Invited Paper
Data acquisition system for ion-selective potentiometric sensors Andrzej Filipkowski, Jan Ogrodzki, Leszek J. Opalski, Radosław Rybaniec, Piotr Z. Wieczorek* Warsaw University of Technology, Institute of Electronic Systems, Nowowiejska Street 15/19, 00-665 Warsaw, Poland ABSTRACT The paper presents an idea and directives on construction of a measurement system for estimation of ions' concentration in water. System presented in paper has been fully designed and manufactured in Warsaw University of Technology in Institute of Electronic Systems. The measurement system works with cheap ion-selective potentiometric sensors. System allows for potentiometric, transient response and voltamperometric measurements. Data fusion method has been implemented in the system to increase the estimation's accuracy. Presented solution contains of many modern electronic elements like 32bit ARM microcontroller, precise operational amplifiers and some hydraulics subsystems essential for chemical measurements. Keywords: WARMER project; potentiometric electrode; ion-selective sensor, Nikolsky-Eisenmann model, ARM CORTEX microcontroller
1. INTRODUCTION Water pollution becomes more and more a serious problem, bringing the risk of environment degradation dangerous for inhabitants of our planet. The 6th European Framework Programme for Research and Development has many projects concerning environment. Project WARMER (WAter Risk Management for EuRope) is a continuation of the FP5 project SEWING (System for European Water monitorING) and has as an objective creation a real-time system for risk management through water monitoring, being universal, reliable, low-cost and easily accessible. Two separate systems will monitor the quality of water: in-situ probes on rivers, lakes and sea and remote probes located on satellites. Many chemical and physical properties of water will be observed. The project WARMER started in September 2006 and is now near its conclusion. Nine institutions from 7 countries are partners, with a SME SYSTEA from Italy as co-ordinator. Two faculties from Warsaw University of Technology (WUT) in Poland are responsible for creation and fabrication of potentiometric chemical sensors and for software and hardware processing of data obtained from these sensors in a reliable and sufficiently accurate way. Institute of Electronic Systems of WUT is responsible for the second problem: software and hardware for data processing. The task is quite involved, as the following properties must be taken into account: − − −
in water samples to be monitored a quite large matrix of different chemical interfering pollutants exists, the accuracy, selectivity and stability of miniaturized and cheap chemical potentiometric sensors is far from being ideal and this must be taken into account, the equipment must work without direct human intervention for a period of at least one month, so the hydraulic, mechanical and controlling parts must avoid, detect and correct automatically any defect or inaccuracy.
Almost three years of work with the project WARMER solved most of these problems, but we are aware that some more improvements are still necessary and possible. The sensors produced in the Department of Analytical Chemistry of WUT are selective for 5 ions, being the most common nutrients pollution in water resources: Na+. K+, Cl+, NH4+ and NO3-. *
[email protected] , phone 48 22 234 7336
Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2009, edited by Ryszard S. Romaniuk, Krzysztof S. Kulpa, Proc. of SPIE Vol. 7502, 750226 · © 2009 SPIE · CCC code: 0277-786X/09/$18 · doi: 10.1117/12.838257 Proc. of SPIE Vol. 7502 750226-1
The selectivity is not ideal and the selectivity coefficients range from 0.1 to 0.001. In the probe several of these sensors are placed and the signals obtained from them must be processed in such a way, as to get information about the density of each polluting ion with accuracy of at least 10%. The problem becomes quite difficult when the densities of interfering ions become meaningful: the dependence of voltage from sensors on logarithm of ion density is then nonlinear and cannot be explicitly formulated in mathematical way. Two methods of measurements try to solve this problem: Standard Addition Method (SAM) and Pre-Calibration Method (PCM). One of them is described in more details in Section 4. These methods solve also the problem of sensors’ offset voltage changing in time and deteriorating the reliability of measurement. The problem of in-situ measurements without direct human intervention is also quite difficult to solve. The samples of monitored water and calibrating liquids are passed automatically to the fluidic system of the probe and their quantity must be very accurate. The bubbles of air must be avoided. Some interesting ideas of mechanical, hardware and software solutions to get better reliability and accuracy of water pollution monitoring are described in Sections 2 and 3. This paper describes the last year achievements of WUT in the project WARMER. A difficult compromise had to be done between reliability and accuracy from one side and simplicity and price from the other. Planned in-the-field measurements made on the river Danube by the Austrian partner of the project will show finally how this compromise has been solved. The text in the subsequent sections is based on the papers presented on the special WARMER session of the 2009 IEEESPIE Joint Symposium in Wilga, Poland.
2. GENERAL CONCEPT OF A MEASUREMENT SYSTEM BASED ON ION-SELECTIVE POTENTIOMETRIC SENSORS 2.1. General concept of system – electrical requirements and circuit implications The main purpose of the system described in this paper is to estimate the concentration of four types of ions in a solution. Ions are ammonium, potassium, nitrate and chloride. The system should in its basic configuration work with potentiometric sensors. Thus the basic principle of operation of the measurement system is based on a precise voltage measurement. The DC voltage which is the subject of interest is observed between the ion selective sensor's contact and the reference electrode (fig. 1.).
Fig.1. Basic concept of the measurement system. Vsrc and Isrc are analog current and voltage sources. Analog subcircuits have separate analog ground – AGND
Ion-selective sensors used in presented basic concept of the system have logarithmic dependence between voltage and ions' concentrations (activities). The shape of voltage vs. solutions concentration curve can be described by a NikolskyEisenmann1 model, or other more precise models1.
Proc. of SPIE Vol. 7502 750226-2
As it was mentioned before four types of ions are the subject of interest, thus the system should contain at least four precise voltage preamplifiers. Outputs of preamplifiers should be buffered and attached to an analogue multiplexer which can be attached to the input of an analog to digital converter. The assumptions2 regarding precision of estimation of ions concentration force at least 12bits of ADC resolution and 5V conversion range. The precise DAC can be a part of a modern sophisticated microcontroller like ARM3, or as a separate chip connected by a standard SPI interface with the main circuit which manages the whole system. The nature of measurements of high impedance DC sources like the ion selective potentiometric sensors implies more handicaps in system's construction. The DC resistance of sensors varies from 5 GΩ up to 30 GΩ4. Such signal sources are highly prone to the electromagnetic fields, electrostatic fields due to very low leakage currents, and low frequency noises. This obstacles are complicating the analogue and digital parts of the system. First of all the preamplifiers (fig.1.) must have an ultra low bias current less than 100 pA. Such currents at constant level did not influence the voltages on sensors during the laboratory tests. Secondly the input resistance of preamplifiers should be significantly larger than the DC input resistance of measurement set consisting of a ion selective structure, solution under tests, and reference electrode. Otherwise the measured voltage might be significantly depreciated by a voltage divider consisting of a ion selective sensor's output resistance and the input resistance of the preamplifier. As it was mentioned, laboratory tests showed that currents less than 100 pA did not significantly influence the voltage readouts from sensors and currents below this value had non-destructive influence on electrical parameters of sensors. However variability of such current, even less than 100 pA (caused by the polarizing current of preamplifier or any other circuit connected to preamplifier's input) might influence the readouts. Thus the preamplifiers must both have very low bias current and low drift (thermal drift) of bias current. In a real system ion selective sensors must be attached to the inputs of preamplifiers by a cable with a ground shielding. Such cable always has its leakage current between signal wire and ground shielding. To reduce this current its necessary to use a triaxial cable with an additional shielding between ground and signal wire. Additional shielding called guard has almost the same potential as the potential on cable's signal wire connected to the ion selective sensor. Guard's potential must be created by the analogue part of the system. Voltage from the output of preamplifier must be additionally buffered and applied to the guard shielding of the cable connecting sensor with analogue electronics (fig.1.). Such wiring solution theoretically reduces the influence of leakage current and electromagnetic fields, however the risk of preamplifiers instability occurs. Measurement system concept described in this paper should also have capabilities of execution of some additional measurement procedures like transient response measurements5, and procedures which are not directly responsible for particular methods of ions' concentration estimation but for sensors' or whole system self diagnostics. Such requirements imply the use of some additional analog circuitry. For the purpose of sensors self diagnosing procedures an ability of DC voltage stimulation of sensors is required4. Once the stimulation is applied to the sensor the ability of current measurement flowing through the reference electrode or the sensor is necessary. This way a DC resistance of sensors can be determined. Sensors with malfunctioning membrane have significantly lower resistance4. The transient response measurement technique might be also very useful in estimation of ions' concentration5. This method requires also a capability of voltage stimulation of sensor (by a pulse signal) and the measurement of current variability in time. 2.2. System requirements concerning hydraulics Sensors which are used in the system described in this paper have high variability of parameters. This parameters are slope of voltage vs. concentration characteristic's, offset voltage, sensitivity and selectivity. Thus in many cases initial sensor calibration must be performed. Calibration requires a set of solutions with known concentrations and known types of ions. I.e. characterization of chloride sensor requires two or three solutions with chloride ions on different levels. In case of determination of such parameter as selectivity (influence of other type of ion called interfering ion) at least three solutions are required with specially chosen levels of concentrations' of ions which are the subject of characterization – main ion and interfering ion (for example potassium). As it is shown in the fig.1. measurement system must contain several hydraulics elements like: valves, pump, vessel with measured solution (working tank), some additional vessels with distilled water and calibrating solutions (called calibrants). Valves and additional vessels are required for sensors' testing and characterization before proper measurements. Pump is needed for filling the hydraulic loop (fig.1.) with calibrating solution or with the solution under tests. Generally the pump can be a peristaltic pump, rotor pump, or a piston pump. However in case of some measurements techniques like Standard Addition Technique6 pump is responsible for dosing precise volumes of solutions to the solution under tests in working tank. Thus if the system is multi-purpose the hydraulic system should be applied
Proc. of SPIE Vol. 7502 750226-3
with a precise micro metering pump. The precision of the pump and its nominal output depends on the volume of measurement loop. In system presented in sec.3. the inaccuracy of pump is less than 20 µl and the volume of loop equals 30 ml.
3. DETAILS OF SYSTEM IMPLEMENTATION 3.1. Detailed block diagram of the system Sensor's Interface
Wireless Interface
RS GP
Sensors
RS-232
Computer Interface
Pump Controller Valves
Human Interface Solutions
Battery Operation
Fig. 2: Block diagram of the system
Fig 2. Shows block diagram of the system designed and built in WUT Institute of Electronic Systems. ARM CORTEXM3 microcontroller programmed in the GNU C environment is responsible to control all the subsystems. Sensor's interface block has ability to take measurements from multiple sensors and also to stimulate them with different types of excitations (voltage signals) for special types of measurements and diagnostics. Hydraulic subsystem is integrated, because of necessity of supplying fluids to the sensors and system's loop. GPRS is used as a wireless communication channel, it's advantages are: artificial unlimited range and ease integration with Internet. Reliable wireless link allows the device to perform environmental measurements without additional human supervision. Communication with PC computer is based on industrial standard, RS-232 interface. However, RS-232 -> USB converter can be easily integrated if needed. PC interface can be used for collection of large amount of data at laboratory. Human communication interface consist of a dedicated microswitch keyboard and LCD alphanumeric 4x16 display. In this way the system can work as a fully autonomic measurement equipment. Furthermore, the system is portable and prepared for a battery operation. 3.2. Electronic subsystems of the measurement system Voltage Current M easure Circuits
Stimulation Circuit
12bit ADC
Serial Peripheral Interface
10bit DAC
ARM CORTEX-M3 microcontroller
USART
General Purpose I/O
Pu m p Wire le s s Co m m u n ica t io n
Va lve s
LCD Dis p la y Ke yb o a rd
Fig. 3: Block diagram of the electronic subsystem
Proc. of SPIE Vol. 7502 750226-4
Fig.3. presents the electronic subsystem. The subsystem can be divided into two parts: analog and microcontroller (digital) circuits. Analog part of the subsystem consists of: • ion-selective potentiometric sensors • interface part: responsible for sensor's voltage and current measurements • stimulation circuit • power supply circuit In general sensors convert solution parameters (concentration) into voltage and current signals. This electric variables have to be amplified and buffered, and it is done by the interface circuit (voltage – current measure circuits in fig.3). The stimulation circuit is used for electrical excitations of the sensors. Stimulations are necessary to allow measurements other than the classic potentiometric measurements, like transient response, voltamperometric etc. The power supply circuit generates voltages for proper operation of all subsystems. Voltages must be stable and noisefree for reliable operation of precision analog electronics. Role of the microcontroller circuit is described in the next subsection. 3.3. Microcontroller – digital subcircuit of the system The STM32F101R8 microcontroller has been used for the design. The µC integrates 32-bit ARM CORTEX-M3 core, 10 KB of RAM and 64 Kilobytes of FLASH program memory3. The microcontroller can operate with clock frequencies up to 32MHz. This facts combined with the 32bit architecture result in powerful performance. The microcontroller has 12bit analog-to-digital converter on board. Converter is able to convert one million of samples from 16 channels at the time of one second. Of course such number of samples is not necessary, however probes have to be averaged. Moreover the period of averaging must be equal to the multiple of mains period. Furthermore µC has very large communication capabilities: integrated triple USART (Universal SynchronousAsynchronous Receiver-Transmitter), SPI (Serial Peripheral Interface), I2C interface and others. Power saving futures, essential for battery operation, are also supported. The microcontroller can be supplied with as low as 30uA of current in a sleep mode. Microcontroller's subsystem is responsible for controlling the whole measurement system; collecting, processing and sending the data farther. The detailed list of implemented functions is as follows: • converting the analog signal from the sensor's interface into the digital representation and to perform further calculations • triggering stimulation circuit and controlling external digital-to-analog converter (DAC) for adjusting electrical stimulation voltage (in transient measurements, voltamperometry, and self-diagnostics) • performing wireless communication (by GPRS modem) and PC communication (RS-232 interface), embedded Universal Synchronous-Asynchronous Receiver-Transmitter (USART) of the microcontroller is used for this task • programming micro metering pump – also done by USART • controlling valves, keyboard and LCD display – it is accomplished by General Purpose I/O ports (pins) of the microcontroller. 3.4. Sensor's interface circuit – analogue subcircuit of the system The interface circuit is responsible for collecting the data from the sensors, and to perform electrical stimulation if necessary. Following types of measurements are supported: 1) potentiometric – this is the classic type of measurements; DC voltage is measured between ion-selective electrode and the reference electrode of the sensor; no stimulation is applied 2) transient response investigation – square wave voltage signal is applied to the sensor; observation (recording) of current response for the stimulation is done; 3) voltamperometry – variable DC voltage signal is applied to the sensor; measurement of current flowing through the reference electrode or the sensors' contact is done.
Proc. of SPIE Vol. 7502 750226-5
S TIMULATOR CIRCUIT
VOLTAGE SHIFT
MOSFET
CONTROL
DAC
uC
++
INA1 1 6
BUFFER
ADC
VOLTAGE MEAS URE CIRCUIT 1 MΩ RES IS TOR
+
+
OP 0 7
BUFFER
ION S ELECTIVE ELECTRODE
CURRENT MEAS URE CIRCUIT REFERENCE ELECTRODE
Fig. 4. : Block diagram of the interface circuit
Fig 4. depicts the designed analogue subsystem. Preamplifier based on INA116 operational amplifier, is constantly amplifying the voltage difference between the ion-selective sensors and reference electrode. This voltage is afterwards buffered, adapted (shifted) and applied to the input of the ADC. The stimulation circuit is generally based on: MOS Field Electric Transistor; voltage shift circuit – using voltage from preamplifier to polarize the transistor properly, and a control circuit – triggering the MOSFET by a digital signal from the µC. Amplitude of the stimulation signal level is also adjustable by the µC through the DAC. The most critical parameter of described interface circuit is it's bias current. As it was mentioned in sec.2 it might influence the measurements' accuracy and also can be dangerous for the sensors structures if it is greater than hundreds of picoamperes. Special attention has been paid to optimize bias in this design. This included special care during phase of preselection of devices and specific PCB design The electronic elements that has been chosen for the purpose of building the interface circuit has been mentioned below: z Voltage Amplifier INA116 has been selected as the preamplifier. This is special instrumentation amplifier circuit, developed especially for electrochemical measurements. It's main feature is ultra low bias current, guaranteed to be as low as 25 fA at room temperature and less than 100fA at 85°C. Input stage Junction FETs, gives better electrostatic discharge immunity than MOSFET solutions. Integrated guard driver, helps reducing leakage current of input cable. z Current Amplifier Current measurement is accomplished by measuring a voltage on 1 MΩ resistance (used for setting the potential of the reference electrode). This voltage is amplified by OP07 precision low noise operating amplifier and after additional buffering supplied to the ADC. z Stimulation circuit Principal feature of this circuit is to not add any current to input bias of the interface. This is accomplished by special use of BF543 MOSFET. Voltage of sensor is being constantly monitored (by preamplifier), and is used to polarize the transistor. Source of the BF543 is directly connected to the ion-electrode of the sensor. When the electrical stimulation is disabled the gate voltage of the BF543 transistor is equal to the voltage on the sensor minus a constant value (subtraction is done by voltage shift circuit). In this way gate-source voltage is kept at the same level all the time, that disables channel in the transistor. The potential on a drain is kept at the same level as on source, so the drain-source voltage is equal to zero. In this way transistor is not conducting current to the sensor because of two reasons: there is no channel and the drain-source voltage is zero. Turning on the stimulation causes : potential on drain is now set to an arbitrary value of control circuit (fig.4.), and
Proc. of SPIE Vol. 7502 750226-6
because it is lower than the potential on the source, the substrate parasitic diode opens. Current from source to drain flows through the diode. This way a voltage of the sensor is set by the stimulation circuit. Stimulation circuit is designed this way that it is able to generate constant voltage and voltage pulses. First type of stimulation is used for determination of static characteristics of the sensor (or voltamperometry), second for transient analyses. Voltage level of the stimulation can be adjusted digitally with 10 bits precision using external DAC. 3.5. Power Supply Circuit System is designed to operate from 12V voltage source, for example standard battery pack. Power supply circuit's (fig.5.) task is to change and convert 12V source into voltages acceptable by other parts of the system. Analogue amplifiers operates with asymmetric voltages, and this fact complicates supply subsystem because third potential (reference potential) has to be generated. The reference potential at half of voltage supply level is called an artificial ground . Digital circuit operates at 3.3V, linear regulator is responsible for voltage decreasing and stabilizing. Step-up converter is used to convert 12V to 24V industrial standard voltage, which supplies pump and valves. Noise reduction techniques have been implemented in the system: decoupling circuits (LC low-pass filters) on supply lines have been used. Analog and digital grounds are connected only at one point and decoupled. Finally RS-232 standard requires higher voltage than supply provides, a MAX3232 charge pump generates proper voltages from 3.3 V potential.
DC/DC St e p -u p Co n v e rt e r Pu m p
Lin e a r Vo lt a g e Re g u la t o rs
Va lv e s
1 2 V Su p p ly Vo llt a g e
Art ificia l Gro u n d Ge n e ra t o r
Dig it a l Circu it
De co u p lin g An a lo g Circu it
Fig. 5. Diagram of the power supply subsystem
3.6. Hydraulic subsystem Substances being measured by the system have liquid nature. The hydraulic subsystem (fig.6.) is responsible for distribution of the fluids into the system. The subsystem consists of: • Micro metering pump capable of dosing precisely selected amounts of fluids in two directions. • set of six valves controlling path of solutions movement • working tank used for mixing fluids • sensor cells, where ion-selective electrodes contact the measured substance Control scheme of emptying the hydraulic system has been described below: 1) all valves are initially closed (no fluid can pass) 2) valve V opens, pump works forward (in direction to the working tank (WT)); fluid from sensor cells is transferred into WT; sensor cells are filled with air (because working tank is not hermetic) 3) valve V closes, valve VI opens, direction of pomp is switched backward; fluid from working tank leaves the system 4) pump stops, all valves are closed
Proc. of SPIE Vol. 7502 750226-7
Filling the sensor cells with proper fluid (water or sample under tests) is done as follows: 1) emptying the system 2) one of valves I-IV opens, pump works forward in WT direction 3) valves I-IV are closed, valve V opens, the direction of the pump is switched backward; sensor cells are being filled by fluid from the working tank 4) the pump stops, all valves are left closed
S o lu t io n I
S o lu t io n II
S o lu t io n III
S o lu t io n IV
I
II
III
IV
VALVE
V S e n s o r Ce lls VI
AIR
PUMP
WAS TE
Wo rkin g Ta n k
Fig. 6. Diagram of the hydraulic subsystem
4. METHOD FOR IONS’ CONCENTRATION ESTIMATION IMPLEMENTED IN THE SYSTEM System presented in details in previous section still remains useless for purpose of estimation of ions concentration unless a numerical method of data processing6 is applied. As it was mentioned in section.2. of this paper each sensor of identical type might have totally different parameters describing its calibration curve. Thus the curve determining the dependency between voltage and main ion concentration (activity) must be recognized. After recognition of the curve – so called calibration – the proper measurement can be done. The idea of the ions' concentration (activities) estimation is depicted in fig.7.
Proc. of SPIE Vol. 7502 750226-8
Fig.7. The idea of model recognition (calibration) and model inversion for estimation of concentration in a sample (water under test)
The calibration curve is recognized according to Nikolsky-Eisenmann model (1). M S z /z U = U 0 + lg( K m ,i ai ) , Km,m=1 (1) zm i =1 For determination of unknowns (1) solutions with known concentrations (activities) are applied. After the calibration curve is characterized (according to model (1)) the formula has to be inverted. This way the unknown concentration (ion activity) corresponding to a measured voltage can be figured out. As it is shown in (1) a full recognition of a calibration curve has to be performed, otherwise voltage readouts and corresponding concentrations might be miscalculated in case of presence of interfering ions'. For the purpose of identification of sensors' sensitivity (sensitivities) to interfering ions at least three solutions with different concentrations of main ion and one interfering ion have to be applied during the calibration process (fig.7.). Also the data from at last one more sensor (for which the interfering ion is the main one) have to be acquisited. This approach can be called a multiparameter approach. Each time during calibration a set of equations describing sensors have to be solved according to voltage readouts corresponding to calibrating solutions' concentrations. This method makes fusion of read-outs from related sensors (i.e. potassium and ammonium sensors). Fig.8. depicts the influence of interfering ions' on concentration's estimation when the model is not fully recognized (the sensors selectivities are omitted). In this case even ideally modeled and calibrated single sensor has an uncertainty of measurement, dependent on its selectivity and activities (concentrations) of interfering ions, only a full recognition of calibration curve according to model (1) gives an opportunity for precise estimation of main ions' concentrations (activities). Sensor response in a binary solution
∑
m
i
read-outs
estimates of the main ion activity Fig.8. Influence of interfering ion on the calibration curve. Two levels of concentrations' of interfering ion are depicted.
Proc. of SPIE Vol. 7502 750226-9
5. CONCLUSIONS AND RESULTS A fully functional system has been designed and built in the WUT laboratory in Institute of Electronic Systems. The concept of analogue circuits has been verified in a real built measurement system with potentiometric sensors attached. The instability of voltage readouts was less than 0.1 mV. Such low variability of readouts gives ability for good recognition of calibration curve2, and proper estimation of measured solutions’ concentrations. The data fusion method based on the calibration has been also implemented in the system, however the method has not been checked in the situation when interfering ions were present in the measured solution. The full functionality of the calibration method has been verified with Systea µLFA7 system, with similar hydraulics concept, but less sophisticated analogue electronics subcircuits. The precision of ions' concentration was surprisingly high – for ammonium ions total inaccuracy was less than 6.5% with fusion of readouts from two sensors, and 2.8% for potassium ions without the sensor's selectivity estimation. It is worth mentioning that µLFA has larger variability of voltage readouts than system presented in this paper. That means that a fully implemented data fusion from two or more sensors in presented system might have significantly smaller inaccuracy of estimations. It is worth mentioning, that presented system is quite inexpensive, despite the micro metering pump, whole set of valves, and electronic parts cost 200$. If the system is used only with calibration method it will not be necessary to use expensive micro metering (dosimetric) pump like Dosca8 pump used in the presented system.
6. REFERENCES 1. J. Ogrodzki , „Diversity of models for ion-selective sensors - behavioral and physical approaches” , XXIV IEEESPIE Joint Symposium on Photonics, Web Engineering, Electronics for Astronomy and High Energy Physics Experiment, Wilga, May 2009 2. J. Ogrodzki , „Modeling of potentiometric sensors sensitive to ions of valency one and two for possible applications in WARMER project”, Proceedings of the SPIE, Volume 6937, pp. 69372I-69372I-8 (2008) 3. STMicroelectronics, „ARM-based 32-bit MCU with 64 or 128 KB Flash, 6 timers, ADC and 7 communication interfaces”, http://www.st.com/mcu/devicedocs-STM32F101R8-110.html 4. P.Z. Wieczorek, L.J Opalski, and J. Ogrodzki, “Electrical properties of potentiometric sensors – an empirical study”, Proceedings of the SPIE, 6937, 0277-786X, (2007) 5. P.Z. Wieczorek, L.J Opalski, “An empirical study of transient responses of potentiometric ion sensors”, Proceedings of the SPIE, Volume 7124, pp. 71240V-71240V-9 (2008) 6. L.J. Opalski, „Software support of multiparametric analysis of water with miniaturized ion-selective potentiometric electrodes – FP6 EU WARMER project perspective”, XXIV IEEE-SPIE Joint Symposium on Photonics, Web Engineering, Electronics for Astronomy and High Energy Physics Experiments,Wilga Poland, May 2009 7. L.J. Opalski , P.Z.Wieczorek, P.Moscetta , L.Sanfilippo, M.Malizia, “Multiparametric microloop flow analyzer with miniaturized ion-selective electrodes for quality analysis of chemical parameters”, Southampton workshop, 30th march 2009 8. DOSCA, „Programmable micro metering pumps ”, http://www.reoterm.com.br/central/Homogeneizadores/micro_metering_pump.pdf
Proc. of SPIE Vol. 7502 750226-10
