actuators B: Chem. (2017), https://doi.org/10.1016/j.snb.2017.09.191

G Model SNB-23282; No. of Pages 6

ARTICLE IN PRESS

6

T. Steinsberger et al. / Sensors and Actuators B xxx (2017) xxx–xxx

site and field applications. The facility is powered and controlled via a 5 V USB port of a laptop. The optimized cell design allowed about 400 highly reproducible alkalinity titrations in oxic surface waters with one wiring board. The lifespan was limited by the mechanical and chemical stability of the pH sensitive RuO2 layer, which is, however, prone to poisoning with reduced compounds that may occur in natural waters under reducing conditions. The easy handling, small size and low power consumption suggests the use of the micro-titrator for automation and application with submersible insitu profilers to provide quasi-continuous recordings of alkalinity in the water column of lakes or for the temporal on-site monitoring of river waters. Acknowledgements We thank Christian Ebi (Eawag Dübendorf) for his support with the circuit board design. Christian Dinkel (Eawag Kastanienbaum) is acknowledged for valuable discussions and technical help. This work was financially supported by SNF Grant 200021 146234. We thank three reviewers for their comments that helped to clarify and improve this paper. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]

W. Stumm, J.J. Morgan, Aquatic Chemis, Wiley, 1996. B. Müller, J.S. Meyer, R. Gächter, Limnol. Oceanogr. 61 (2016) 341–352. R. Carignan, Nature 317 (1985) 158–160. D.P. Schrag, J.A. Higgins, F.A. Macdonald, D.T. Johnston, Science 339 (2013) 540–543. F. Segara, T. Kobayashi, T. Tajima, H. Ijyuin, I. Yoshida, D. Ishii, K. Ueno, Anal. Chim. Acta 261 (1992) 505–508. Q. Li, F. Wang, Z.A. Wang, D. Yuan, M. Dai, J. Chen, J. Dai, K.A. Hoering, Environ. Sci. Technol. 47 (2013) 11139–11146. M.G. Afshar, M.L. Tercier-Waeber, B. Wehrli, E. Bakker, Geochem. Perspect. Lett. 3 (2017) 85–93. M.G. Afshar, G.A. Crespo, E. Bakker, Angew. Chem. 127 (2015) 8228–8231. B. Karlmark, P. Jaeger, H. Fein, G. Giebisch, Am. J. Physiol. − Renal Physiol. 242 (1982) F95–F99. W. Olthuis, J. Luo, B.H. van der Schoot, J.G. Bomer, P. Bergveld, Sens. Actuators B: Chem. 1 (1990) 416–420. O.T. Guenat, W.E. Morf, B.H. van der Schoot, N.F. de Rooij, Anal. Chim. Acta 361 (1998) 261–272. C. Colombo, T. Kappes, P.C. Hauser, Anal. Chim. Acta 412 (2000) 69–75.

[13] B. Gaida, K. Assmann, Technologie der Galvanotechnik, Eugen G. Laulgau, Leuze, 1996. [14] F.A. Lowenheim, Modern Electroplating, Wiley, New York, 1974. [15] G. Gran, Acta Chem. Scand. 4 (1950) 559–577. [16] G. Gran, A. Johansson, S. Johansson, Analyst 106 (1981) 1109–1118. [17] B. van der Schoot, P. van der Wal, N. de Rooij, S. West, Sens. Actuators B: Chem. 105 (2005) 88–95. [18] P. Kurzweil, Sensors 9 (2009) 4955.

Biographies Thomas Steinsberger received his MSc from the department of geology at LMU Munich, Germany. Currently, he is a PhD student at ETH Zürich and Eawag in Kastanienbaum, Switzerland, working in the field of aquatic geochemistry of freshwater lakes. Patrick Kathriner In 2010, Patrick Kathriner finished his training as a Swiss Federal Laboratory Assistant. He completed his diploma as Swiss Federal Laboratory Technician in Life Science in 2013. Since then, he works as laboratory technician at the Swiss Federal Institute of Aquatic Science and Technology (Eawag). Philipp Meier graduated in Environmental Engineering from ETH Zurich. After his PhD at the Institute of Environmental Engineering at ETH, he worked as a postdoctoral researcher at the University Laval in Quebec City, Canada, developing tools for economic optimization of hydropower operation. After a position as software developer at the University College London and the University of Manchester, he joined Eawag in 2014 as project scientist working on strategies for minimizing ecological impacts of hydropower production. Peter C. Hauser carried out his undergraduate studies in Switzerland and then obtained an MSc at the University of British Columbia (UBC) under Prof. M. W. Blades (1985), followed by a PhD at LaTrobe University (Melbourne, Australia) under Prof. R. W. Cattrall (1988). Following a lectureship at Auckland University (New Zealand) in 1996 he took up his current position as Associate Professor at the University of Basel. His research interests in the analytical sciences have always included electronic aspects and he has been designing electronic analytical devices since the 1980s. Beat Müller received his MSc from the department of chemistry at ETH Zürich, Switzerland. After his PhD at Eawag, the Swiss Federal Insttute of Science and Technology under the guidance of Profs. Laura Sigg and Werner Stumm, and a Post Doc with Prof. Peter Hauser at the University of Auckland, New Zealand, he developed and applied chemical sensors at the sediment-water interface of natural surface waters to determine fluxes of compounds. Presently, he is a senior scientist at the department of surface waters at Eawag working in the field of aquatic geochemistry of freshwater lakes and rivers.

Please cite this article in press as: T. Steinsberger, et al., A portable low cost coulometric micro-titrator for the determination of alkalinity in lake and sediment porewaters, Sens. Actuators B: Chem. (2017), https://doi.org/10.1016/j.snb.2017.09.191