JINSONG YU. Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy. Thesis Adviser: Dr

DEVELOPMENT OF MICROFABRICATED ELECTROCHEMICAL SENSORS FOR ENVIRONMENTAL PARAMETER MEASUREMENTS APPLICABLE TO CORROSION EVALUATION AND GASEOUS OXYGEN ...
Author: Bethanie King
1 downloads 0 Views 3MB Size
DEVELOPMENT OF MICROFABRICATED ELECTROCHEMICAL SENSORS FOR ENVIRONMENTAL PARAMETER MEASUREMENTS APPLICABLE TO CORROSION EVALUATION AND GASEOUS OXYGEN DETECTION

by JINSONG YU

Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy

Thesis Adviser: Dr. Chung-Chiun Liu

Department of Chemical Engineering CASE WESTERN RESERVE UNIVERSITY

MAY, 2008

CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Jinsong Yu _____________________________________________________

Ph.D. candidate for the ______________________degree *.

Chung-Chiun Liu

(signed)_______________________________________________ (chair of the committee)

________________________________________________

Uziel Landau

________________________________________________

Heidi Martin

________________________________________________

Kenneth Loparo

________________________________________________

________________________________________________

March 31th, 2008

(date) _______________________

*We also certify that written approval has been obtained for any proprietary material contained therein.

Dedication This work is dedicated to my wife Lina Li for her continuous support and understanding and to my new born daughter Amy Yue Yu who has made my life a much more meaningful one

Table of Contents Table of Contents…………………………………………………………………………iv List of Tables……………………………………………………………………………viii List of Figures………………………………………………………………………….....ix Acknowledgements………………………………………………………………………xii Abstract………………………………………………………………………………….xiii Chapter 1 Introduction and Motivation………………………………………………..1 Chapter 2 Sensor Design and Fabrication…………………………………………….10 2.1 Fabrication Techniques……………………………………………………....11 2.1.1 Photolithography…………………………………………...11 2.1.2 Sputtering: a method of physical vapor deposition………...19 2.1.3 Lift-off processing…............................................................22 2.1.4 Thick film pringting technology…………………………...24 2.2 Sensor Fabrication Process…………………………………………………..28 2.2.1 Thin film high temperature oxygen sensor………………...28 2.2.2 Thin film conductivity sensor……………………………...33 2.2.3 Thick film pH sensor and dissolved oxygen sensor……….35 2.3 Summary…………………………………………………………………......40

iv

Chapter 3 AC Impedance Measurement of Conductivity Sensor…………………...42 3.1 Introduction to AC Impedance Measurement………………………………..42 3.2 Fundamentals of AC Impedance Technique…………………………………44 3.3 Equivalent Circuit……………………………………………………………48 3.4 Experimental…………………………………………………………………53 3.5 Results and Discussion………………………………………………………61 3.5.1 Determination of the solution resistance…………………………..61 3.5.2 Bulk resistivity/conductivity and cell constant calculation………..69 3.6 Summary…………………………………………………………………….75 Chapter 4 Open Circuit Potential pH Measurement………………………………..77 4.1 Introduction…………………………………………………………………77 4.2 Principle of Pd/PdO pH sensing…………………………………………….80 4.3 Experimental………………………………………………………………..83 4.4 Results and Discussion……………………………………………………..86 4.5 Summary……………………………………………………………………90 Chapter 5 Amperometric i-t Oxygen Concentration Measurement….....................93 5.1 Introduction………………………………………………………………...93

v

5.1.1 Dissolved oxygen measurements…………………………………..93 5.1.2 Microbabricated oxygen sensor……………………………………96 5.2 Fundamentals………………………………………………………………...98 5.2.1 Cyclic voltammetry………………………………………………..98 5.2.2 Potentiometric measurements…………………………………….102 5.2.3 Amperometric i-t measurements………………………………….104 5.3 Experimental………………………………………………………………..105 5.4 Results and Discussion……………………………………………………..108 5.5 Summary…………………………………………………………………...122 Chapter 6 Calibration of Platinum RTD and Evaluation of High Temperature Oxygen sensor………………………………………………………………………....125 6.1 Introduction………………………………………………………………...125 6.2 Fundamentals……………………………………………………………....130 6.2.1 Resistance Temperature Detector………………………………..130 6.2.2 Principle of the limiting current amperometric measurements….133 6.2.3 Principle of the potentiometric measurements…………………..134 6.2.4 Mechanism of oxygen transport………………………………....135 6.3 Experimental……………………………………………………………….138 vi

6.4 Results and Discussion……………………………………………………..140 6.5 Summary……………………………………………………………………159 Chapter 7 Summary of This Study and Recommendations for Future Work…….163 7.1 Summary of This Study…………………………………………………….163 7.2 Recommended Future Work………………………………………..............166 Bibliography…………………………………………………………………………...168

vii

List of Tables Table 2-1 Deposition time for sputtering of materials…………………………………..21 Table 2-2 Ink information for thick film printing process………………………………27 Table 4-1 Comparison of meter measured values and nominal values of pH buffer solutions…………………………………………………………………………………87 Table 5-1 Peak currents at various reduction potentials for dissolved oxygen sensor…109 Table 5-2 Summary of effective oxygen diffusion coefficient…………………………121 Table 6-1 Important features of various types of temperature sensing elements………130 Table 6-2 Summary of heater voltage and corresponding temperature………………..144

viii

List of Figures Figure 2-1 Comparison of light field and dark field…………………………………….14 Figure 2-2 Schematic of thin film microfabrication process……………………………23 Figure 2-3 Illustration of the thick film printing process………………………………..26 Figure 2-4 Fabrication process of high temperature oxygen sensor…………………….28 Figure 2-5 High temperature oxygen sensor side view………………………………….29 Figure 2-6 Silicon wafer after all the sputtering processes……………………………...32 Figure 2-7 Diced sensor chip on packaging material……………………………………32 Figure 2-8 Packed testing device for high temperature oxygen sensor………………….33 Figure 2-9 Conductivity sensor design and wafer layout………………………………..34 Figure 2-10 Silicon wafer for conductivity sensor before dicing and packaging………..34 Figure 2-11 Design and wafer layout for pH sensor…………………………………......37 Figure 2-12 Design and dimension of the dissolved oxygen sensor…………………......38 Figure 2-13 Pd/PdO based solid state pH sensor………………………………………...39 Figure 2-14 Three-electrode configuration dissolved oxygen sensor with RTD………...39 Figure 3-1 Frequency and pahse response……….……………………………………....43 Figure 3-2 Sinusoidal voltage and the current/voltage relationship……………………..44 Figure 3-3 Typical Nyquist plot of an electrochemical system..………………………...47 Figure 3-4 Typical Bode plot of an electrochemical system…………………………….47 Figure 3-5 Three-component equivalent circuit…….. ………………………………….50 Figure 3-6 Sample of the Nyquist plot…………………...……………………………...50 Figure 3-7 Sample of equivalent circuit of a complex system…….…………………….51 Figure 3-8 Sample of equivalent circuit of a conductance cell………………………….53 Figure 3-9 Testing cell design for conductivity sensor...………………………………..54 Figure 3-10 Conductivity sensor and assembled testing cell…………………...……….54

ix

Figure 3-11 Typical AC impedance of 4mm NaCl solutions (a) (b) (c)………………...55 Figure 3-12 AC impedance of 8mm NaCl solutions………………………….………....57 Figure 3-13 AC impedance of NaCl solutions at fixed electrode distance (a) (b) (c)......58 Figure 3-14 AC impedance of 8mm thick 0.001M NaCl solution (a) (b) (c) …………..60 Figure 3-15 Sample of real part impedance as a function of frequency………………...63 Figure 3-16 Real resistance for an 8mm thick layer…………………………………….64 Figure 3-17 Real resistance vs. solution thickness for pure NaCl solution……………..64 Figure 3-18 Illustration of solution thickness effects on real resistance………………..65 Figure 3-19 Comparison of resistance in NaCl solutions (a) (b) (c)……………………67 Figure 3-20 NaCl concentration effects on 4mm particulate layer……………………..68 Figure 3-21Analogy of solution/particulate layer to a conventional resistor…………...69 Figure 3-22 Resistance plots for particulate layer saturated with NaCl solutions (a) (b) (c)………………………………………………………………………………………..70 Figure 3-23 Conductivity of 0.001M KCl solution as a function of solution thickness…72 Figure 3-24 Cell constant as a function of solution thickness for KCl solutions……….74 Figure 4-1 Cyclic voltammetry of Pd/PdO based pH sensor……………………….…..85 Figure 4-2 Open circuit potential of Pd/PdO based pH sensor…………………………85 Figure 4-3 Rate of increase in open circuit potential measurements.…………………..86 Figure 4-4 Open circuit potential measurements of pH buffer solutions……………….88 Figure 4-5 Open circuit potential test of pH buffer solutions….……………………….88 Figure 4-6 Collective first run calibration curve for 9 different pH sensors.……….......89 Figure 4-7 Calibration curve of Pd/PdO based pH sensors……………………………..89 Figure 5-1 Linear Potential sweep for cyclic voltammetry……………………………..99 Figure 5-2 Cyclic Coltammogram…………………………………………………….100 Figure 5-3 Schematic of an electrochemical cell……………....……………………..102

x

Figure 5-4 Cyclic voltammetry of dissolved oxygen sensor...…………………………107 Figure 5-5 Real time amperomatric i-t curve at -0.5v for dissolved oxygen sensor...…107 Figure 5-6 Cyclic voltammograms of dissolved oxygen sensor……...………………..108 Figure 5-7 Effects of reduction potential on peak currents……….……………………110 Figure 5-8 Amperometric i-t response of the oxygen sensor to oxygen concentration cycling………………………………………………………………………….……....111 Figure 5-9 Comparison of amperometric i-t measurements at -0.55v and -0.6v...…….112 Figure 5-10 Reproducibility of first run amperometric i-t measurements at -0.55v for 3 different dissolved oxygen sensors………………………..…………………...……....113 Figure 5-11 Reproducibility of amperometric i-t measurements at -0.55v for a single dissolved oxygen sensor chip………………………………………..………………....114 Figure 5-12 Top view and side view of testing device assembled (a) (b).....……….....115 Figure 5-13 Amperometric i-t measurements of oxygen testing device assembled.......116 Figure 5-14 Rate of change in current for amperometric i-t measurements…………...117 Figure 5-15 Calibration curve for oxygen sensor chips (a) (b) (c)….………………....118 Figure 5-16 Cottrell plots for dissolved oxygen sensors (a) (b) ….……………………120 Figure 6-1 Schematic of an electrochemical cell………...………………………..…...134 Figure 6-2 Schematic of perovskite structure..………………………………………..136 Figure 6-3 Schematic of RTD elements……….……………....………………………140 Figure 6-4 RTD element resistance vs. temperature for big sensor...………………....141 Figure 6-5 RTD element resistance vs. temperature for small sensor...........................142 Figure 6-6 Heater voltage and corresponding RTD response for big sensor…...…….143 Figure 6-8 Typical open circuit potential results…………………………….……….145 Figure 6-9 First derivative result of a typical OCP measurements………….....…….146 Figure 6-10 Temperature effect on sensor response...…..…………………...………147 Figure 6-11 Big sensor amperometric element OCP at various temperatures (a) (b) (c)……………………………………………………………………………………...148 xi

Figure 6-12 Temperature effects on OCP measurements of amperometric element in air (a) (b)…………………………………………………………………........………....150 Figure 6-13 Typical OCP measurements for the oxygen sensors (a) (b) (c) (d) (e) (f)…………………………………………………………………………………......151 Figure 6-14 Calibration curves for the oxygen sensors (a) (b) (c) (d) (e) (f) (g) (h)...155

xii

Acknowledgements

I would like to thank my thesis advisor, Professor Dr. Chung-Chiun Liu for his guidance, encouragement, and patience through all these years. His supervision and teachings has been an important asset to the successful completion of this work. Further thanks go to Dr. Uziel Landau, Dr. Heidi Martin, and Dr. Kenneth Loparo for serving on my defense committee. I would also like to achnowledge the support and help of my labmates and coworkers in the Electronics Design Center (EDC) and the department of Chemical Engineering at Case Western Reserve University. Special thanks go to Meijun Shao, Laurie Dudik, and Shubin Yu for their technical advice, great skills, and help with my research work. Finally, I would like to acknowledge my family and my friends for their support through out my thesis work. Funding for this work: NASA Grants, DOE Grants, Delta Environmental and Educational Foundation

xiii

Development of Microfabricated Electrochemical Sensors for Environmental Property Measurements Applicable to Corrosion Evaluation and Gaseous Oxygen Detection

Abstract by Jinsong Yu Solid state electrochemical sensors have demonstrated their usefulness for measuring the physical and the chemical properties of the electrolytes in both liquid and gaseous

phases.

These

electrochemical

sensors

provide

rapid

response

to

property/concentration changes with high sensitivity, excellent reproducibility, and good long-term stability. Using thin and thick film microfabrication techniques, these solid state electrochemical sensors can be further improved and miniaturized to incorporate additional advantages such as high resolution, small feature size (µm or even nm), small required sample volume, small ohmic potential drop, improved signal to noise ratio, and reduced fabrication cost.

Advanced analytical and computational methods for the evolution of the environment on metal surfaces require data for the environmental properties of thin layers of moisture, moist particulates, and deposits that will affect the corrosion performance of metals. The advanced sensor technology development and application of the sensors reported in this study address these needs. The sensor development and findings reported here are important complements to a multi-investigator program that strives for increased scientific understanding, enhanced process models, and advanced technologies for longterm corrosion performance.

xiv

In the first part of this study, microfabricated electrochemical sensors for environmental parameter measurements have been developed and characterized, including Pt based thin film electrolyte solution conductivity sensor, Pd/PdO based thick film pH sensor, and Au based thick film dissolved oxygen concentration sensor. These sensors utilized AC impedance, potentiometric open circuit potential decay, and amperometric i-t techniques for conductivity, pH, and dissolved oxygen concentration measurements, respectively.

In the second part of this study, a multi-component, multi-layer structured thin film high-temperature gaseous oxygen sensor incorporating two major types of gaseous oxygen sensing principles has been developed and evaluated. This sensor is capable of performing high-temperature potentiometric and amperometric measurements for gaseous oxygen concentration monitoring. Silicon-based micro fabrication processes are used in this study including photolithography patterning, thin film sputtering metallization, thick film printing technique, lift-off process, and others. The sensors developed are geometrically well defined, reproducible, mechanically robust, and potentially low cost. The test results in this study indicated that the Pt conductivity sensor can measure the conductivity of particulate layer saturated with electrolyte solutions; the Nernstian Pd/PdO pH sensor can measure electrolyte solution pH value in the alkaline region; the Au dissolved oxygen sensor can measure oxygen concentration in solution and the effective oxygen diffusion coefficient in solution and particulate layer; and the hightemperature oxygen sensor can measure gaseous oxygen concentration.

xv

Chapter 1 Introduction and Motivation Corrosion has been a subject that scientists have tried to understand and control ever since they started to use metal objects.

[1]

Corrosion is defined as a process of

deterioration or degradation of metal components due to contact with a corrosive environment. Metallic corrosion is an electrochemical process. The metal’s deterioration or degradation can be concentrated locally forming a pit or a crack, or it can extend across a wide surface area. The corrosion behavior of a metal is determined by the combination of inherent corrosion resistance of the metal and the corrosivity of the environment. This work has focused on determining the properties of a corrosive environment comprised of a thin layer of electrolyte or a layer of moist particulate on the metal surface. The quantity, distribution, and chemical composition of the moisture are controlling parameters of the corrosion behavior of the metal. In order to simulate the corrosion behavior of a metal using the advanced analytical and computational methods, it is essential to obtain information for the key environmental properties of a thin layer of moisture, moist particulates, and deposits that will affect the corrosion process. The advanced sensor technology development and application of those sensors reported in this study address these needs. Environmental factors that affect corrosion process [2] include: Hydrogen-ion concentration (pH) in the solution Oxygen concentration in solution adjacent to the metal

1

Specific nature and concentration of other ions in solution Transport through and rate of flow of the solution in contact with the metal Ability of environment to form a protective film on the metal Temperature Cyclic stress (corrosion fatigue) Interface of dissimilar metals that affects localized corrosion This study focuses on several of these factors that are important to the increased understanding of corrosion processes for metals covered by a thin layer of particulate. The key parameters investigated in this study are the electrolyte resistivity/conductivity, the solution pH value, and the oxygen concentration and effective diffusion coefficient. These environmental parameters can affect directly the corrosion behavior/progress of metals covered by thin layers of electrolyte and layers of particulate or deposits. Data obtained from the electrochemical measurements using the sensors developed in this study will supplement the information from the electrochemical corrosion tests and provide necessary property data for computational modeling of corrosion processes. The sensors can be used individually or in a multi-sensor arry. This sensor development is an important part of a multi-investigator project to increase the understanding and the technical basis for the prediction of corrosion damage evolution over long periods of time at the proposed Yucca Mountain repository for the disposal of nuclear waste.

[3]

These sensors may facilitate the measurements of the

conductivity, the pH value, and the oxygen concentration and diffusivity of solutions that mimic the conditions on the waste packages after long times at the proposed repository,

2

i.e. the metal surface may be covered by a moist layer of particulate. The Yucca Mountain site was recommended to be a geological repository for commercial spent nuclear fuel and high-level radioactive waste. A major component of the long-term strategy for safe disposal of nuclear waste is to completely isolate the radionuclides. Corrosion is a primary determinant of waste package performance at the proposed Yucca Mountain repository and will control the delay time for radionuclide leakage from the waste package. [3] The processes of the fabrication of these electrochemical sensors employ silicon based microfabrication techniques. Microfabrication is the term to describe processes of fabrication of miniature structures. [4] It is a collection of technologies of making devices with micro or even nano feature resolutions. Advances in microelectromechanical systems (MEMS) and other nanotechnology have extended the scope and techniques of microfabrication. Sequential processes such as photolithography, thin film deposition, and lift-off process for thin film metallization and paste printing and firing for thick film printing for thick film metallization are performed to fabricate a microdevice.

[4]

For

multi-layer structured devices, these processes are used to fabricate one layer a time. Microfabricated electrochemical sensors possess some advantages such as high resolution, small feature size (µm or even nm), small required sample volume, small ohmic potential drop, improved signal to noise ratio, and reduced cost (one of the benefits of batch production). These advantages have made microfabricated electrochemical sensors suitable for this study. In this research, four sensors were fabricated using either thin film metallization or thick film printing. Among them, silicon-based solid state resistivity sensors were designed, constructed, and tested. The design and usage of such

3

conductivity sensors provided a simple yet effective means for measuring solution resistance, which would be critical to the modeling of corrosion progress. Electrode material was platinum and the process used for sensor fabrication was thin-film metallization technique. Two configurations were designed, one consisted of multiple parallel electrodes that were each 0.2mm in width and 3000Å in thickness separated by a 1mm gap between two nearest electrodes; the other consisted of several pairs of electrodes that were each 0.1mm in width with various spacing 0.03mm, 0.05mm, 0.08mm, 0.1mm, and 0.2mm. Two electrode configuration A.C. impedance measurements were conducted on NaCl solutions (0.1M, 0.01M, and 0.001M) with/without Al2O3 particulate (averaging 3µm in diameter) and sand particles (with mean diameter of 300µm) to extract solution resistivity. KCl solutions (0.1M, 0.01M, and 0.001M) were also tested to obtain the cell constants for the bulk solution resistivity measurements. A.C. impedance technique measured the frequency response of a testing media and could be used to investigate the conductivity properties of low or high conductive materials. [5-7] Alumina-based solid state pH sensors were designed, constructed, and tested. The fabrication employed thick film printing metallization. Two types of a three-electrode configuration solid state pH sensor had been designed. One consisted of a circular working electrode; while the other consisted of all rectangular electrodes. Through subsequent preliminary tests, it was determined that the rectangular configuration should utilized for this study. The decision was based on the fact that the rectangular configuration provided much better stability and reproducibility. The working and counter electrodes were made of palladium, while the reference electrode was a standard

4

thick film silver ink which was subsequently electrochemically chloridized after the sensor had been fabricated. Each electrode had a width of 0.5mm and the sensing area was about 1.75mm2 and the distance between two electrodes was 1mm. Open circuit potential measurements were conducted in pH buffer solutions with nominal pH values of 6, 8, 9, 10, 11, 12, and 13. A solid state solution oxygen sensor was also fabricated on alumina substrate. The fabrication of this sensor was carried out using thick film printing metallization process. This was a three-electrode configuration sensor with a resistance temperature detector (RTD) on the side; each working electrode had a length of 7.5mm and a width of 2mm and all electrodes were spaced at 1mm apart. The working and counter electrodes were made of gold while the reference electrode was a thick film silver film which was subsequently electrochemically chloridized after the sensor had been fabricated. Cyclic voltammetry technique was used to determine the holding voltage for the amperometric current-time (i-t) measurements, which was set to -0.55v. Steady state oxygen reduction currents at this voltage were used to measure the solution oxygen concentration. Oxygen diffusion coefficient in 0.01M KCl solution, 1mm particulate in 0.01M KCl, and 10mm particulate in 0.01M KCl were extracted from i-t curves at values of 9.34x10-5 cm2/s, 5.06x10-5 cm2/s, and 3.15x10-5 cm2/s, respectively. The other part of this study involved the design, the fabrication, and the testing of a high temperature solid state gaseous oxygen sensor, which is not related directly to corrosion studies. Measurements of gaseous oxygen have been very important to a variety of applications such as environmental control, controlled atmospheres, emission control, and combustion optimization.

[8-10]

In the early 19th century, Nernst suggested a

5

nearly pure oxygen ion conducting material when he observed that oxygen could evolve at an anode when a DC voltage was applied on a mixed oxide such as the yttriastabilized-zirconia (YSZ).

[11]

Years later Wagner established a theory for the

electromotive force of a solid electrolyte cell.

[12]

Weissburt and Ruka fabricated a solid

electrolyte oxygen sensor 4 years later. [12] A practical oxygen sensor for automobiles was made in 1976 by Bosch Co. Two major types of oxygen sensors have been developed: amperometric and potentiometric. [9] Amperometric sensors possessed high sensitivity as they relate oxygen activity (concentration) linearly to the current output of the sensor whereas the potentiometric sensors were semi-logarithmic. Although amperometric oxygen sensors could be attractive, issues such as the mechanical robustness, the reliability, size, and the power consumption for the sensor needed to be addressed prior to practical applications. [9]

Meanwhile, YSZ based potentiometric oxygen sensors were well developed. One of

the examples was the oxygen sensor in the automobile which controlled the air-to-fuel ratio for maximum fuel efficiency and minimum pollutants exhausted. In this dissertation, thin film microfabrication of a solid state oxygen sensor was carried out and assessed based on the silicon-based MEMS technology. The sensors developed were geometrically well-defined, reproducible, mechanically robust, and potentially low cost. Each sensor contained two separate units, one intended to operate in an amperometric mode whereas the other in a potentiometric mode. The goal of this research was to develop a layered structure of the sensor using thin film metallization technique minimizing the size of the sensor and exploring the possibility of combining amperometric pumping with potentiometric calibration to add an additional degree of freedom and to establish an

6

environment with well-controlled oxygen concentrations. Each sensor contained one heater and a platinum resistance temperature detector layer covered by one silicon dioxide insulation layer. Then a platinum contact electrode layer was deposited on this insulation layer. On the potentiometric side, a nickel layer was then deposited followed by an YSZ layer and another platinum contact electrode layer; on the amperometric side, YSZ layer was deposited followed by a diffusion barrier layer and another platinum contact electrode layer. This sensor structure was then insulated by a mixture of silicon dioxide and aluminum oxide so that only the selected sensing area was exposed to the environment to be tested. Two different sizes of sensing area were designed as well to explore the influence of area on the sensitivity of these oxygen sensors. The sensors developed in this work were fabricated at the Electronics Design Center of Case Western Reserve University and the measurements of the AC impedance, the open circuit potential, the cyclic voltammetry, and the amperometric i-t technique were carried out for the evaluation and calibration of these sensors. In this dissertation, the sensor design and fabrication procedures are reviewed in Chapter 2. The experimental method for AC impedance measurements and the testing results are described in Chapter 3. Chapter 4 covers the open circuit measurements of solution pH values and results. In Chapter 5, amperometric solution oxygen concentration and diffusion coefficient measurements and results are discussed. The performance of gaseous oxygen sensors and the calibration and the performance of the platinum RTD for high temperature oxygen sensors are illustrated in Chapter 6. Chapter 7 summarizes this research work and makes recommendations for future work.

7

References: [1] http://www.corrosion-doctors.org/principles/theory/html

[2] H. Norton, Sensor and analyzer handbook, Prentice, Hall Inc., 1982

[3] FY06-67: Research summaries fiscal years 2006-2007, office of the Chief Scientist, Science, Technology, and Management, US Department of Energy, Office of Civilian Radioactive Waste Management, Washington, DC, DOE/RW-0594 Also see: http://www.ocrwm.doe.gov/science/targeted_thrusts/matperf_targetedthrusts.shtml

[4] J. Li, “Development of a microfabricated sensor array for oil evaluation”, Ph.D. dissertation, 2005

[5] Gamry Inc., “Electrochemical impedance spectroscopy theory: a primer”, 2005 [6] R. Robinson and R. Stokes, “Electrolyte solutions”, 2nd ed., Butter Worths, 1959 [7] S. Refaey and G. Schwitzgebel, “Electrochemical impedance spectroscopic investigation of dissolution, passibation and pitting corrosion of tin in Na2CO3 solution and the effect of Cl- and I- ions”, Applied Surface Science, 135 (1998) 243-253 [8] S. Sotiropoulos and K. Wallgren, “Solid-state microelectrode oxygen sensors”, Analytica Chimica Acta, 388 (1999) 51-62 [9] S. Yu, Q. Wu, M. Tabib-Azar, and C. Liu, “Development of a silicon-based yttria-stabilized-zirconia (YSZ) amperometric oxygen sensor”, Sensors and Actuators B, 85 (2002) 212-218 [10] S. Badwal, M. Bannister, F. Ciacchi, and G. Hooshmand, “Response rate techniques for zirconia-based Nernstian oxygen sensors”, J. Appl. Electrochem., 18 (1988) 608-613 [11] R. Kocache, J. Swan, and D. Holman, “Applications of oxygen-ion-conducting solid electrolytes” in “Solid State Gas Sensors”, Philadelphia, 1987

8

[12] T. Seiyama, “Current state and future outlook” in “Chemical Sensor Technology”, Vol. 1 (1988) Kodansha Ltd.

9

Chapter 2 Sensor Design and Fabrication A conductivity sensor and a gaseous oxygen sensor were fabricated at the Electronics Design Center at Case Western Reserve University and used for measuring the solution conductivity and the gaseous oxygen concentration, respectively. The construction of these sensors was through the silicon-based thin film microfabrication techniques. A solid-state metal/metal oxide pH sensor and a solution oxygen concentration sensor were produced using the thick film printing technique. Both the thin film and the thick film metallization techniques are branches of microfabrication. In manufacturing, deposition (such as vapor deposition) and removal (such as etching and lift-off) are the two primary processes for the thin-film metallization. Deposition promotes cohesion between particles and therefore creates a pattern while removal destroys cohesion between particles and thereby enables precision shaping.

[1]

Microfabrication refers to

such processes that deal with features that are of micro or even nano meter critical dimensions

[2]

. Several advantages of the microfabrication technique for electrochemical

sensor applications can be summarized as follows: [1] •

Very small feature sizes (µm or nm) enable sensor miniaturization



Sensing elements are small in size and therefore require a relatively small sample volume



Very low Faradiac current of an electrochemical sensor results in relatively small ohmic potential drop even for highly resistive samples

10



The reduction of electrode size brings an increase in the limiting current, resulting in improved signal-to-noise ratio



Microelectrodes enable a relatively fast time-response which allows the monitoring of fluctuating signals and the rapid recording of polarization curves

2.1

Fabrication Techniques All of the sensors were designed and fabricated at the Electronics Design Center

of Center at Case Western Reserve University. Thin-film sensors were fabricated on silicon wafers while thick-film sensors were printed on highly polished alumina substrates. The techniques used included photolithography, physical vapor deposition (sputtering), and lift-off pattern-realization. The following sections describe these processes in detail. 2.1.1

Photolithography Both microelectronic fabrication and micromachining start with lithography, a

technique first invented by a German map inspector, Aloys Senefelder, in 1796. Lithography is a technique used to transfer copies of a master pattern onto the surface of a solid substrate, such as silicon wafer or alumina substrate (for centuries this technique has been used to reproduce artwork). The word “lithography” itself (lithos and práphein in Greek means “stone” and “to write,” respectively) literally refers to writing on rocks. [2]

Currently the most widely applied form of lithography for the semiconductor and the micromachining industry is photolithography, based on the invention of Niécphore

11

Niépce in 1822, in which a glass plate coated with bitumen (a light sensitive substance) was used to reproduce an etched print on oiled paper

[2]

. Other forms of lithography

include charged particle (such as electrons and ions) lithography and x-ray lithography. After more than one century of development, photolithography patterning has become a standard process with resolution of submicron accuracy

[3, 4]

. With the continuous

improvement in resolution, photolithography has outpaced and prevented the extensive adaptation of alternative lithography techniques such as x-ray lithography [2]. Photolithography has become the most widely used form of lithography, capable of providing accurate resolution (down to micrometers in dimension) and enabling exposures of a series of successive patterns, provided that a well-defined positioning methodology is applied. A complete photolithography process normally contains the following steps: [2] 1.

Mask Making

A photomask normally refers to the stencil that can be used repeatedly to generate a desired pattern on photoresist-spin-coated wafers or substrates. Usually a photomask is made of either quartz plate (which is transparent to deep ultraviolet (UV) light) or optically flat (or at least nearly flat) glass (which is transparent to near-UV light instead) with a metal (commonly chromium) absorber pattern (such as metal which is opaque to UV light). Depending on the polarity of the mask, when a masked substrate is exposed to UV radiation, a light field or dark field image will be transferred onto the substrate surface.

12

In this work, Autodesk’s AutoCAD LT 2000 software was used to design the patterns of the masks. The electronic file was then sent for mask fabrication to a mask making vendor (Photo Sciences, Inc., Torrance, California). There are eight masks for the fabrication of the high-temperature, gaseous oxygen sensor and two masks for the conductivity sensor. Each mask was made on a 4-inch glass plate. The eight masks for the high-temperature, gaseous oxygen sensor were used for the deposition of platinum heater and temperature detector, silicon dioxide insulator, platinum contact electrode, nickel/nickel oxide reference electrode, YSZ electrolyte, another platinum contact electrode, lanthanum strontium manganese oxide (LSM) diffusion barrier, mixed silicon dioxide and aluminum oxide insulator, respectively. The two masks for the conductivity sensor were for the deposition of platinum electrodes and the silicon dioxide insulator. 2.

Spin Coating of Photoresist and Soft Baking

Spin coating of photoresist is the first step of the photolithography process. In this step, a thin photoresist layer, made of an organic polymer that is sensitive to UV radiation, is spin-coated onto the substrate surface. Normally, a photoresist can be divided into three components: a base resin (an organic polymer) which is sensitive to UV radiation and changes chemical/physical structure when exposed, a sensitizer which controls the rate of the polymeric phase reaction, and a casting solvent which provides a homogeneous means for spin coating. positive and negative.

[3]

[2]

There are two categories of photoresists,

In the case of a positive resist, the polymer structure will be

weakened when exposed to the UV light and become more soluble in a developing solution. When subjected to a developing solution, the exposed, positive photoresist will be washed away by the developer, exposing the bare materials that are underneath.

13

Therefore, positive photoresists are always used with light field masks. On the other hand, when a negative resist is exposed to the UV light, the polymer structure is strengthened and will become less soluble in the developer. When subjected to a developing solution, the negative photoresist that is not exposed will be washed away by the developer. Therefore, negative photoresists must be used with dark field masks, which contain the inverse of the pattern to be transferred. Figure 2-1 illustrates the comparison of light field and dark field. [2] (the black area is the dark field)

Figure 2-1

Comparison of light field and dark field for forming the “I” pattern [2]

In this study, a Specialty Coating System’s P6204-A spinner was used. During a spin coating process, the substrate was placed on the wafer plate of the spinner and a vacuum chuck was applied to hold the substrate in place. The photoresist purchased from Shipley Co., Newton, MA (#S1818 positive photoresist) was then dispensed onto the substrate surface using a syringe with controlled volume. In the next step, the substrate was subjected to a high speed spin process for a selected time span to produce a uniformly distributed thin film of the photoresist. Two factors, a well-controlled thickness and good uniformity of the phtoresist thin film were of primary importance to

14

the success of the spin coating process, specifically the line-width reproducibility and subsequent development. It was determined that the thickness of the photoresist film was dependent on the spin speed, the photoresist concentration, and the molecular weight of the organic polymer

[2]

. In this study, a photoresist spin coating process was established

with the spin rate set to 2200rpm and 30 seconds selected for the spin time. Following the spin coating process, a soft-baking procedure was carried out in an oven set at 90°C for 10 minutes so that the remaining solvent could be removed. This procedure also released any possible internal stress inside the photoresist thin film. The soft-baking procedure was a critical step in photolithography since the photoresist must be soft-baked before it could become photosensitive. An additional benefit of the softbaking was the enhanced adhesion of the photoresist thin film to the substrate surface. 3.

Alignment and Exposure

The next step in the photolithography, after soft-baking of photoresist film, was the mask alignment in which the mask was aligned with the wafer/substrate and exposed to a UV light source transferring the pattern onto the substrate surface. When multiple masks were used, each subsequent mask after the initial mask must align with the previous pattern. The instrument used for this purpose was a mask aligner. The precise alignment of patterns was of great importance to photolithography process. Thus, in addition to the desired pattern, alignment marks were included in the mask design. Usually the dimensions of the alignment marks were slightly larger than those on the previous mask so that the previous alignment marks could be observed through the optical system of the aligner. To ensure the precision in pattern positioning,

15

the difference in the sizes of the marks on two successive masks must be designed to be smaller than the critical dimensions of the patterns. After the accurate alignment of the mask and the pattern on the wafer/substrate had been established, the photoresist thin film was exposed through the mask to high intensity UV light. Three basic exposure methods have been commonly utilized: contact, proximity, and projection. In contact exposure, the resist-coated wafer would be in direct physical contact with the mask. Although a very high resolution could be achieved using this method, defects in the pattern could be generated, due to the possible damage of the mask by the direct contact of the wafer and the mask. In proximity exposure, a small gap separated the wafer and the mask, usually 10-25 microns in width. The advantage of this method would be that the defects can be minimized, but the disadvantage would be that the diffraction of the UV light resulted in reduced resolution. In projection exposure, the image of the pattern on the mask was projected onto the photoresist thin film by an optical system. With this method, contact between the wafer and the mask was completely avoided, yielding a resolution not quite as high as the contact exposure method due to the diffractions and the imperfections in the lens that were used. [4] In this research, an ABM high resolution mask aligner was used with a contact exposure method in order to achieve 1:1 ratio of pattern transferring. This ABM aligner consisted of 5 major modules (subsystems): [5] 1)

Mask-to-wafer alignment module:

16

This module controlled the substrate movement. The motions of the substrate in the X, Y, Z and the Theta (angular) directions were precisely controlled using 4 controlling knobs. 2)

Alignment optics A Zeiss binocular microscope was used for this module. The splitfield

images could be viewed on the monitor. 3)

Control panel Electrical and pneumatic controls of all subsystems were enabled through

this panel. 4)

UV light source A 350W mercury arc lamp was used as the light source for this aligner.

The wavelength of the UV light employed was 365nm. 5)

Lamp power controller This module provided the control of the UV light intensity.

Before the mask was aligned, the wafer/substrate was first positioned onto the substrate plate of the alignment module and a vacuum chuck was used to hold the wafer/substrate in place. The mask was then placed on the top of the mask holder and vacuum clamped as well. Once the mask and the wafer/substrate were both correctly positioned and securely held, The Z direction control knob raised the wafer/substrate so that it was in contact with the mask. The wafer/substrate was then locked in a planar

17

position to the mask by pressing the planarization function button. Following that, in order to allow the observation of the patterns for alignment as well as the movement of the wafer/substrate, the mask and the wafer/substrate were separated while maintaining the focus for both the mask and the wafer/substrate in the aligner optics system. After the alignment was achieved through X, Y, and the angular knobs, the wafer/substrate was raised again to reach the contact position with the mask and sealed by vacuum to establish direct contact of the mask and the wafer/substrate. Lastly, the exposing system was then turned on for exposure. The light intensity and the exposure time determined the incident energy of the exposure. In this system, a 5 second exposure time was selected and the quality of the patterns transferred was both consistent and satisfactory. 4.

Post-Exposure Development

During the post-exposure development step, the latent resist image formed during the exposure step was transformed into a relief image, which served as a mask for the subsequent steps after the exposed (positive) or the unexposed (negative) photoresists were selectively dissolved. In this study, the most widely used wet development technique was employed. A Shipley microposit development 351 was diluted with water in a 1:4 ratio to form the development solution. The exposed wafer/substrate was immersed in a bath of such a solution for 60 seconds and then rinsed in a de-ionized water bath. 5.

Post-Baking

The last step of the photolithography process was post-baking. In this study, the developed wafer was put in a soft-bake oven at 90C for 10 minutes. The residual

18

developing solvent and water were removed by this process. This post-baking process improved the adhesion of the photoresist to the wafer surface. 2.1.2

Sputtering: A Method of Physical Vapor Deposition Several techniques could be used to form high performance solid state thin film

materials, including chemical vapor deposition (CVD) and physical vapor deposition (PVD). In a typical CVD process, the wafer/substrate would be exposed to one or more volatile precursors, which reacte and/or decompose on the substrate surface to produce the desired deposit. Microfabrication processes widely used CVD to deposit materials in various crystalline structures, including: monocrystalline, polycrystalline, amorphous, and epitaxial. [6] Physical vapor deposition deposits thin films one atom (or molecule) at a time onto various surfaces (e.g., onto semiconductor wafers). The coating source was physical (i.e. solid or liquid) rather than chemical as in CVD. It would be fundamentally a vaporization coating process. In the process there were three steps: first the material to be deposited was converted into a vapor through physical means (such as heating or electron beam bombarding); then the vapor from the source was transferred to the wafer/substrate through high vacuum; lastly, the vapor condensed onto the wafer/substrate surface forming a thin film with controllable thickness. If reactive deposition was involved, the vapor would react with a gaseous co-deposited material to form a thin film made of compound material.

[7]

Variants of PVD included cathodic arc deposition, evaporative

deposition, electron beam physical vapor deposition, pulsed laser deposition, and sputtering deposition. All PVD techniques employ the same three steps even though the approaches utilized to vaporize and deposit the material are the distinguishing characteristics.

[2-4]

Sputtering deposition was selected in this study to utilize the 19

sputtering instrument readily available in the EDC. In sputtering, the material to be deposited was in the form of a disc and referred to as a target. Atoms/molecules in the vapor phase would be ejected from the target by bombardment of high-energy ions (commonly argon ions) created in a plasma. The sputtered atoms are drawn through a high vacuum region, reaching the substrate where they condense to form a thin film. Despite negative aspects of sputtering such as the process complexity and the relatively low deposition rate, several advantages have made sputtering the most widely used PVD technique: 1. A large surface area target could be used in sputtering decreasing the possible shadow effect; 2. The thickness of the formed thin film could be easily controlled through the adjustment of the deposition time; 3. When alloy or compound targets are used, the decomposition is less significant than in evaporation, making it easier to control the composition of the thin film deposited; 4. The substrate surface could easily be sputter-cleaned using in-milling in vacuum before the thin film is deposited enhancing the adhesion of subsequent depositions; 5. Unlike electron beam evaporation, x-ray damage to the device could be avoided. In this study, a Commonwealth 1130 IBC sputtering machine was used for the sputtering deposition of the thin film layers on the wafer/substrate. The vacuum chamber

20

was isolated and a vacuum pump was used to reduce the pressure to 2x10-6 torr (base pressure). When this base pressure was reached, argon gas was introduced into the chamber raising the pressure to the work pressure (2x10-4 torr). The next step was to clean the substrate surface using a 90-second plasma etching with the beam voltage and current set at 500V and 7.5mA, respectively. To enhance the adhesion to the wafer/substrate for metal depositions such as Au and Pt, a thin titanium layer with the thickness of 50Å was first deposited. The working pressure was maintained at 2x10-4 torr throughout the whole sputtering process. When oxides (such as silicon dioxide and aluminum oxide) were deposited, the gas composition would be changed to a mixture of air and argon at the ratio of 1:9. In this case, the beam voltage and current would be set to 800V and 15mA, respectively. The rate of deposition varied with the material to be deposited. Generally, the metal deposition rate was greater than that for oxides. Table 2-1 lists several typical materials and their corresponding deposition time using the IBC machine for a film of 3000Å thickness. Table 2-1

Deposition time for sputtering of materials

Material

Ti

Au

Pt

Ag

Al2O3

Ni

YSZ

LSM

Thickness (Å)

50

3000

3000

3000

3000

3000

3000

2000

Deposition time

11

58

79

32.5

296

90

357

208

(minutes)

21

2.1.3

Lift-off Processing After a thin film of desired material had been blanket-deposited onto a

wafer/substrate, the excess material that covers the photoresist needed to be removed so that desired patterning could be achieved. Subtractive processes used in micromachining and microfabrication include mask-based processes such as wet and dry etching and maskless processes such as focused ion-beam etching (FIB), laser machining, and electrochemical discharge machining (EDM). Among these processes, lift-off processing was a simple, easy method that was commonly used. During the lift-off process, the photoresist under the film would be removed with solvent starting at the edge of the unexposed photoresist, and dissolving laterally taking the deposited film with it, and eventually leaving only the film which was deposited directly on the substrate forming the desired pattern. [8] Depending on the type of lift-off process used, patterns could be defined with extremely high fidelity and very fine geometries. Any deposited film could be lifted-off, provided: [8] 1. Temperatures did not exceed the point at which the photoresist may be burned; 2. The film quality was not absolutely critical; 3. Good adhesion of the deposited film was achieved on the substrate; 4. The film could be easily wetted by the solvent; 5. The film was thin enough and/or grainy enough to allow solvent to seep underneath;

22

6. The film was not elastic and was thin and/or brittle enough to tear along adhesion lines. Figure 2-2 shows the schematic for the above sequential steps including the photolithography process, the sputtering step, and the lift-off process for all layers that were deposited. In this study, acetone was chosen as the solvent for lift-off process which required 3-4 hours for the dissolution of the remaining photoresist post-exposure. Later, the wafer/substrate was placed in a hard-bake oven to remove the remaining solvent. After the wafer/substrate was taken out of the oven and allowed to cool down, it was ready for the deposition of the next layer, if necessary.

Figure 2-2

Schematic of thin film microfabrication process [9]

23

2.1.4

Thick Film Printing Technique When thick films were deposited to fabricate sensors, it was a common practice to

use thick film printing (or screen printing), an additive process in which a thick film (10 – 50 µm) would be deposited on a substrate through a mask.

[10, 11]

Figure 2-3-a shows a

schematic of the process. During the thick film printing process, a paste, or ink, was first placed on a mask, and a squeegee moved across the top of the mask pushing the ink through holes in the mask onto the substrate. Following this step, a process known as curing would be performed, where the substrate was heated to remove solvents from the ink. Figure 2-3-b shows more details of the process

[2]

. The detailed process involves

several steps. First AutoCAD is used to produce the design to be printed on the substrate. Next a transparency was used to pattern a red photoemulsion made by the Electronics Design Center. The areas that were to be covered by metal on the substrate were left open; the areas where no material would be deposited were covered with the emulsion. The ink was placed in front of the squeegee. A gap would be maintained between the substrate and the mask so that when the squeegee began to move, the mask was pressed down to make contact with the substrate. Once the squeegee moved beyond the open area of the mask, the mask snapped back up to its normal position, leaving the ink on the substrate. The major parameters for the actual process would be the squeegee pressure, the separation of the mask and substrate, and the height of the squeegee. Other non-process parameters were the ink properties and the type of mask. [11] The final product would be a substrate with a film between 5 – 30µm in height. The film would have a resolution, or lateral accuracy of the pattern transfer, of 100µm.

24

The mask may either be a screen or a stencil. A screen would be simply a mesh of fine stainless steel wires onto which a patterned emulsion had been adhered to one side. A stencil was a solid piece of stainless steel that had the pattern cut out of it, leaving holes in the metal. The screens used to print the electrodes were a 400, 325 or 280 mesh stainless steel made by the Electronics Design Center at Case Western Reserve University. The stencil used to deposit the carbon layers was made by Hybrid Screen Technologies (Northridge, CA). A 0.002 inch thick piece of stainless steel was etched to have open areas and then bonded to a screen frame forming the stencil. The metal was etched according to a supplied AutoCAD drawing. The thick film printer used for fabrication of the solid state pH sensors and solution oxygen concentration sensors was a TF-100 Semi-automatic Screen Printer made by the MPM Corporation (Franklin, MA). The main parameters of the process that could be changed between layers were the mask type (screen or stencil), the snap-off height (distance between the screen and substrate), the squeegee pressure, and the squeegee downstop (distance from squeegee to the top of the mask).

25

a

b

Figure 2-3: Illustration of the thick film printing process [2]. a) Schematic of screen-printing; b) Details of screen-printing process The substrate for sensor fabrication was a 96% alumina substrate made by Coors (Golden, CO). The counter and reference electrodes were made from standard thick film metal inks. The working and counter electrodes consisted of a standard thick film ink, Pd for pH sensors and Au for solution oxygen sensors, respectively. The reference electrode was a standard, thick-film silver ink which was then electrochemically chloridized after the sensor had been fabricated. Please refer to section 2.1.2 for details. Table 2-2 lists the specifications for the inks used in this study. The screens used for the electrode fabrication were 325 meshes in size, while the screen used for the insulation layer was 280 meshes in size.

26

Table 2-2 Ink information for thick film printing process Ink information

Ink type

Manufacturer

Working electrode (Au)

D-8843

Electro Science

Counter electrode (Au)

D-8843

Electro Science

Working electrode (Pd)

A4855

Englehard

Counter electrode (Pd)

A4855

Englehard

Reference electrode (Ag)

D-9912-G

Electro Science

Insulation

4612

Electro Science

27

2.2

Sensor Fabrication Process

2.2.1

Thin film high temperature oxygen sensor The complete fabrication process for the high-temperature oxygen sensor is

shown in figure 2-4 and the side view is shown in figure 5.

1. Pt deposition

5. YSZ deposition

2. Al2O3 deposition

6. LSM deposition

3. Pt deposition

7. Pt deposition

4. Ni deposition

8. Al2O3 deposition

Figure 2-4 Fabrication process of high temperature oxygen sensor

28

Figure 2-5 High temperature oxygen sensor side view Each step involved photolithography for pattern transfer, sputtering for material deposition, and lift-off for excess material removal. A detailed description of such fabrication process is given as follows: 1. Formation of silicon dioxide layer A commercial silicon wafer was used. The wafer first went through a RCA cleaning process to remove residue oxide layer. After RCA cleaning, the wafer was placed in a 4-stack 4500 furnace made by MRL Industry Inc. for oxide layer growth. In order to form a 3000Å thick silicon dioxide layer, the temperature was set to 1050°C and the wafer exposed for 10 minutes to dry O2, 30 minutes to wet O2 (mixture of H2 and O2), and 10 minutes dry O2 or N2 processing. The detailed RCA cleaning steps are: a.

1:20 HF/de-ionized water cleaning for 5 seconds followed by de-ionized

water bath;

29

b.

5:1:1 RCA1 solution (de-ionized water-NH4OH-H2O2) cleaning for 15

minutes followed by de-ionized water bath; c.

5:1:1 RCA2 solution (de-ionized water-HCl-H2O2) cleaning for 15

minutes followed by de-ionized water bath; d.

1:20 HF/de-ionized water cleaning for 5 seconds followed by de-ionized

water bath. 2. Deposition of Platinum In this step, the platinum heater and temperature detector pattern was transferred onto the wafer. Then sputtering deposition of 3000Å thin film was performed at 2x10-4 torr of argon with beam voltage and current valued at 500V and 7.5mA, respectively. The sputtering time was about 79 minutes for these settings. 3. Deposition of Aluminum oxide (insulation layer 1) This step used the photolithography technique illustrated previously in section 2.1. The working pressure for Al2O3 sputtering was set to 2x10-4 torr of 90% Argon + 10% air. The beam voltage and current were 800V and 15mA, respectively. The sputtering deposition time the Al2O3 layer of 3000Å was approximately 296 minutes. 4. Deposition of Platinum contact electrode (bottom) 5. Deposition of nickel reference electrode for the potentiometric unit The beam voltage and the current used for the Ni deposition was 800V and 15mA, respectively. The operating pressure was also 2x10-4 torr of argon and the sputtering time

30

was approximately 90 minutes for a thickness of 3000Å. The Ni thin layer was partially oxidized to form a mixture of Ni and NiO. 6. Deposition of YSZ electrolyte The beam voltage and current used for YSZ sputtering was also 800V and 15mA, respectively. The operating pressure was 2x10-4 torr of 90% argon + 10% air and the sputtering time was approximately 357 minutes to form a thin layer of YSZ of 3000Å. 7. Deposition of lanthanum strontium manganese oxide (LSM) electrolyte for the amperometric unit The beam voltage and current used for LSM deposition was 800V and 15mA, respectively. The operation pressure was also 2E-4 torr of 90% argon + 10% air and the sputtering time was approximately 208 minutes for a layer of 2000Å thin film. 8. Deposition of platinum contact electrode (top) 9. Deposition of aluminum oxide (insulation layer 2) The sequential processing steps illustrated in figure 2-5 symbolized the completion of the fabrication of the prototypes of the high-temperature oxygen sensor. The silicon wafer after the sputtering deposition processes is shown in figure 2-6. The whole patterned wafer was then cut into individual sensor chips using a Model 1100 wafer dicing saw from Micro Automation. To protect the surface, a thin layer of photoresist was added until after dicing. Figure 2-7 shows a picture of the individual sensor chip attached to the packaging material (Aluminum oxide substrate with Au

31

contact pads). Subsequently, wire bonding and soldering were performed. Additional packaging is applied; figure 2-8 shows the whole package.

Figure 2-6 Silicon wafer after all the sputtering processes

Figure 2-7 Diced sensor chip on packaging material

32

Figure 2-8 Packed testing device for high temperature oxygen sensor 2.2.2

Thin film conductivity sensor The thin film conductivity sensor fabrication processing was similar to that of the

oxygen sensor but involved only two layers, one platinum electrode layer followed by an Al2O3 insulation layer. Photolithography, sputtering, and lift-off processes were performed to transfer the pattern shown in Figure 2-9 onto a silicon wafer. The individual sensor chip after dicing and soldering is shown in Figure 2-10. The overall size of the individual sensor was 21.8mm in length and 21mm in width for the parallel electrode sensor (green lines), and 27.2mm in length and 22.1mm in width for the pair-electrode sensor (red pairs). Detailed dimensions of these sensors are shown in chapter 3.

33

Figure 2-9 Conductivity sensor design and wafer layout (the parallel-line structure was tested for this study; the pair structure was intended for point to point resistance study, which was not carried out in this study)

Figure 2-10 Silicon wafer for conductivity sensor before dicing and packaging (consisted of 6 parallel structure sensors and 3 pair structure sensors)

34

2.2.3

Thick film pH sensor and dissolved oxygen sensor The fabrication procedure for the thick-film printing process could be given as

follows: 1. Cleaning of the aluminum oxide substrate The substrates used for thick film printing in this study were 4 inch round, highly polished aluminum oxide discs. The substrate was immersed in acetone followed by methanol and water to remove any organic impurities and then rinsed in de-ionized water. The substrate was then put in a hard-bake oven for 3-4 hours with at 120°C to remove water and other possible residual solvents. 2. Printing the silver electrode followed by curing in oven at 110°C for 10 minutes 3. Printing the palladium (pH sensor) or gold (solution oxygen sensor) working and counter electrodes followed by curing in the oven at 110°C for 10 minutes 4. Firing the substrate in a furnace at peak temperature of 850°C for 10 minutes 5. Printing the insulation, curing in the oven at 110 C for 10 minutes. 6. Firing the substrate in a furnace at peak temperature of 850°C for 10 minutes 7. Chloridizing the silver electrode. The chloridization of silver electrodes was performed in a 0.1M HCl solution using a standard electrochemical procedure. The reaction can be described by equation 2.1:

35

2 Ag + 2 HCl = H 2 + 2 AgCl

(2.1)

A Model 273 EG&G potentiostat set at the galvanostatic mode was used. First, the potentiostat was set up as follows: a. Two sweep voltage switches and AC power were “OFF” b. Mode was set to “POT” c. Electrode mode was set to “DUMMY” d. K2 electrode offset voltage was set to “OFF” e. K1 electrode offset voltage was set to “POSITIVE” f. Current converter per voltage was set to “MAXIMUM” g. Voltage was set to 1V Next, the CE and REF outputs were connected, while the reference output was connected to a platinum screen and the K1 output was connected to the silver electrode. The thick film silver electrode surface was first polished using fine sand paper and rinsed with de-ionized water. The platinum screen and the sensor chip were immersed in the 0.1M HCl solution. Then the AC power was turned on and the electrode mode was switched from “dummy” to “normal”. The surface was then further cleaned electrochemically by switching K1 offset voltage from “Positive” to “Negative” three times. At the end of the process, K1 offset voltage was set to “Positive” and the chloridization process was run for 1 minute. Finally, the electrode mode was switched back to “dummy” and the sensor electrodes were rinsed with de-ionized water.

36

The designs for pH sensors and solution oxygen sensors are shown in figures 2-11 and 2-12, respectively. Figures 2-13 and 2-14 show the pictures of individual sensors after dicing for the pH sensor and oxygen sensor, respectively. Dicing in this context referred to the separation of individual sensor chips from each other using a rotating diamond-coated blade.

Figure 2-11 Design and wafer layout for pH sensor 37

Figure 2-12 Design and dimension of solution oxygen sensor

38

Figure 2-13 Pd/PdO based solid state pH sensor

Figure 2-14 Three electrode configuration solution oxygen sensor with RTD

39

2.3

Summary AutoCAD was utilized for the design of sensor patterns. Electronic files of these

designs were used to fabricate photo masks, screens, and/or stencil for patterns transfer. Silicon wafers were used to fabricate the thin-film high-temperature oxygen sensor and thin film conductivity sensor. Positive photoresist (#S1818 from Shiplye Co.) was used in photolithography for patterning. Deposition of Ti, Pt, Al2O3, Ni, YSZ, and LSM was achieved using sputtering. A thin titanium layer (50Å) was first deposited on the wafer to enhance the adhesion between wafer and platinum. The thickness of metal and oxide films was approximately 3000Å with the exception of LSM which was set to 2000Å. Aluminum oxide substrate was used to fabricate solid-state pH sensor and solution oxygen sensor sensors. The deposition of Pd, Au, and Ag was achieved using thick-film printing technique. The Ag film was further chloridized electrochemically to form a Ag/AgCl reference electrode. All sensors were diced, bonded (wire bonding or soldering), and packed into individual testing devices.

40

References: [1] Q. Wu, K. Lee, and C. Liu, “Development of chemical sensors using microfabrication and micromachining techniques”, Sensors and Actuators B 1-6 (1993) 13-14 [2] M. Madou, “Fundamentals of microfabrication: the science of miniaturization”, 2nd ed., CRC Press, 2002 [3] S. Wolf and R. Tauber, “Silicon processing for the VLSI era”, Vol. 1 2nd ed., Lattice Press, 2000 [4] J. Helbert, “Handbook of VLSI microlithography”, 2nd ed., Noyes Publications, 2001 [5] ABM Inc., “Operation manual for ABM mask alignment & exposure system” [6] http://en.wikipedia.org/wiki/chemical_vapor_deposition.html [7] http://en.wikipedia.org/wiki/physical_vapor_deposition.html [8] http://snf.stanford.edu/process/lithography/lift-off.html [9] J. Li, “Development of a microfabricated sensor array for oil evaluation”, Ph.D. dissertation, 2005 [10] M. Albareda, A. Merkoçi, and S. Alegret, “Configurations used in the design of screen-printed enzymatic biosensors” A Review, Sensors and Actuators B-chem., 69 (2000) 153-163 [11] J. Shen, L. Dudik, and C. Liu, “An iridium nanoparticles dispersed carbon based thick film electrochemical biosensor and its application for a single use disposable glucose biosensor”, Sensors and Actuators B 125 (2007) 106-113

41

Chapter 3 A. C. Impedance Measurement of Conductivity Sensor 3.1 Introduction to AC Impedance Measurement One of the well-developed branches of AC theory is the electrochemical impedance theory, which describes the response of a circuit to an alternating excitation current or voltage as a function of frequency. An analogy has been established between an electrochemical system and an equivalent circuit. The advantage of such an analogy is that the well-established AC circuit theory can be used to characterize an electrochemical system of interest in terms of its equivalent circuit

[1]

. For instance electrochemical cells

with slow electrode kinetics (which in turn means slow chemical reactions) can exhibit impeded electron flows. Thereby an analogy may be established using an AC circuit composed of resistors, capacitors, and inductors that also impede the flow of electrons. Such a simple equivalent circuit model can frequently be a good approximation of a real system and data can often be fitted yielding results of reasonable accuracy [1]. The ratio of voltage to current V/I for a resistor is defined as its resistance. When voltage and current are in phase, their ratio (resistance) does not depend on frequency. However more interest has been directed to the cases where the ratio of voltage to current does depend on the frequency and there is a phase difference. Impedance is the general name for the ratio of voltage to current. It has been given the symbol Z. Resistance is a special case of impedance

[2]

. Figure 3-1 shows the impedance of components: resistors,

capacitors, and inductors as a function of frequency.

42

Figure 3-1 Frequency and phase response of resistor, capacitor, and inductor [1] The technique where cell or electrode impedance is plotted vs. frequency is commonly called electrochemical impedance spectroscopy (EIS). This technique can be used to investigate a variety of materials and chemical mechanisms

[3]

. The principal

advantage of EIS is that a purely electronic model can be used to represent an otherwise complex electrochemical cell. EIS techniques use very small excitation amplitudes, often in the range of 5 to 10 mV for perturbation purposes. Excitation waveforms of this amplitude cause only minimal perturbation of the electrochemical test system, thereby minimizing errors caused by the measurement technique 43

[2, 3]

. Compared to other

conventional electrochemical techniques, EIS works close to equilibrium potential; thus it has become unnecessary to know the details of the current-potential response curve over large overpotential range. This makes treating kinetics and diffusion phenomenon much simpler. 3.2 Fundamentals of AC Impedance technique A pure sinusoidal voltage excitation signal normally can be expressed as equation 3.1: e = E sin ϖ t

(3.1)

Where e is the potential of the excitation voltage, ω is the angular frequency, E is the amplitude of AC excitation, and t is time.

Figure 3-2 Sinusoidal voltage and the current/voltage relationship for resistor, capacitor, and inductor [4, 5] As can be seen in figure 3-2, the related current signal, i, is not necessarily in the same phase with e even though it has the same frequency. The rotating vectors of i and e

44

are separated by a phase angle, Φ. In return, the current, i, can be shown as equation 3.2 where I is the amplitude of the AC current:

i = I sin(ϖ t + φ )

(3.2)

Depending on the load type, the phase angle Φ can be 0 for a pure resistor R, π/2 for a pure capacitor, and –π/2 for a pure inductor. On the vector diagram, it is more convenient to represent phasors in terms of complex notation: imaginary along the ordinate and real along the abscissa. A more complex system is a resistor R and a capacitor C connected in series. Equation 3.3 gives the definition of impedance where j = −1 :

Z = R− j

1 ϖC

(3.3)

In many cases, impedance is often defined by the follow equation 3.4 [3]: Z = Z Re + jZ Im

(3.4)

In electrochemistry, the imaginary impedance is almost always capacitive and therefore negative. With such a definition of Z, generally only positive values for ZIm are utilized and thus impedance plots appear naturally in the first quadrant [3]. Two of the most commonly used plots to display the variations of impedance with frequency are Bode plot and Nyquist plot

[3]

. A Bode plot, named after Hendrik Wade

Bode, is usually a combination of a Bode magnitude plot and a Bode phase plot. In a Bode plot, log(|Z|) and Φ are both plotted against log(ω), where Φ is the phase angle and ω is the angular frequency (2π times the frequency in Hz). Φ equals to 0 for a pure

45

resistance while Φ equals to π/2 for a pure capacitance. A Nyquist plot is represented by a graph in polar coordinates in which the gain and phase of a frequency response are plotted. The plot of these phasor quantities demonstrates the phase as the angle and the magnitude as the distance from the origin. The Nyquist plot is named after Harry Nyquist. A Nyquist plot displays ZIm vs. ZRe for different values of ω/frequency [6]. Figure 3-3 and 3-4 show examples of a typical Nyquist plot and Bode plot

[6, 7]

. The measured cell

impedance data is a function of frequency. The Nyquist plot for a purely charge-transfercontrol system is a semicircle, as shown in figure 3-3, where the charge transfer resistance can be derived from the diameter of the semicircle

[13]

. When both diffusion

and kinetic control are involved, a more complex Nyquist plot will occur, which can be illustrated in Figure 3-5. The circular portion corresponds to the charge transfer control while the linear part relates to the diffusion control. In a real system, the measured semicircle may have a center below the x-axis determined by the electrolyte/electrode interfacial properties such as the roughness of a polycrystalline electrode, the inhomogeneous reaction rates across the electrode surface, or the non-uniformity of current distribution.

46

Figure 3-3 Typical Nyquist plot of an electrochemical system [5]

Figure 3-4 Typical Bode plot of an electrochemical system [4]

47

3.3 Equivalent Circuit In principle, the response of any system to an electrical excitation signal can be represented by that of an equivalent circuit constructed of resistors, capacitors, and inductors.

[12]

Unfortunately, it is too difficult to find an exact equivalent circuit for an

actual electrochemical solution system due to the extremely complex nature of the electrode-solution interface. However, for practical utilization, many models that are available can still be very useful and yield reasonably accurate results. An example of a three-electrode cell is shown in figure 3-5.

[12]

The working electrode and auxiliary

electrode are deemed as faradic impedance parallel to a capacitor; whereas the reference electrode is represented by a battery in series with a resistance and an associated capacitance. Also a resistance is associated with the solution contacting all electrodes. It is obvious that the solution will not behave as a simple resistor and the threeelement equivalent circuit for working electrode is too simplistic. The capacitance corresponding to the interface depends on the applied potential is thereby not linear. Despite all these limitations, such an equivalent circuit model functions well in many circumstances. Rules analogous to those applicable to resistors can also be employed to analyze impedances for more complex circuits. For impedances in a series the total impedance is simply the sum of the individual values while for impedances in parallel, the inverse of the total impedance equals to the sum of reciprocals of individual values. [3]

48

According to Allen J. Bard

[3]

, for low frequency limit, the plot of ZIm vs. ZRe for

impedance measurement in a solution should be linear with an unit slope as shown by equation 3.5 [3]: Z Im = Z Re − RΩ − RCt + 2σ 2Cd

(3.5)

Where RΩ is the solution resistance, Rct is the charge transfer resistance, Cd is the double-layer capacitance, and σ is a constant to be determined from experiments. The extrapolated line intersects at the real axis at

. Such a linear correlation

of ZIm vs. ZRe is a characteristic of a diffusion controlled electrode process.

For a high frequency limit, a circular plot centered at Z Re = RΩ + with a radius of

( Z Re − RΩ −

RCt and ZIm=0 2

RCt can be shown in equation 3.6 [3]: 2

RCt 2 R 2 ) + Z Im = ( Ct ) 2 2 2

(3.6)

The actual impedance plot for a complex mixed control electrochemical system may be the combination of the features of these two limiting cases illustrated by figure 36 [8], corresponding to an equivalent circuit shown by figure 3-7.

49

Figure 3-5 Three component equivalent circuit [12]

Kinetic control

Mass transfer control

Figure 3-6 Sample of Nyquist plot of a mixed control complex electrochemical system [8] 50

Figure 3-7 Sample equivalent circuit of a complex electrochemical system

EIS has been proven to be a powerful tool that can be applied to various electrochemical systems involving situations such as corrosion, electrodeposition, and semiconductor electrodes

[9]

. One of the applications of EIS is to obtain electrolyte

resistance. This is particularly of interest to this work since the measurement of solution resistance is of great importance to the study of current distributions between anodes and cathodes (crucial information for corrosion processes). To achieve this goal, a two terminal configuration was used: the counter electrode (CE) and reference electrode (RE1) of the potentiostat are connected together to one electrode, while the working electrode (WE) and the working sense electrode (RE2) of the potentiostat are connected to another electrode. An AC voltage of small amplitude (5mv) is applied and a pre-set range of frequency is scanned. According to Robinson

[10]

, the equivalent circuit for a conductance cell can be

illustrated by figure 3-8, where Rs is the solution resistance, Cdl is the double layer capacitance, C0 is the capacitance caused by electrodes and their leads, Rct is the

51

polarization resistance, and W is the Warburg impedance. According to this equivalent circuit, the Nyquist plot at high frequency (near 100 KHz) is determined by solution resistance Rs and electrode capacitance C0, while the lower intercept cannot be considered simply the solution resistance Rs. A complex number is obtained when AC impedance measurements are taken, the real part of which can be used to extract solution resistance according to equation 3.7 where R and R∞ are resistances, and f is the AC frequency [10, 11]:

R = R∞ + af



1 2

+ b( f



1 2 2

)

(3.7)

In this case, by extrapolating frequency to infinity, the solution resistance can be decided. For the bright platinum electrodes that are used in this study, af



1 2

is negligible

according to Robinson. In fact, NIST uses this equation as well to produce standard electrolyte conductivity references. They obtained solution resistances by plotting real part of the impedance data vs. f-1 and extrapolating the frequency to infinity

[11]

. This

methodology is followed in this study to measure the resistances of the electrolytes of interest under various conditions which will be discussed in detail in later chapters.

52

Figure 3-8 Sample equivalent circuit of a conductance cell [10] 3.4 Experimental In this study, a 1287 electrochemical interface (potentiostat) from Solartron Inc. was used to control the AC voltage measuring the current output whereas a 1255 Frequency analyzer was used for monitoring the frequencies. In this study, an AC voltage of 5mV magnitude was applied to the electrodes that were merged in a solution in a custom-built testing cell shown in figure 3-9. All experiments started with an initial frequency of 1MHz and ended with a final frequency of 0.1Hz. A signal proportional to the resistive component of the cell impedance was produced and recorded automatically.

53

Figure 3-9 Testing cell design for conductivity sensor

Figure 3-10 Conductivity sensor and testing cell as assembled 54

AC Impedance measurements were carried out in three different NaCl solutions through testing set up as shown in figure 3-10. Pure NaCl solution, NaCl solution with sand particulates (around 300µm in average size) and Al2O3 powders (3µm in size). When pure solution was tested, a medical grade syringe was used to add a specific volume of solution into the testing cell so that the solution thickness was controlled. The thickness of solution above the conductivity sensor was measured separately. For particulate measurements, particles were mixed with a NaCl solution and added into the testing cell. Excess NaCl solution was subsequently removed, leaving only particulate layer saturated with NaCl solution inside the testing cell. Three different NaCl concentrations (0.1M, 0.01M, and 0.001M) were tested. In order to calculate the cell constant of the testing cell, KCl solutions (0.1M, 0.01M, and 0.001M) were used. Typical AC impedance data Nyquist plots for different solutions and electrode distances were shown in Figure 3-11. All experiments were carried out at room temperature.

-75000

Z''

-50000

-25000

0 0

25000

50000

75000

Z'

Figure 3-11 (a) AC Impedance of 4mm thick 0.001M NaCl solution at various electrode distances: impedance increases as the electrode distance increases

55

-75000

Z''

-50000

-25000

0 0

25000

50000

75000

Z'

Figure 3-11 (b) AC Impedance of 4mm thick 0.01M NaCl solution at various electrode distances

-600

-500

Z''

-400

-300

-200

-100

0 300

400

500

600

700

800

900

Z'

Figure 3-11 (c) AC Impedance of 4mm thick 0.1M NaCl solution at various electrode distances

56

Figure 3-12 shows that the impedance of the NaCl solutions increases with the decrease in in the NaCl solution concentration. The data shown in figure 3-12 were taken at maximum solution thickness in this study (8mm). Figure 3-12 further displays the comparison of impedance data for three solutions at the same electrode distance and solution thickness (8mm). Figure 3-11 illustrates the expected phenomenon that the impedance increases as the distance between the two electrodes increases. Furthermore, for the relatively more conductive 0.1M NaCl solution, the semicircle in figure 3-12 does not appear for short electrode distance measurements; while for the relatively more resistive 0.001M NaCl solution there was a clear semicircle presence for the same testing configuration. Figure 3-13 shows the AC impedance as a function of solution thickness for fixed electrode distance tests. Impedance decreases as the solution thickness increases. This can be explained because as the solution thickness increases, more pathways became available to conduct current. Thus the resistance should decrease.

-50000

-30000

-40000

-20000

Z''

Z''

-30000

-20000 -10000 0.1M NaCl solution

0.1M NaCl solution

0.001M NaCl solution

0.001M NaCl solution

-10000 0.01M NaCl solution

0.01M NaCl solution

0

0 0

10000

20000

30000

0

10000

20000

30000

40000

50000

Z'

Z'

Figure 3-12 AC Impedance of 8mm 0.1M, 0.01M, and 0.001M NaCl solutions: impedance increases with the decrease in solution concentration

Left: 1mm; right: 10mm

57

-75000

-30000 1mm thickness

2mm thickness

-50000

Z''

Z''

-20000

1mm thickness

-25000

-10000

2mm thickness 4mm thickness

4mm thickness

8mm thickness

8mm thickness

0

0 0

10000

20000

0

30000

25000

50000

75000

Z'

Z'

Figure 3-13 (a) AC Impedance of 0.001M NaCl solution at fixed electrode distance: impedance increases with the decrease in solution thickness Left: 1mm; right: 10mm

-15000

-4000

-3000

Z''

Z''

-10000

-2000

-5000 -1000

8mm thickness

1mm thickness 1mm thickness

2mm thickness

2mm thickness

8mm thickness

0

0 0

1000

2000

3000

4000

0

5000

10000

Z'

Z'

Figure 3-13 (b) AC Impedance of 0.01M NaCl solution at fixed electrode distance Left: 1mm; right: 10mm

58

15000

-2500

-200

-2000

-150

-1500

Z''

Z''

-250

-1000

-100 1mm thickness

2mm thickness

-500

-50

1mm thickness

4mm thickness 8mm thickness

2mm thickness

0 350

4mm thickness 8mm thickness

0 400

450

500

550

0

600

500

1000

1500

2000

2500

Z'

Z'

Figure 3-13 (c) AC Impedance of 0.1M NaCl solution at fixed electrode distance Left: 1mm; right: 10mm

The presence of particulates can greatly increase the resistance of the tested layer, compared to the case of pure solutions. Figure 3-14 showed the comparison of the AC impedance data for 0.001M NaCl solution, 0.001M NaCl solution with 3µm Al2O3 powders, and 0.001M NaCl solution with 300µm sand particles. The impedance results of NaCl solutions 0.01M and 0.1M were also shown in figure 3-14. At the same electrode distance and thickness, the impedance of the sand particulate layer was greater than that of Al2O3 powder layer; while the Al2O3 powder layer possessed greater impedance than that of pure solution. The test results for three solution concentrations were similar. Such a phenomenon could be attributed to the increase in particle sizes from 3µm (Al2O3) to 300µm (sand). The presence of non-conductive solids would block the current pathway and induce disturbance to the current flow, the combined result of these two effects was reduced conductivity/increased resistance. Bigger particles caused more disturbances to

59

the current flow and induced narrower current pathways; therefore, the overall resistance became larger.

-100000

-200000

-75000

Z''

-150000

Z''

-50000

-100000

-25000

-50000

Al2O3 particulate layer

sand particulate layer

sand particulate layer

Al2O3 particulate layer

pure NaCl solution

pure NaCl solution

0

0

0

25000

50000

75000

100000

0

50000

100000

Z'

150000

200000

Z'

Figure 3-14 (a) AC Impedance of 8mm thick 0.001M NaCl solution, 0.001M NaCl solution with Al2O3 powders, and 0.001M NaCl solution with sand particles at fixed electrode distance (Left: 1mm; right: 10mm): impedance for sand particulate layer is the highest

-15000

-25000

-20000

-10000

Z''

Z''

-15000

-10000 -5000

-5000

sand particulate layer

Al2O3 particulate layer

sand particulate layer

Al2O3 particulate layer

pure 0.01M NaCl solution

pure 0.01M NaCl solution

0

0 0

5000

10000

15000

0

Z'

5000

10000

15000

20000

25000

Z'

Figure 3-14 (b) AC Impedance of 8mm thick 0.01M NaCl solution, 0.01M NaCl solution with Al2O3 powders, and 0.01M NaCl solution with sand particles at fixed electrode distance (Left: 1mm; right: 10mm)

60

-1500

-2500

-2000

-1000

Z''

Z''

-1500

-1000

-500

-500

sand particulate layer

Al2O3 particulate layer

sand particulate layer

Al2O3 particulate layer

pure 0.1M NaCl solution

0

0

0

500

1000

1500

pure 0.1M NaCl solution

0

Z'

500

1000

1500

2000

2500

Z'

Figure 3-14 (c) AC Impedance of 8mm thick 0.1M NaCl solution, 0.1M NaCl solution with Al2O3 powders, and 0.1M NaCl solution with sand particles at fixed electrode distance (Left: 1mm; right: 10mm)

3.5 Results and Discussion 3.5.1 Determination of the solution resistance One of the applications of Electrochemical Impedance Spectroscopy (EIS) is to obtain the electrolyte resistance/conductance. In this study, the two-terminal configuration impedance measurements were carried out. The Nyquist diagrams consisted of two parts: one semicircle followed by a near linear line. As discussed in previous chapters, the semicircle section represented the charge transfer predominant region, whereas the near linear line represented the mass transfer predominant region. The equivalent circuit for a conductance cell can be illustrated by figure 3-8, where the solution resistance Rs can be extracted. According to this equivalent circuit, the Nyquist plot at high frequency (near 100 KHz) is determined by the solution resistance Rs and the electrode capacitance C0, while the lower intercept cannot be considered simply the solution resistance Rs. A complex number is obtained when AC impedance measurements are taken, the real part of which is used to extract the solution resistance

61

according to equation 3.7 where R is the diffusion control resistance, R∞ is the solution resistance, f is the AC frequency, a and b are constants:

R = R∞ + af



1 2

+ b( f



1 2 2

)

(3.7)

Figure 3-15 shows examples of the real resistance part of the impedance (real resistance hereafter) as a function of AC frequency. Figure 3-16 shows the influence of the solution concentration on the real resistance. The real resistance increases with the decrease in the solution concentration. A 10 times decrease in concentration only resulted in 7-8 times increase in resistance. This could be explained by as the dissociation of ions should be taken into account. For more concentrated solution, Na+ and Cl- ions may not dissociate completely since they had a better chance to recombine with each other. Thus not all the ions could contribute to current conduction, resulting in reduced apparent conductivity. The resistance data corresponding to the linear part of the Nyquist diagrams were plotted according to equation 3.7. The solution resistance was recorded by extrapolating frequency to infinity. In this study, AC impedance measurements were made over a broad frequency range (from 1MHz to 0.1Hz). In order to calculate the solution resistance, data of the near linear part of Nyquist diagrams that corresponded to a frequency on the order of 1000 Hz was used for the resistance/conductivity measurements. The solution resistance was recorded at various electrode distance and solution/particulate layer thickness. Figure 3-17 shows a sample of the solution resistance as a function of the solution thickness for various electrode distances, while figure 3-18 shows a sample

62

solution resistance as a function of the electrode distance for various solution thicknesses. It clearly shows that the solution resistance increases with the increase in electrode distance but decreases with the increase in solution thickness. These two phenomena are further illustrated in the following chapter 3.4.2.

Figure 3-15 Sample of real part of impedance as a function of frequency: resistance increases with the decrease in frequency

63

3.5E+04

AC impedance resistance (Ω)

0.001M electrode distance 2.1mm 0.001M electrode distance 4.2mm

3.0E+04

0.001M electrode distance 8.4mm 0.01M electrode distance 2.1mm 0.01M electrode distance 4.2mm

2.5E+04

0.01M electrode distance 8.4mm 0.1M electrode distance 2.1mm 0.1M electrode distance 4.2mm

2.0E+04

0.1M electrode distance 8.4mm

1.5E+04 1.0E+04 5.0E+03 0.0E+00 1.E+06

1.E+05

1.E+04

1.E+03

1.E+02

Frequency (Hz)

Figure 3-16 Real resistance for an 8mm thick layer 3µm Al2O3 particulate saturated with NaCl of 0.001 M, 0.01 M and 0.1M concentrations.

90000

electrode distance 1mm electrode distance 2mm electrode distance 4mm electrode distance 5mm electrode distance 8mm electrode distance 10mm

80000

70000

Resistance

60000

50000

40000

30000

20000

10000

0 1

2

4

8

solution thickness

Figure 3-17 Real resistance vs. solution thickness for pure NaCl solution at various electrode distances: resistance decreases with the increase in solution thickness

64

70000

60000

1mm 0.001M NaCl solution 1mm 0.01M NaCl solution 1mm 0.1M NaCl solution 8mm 0.001M NaCl solution

50000

Resistance (ohm)

8mm 0.01M NaCl solution 8mm 0.1M NaCl solution

40000

30000

20000

10000

0 1

2

3

4

5

6

8

Electrode distance (mm)

Figure 3-18 Illustration of solution thickness effect on real resistance as a function of electrode distance (0.001M NaCl solution with/without particulates)

From the AC impedance tests conducted in this study, it was concluded that the resistance behavior of particulate layer with saturated conducting electrolyte solutions was similar to the pure solution, only with increased resistance. AC impedance tests in

65

this study indicated that the particulate layer consisting of sand particles (ranging from 75µm to 355µm, average size is about 300µm) with saturated NaCl solutions had a much greater resistance than that of the particulate layer consisting of Al2O3 powders (nominal particle size is 3µm) with saturated NaCl solutions. The explanation could be that the bigger the particle size, the more difficult it will be for current to bypass such particles, thereby the higher the apparent resistance. Figure 3-19 shows the comparison of resistances of pure NaCl solution, particulate layer consisting of sand particles with saturated NaCl solution, and particulate layer consisting of Al2O3 powders with saturated NaCl solution. As the solution concentration increases, the difference in the increase of resistance decreases. Figure 3-20 shows the comparisons of these conditions for 4mm thick sand particulate layer saturated with different NaCl concentrations. Detailed exploration could be carried out through modeling tools such as Molecular Dynamics simulation. However, this was not included in this study.

66

160000

140000

resistance

120000

sand particulate in 0.001M solution (8mm)

100000

Al2O3 particulate in 0.001M NaCl solution (8mm)

80000

pure 0.001MNaCl solution (8mm)

60000

40000

20000

0 1

2

3

4

5

6

8

el ectr ode di st ance

Figure 3-19 (a) Comparison of resistance in 0.001M NaCl solution: sand particulate increases resistance dramatically

20000

18000

16000

pure 0.01M NaCl solution

resistance

14000

12000

sand particulate in 0.01M NaCl solution 10000

Al2O3 particulate in 0.01M NaCl solution

8000

6000

4000

2000

0 1

2

3

4

5

6

8

electrode distance

Figure 3-19 (b) Comparison of resistance in 0.01M NaCl solution

67

2500

2000

pure 0.1M NaCl solution Resistance (ohm)

1500

sand particulate in 0.1M NaCl solution 1000

Al2O3 particulate in 0.1M NaCl solution 500

0 1

2

3

4

5

6

8

electrode distance (mm)

Figure 3-19 (c) Comparison of resistance in 0.1M NaCl solution

180000

0.001M/sand

160000

0.01M/sand 140000

0.1M/sand

Resistance (ohm)

120000

100000

80000

60000

40000

20000

0 1

2

3

4

5

6

8

11

Electrode distance (mm)

Figure 3-20 NaCl concentration effects on 4mm particulate layer resistance: resistance is smallest for the highest concentration

68

3.5.2 Bulk resistivity/conductivity and cell constant calculation As shown in the previous section, the solution resistance calculated through the AC impedance measurements depends on the concentration of the electrolyte (NaCl in this case) (which determines the bulk resistivity ρ), the distance between two testing electrodes, and the thickness of the solution/particulate layer saturated with the electrolyte solution. The overall resistance decreased with the increase in electrolyte concentration and the thickness of the solution/particulate layer (height h), but increased with the increase in the distance between two testing electrodes (length l). It should also depend on the width (w) of the testing cell, which remained a constant (9mm) in this study. In

order

to

explore

the

methodology

for

calculating

the

bulk

resistivity/conductivity of a particulate layer saturated with electrolyte solution, an analogy was made between the solution layer in the testing cell and a resistor which had a resistivity ρ (inverse of conductivity σ), a length of l, a height of h, and width of w as shown in figure 3-21. The resistance of such a resistor can be determined by equation 3.8:

R=ρ

l h∗w

(3.8)

Figure 3-21 Analogy of solution/particulate layer to a conventional resistor

69

A Plot of resistance vs.

l at various heights for NaCl solution with sand h∗w

particulates was shown in figure 3-22. Figure 3-22 shows that for each thickness, the R vs.

l plot was near linear with the slope (corresponding to the resistivity) increasing h∗w

with the increase in thickness. This supported the feasibility of the methodology as described previously. In order to simplify the investigation of this methodology, KCl solutions (0.1M, 0.01M, and 0.001M in concentration) were used as the testing media. The choice of KCl solutions was based on the fact that KCl solutions are standard solutions for calibrating conductivity vessels [13, 14]. Conductivity data of KCl solutions of various concentrations were available

[14, 15]

. In this study, KCl solutions were prepared

by dissolving commercially available KCl powders into de-ionized water forming a 1000ml solution. The weights of the KCl powers were controlled to obtain specific concentrations.

200000

180000

y =74819x +87088 R2 =0.8993

160000

y =17569x +81919

y =14528x +66818

R2 =0.9182

R2 =0.9834

y =32716x +75502

Resistance (ohm

R2 =0.9454 140000

120000

100000

height=1 height=2

80000

height=4 height=8

60000

Linear (height=1) Linear (height=2)

40000

Linear (height=4) Linear (height=8)

20000

0 0

1

2

3

4

5

6

7

8

9

l/ wh ( 1/ mm)

Figure 3-22 (a) Resistance plot for 0.001M NaCl solution saturated sand particulate: R is near linear with l/(h*w), the slope increases with the increase in solution thickness

70

30000

y = 3609.1x + 9001.8 R2 = 0.9843

25000

y = 2439x + 8002.8 R2 = 0.9964 y = 4356.2x + 10009 R2 = 0.9631

Resistance (ohm)

20000

y = 8204.1x + 10038 R2 = 0.9501 15000 height=1 height=2 height=4

10000

height=8 Linear (height=2) Linear (height=1) Linear (height=4)

5000

Linear (height=8)

0 0

1

2

3

4

5

6

7

8

9

l/wh (1/mm)

Figure 3-22 (b) Resistance plot for 0.01M NaCl solution saturated sand particulate

3500

3000 y = 368.33x + 1141 R2 = 0.9506 y = 251.38x + 1063.5 R2 = 0.9622

2500

Resistance (ohm)

y = 456.14x + 1160.6 y = 762.93x + 1200.3

R2 = 0.9257

R2 = 0.8565 2000

1500

height=1 height=2 height=4 height=8

1000

Linear (height=1) Linear (height=2) Linear (height=4)

500

Linear (height=8)

0 0

1

2

3

4

5

6

7

8

9

l/wh (1/mm)

Figure 3-22 (c) Resistance plot for 0.1M NaCl solution saturated sand particulate

71

Figure 3-22 indicates that the resistivity/conductivity measured through this methodology was a function of solution/particulate layer thickness for this testing cell design. This was illustrated by figure 3-23. In order to be consistent with the standard conductivity calibration procedure, the conductivity was used which was simply the inverse of the resistivity.

0.00025

conductivity (s)

0.0002

0.00015

0.0001

0.00005

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

thickness h (cm)

Figure 3-23 Conductivity of 0.001M KCl solution: conductivity decreases with the increase in solution thickness

Modeling of corrosion phenomenon often required the use of bulk conductivity. In this study, a cell constant methodology was utilized to facilitate the measurement of the bulk conductivity for the electrolyte solution/particulate layer saturated with electrolyte solution. The solution thickness specific conductivity could be correlated with

72

the bulk conductivity through cell constant c. cell constant was defined as the ratio of the solution thickness specific conductivity ks and bulk conductivity kbulk as shown by equation 3.9:

c=

ks kbulk

(3.9)

The bulk conductivity data of KCl solutions used in this study were obtained using a commercial conductivity meter. The conductivities used in this study were 12.86ms/cm, 1.409ms/cm, and 0.147ms/cm for 0.1M, 0.01M, and 0.001M KCl solutions, respectively. These conductivity data were comparable with NIST standard reference materials data

[15]

. Figure 3-24 shows the calculated cell constant as a function of the

solution thickness for the three different KCl solutions. Within the allowed experimental errors, the cell constant plot of these three solutions fell into one curve, indicating that the cell constant was not a function of the solution concentration. This fact supported the feasibility of using such a conductivity cell to measure the bulk conductivity of an unknown solution/particulate layer. Figure 3-24 also shows that the cell constant decreases as the solution thickness increases. This could be understood as demonstrated in equation 3.10:

c=

k kbulk

=

1 1 l = ∗ kbluk ∗ ρ kbulk ∗ R w ∗ h

(3.10)

Despite the fact that the resistance R decreased (which led to the increase in cell constant c) with the increase in solution height h as shown in figure 3-17 previously

73

(which led to the decrease in cell constant c), the combined effect was that the cell constant c decreased with the increase in solution height h.

1.4

1.2 0.1M KCl 1

0.01M KCl

cell constan

0.001M KCl 0.8

0.6

0.4

0.2

height of solution (cm) 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Figure 3-24 Cell constant as a function of solution thickness for 0.1M, 0.01M, and 0.001M KCl solutions: cell constant does not depend on the solution concentration and decreases with the increase in solution thickness

74

3.6 Summary 1. A.C. Impedance technique was a good method for measuring the resistances of an electrolyte solutions/particulate layer saturated with an electrolyte solution because an equivalent circuit could be established to analyze the mass transfer controlled impedance. 2. In order to measure solution resistance, the mass transfer resistance portion of the Nyquist plots of A.C. impedance (real part) was utilized by extrapolating the A.C. frequency to infinity. By using a microfabricated thin film electrochemical sensor, the sensitivity of the measurements was improved. 3. The impedance measurements of pure solutions and particulate layers saturated with solutions showed similar behavior, indicating that this technique was applicable to corrosion studies for Yucca Mountain project. The test results showed that the resistance increased with the decrease in solution concentration and the increase in electrode distance, but decreased with the increase in solution/particulate layer thickness. 4. Thin layer resistivities of solutions/particulate layers were extracted from resistance measured in this study by plotting resistance against l/wh. The slopes of these plots were deemed as the resistivities according to an analogy to resistors. Resistivities were found to increase with the increase in solution/particulate layer thickness. 5. Cell constants were calculated as ratios of thickness specific conductivities and bulk conductivities. Cell constants decreased with the increase in solution/particulate layer thickness. The cell constants were useful for measuring bulk conductivities of unknown solutions, important parameters for corrosion behavior modeling.

75

References: [1] http://www.physclips.unsw.edu.au/jw/ac.html#impedance

[2] http://www.fuelcell-magazine.com/eprings/free/agilentfeb03.pdf [3] A. bard and L. Faulkner, “Electrochemical methods: fundamentals and applications”, 2nd edition, 2001 John Wiley & Sons, Inc.

[4] http://www.yorogawa.com/tm/ty/tm-tr0605_01.html

[5] http://en.wikipedia.org/wiki/image:angular velocity.svg

[6] http://en.wikipedia.org

[7] http://www.ciks.cbt.nist.gov/garbocz/papers3/nodes3.html

[8] http://www.gamry.com/app_notes/EIS_primer/EIS_primer.html

[9] S. Refaey and G. Schwitzgebel, “Electrochemical impedance spectroscopic investigation of dissolution, passibation and pitting corrosion……”, Applied Surface Science, 135 (1998) 243-253 [10] R. Robinson and R. Stokes, “Electrolyte solutions” 2nd edition, 1959, London: Butter Worths

[11] R. Jameel, NIST Special Publication 260-142. 2000, National Institute of Standards and Technology

[12] P. Kissinger and W. Heineman, “Laboratory techniques in electroanalytical chemistry”, Marcel Dekker Inc., 1984

[13] http://www.kayelaby.npl.co.uk/chemistry/3_9/3_9_1.html

[14] http://www.radiometer-analytical.com/en_meterlab_molar.asp

[15] http://ts.nist.gov/MeasurementServices/ReferenceMaterials/upload/260-142-2ndVersion.pdf 76

Chapter 4 pH Measurement Based on Open Circuit Potential Approach

4.1 Introduction The pH value is one of the many important common parameters both in a laboratory and in an industrial application because many chemical and biological processes depend on pH. The determination of pH values has been one of the most important tasks in analytical chemistry. [1] Both the solubility of many chemicals and the rate of electrochemical reactions are dependent on the solution pH.

[2]

For the modeling

of corrosion processes, it is essential to know the pH value in the environment. In order to evaluate the long-term stability of the barriers for the Yucca Mountain Project, it is necessary to monitor the pH changes in the testing environment. The term pH comes from the combination of p (which means power) and H (which symbolizes element hydrogen) [3]. In aqueous solutions, equation 4.1 describes the equilibrium between the water molecule (H2O), the hydronium ion (H+), and the hydroxyl ion (OH-): H 2O R H + + OH −

(4.1)

A pH value describes the degree of acidity or alkalinity of a solution with a measure on a scale of 0 to 14. The formal definition of pH is the negative logarithm of the hydrogen ion activity (α), [4] which is defined as in equation 4.2. Activity was used in equation 4.2. However, for practical usage, hydrogen ion concentration is often used to replace its activity. The results of such a substitution are reasonable and acceptable when the proton is in dilute concentration range.

77

pH = − log(α H + )

(4.2)

Many electrochemical and non-electrochemical methods have been used to measure pH values, tracing to 1889, when Arrhenius proposed using catalytic measurements for pH detection.

[4]

Another conventional pH measuring technique is

colorimetry (based on the change of color of indicator reagents (organic acid-base systems that are absorbent in the visible range)). The widely used pH strips are an example. Optical sensors had also been developed for pH sensing

[16]

. Details of several

techniques for pH sensing are summarized in section 4.2. In recent years, pH value is most often measured electrochemically by using electrochemical reactions involving H+ or OH- ions. The traditional pH measuring circuit consists of two electrodes immersed into the fluid (whose pH value is to be determined), one pH sensing electrode and one reference electrode. Glass electrodes, hydrogen electrodes, organic redox-system based electrodes, and metal/metal oxide electrodes can all function as the sensing electrode

[5 & 6]

. A glass electrode is the commonly used pH

electrode; the principle is based on the discovery of German chemist Fritz Haber in 1901 that the voltage changes with solution acidity [5]. Although such a glass pH electrode is a well-established laboratory instrument, it is not easily adaptive for integration onto a micro sensor array. A hydrogen electrode is a standard for all pH measuring methods, in which the activity of the hydrogen ions is potentiometrically determined. However, this method requires the handling of hydrogen gas, which makes it unpractical for daily usage. The use of organic redox-system based electrode has its reproducibility issue. Metal/metal oxide coupling for pH sensing has been widely explored as well.

78

[6]

[7, 8, and 9]

Some metal/metal oxide based pH electrodes possess advantages including easy fabrication, excellent sensitivity, good reproducibility, and good long term stability. They can also function at elevated temperature covering a wide range of pH values. Pt/PtO2, W/WO3, Sn/SnO2 Pb/PbO2, Ru/RuO2, Ir/IrO2, Sb/Sb2O3, and Pd/PdO have been investigated as pH sensor materials [10]. Shuk et al. reviewed newer metal-oxide-type pH sensors

[17]

. Among them, Ir/IrO2 based pH electrodes were attractive, demonstrating

good stability over a wide pH range and fast response even at high temperature or in nonaqueous solutions. However, it was reported by Glab that the performance (in term of sensitivity) of Ir/IrO2 pH sensors depended highly on the process of their fabrication. [11] It also often displayed potential drift, which caused errors in continuous pH monitoring [15]

. A good alternative choice was a Pd/PdO based pH sensor

[10, 12-14]

. Grubb et al.

reported in 1980 that a Pd/PdO wire pH electrode was stable for more than 6 years and covered pH values from 3 to 11

[12]

. Shao et al. also published their work on a

microfabricated palladium based pH sensor, the measurement of which was based on the on-site pre-cleaning step followed by a potential decay process.

[9]

This approach

provided a reproducible, linear pH calibration response over a wide pH range. Another advantage of this approach was the feasibility of miniaturization, which made it possible for the sensor to be integrated onto a sensor array. [9] Palladium and its alloys had been widely studied for their unique ability to selectively adsorb and transfer hydrogen.

[18]

The interaction between palladium and

hydrogen and the semiconductor feature of palladium oxide had made palladium and palladium oxide attractive candidates for a variety of applications, ranging from medical,

79

chemical, to electronic applications. One of these applications would be serving as a pH sensor. [10, 12, 19, and 20] Various techniques could be utilized to produce palladium/palladium oxide electrodes for pH sensing, such as sputtering, electrochemical growth, and thermal preparation. In this study, the thick-film printing technique was applied to form a palladium layer (please refer back to Chapter 2 for details), which could be electrochemically oxidized in situ to form the Pd/PdO coupling for pH sensing. Thick film printing of Pd/PdO pH sensors was a developed process in the EDC at Case Western Reserve University with low cost, close to theoretical sensitivity, good stability, and good reproducibility. 4.2 Principle of Pd/PdO pH Sensing Many detecting techniques for pH sensing and monitoring can be used. They can be generally divided into two categories: non-electrochemical methods and electrochemical methods

[1]

. Depending on the detecting methods, the principles can be

vastly different. The examples of non-electrochemical pH sensors are [2]: 1. Optical-fiber based pH sensors These sensors use organic dye molecules with pH-dependent spectral properties. The surface-adsorption of light by the organic dye occurs as part of the evanescent wave that penetrates the fiber core, internally reflecting at the interface of the core and cladding or at the end. The depth of penetration of light is approximately the wavelength, which can be detected by the sensing optical fiber.

80

2. Mass balance based pH sensors These sensors function through utilizing the mass-changing pH-dependent hydrogels, which are coupled with a magneto-elastic sensor. The resonance frequency of the magneto-elastic sensor depends on the applied mass loads, which can be used to quantify the pH values. 3. Nano-structured cantilever based pH sensors These sensors use a cantilever structure that are coated with a selected polymer (for example, poly(methacrylic acid PMAA) with poly(ethylene glycol) dimethacrylate). The polymer network expands once the pH value exceeds the pKa of the polymer, resulting in a reversible surface stress change, which causes the cantilever to bend. The degree of bending depends on pH values. 4. pH-imaging based pH sensors. The proton-sensitive surface such as Si3N4/SiO2/Si structure can function as a light addressable potentiometric sensor. This photocurrent characteristic of silicon can be utilized to obtain acid or base spatial resolutions, which can be converted to pH-images and thereby to achieve pH measurements [2]. Electrochemical methods are more commonly used than the non-electrochemical methods. As described previously, electrochemical methods can be divided into several subcategories, such as Pt-hydrogen electrode based pH sensors, organic redox electrode based pH sensors, metal/metal oxide electrode based pH sensors, liquid and gel

81

membrane electrode based pH sensors, glass electrode based pH sensors, and pH-field effect transistor (FET) based pH sensors

[1 & 2]

. pH sensors that use measurements other

than potentiometric measurements have been reported, for example, Sheppard et al. reported a microfabricated conductimetric pH sensor

[21]

. In their report, the electrical

impedance (which was converted into conductivity) of the pH-responsive hydrogel was used for pH value determination. But the majority of the electrochemical pH sensors function based on the Nernst equation as described below. This potentiometric principle of pH detection was followed by this study. Arrhenius’ ionic theory revolutionized the understanding the behavior of acids and bases. The basic concept was that electrolytes in aqueous solutions will dissociate into ions of opposite electrical charges. Weak acids and bases could dissociate only partially. Taking an acid as an example, the dissociation of an acid can be described by equation 4.3: H n A ⇔ nH + + An −

(4.3)

According to Fog and Buck [10], the surface mechanism of a metal/metal oxide pH electrode can be expressed by equation 4.4 [22]: MOx + 2δ H + + 2δ e − ⇔ MOx −δ + δ H 2O

(4.4)

The electrode potential in this case can be written as equation 4.5 [22]:

µs µ 1 RT ln(α Hl + ) φ − φ = ( )( µ Hl + + µes− + O − H 2O ) = " cons tan t "+ F 2 2 F l

s

l

82

(4.5)

Where M was metal, µ represented chemical potential, Φ was Galvani potential, F was the Faradiac constant (96485C), R is the molar gas constant (8.314/mol/K), T is the absolute temperature (K), and s and l were solid phase and liquid phase, respectively. In this study, Palladium was selected as the pH electrode material, and equation 4.3 can be re-written as equation 4.6 accordingly (x = 1 and δ = 1): PdO + 2 H + + 2e − ⇔ Pd + H 2O

(4.6)

The Nernst equation can be expressed as equation 4.7 (replacing Φ with E):

E = E° −

[ Pd ] ∗ [ H 2O] RT 2.303* RT ln( ) ≈ E′ − [ pH ] + 2 2 F [ PdO] ∗ [ H ] F

(4.7)

According to equation 4.7, the potential vs. pH plot of this Palladium/Palladium oxide electrode should yield a straight line with a Nernstian slope, representing the sensitivity of this pH electrode. The value of this pH sensor depends on the temperature as shown in equation 4.7. For example, if the temperature is 25°C, the Nernstian slope calculated should be around 59.2mv/pH. It could also be deduced from equation 4.6 that the value of the potential of this Pd/PdO based pH electrode should decrease with the increase in pH value. 4.3 Experimental A CHI660C electrochemical workstation (potentiostat) from CH Instruments, Inc. was used to control AC voltage and measure the current output. Characterization of Pd/PdO based pH sensors were carried out by submerging soldered and insulated sensor

83

chips in pH buffer solutions with different pH values in a beaker at ambient environment. The testing was conducted in a three-electrode-configuration fashion, with Pd working electrode, Pd counter electrode, and Ag/AgCl reference electrode. A picture of the pH sensor was shown in chapter 2 (figure 2-13). The Pd working electrode was first cleaned in situ in the pH buffer solution using cyclic voltammetry for 999 cycles (-0.6v to 1.2v vs. Ag/AgCl reference electrode). Then the working electrode was held at +0.9v for 60s and later stepped to -0.4v for 90s. Lastly, the open circuit potential decay of the working electrode was recorded as a function of time. When it reached equilibrium, the value of open circuit potential was used to represent the pH value of the testing medium. A calibration curve for the Pd/PdO based pH sensors in this study was established using open circuit potential data at near equilibrium. Figure 4-1 shows a typical cyclic voltammogram result of a Pd working electrode in buffer solution with a nominal pH value of 6. The voltage window (the voltage range within which the voltage was scanned) and the sweep rate of voltage are important parameters afeecting the shape of the cyclic voltammogram. Cyclic voltammetry is a very useful electrochemical technique, and will be more fully illustrated in Chapter 5. In this part of the study, cyclic voltammetry was used only to clean in situ the working electrode; therefore, detailed cyclic voltammetry study and data analysis are not summarized at this point. Figure 4-2 shows the typical open circuit potential measurement of Pd/PdO based thick film pH sensor (pH = 6) before open circuit potential measurements. When the open circuit potential measurements were carried out, the working electrode was first held at -0.4v before data were taken, therefore the open circuit potential increased with time until it reached equilibrium.

84

Figure 4-1 Cyclic voltammetry of Pd-based pH sensor in pH=6 buffer solution: potential is vs. Ag/AgCl reference electrode and the peak at near -0.3v is due to the reduction of PdO

Figure 4-2 Open circuit potential of Pd-based pH sensor in pH=6 buffer solution: open circuit potential increases with time and reaches near equilibrium in less than 300s

85

4.4 Results and discussion For the open circuit potential (OCP) to reach true equilibrium, several minutes was necessary. For practical purposes, the response time of such pH sensors should ideally be short. An alternative approach was to read the data when the rate of change in open circuit potential became negligible. The experiments showed that the rate of increase in the open circuit potential became negligible after merely 120 seconds. In order to better illustrate this phenomenon, the first order derivative of the open circuit potential is shown in Figure 4-3. Figure 4-3 also proves that the Pd/PdO based thick film pH sensor developed in this study displayed fast response time. Accordingly, the open circuit potential tests were taken at 300 seconds for data regression when establishing the pH sensor calibration curve.

Figure 4-3 Rate of increase in open circuit potential measurement of Pd-based pH sensor in pH=6 buffer solution: very fast response, after only 120s the increase in potential is negligible

86

The pH buffer solutions were purchased from Fisher Scientific, Inc. ( nominal pH values of 6, 8, 10, 11, 12, and 13). These pH values were first calibrated using a commercial pH meter and the comparison of nominal values and actual meter readouts are shown in table 4-1. The pH calibration curve used the meter measured values instead of the nominal values of pH buffer solutions. Dilute KCl (0.002M) was added into the buffer solutions increasing the solution conductivity to obtain better measurements. Table 4-1 Comparison of meter measured values & nominal values of pH buffer solutions Nominal value

6

8

10

11

12

13

Measured value

5.92

7.87

10.09

10.81

11.63

12.49

The typical open circuit potential measurements of Pd/PdO based pH sensors in this study are shown in figure 4-4 (sensor A). Figure 4-5 shows the testing results of a different sensor chip (sensor B) to illustrate the reproducibility among individual sensors. The open circuit potential values decreased with the increase in pH values, in agreement with the prediction of Nernst equation (equation 4.7).

87

Figure 4-4 Open circuit potential measurements of pH buffer solutions for sensor A: potential increases with the decrease in pH value

Figure 4-5 Open circuit potential tests of pH buffer solutions for sensor B: potential increases with the decrease in pH value

88

To better illustrate the reproducibility of thick film Pd/PdO based pH sensors in this study, the collective calibration curve of nine sensors is shown in figure 4-6. Also the collective calibration curve for one sensor (4 runs) is shown in figure 4-7.

Figure 4-6 Collective first run calibration for 9 different pH sensors: OCP is in linear with pH values

Figure 4-7 Calibration curve of one Pd/PdO based pH sensor (4 runs): OCP is in linear with pH values

89

The sensitivity of the thick film Pd/PdO based pH sensors was about 50mV/pH at room temperature, which was represented by the Nernstian slopes in figure 4-6 and figure 4-7 (49mV/pH and 50mV/pH, respectively). It was in good agreement with the theoretical value (59.2mV/pH) predicted by equation 4.7. The sensitivity of this Pd/PdO based pH sensor was comparable to that of RuOx (51mV/pH) and IrOx (49mV/pH).

[14]

The discrepancy between the tested sensitivity and theoretical sensitivity could be attributed to several factors. For example, the actual temperature of the testing solution might be quite low relative to room temperature. Another reason was that the thick-film printed Pd layer was not composed of 100% palladium; it also contained binding reagent residues. 4.5 Summary Thick film printed Pd/PdO based pH sensors were fabricated and tested. The results indicated that this Pd/PdO-based pH sensor could be used for pH value measurements over a wide pH range (6-13) with near-linear dependence of pH, fast response time, excellent reproducibility, and satisfactory sensitivity. Cyclic voltammetric in situ pre-cleaning of the working electrode and open circuit potential decay measurements were used in this study. A good linear dependence of open circuit potential on pH was exhibited. The Nernstian slope that represented the sensitivity of this pH sensor was obtained from the experimental results with a value of 50mV/pH. It was in relatively good agreement with the theoretical Nernstain sensitivity values.

90

References: [1] W. Vonau and U. Guth, “pH monitoring: a review”, J. Solid State Electrochem., 10 (2006) 746-752

[2] Y. Miao, J. Chen, and K. Fang, “New technology for the detection of pH”, J. Biochem. Biophys. Methods 63 (2005) 1-9

[3] Sorenson SPL, “Enzyme studies II: the measurement and meaning of hydrogen ion concentration in enzymatic processes”, Biochem. Z 21 (1909) 131-200

[4] S. Arrhenius, Z Phys. Chem. 4 (1889) 226

[5] H. Norton, Sensor and analyzer handbook, Prentice, Hall Inc., 1982

[6] R. Buck, S. Rondinini, AK Covington, FGK Baucke, CMA Brett, and MF Camoes, “Measurement of pH: definition, standards, and procedures”, Pure Appl. Chem. 74 (2002) 2169-2000

[7] C. Liu, D. Bocchicchio, P. Overmyer, and M. Neuman, “A palladium-palladium oxide miniature pH electrode”, Science, vol. 207, 11 Jan. 1980

[8] A. Fog and R.P. Buck, “Electronic semiconducting oxides as pH sensors”, Sensors and actuators, 5, 137-146, 1984

[9] M. Shao, X. Xing, and C. Liu, “pH measurements based on a palladium electrode”, Electroanalysis, 6, 245-249, 1994

[10] S. Glab, A. Hulanicki, G. Edwall, and F. Ingman, “Metal-metal oxide and metal oxide electrodes as pH sensors”, Crit. Rev. Anal. Chem. 21 (1989) 29-47

[11] W. Grubb, “Palladium-palladium oxide pH electrodes”, Anal. Chem. 5 (1980) 273-276

[12] VA Karagonis, CC Liu, MR Neuman, LT Romankiw, PA Leary, and JJ Cuomo, “A Pd-Rd film potentiometric pH sensor” IEEE Trans. Biomed. Eng. BME-33 (1986) 113-116

91

[13] K. Kreider, M. Tarlov, and J. Cline,“Sputtered thin-film pH electrodes of platinum, palladium, ruthenium, and iridium oxides”, Sensors and Actuators B 28 (1995) 167-172

[14] M. Wang, S. Yao, and M. Madou, “A long-term stable iridium oxide pH electrode”, Sensors and Actuators B 81 (2002) 313-315

[15] A. Safavi and H. Abdollahi, “Optical sensor for high pH values”, Analytica Chimica Acta 367 (1998) 167-173

[16] P. Shuk, KV Ramanujachary, and M. Greenblatt, “New metal-oxide-type pH sensors”, Solid State Ionics 86-88 (1996) 1115-1120

[17] T. Yang and S. Pyun, “Hydrogen adsorption and diffusion into and in palladium”, Electrochimica Acta. 41 (6) (1996) 843-848

[18] C. Herard, P. Bowen, J. Lemaitre, and J. Dutta, “Chemical synthesis and characterization of nanocrystalline palladium oxide”, Nanostructured Materials 6 (1995) 313-316

[19] M. Jeong, CH Pyun, and IH Yeo, J. Electrochem. Soc. 140 (1993) 1986

[20] N. Sheppard, M. Lesho, P. McNally, and A> Francomacaro, “Microfabricated conductimetric pH sensor”, Sensors and Actuators B 28 (1995) 95-102

[21] C. Pan, J. Chou, T. Sun, and S. Hsiung, “Development of the tin oxide pH electrode by the sputtering method”, Sensors and Actuators B 108 (2005) 863-869

92

Chapter 5 Amperometric i-t Oxygen Concentration Measurement

5.1

Introduction

5.1.1 Dissolved Oxygen Measurements A major component of the long-term strategy for safe disposal of nuclear waste is to completely isolate the radionuclides from penetrated packages. Corrosion is a primary determinant of waste package performance at the proposed Yucca Mountain repository and will control the delay time for radionuclide transport from the waste package. Corrosion is the most likely degradation process that will determine when packages will be penetrated and also the shape, size, and distribution of those penetrations. Thus, corrosion resistance is important to the long-term performance [1]. The corrosion performance of a metal is determined by the inherent corrosion resistance of the metal and the corrosivity of the environment. The amount, distribution, and chemical composition of the moisture on waste packages are controlling parameters of corrosion performance. Advanced analytical and computational methods for the evolution of the environment on metal surfaces require data for the properties of thin layers of moisture, moist particulates, and deposits that will affect the corrosion performance of metals. The advanced sensor technology development and application of those sensors reported here address these needs. The continuous measurement of dissolved oxygen is an essential task to not only the study of corrosion process, but also many technological processes such as environmental analysis, waste management, food processing, and medical applications. L.C. Clark Jr. introduced the amperometric principle of an oxygen sensor in 1956 [2]. The

93

Clark oxygen sensor consisted of a sensing platinum electrode (cathode) and a reference silver electrode (anode) enclosed with an oxygen permeable membrane. Based on his approaches, various oxygen sensors have been developed. Generally, sensors for dissolved oxygen measurements that can be classified into two categories: optical and electrochemical. The optical oxygen sensors typically use organic materials and are equilibrium devices. The function of these sensors is based on either the changes in the absorbance of molecules that bind reversibly with oxygen, or on the ability of oxygen to quench the phosphorescence or the fluorescence of aromatic species such as anthracene, pyrene derivatives, and porphyrins

[3]

. The electrochemical

oxygen sensors include the Clark style amperometric sensors [2], galvanic oxygen sensors [4, 5]

, and solid-state electrolyte potentiometric sensors [6]. Other approaches such as iodine

titration have been used. Although the optical sensors possess high sensitivity and stability [7], they are not suitable for continuous dissolved oxygen concentration monitoring, especially when the transient oxygen change measurements are involved. Thereby, the electrochemical alternatives are much more widely employed in both laboratory studies and industrial applications. The electrochemical measurement of dissolved oxygen can be achieved using amperometric, potentiometric, and resistive approaches

[2, 6, and 8]

, depending on the

testing conditions and the oxygen concentrations. The principle of the amperometric oxygen sensor is measuring the reductive current of oxygen at the cathode at a given potential (relative to the potential of the reference electrode) (-800mV for example with an Ag/AgCl reference electrode [2]). The

94

potential at which the amperometric tests are taken depends on the type of electrode material used to construct the working electrode

[10]

and the test environment. This

potential can be determined through experiments, normally through cyclic voltammetry. A low potential value is favorable in order to reduce the influence of reduction processes of other species. The reduction current is typically in linear relationship with the concentration of dissolved oxygen. Typically a three electrode configuration is employed in this method, with noble metals such as platinum and gold as the working and counter electrodes and either an external reference electrode or an embedded Ag/AgCl reference electrode. The detailed measuring mechanism will be discussed in the following section 5.2. The potentiometric oxygen sensor measures the potentials of the reversible oxygen electrode against a reference gas or reference materials such as metal-metal oxide mixtures or the mixed potentials resulted from metallic corrosion and simultaneous oxygen reduction

[1 & 2]

. These measurements consequently exhibit a Nernstian behavior

as described in chapter 4. In this case, the potential is in logarithmic relationship with the concentration of the dissolved oxygen. Section 5.2 will describe such a mechanism in more detail. For the resistive oxygen sensor, the measurement of the oxygen concentration is achieved through measuring the resistance/conductance of a specific solid solution. This solid solution possesses large quantity of oxygen vacancies. The resistance of these materials decreases as the concentration of oxygen vacancy increases. Meanwhile, the oxygen vacancy concentration decreases with the increase in the concentration of the oxygen. Thereby, the resistance of the solid solution increases with the increase in

95

oxygen concentration. Such a resistive-type oxygen sensing is achieved only in elevated temperatures, rendering it unsuitable for ambient measurements. Semiconductors can also be utilized as resistance-type oxygen sensors

[11]

. This type of sensing mechanism is not

discussed further in this study. Depending on the concentration of dissolved oxygen, either the amperometric method or the potentiometric method can become favorable. This is determined by the sensitivity of these methods. As stated before, in the amperometric method, the oxygen concentration is represented by the reductive current and in linear relationship with the current. Thus for rich (high concentration) oxygen measurements, amperometric approach is better suited due to its linear sensitivity. However, for lean (low concentration) oxygen measurements, the logarithmic sensitivity of potentiometric method becomes an advantage over the linear counterpart

[9]

because a small change in

the oxygen partial pressure can lead to a large logarithmic change in the potential. 5.1.2 Microfabricated Oxygen Sensors The demands of automotive, environmental, and process control applications on chemical sensors have driven the advancement of the fabrication techniques of these sensors in recent years. These chemical sensors can be characterized by three parameters: the sensitivity that represents the ability to quantify the concentration, the selectivity that shows the ability to detect specific species free from interference, and the response time. In addition, long term stability, power consumption, and reversibility are also needed to be considered

[9]

. It is also desirable that the fabrication techniques for these sensors can

96

bring advantages such as reduced size, small sample consumption, geometrically welldefined elements, and high uniformity [12]. Advancements in microfabrication and micromachining techniques in recent years have shown great promise for the manufacture of electrochemical devices with small critical dimensions, high sensitivity, low energy consumption, and modest cost. In addition, these techniques have demonstrated the capability of producing geometrically well-defined, identical, and highly reproducible microstructures in large batch quantities [13]

. The micro-electrochemical devices produced have the same electrochemical behavior

as the bulk macroscopic counterparts, and the micro-devices will have better sensitivity and shorter response time. These microfabricated devices are also ideal for integration due to their reduced sizes. Therefore, microfabrication has become a widely used technique for the manufacture of chemical sensors and biosensors for various applications such as environmental monitoring and dissolved oxygen measurements [15]. Two of the most widely used microfabrication techniques are thin film metallization and thick film printing. The former technique is capable of producing electrochemical devices such as chemical sensors with feature dimensions at micro and even sub-micro levels. The latter one is commonly used to manufacture electrochemical devices that are hundred-microns in sizes. The selection of these two techniques depends on the specific requirements of the sensors to be made since the significance of the miniaturization of sensors depends on the particular application. In this study, thick film printing technique was used for the microfabrication of dissolved oxygen sensors according to the consideration of their applications in the Yucca Mountain Nuclear Waste Barrier Project.

97

5.2 Fundamentals 5.2.1 Cyclic voltammetry Amperometric oxygen sensors are based on the measurement of oxygen reduction current at the working electrode at a given potential. In order to achieve such amperometric measurements, the potential for the working electrode needs to be determined first. This is typically done through executing cyclic voltammetry tests. Cyclic voltammetry has become a popular technique for initial electrochemical studies of an un-identified system. It has proven to be a very useful tool for obtaining information about complicated electrode reactions potentiodynamic

electrochemical

[16]

. Cyclic voltammetry is a type of

measurement.

In

order

to

obtain

a

cyclic

voltammogram, the voltage is varied in a solution (test medium) and the change in current is measured with respect to the change in voltage. Redox properties of chemicals can be studied by cyclic voltammetry. [17]. In a cyclic voltammetry experiment a potential is applied to the system, and the faradaic redox current response is measured. Figure 5-1 showed a typical linear potential ramping as a function of time. The current response over a potential window starting to a pre-set value with two pre-defined limits is measured. At one potential limit (commonly referred to as a switching potential), the potential scan is reversed to the opposite direction (polarity) until reaches the other limit. The same potential window is then scanned in the opposite direction again. This mode of operation is defined as “cyclic” operation. During such a cyclic process, species formed by oxidation on the forward scan can be reduced on the reverse scan. Cyclic voltammetry is a fast and simple method to obtain an estimation of the redox potential for amperometric measurements [17].

98

Figure 5-1 Linear potential sweep for cyclic voltammetry [17] Three-electrode method (a reference electrode, a working electrode, and a counter electrode/auxiliary electrode) is mostly used for cyclic voltammetry as the majority of experiments are conducted in a solution. In cyclic voltammetry the electrical potential of reference normally does not change during the measurements. In order to ensure sufficient conductivity of the testing solution, supporting electrolyte such as KCl is often added as a common practise. The combination of the solvent, supporting electrolyte and the specific working electrode material determines the range of the potential

[17]

. The

potential is measured between the reference electrode and the working electrode and the current is measured between the working electrode and the counter electrode. The experimental results are then plotted as current (i) vs. potential (E). The typical cyclic voltammogram is illustrated by figure 5-2. As the waveform shows, the forward scan produces a current peak for an analyte that can be either oxidized or reduced through the 99

range of the potential. In this example, we will choose to reduce the analyte first. The current will increase as the potential reaches the reduction potential of the analyte, but then falls off as the concentration of the analyte is depleted at the electrode surface. As the applied potential is reversed, it will reach a potential that will reoxidize the product formed in the first reduction reaction, and produce a current of reverse polarity from the forward scan. This oxidation peak will usually have a similar shape to the reduction peak.

Figure 5-2 Cyclic voltammogram [17] As a result, information about the redox potential and electrochemical reaction rates of the compounds are obtained. If the electronic transfer at the surface is fast and the current is limited by the diffusion of species to the electrode surface, then the current

100

peak will be proportional to the square root of the scan rate. The peak current can be determined by equation 5.1[16]: i p = cn3/ 2 ADO1/ 2CO* v1/ 2

(5.1)

where c is a constant, n is the number of electrons transferred, A is the working electrode surface area, Do is the diffusion coefficient of the analyte, CO* is the bulk concentration, and v is the scan rate. The voltage at which the peak current occurs can be used subsequently for amperometric measurements. For a reversible wave, such a peak potential is independent of the scan rate as well as the bulk concentration. The current peak value can technically be used to correlate with the concentrations of the analyte. If the potential window is sufficiently broad, the reductive and oxidisive currents may reach a diffusion-controlled equilibrium limit, determined by the concentration gradient across the double layer on the surface of the working electrode. This diffusion-limited current is determined by a combination of Fick’s and Faraday’s Laws as shown in eqaution 5.2 [18]:

iL = −nAFDO

CO*

(5.2)

δ

where iL is the limiting current, n is the number of electrons transferred, A is the surface area of the working electrode, F is the Faraday constant, DO is the diffusion coefficient of the analyte, CO* is the bulk concentration, and δ is the thickness of the double layer. In accordance to standard electrochemical conventions, the reductive cathodic current is taken to be negative.

101

In theory, the peak currents and the limiting currents can be used to deduce the bulk dissolved oxygen concentrations, however, such a cyclic voltammetry testing can not be used for real-time or on-line monitoring of dissolved oxygen. Therefore in this part of the study, cyclic voltammetry is used to determine the working electrode potential for amperometric oxygen reductive current measurements. 5.2.2 Potentiometric measurements As stated in chapter 5.1, depending on the concentration of the dissolved oxygen, either amperometric method or potentiometric method can become favorable. For a rich (high concentration) oxygen measurement, amperometric approach with linear sensitivity is a better choice. On the other hand, for lean (low concentration) oxygen measurements, the logarithmic sensitivity of potentiometric method becomes an advantage over the linear counterpart [9]. When the potentiometric measurements are carried out, the electromotive force (EMF) correlates to the oxygen concentration. Figure 5-3 shows an example of an electrochemical cell:

POII2

Oxygen ionic conducting electrolyte

POI2

Figure 5-3 Schematic of an electrochemical cell

102

The EMF induced by the oxygen concentration gradient across the oxygen ionic conducting electrolyte at a steady state can be represented by the Nernstian equation 5.3 [19]

similar to that of a pH sensor:

EMF ∝

POII RT ln( I2 ) nF PO2

(5.3)

where POI2 and POII2 are the oxygen partial pressures on the two sides of the oxygen ionic conducting electrolyte respectively, R is the gas constant, T is the temperature, n is the number of electrons transferred, and F is the Faraday constant. A common practice is to establish the oxygen partial pressure at one side to be a constant with a reference gas (such as air) or a solid state reference (such as a mixture of metal and metal oxide). In this case, equation 5.3 can be further simplified to show a typical logarithmic relationship between the EMF and the oxygen partial pressure to be determined as shown in equation 5.4: EMF = a ln( POII2 ) + b

(5.4)

The monitoring of the dissolved oxygen for corrosion process often involves of solutions. A reference oxygen concentration that is essential to potentiometric approaches is difficult to establish in a solution environment. In addition, most of the oxygen solid ionic conducting electrolytes for the potentiometric approaches can function well only at elevated temperatures. Meanwhile, the corrosion process of interest in this study mostly takes place in an ambient environment. Although organic membranes such as Nafion can serve as the oxygen ionic conducting electrolyte at relatively low temperatures, the particulate nature of the layers of interest to the corrosion study in this work makes the 103

organic membrane approach unpractical. Also, one of the goals in this study is to obtain the effective oxygen diffusion coefficients under various conditions. This requires transient studies of oxygen diffusion process, which is a very difficult task to achieve for potentiometric measurement. Therefore the amperometric approach was selected for the dissolved oxygen measurements. 5.2.3 Amperometric i-t measurements An amperometric oxygen sensor reduces oxygen at the working electrode surface (cathode), which is held at a reduction potential versus the reference electrode. The working electrode potential is normally determined through cyclic voltammetry tests. Such a potential can either be the peak voltage where peak reductive current occurs or a voltage where diffusion controlled limiting current takes place. The oxygen reduction at a cathode can have two possible mechanisms, one involves two-electron reactions with hydrogen peroxide as an intermediate as shown in equation 5.5 and 5.6 [20]: O2 + 2 H 2O + 2e − = H 2O2 + 2OH −

(5.5)

H 2O2 + 2e − = 2OH −

(5.6)

The other mechanism is a four-electron direct oxygen reduction process, which can be shown as equation 5.7 [20]: O2 + 2 H 2O + 4e − = 4OH −

(5.7)

104

On the counter electrode, a metal oxidization reaction will take place. In this study, Au is selected as the material for the counter electrode for its superb stability. The anodic reaction can be represented by equation 5.8: Au + 2OH − = AuO + 2e− + H 2O

(5.8)

All these reactions on both electrodes are kinetically fast, therefore, the steady state is only determined by the oxygen diffusion rate through the double layer boundary or the oxygen permeable membrane. The transport of oxygen molecules to the cathode is a well known diffusion-limited process. At the cathode surface, the oxygen concentration can be driven to zero, resulting in a limiting diffusion current determined by the oxygen concentration gradient. According to equation 5.5, 5.6, and 5.7, hydroxyl ions can accumulate at the cathode surface under the coverage of the membrane (if membrane is used). This is confirmed by Hale et al. through their simulation study

[21]

. This

accumulation of the OH- ions can lead to an increase in pH value, which will affect the local corrosion process. To minimize the potential interference between the pH measurements and the dissolved oxygen measurements, no permeable membrane was used in this study. 5.3 Experimental In this study, oxygen sensor tests were carried out in a prepared 0.1M NaClO4 + 0.002M KCl solution with Air/N2 bubbling in order to control the solution oxygen concentration. Preliminary oxygen sensor calibration was achieved through submerging a packed oxygen sensor chip into the test solution and subsequently carrying out the cyclic voltammetry and the amperometric measurements. The gas bubbling was placed at the

105

back side of the sensor chip in order to minimize any testing error caused by the gas bubbles in the solution and at the surface of the solution, respectively. Appropriate flow rates of purified air and N2 gases were coordinated using a MKS Mass-flo Controller System from MKS Instruments Inc. to control the oxygen concentration. A CHI660C electrochemical workstation (potentialstat) from CH Instruments, Inc. (Austin, TX) was used to perform the three-electrode cyclic voltammetry tests and to measure the electrochemical oxygen reduction current outputs at the amperometric i-t mode. A typical cyclic voltammetry of oxygen sensor in test a solution was shown in figure 5-4. Considering the reductive nature of the cathode reaction, the voltage scan window was set from -0.2v to -0.6v and a voltage scan rate of 50mv/s was selected. Based on the results of the cyclic voltammetry, the reduction voltage for i-t measurements was set at -0.50v or -0.55v (-0.5v was the peak voltage, while -0.55v was the limiting current voltage). A typical i-t curve is shown in figure 5-5, where the current follows continuously as the oxygen concentration changes in nearly an instantaneous manner.

106

Figure 5-4 Cyclic voltammogram of oxygen sensor: limiting current happens from – 0.55v

Figure 5-5 Real-time amperometric i-t response to oxygen concentration step change for oxygen sensor: reduction current increases with the increase in oxygen concentration

107

5.4 Results and Discussion Figure 5-6 showed the cyclic voltammetry results of the oxygen sensor with different Air/N2 combinations (corresponding to different oxygen concentrations). The testing results were consistent with peak and limiting voltage between -0.6v and -0.45v. Therefore, the reduction potential against Ag/AgCl reference electrode for the amperometric tests was set at -0.55v. To demonstrate the versatility of the oxygen sensor used in this study, peak currents at -0.5v, -0.55v, and -0.6v were summarized in table 5.1. It cleared showed that the peak currents could be used to represent the bulk oxygen concentration at steady-state. To be consistent, the amperometric i-t measurements were taken at -0.55v working electrode potential in this study. Forward referred to the process with increasing oxygen concentration in table 5-1, meanwhile reverse referred to the process with decreasing oxygen concentration.

Figure 5-6 Cyclic Voltammograms of oxygen sensor at various oxygen concentrations (Pure air for 21% oxygen and pure N2 for 0% oxygen): limiting current increases with the increase in oxygen concentration

108

Table 5-1 Peak currents at various reduction potentials for oxygen sensor Oxygen

Current at -0.5v/µA

Current at -0.55v/µA

Current at -0.6v/µA

Forward

Reverse

Forward

Reverse

Forward

Reserve

21%

-22.96

-24.32

-21.51

-23.44

-20.27

-23.44

15.75%

-16.49

-17.39

-16.11

-17.1

-15.63

-17.44

10.5%

-10.75

-11.44

-10.75

-11.73

-10.74

-11.73

5.25%

-5.87

-6.35

-6.10

-5.78

-6.34

-7.31

0%

-1.47

-1.71

-1.49

-1.45

-1.73

-2.00

concentration

The linear response of the peak current with the oxygen concentration was shown in figure 5-7. The slopes of the lines represented the sensitivity of the amperometric reduction current measurements at -0.5v, -0.55v, and -0.6v, respectively. According to the cyclic voltammetry test results, peak current potential (which was about -0.5v) measurements should exhibit the best sensitivity. This was confirmed by the sensitivities shown in figure 5-7 (1.07µA vs. 1.03µA vs. 1.01µA).

109

0 0

0.05

0.1

0.15

0.2

0.25

-5

peak current (uA)

-10

-15

-20

-0.5v -0.55v -0.6v Linear (-0.5v) Linear (-0.6v) Linear (-0.55v)

-25

-30

Oxygen concentration

Figure 5-7 Effects of reduction potential on peak currents for the oxygen sensor: current response to oxygen concentration change similarly for all three potentials with good linear relationship

Figure 5-8 showed the real time amperometric i-t response of the oxygen sensor with oxygen concentration step changes. It demonstrated that the time response of amperometric measurements with oxygen concentration change was almost instantaneous, which proved the feasibility of using these sensors for on-line oxygen concentration monitoring. The forward and backward currents were within reasonable range for several oxygen-concentration-cycling although the absolute values showed slight changes for different cycles. This indicated there was presence of some hysterisis. This could be attributed to the fouling of the working electrode surface caused by the reactions of impurities in the test solution.

110

Figure 5-8 Amperometric i-t response of the oxygen sensor to oxygen concentration cycling: good reproducibility for several cycles without noticeable deterioration of performance

Figure 5-9 showed the comparison between two amperometric i-t measurements at -0.55v and -0.6v, respectively. The two curves were reasonably close to each other, indicating that both peak current potential and limiting current potential were all capable of being used as the working potential of the oxygen reduction measurements. In this study, the subsequent amperometric i-t measurements were carried at -0.55v reduction potential.

111

Figure 5-9 Comparison of amperometric i-t measurements at -0.55v and -0.6v: very similar performance

Figure 5-10 and figure 5-11 showed the first amperometric i-t measurements on several sensor chips and several repeated tests on one single sensor chip, respectively. Figure 5-10 showed that the sensitivities of 3 different sensor chips were reasonably close to each other. On the other hand, figure 5-11 showed that although each individual run on one sensor showed excellent linear response, decrease in sensitivity for one single sensor chip was noticed in figure 5-11. This suggested that initial oxygen concentration cycling might become necessary to obtain stable oxygen concentration sensing.

112

I vs. O2%, #1 held at -0.55V

I vs.O2%, #2 held at -0.55V 0

-10 0

5

10

15

20

25

I / uA

I / uA

0 -20 -30

-10 0

5

10

25

20

25

-20 -40

O2%

O2%

I vs. O2%, #3 held at -0.55V

I vs.O2%, #1, 2, 3 combined held at -0.55V

0

0 0

5

10

15

20

25

-10 I / uA

I / uA

20

-30

-40

-10

15

-20

0

5

10

15

-20

-30

-30

-40

-40 O2%

O2%

Figure 5-10 Reproducibility of first run amperometric i-t measurements at -0.55v for 3 individual oxygen sensors: all three sensors showed very good linear relationship (R2 more than 0.99) between current and oxygen concentration with good reproducibility

113

I vs.O2%,held at -0.55V, 1nd run

I / uA

0 -10 0

5

10

15

20

25

i = -1.5827C- 0.31 R2 = 0.9994

-20 -30 -40

O2%

I vs.O2%, held at -0.55V, 3rd run

I vs.O2%, held at -0.55V, 5th run

0

-10

0

5

10

15

20

25

I / uA

I / uA

0

i = -0.9793C- 0.188 R2 = 0.9999

-20 -30

-10

0

5

10

15

20

25

20

25

i = -1.2852C + 0.1591 R2 = 0.9978

-20 -30

O2%

O2%

I vs. O2%, held at -0.55V, 4th run

I vs.O2%, held at -0.55V, 5th run

0 0

5

10

15

20

25

0

I / uA

I / uA

-10

-20

i= -0.9769C - 0.8589 R2 = 0.9961

-10 -20

0

5

10

15

i = -0.9793C- 0.188 R2 = 0.9999

-30 -30

O2% O2%

Figure 5-11 Reproducibility of amperometric i-t measurements at -0.55v for a single oxygen sensor chip: sensor displays excellent performance for several tests with excellent linear relationship between current and oxygen concentration, however, sensitivity does decrease after multiple runs

Besides obtaining the calibration curve for the thick film printed oxygen sensors, measuring the effective oxygen diffusion coefficient under different conditions was desired in this study. In order to achieve this goal, a custom built testing chamber was designed and fabricated; the assembled testing device was shown in figure 5-12. The testing chamber was made of transparent poly-carbonate with a thickness of 5.5mm. The

114

inner dimensions of the testing chamber were 46mm in width, 46mm in length, and 60mm in height. The cover of the chamber had two holes driven through, one for gas inlet and the other one for gas outlet. In this device, the oxygen sensor chips were laid flat at the bottom, covered by either 0.01M KCl solution or particulate layer saturated with 0.01M KCl solution with controllable height/thickness. Multiple sensor chips were used to facilitate the oxygen concentration mapping in 2-D if necessary. Such a design also made easy the reproducibility test. For sensor calibration, gas with controlled oxygen concentration was bubbled into solution for 20 minutes, then amperometric i-t measurements at -0.55v were carried out. For the effective diffusion coefficient measurements, 21% oxygen was used. The typical amperometric i-t measurement results at various oxygen concentration conditions were shown in figure 5-13. The current reached near-steady-state in less than 2 minutes, as shown in figure 5-14. Typical sensor calibration curve obtained from this approach was shown in figure 5-15, which was similar to that was produced previously in beakers.

Figure 5-12 (a) Top view of assembled testing device for oxygen diffusion coefficient measurements

115

Figure 5-12 (b) Side view of assembled testing device for oxygen diffusion coefficient measurements

Figure 5-13 Amperometric i-t measurements of assembled oxygen testing device: reduction current increases with the increase in oxygen concentration

116

Figure 5-14 Rate of change in current for amperometric i-t measurement: very fast response, reaches near equilibrium in less than 100s

To calculate the effective oxygen diffusion coefficient, amperometrix i-t measurements were carried out with sensor chips placed under 3mm 0.01M KCl solution, 1mm sand particulates saturated with 0.01M KCl solution, and 10mm sand particulates saturated with 0.01M KCl solution. Typical calibration curves of the oxygen sensors under these three conditions were shown in figure 5-15. Tests results showed that the sensitivity decreased dramatically with the presence of particulates. It went down from 0.35µA for pure solution to 0.08µA for 1mm particulate layer and an even smaller 0.04µA for 10mm particulate layer.

117

0

I / uA

0

5

10

15

20

25

-4

i = -0.3514C - 0.402 2 R = 0.991

-8 O2%

Figure 5-15 (a) Calibration curve for oxygen sensor chips under 3mm 0.01M KCl solution: good linear relationship between reduction current and oxygen concentration

0

I / uA

0

5

10

15

20

25

-2 i = -0.0792C - 0.1098 R2 = 0.9915

-4

O2%

Figure 5-15 (b) Calibration curve for oxygen sensor chips under 1mm particulate in 0.01M KCl solution: good linear relationship between reduction current and oxygen concentration

118

0 0

5

10

15

20

I / uA

-0.5 i = -0.0398C - 0.9934 R2 = 0.9642

-1

-1.5

-2 O2%

Figure 5-15 (b) Calibration curve for oxygen sensor chips under 10mm particulate in 0.01M KCl solution: acceptable linear relationship between reduction current and oxygen concentration

To calculate the effective oxygen diffusion coefficient, the Cottrell Equation was applied as shown in equation 5.9:

i (t ) = nFAC0

D πt

(5.9)

where in this study n=4, F=96500 coulombs, A=0.15cm2, and C0=0.281E-6mol/cm3. According to equation 5.9, the I vs. t-1/2 plot yelled a straight line and the slope of which could be used to calculate the diffusion coefficient D in the following equation 5.10: i (t ) = at −0.5 + b

(5.10)

119

The slope a is a function of the diffusion coefficient as determined by equation 5.11 following numerical calculation utilizing the values of those constants (n, F, A, and C0):

a = 0.0918D1/ 2

(5.11)

For this purpose, the initial current response with time was used for the i-t-1/2 plots, as shown in figure 5-16 for the three conditions tested in this study.

8.00E+01 7.00E+01 6.00E+01

i = 28.597t-0.5 - 3.0234 R2 = 0.9465

I ( uA )

5.00E+01 4.00E+01 3.00E+01 2.00E+01 1.00E+01 0.00E+00 0

0.5

1

1.5

2

2.5

3

t ( s ) -1/2

Figure 5-16 (a) Cottrell plot for oxygen sensors in 3mm 0.01M KCl solution: good linear relationship between reduction current and t-0.5

120

6.00E+01

5.00E+01

I ( uA )

4.00E+01 i = 21.29t-0.5 - 3.0142 R2 = 0.9644

3.00E+01

2.00E+01

1.00E+01

0.00E+00 0

0.5

1

1.5

2

2.5

3

t ( s ) -1/2

Figure 5-16 (b) Cottrell plot for oxygen sensors in 1mm 0.01M KCl solution: good linear relationship between reduction current and t-0.5

Table 5-2 summarized the calculated effective oxygen diffusion coefficient under the three testing conditions. It showed that oxygen diffusion coefficient decreased with the presence of particulates depending on the thickness of the particulate layer. Table 5-2 Summary of effective oxygen diffusion coefficient under various conditions

Diffusion

3mm 0.01M KCl

1mm particulate layer

10mm particulate layer

9.34x10-5 cm2/s

5.06x10-5 cm2/s

3.15x10-5 cm2/s

Coefficient

121

5.5 Summary In this part of the study, thick film printed Au based dissolved oxygen sensors were fabricated and tested. The oxygen reduction currents were measured through amperometric i-t measurements at a reduction potential of -0.55v, which was determined by cyclic voltammetry experiments. The amperometric i-t measurements yielded a linear dependence of the current on the concentration of the dissolved oxygen. The slopes of the i-t plots were considered the sensitivity of the oxygen sensors developed and tested in this study. These sensors exhibited good reproducibility, fast response time, and satisfactory sensitivity. Transient amerpometric i-t studies were also carried out to determine the effective oxygen diffusion coefficient under various conditions. Testing results showed that the oxygen diffusion coefficient decreased dramatically when there was the presence of particulate layer.

122

References: [1] http://www.ocrwm.doe.gov/science/targeted_thrusts/matperf_targetedthrusts.shtml [2] L. Clark, “Monitoring and control of blood and tissue oxygen tensions” Trans. Am. Soc. Art. Internal Organs, 2 (1956) 41-49 [3] K. Mancy, D. Okun, and C. Reiley, J. Electroanal. Chem. 4 (1962) 65-92 [4] H. Ogino and K. Asakura, “Development of a highly sensitive galvanic cell oxygen sensor” Talanta 42 (2) (1995) 305-310 [5] L. Nei and R. Compton, “Novel approaches to galvanic oxygen analysis” Analytical Communications 33 (1996) 319-321 [6] R. Meruva and M. Meyerhoff, “Potentiometric oxygen sensor based on mixed potential of cobalt wire electrode” Analytica Chimica Acta 341 (1997) 187-194 [7] G. John, I. Klimant, C. Wittmann, and E. Heinzle, “Integrated optical sensing of dissolved oxygen in microtiter plates: a novel tool for microbial cultivation” Biotechnology and Bioengineering, 81 (7) (2003) 829-835 [8] W. Cao, O. Tan, W. Zhu, B. Jiang, and C. Reddy, “An amorphous-like solid solution system for low temperature resistive-type oxygen sensing” Sensors and Actuators B 77 (2001) 421-426 [9] J. Hendrikse, W. Olthuis, and P. Bergveld, “The MOSFET as an oxygen sensor: constant current potentiommetry” Sensors and Actuators B 59 (1999) 35-41 [10] L. Nei and R. Compton, “An improved Clark-type galvanic sensor for dissolved oxygen” Sensors and Actuators B 30 (1996) 83-87 [11]X. Wu, X. Wu, M. Tian, S. Zhang, and Y. Wang, “Research on n+p type oxygen sensor”, Solid State Electronics 46 (2002) 97-101

123

[12] H. Suzuki, “Advances in the microfabrication of electrochemical sensors and systems” Electroanalysis 12 (9) (2000) 703-715 [13] C. Liu, “Development of chemical sensors using microfabrication and micromachining techniques” Materials Chemistry and Physics 42 (1995) 87-90 [14] J. Currie, A. Essalik, and J. Marusic, “Micromachined thin film solid state electrochemical gas sensors” Sensors and Actuators B 59 (1999) 235-241 [15] H. Suzuki, “Microfabrication of chemical sensors and biosensors for environmental monitoring” Materials Science and Engineering C 12 (2000) 55-61 [16] A. Bard and L. Faulkner, “Electrochemical methods: fundamentals and applications” John Wiley & Sons Inc. 2nd Edition, 2001 [17] http://en.wikipedia.org/wiki/Cyclic_voltammetry [18] C. Vallieres, J. Gray, S. Poncin, and M. matlosz, “Potentiometric detection of oxygen based on the mixed potential of zinc” Electroanalysis 10 (3) (1998) 191-197 [19] W. Maskell, “Inorganic solid state chemically sensitive devices: electrochemical oxygen gas sensors” J. Phys. E. Sci. Instrum. 20 (1987) 1156-1168 [20] B. Sohn and C. Kim, “A new pH-ISFET based dissolved oxygen sensor by employing electrolysis of oxygen” Sensors and Actuators B 34 (1996) 435-550 [21] J. Hale and M. Hitchman, “Some considerations of the steady state and transient behavior of membrane covered dissolved oxygen detectors” J. Electroanal. Chem. 107 (1980) 281-294

124

Chapter 6 Calibration of the Platinum RTD and Evaluation of the High Temperature Oxygen Sensor

6.1 Introduction Substantial research and development efforts had been directed to address the environmental issues in the past two decades. This had driven the research on developing gas sensors and their applications in the emission controls for various industrial sectors. Among these gas sensors, oxygen gas sensor played a key role in many applications including combustion optimization, automobile exhaust quality control, biological and food processing, industrial boilers, and steel and cement industries Ramamoorthy et al.

[1]

. According to

[1]

, the predominant use of oxygen gas sensors had been in the

control of the air-fuel mixture in an automobile combustion engine and in the exhaust emission control systems. Furthermore, in many fields of science and technology, oxygen and the measurement of its concentration also played a key role, such as environmental control and modified atmospheres (such as greenhouses and respiratory gases)

[2]

. In

many aspects, the monitoring of gaseous oxygen would be as important as that of dissolved oxygen. The concentration/partial pressure of gaseous oxygen in an environment can be electrochemically determined using various measuring principles, depending on the conditions such as the temperature and the absolute concentrations (rich or lean). Clark introduced the amperometric principle of an oxygen sensor in 1956 [3]. The Clark oxygen sensor consisted of a sensing platinum electrode (cathode) and a reference Ag/AgCl electrode (anode) enclosed with an oxygen permeable membrane. The output of this amperometric oxygen sensor varied linearly with the oxygen concentration. Since then, 125

many oxygen sensors had been developed and reported. Potentiometric (electromotive force (EMF)) determination of the oxygen concentration was another major principle and the sensor output of this type of sensor depended logarithmically with the oxygen partial pressure

[1]

. This potentiometric measurement was based on the Nernst equation as

discussed in chapter 5. As discussed in chapter 5 previously, the principle of the amperometric oxygen sensor measured the reductive current of the oxygen at the cathode at a given potential (relative to the potential of the reference electrode). The potential at which the amperometric tests were undertaken would be based on the type of electrode material that was selectively used to construct the working electrode

[4]

. This potential could be

determined through experiments, normally through the cyclic voltammetry tests. The potentiometric oxygen sensor needed reference oxygen potential. This reference oxygen potential can be established by using a reference gas, reference materials such as metal-metal oxide mixtures, or the mixed potentials resulted from metallic corrosion and simultaneous oxygen reduction. Therefore the potentiometric measurement exhibited a Nernstain behavior as described in chapter 4. For any application involved of the high temperature measurements of oxygen concentration, ceramic based oxygen gas sensors were the practical choice

[1]

. Yttria-

stabilized-zirconia (YSZ) was one of the most widely used ceramic materials for high temperature gaseous oxygen sensor application, owning to its unique oxygen ionic conductivity at elevated temperatures. With appropriate modification, oxygen gas sensors based on YSZ could function on either an amperometric mode or a potentiometric mode

126

[2, 5, and 11]

. The potentiometric oxygen gas sensors were ideal for the air-to-fuel ratio

monitoring close to the combustion stoichiometry, which was in the lean oxygen region [5]

. On the other hand, when the environment was in the rich oxygen region such as the

operation of gasoline piston engines where excess combustion air was desired, potentiometric sensor was insufficient to monitor the oxygen concentration change due to its logarithmic dependence nature. Thus a linear dependence of the amperometric sensor would become a better choice under this situation. Therefore, depending on the concentration of dissolved oxygen in the test medium, either the amperometric method or the potentiometric method could be selected and assessed. As stated before, in amperometric method, the oxygen concentration is represented by the reductive current and it is in a linear relationship with the current. Thus for the rich (high concentration) oxygen measurement, amperometric approach is better suited due to its linear sensitivity. However, for lean (low concentration) oxygen measurement, the logarithmic sensitivity of potentiometric method becomes an advantage over the linear counterpart

[6]

. In this

study, our sensor chip with two separated sensing elements were designed and constructed side by side with a shared heater made of platinum. One sensing element was intended to function at the amperometric mode, whereas the other was intended to function at the potentiometric mode. This design would ideally enable this sensor to provide sensitive oxygen concentration measurement in both lean and rich oxygen regions. An additional function of this sensor structure was that the electrochemical pumping of oxygen using the amperometric element could be used to adjust the oxygen concentration in a confined space to a desired level monitored by the potentiometric element.

127

When amperometric mode of operation was utilized, the oxygen concentration would likely be measured by the limiting current determined by the diffusion rate of the oxygen molecules through a selected ceramic layer. This ceramic was commonly defined as the diffusion barrier. It could be either a porous ceramic layer (such as Al2O3 or SiO2), or a dense ceramic layer that could selectively transport oxygen at elevated temperatures (such as YSZ or a perovskite material)

[5 & 7]

. Perovskite materials possessed unique

properties that would be discussed later in this chapter. The amperometric oxygen gas sensors that employed a dense diffusion limiting barrier possessed an additional benefit as the sensor utilized the unique ability of the barrier to selectively transport oxygen, thereby minimized the influence of other reactive gaseous species. When potentiometric mode was applied for the measurement of oxygen partial pressure, it often would require an oxygen concentration reference due to its Nernst behavior nature as discussed in previous chapters. This reference could be either an external gas with constant oxygen concentration, or an internal reference consisted of a mixture of selected metal and its oxide. Although an external gas flow to function as a reference for the potentiometric oxygen gas sensors could be easily established for the laboratory-level studies, it would become very difficult or even impractical to achieve when high temperature applications such as air-to-fuel control were concerned [8-10]. Thus an internal solid state reference was preferred when the miniaturization of the oxygen sensor was concerned. Therefore, a solid-state internal reference was employed in this study. Nickel-nickel oxide mixture was selected in this study because it was stable [5] and it was easy to be constructed using microfabrication process such as the sputtering metallization technique.

128

Solid state ceramic based gaseous oxygen sensors for automobile application for example, needed to function at an elevated temperature. Therefore, either the temperature of the environment needed to be tailored towards the functional temperature of the ceramic material that was used to construct the oxygen sensor, or there would have a need to include a built-in heater raising the temperature of the sensing elements and a built-in temperature detector to monitor the temperature. When the fabrication of a microsensor was involved, the latter choice was preferred. Different microfabrication techniques could be used to fabricate the temperature sensing-elements, such as thermocouple, thermistor, transistor, and resistance temperature detector (RTD). Table 61 summarizes the important features of these temperature-sensing elements [12]. However, the selected technique must be able to provide a continuous electrical signal and would be easy to be integrated into a sensor array for this study. In this study, a built-in heater and a built-in RTD temperature detector made of sputtered platinum thin film structure were employed. The principle of this temperature-sensing element was based on the property of metal to change its electrical resistance with the change of temperature [13]. A detailed discussion of this technique was presented in the following section. In this section, the properties of the perovskite materials were also introduced, as well as a brief discussion of the principles of the amperometric and the potentiometric measurements employed for oxygen concentration/partial pressure sensing. In this study, thin film metallization microfabrication technique was employed for the fabrication of the high temperature oxygen gas sensors. The advantages of the microfabrication of miniaturized sensors were discussed in chapter 5, please refer to the previous sections when needed.

129

Table 6-1 Important features of various types of temperature-sensing elements Thermocouples

Thermisors

Transistors

Pt RTDs

Temp. range (°C)

-270 to +3500

-80 to +180

-50 to +180

-260 to +1000

Accuracy of

Problematic

High

Medium

High

High

Medium

Medium

Medium

Integration

Not in standard

Not in standard

Yes

Yes

feasibility

process

process

Sensitivity

Low

High

High

Low

Linearity

Good

Non-linear

Good

Good

Electric signal type

Voltage

Resistance

Voltage

Resistance

absolute temp. measurements Accuracy of small temp. differences

6.2 Fundamentals 6.2.1 Resistance temperature detector (RTD) Metals exhibit a change in their electrical resistances when subjected to a change in temperature

[13]

. The resistance of many metals (such as iron, copper, and aluminum)

increases with the increase of temperature over a wide temperature range at a small rate (0.3%/°C)

[14]

. Appropriate patterning of such a metal to form a thin film can yield a

significant change in resistance, and the resulted device has been known as a resistance

130

temperature detector RTD. A RTD offers better accuracy and better long-term stability than a thermocouple by comparison. According to physics, the resistance of a resistor, R, is directly proportional to its length L, and inversely proportional to its cross-sectional area A, as determined by equation 6.1:

R=ρ

L A

(6.1)

where ρ is the resistivity of the material of which the resistor is made. The inverse of the electric conductivity of the material, σ, can be expressed by equation 6.2:

ρ=

1

(6.2)

σ

The electric conductivity σ can be described by equation 6.3, assuming that the simple electron-gas model of Drude can be applied:

σ = ne2

τ

(6.3)

m

where n is the number of electrons per unit volume, e is the electron charge, τ is the relaxation time, and m is the electron mass. In equation 6.3 only τ is a function of temperature. The ability of a metal to conduct electricity cn be described physically as the ability of the free electrons traveling through the periodic lattice of the metal. Therefore, the resistivity can be induced by the imperfections of the lattice

[15]

. Foreign atoms and

defects such as vacancies and grain boundaries can induce additional scattering of the 131

electrons, and hence, increase in the resistivity. Emission or absorption of a quantum lattice vibrational energy, a phonon, can also increase electron scattering

[15]

. Assuming

the relaxation times of each scattering mechanisms are independent of each other, the overall relaxation time can thereby be described in equation 6.4 and the total resistivity can be simply deemed as the sum of each individual resistivity according to the Matthiessen’s rule

[16]

. Only the electron-photon scattering mechanism depends on the

temperature as expressed in quation 6.4. 1

τ

=

1

+

1

τ1 τ 2

+ ...

(6.4)

The resistance of a metal, within a limited temperature range, increases linearly with the temperature and is governed by equation 6.5:

ρt = a(t − t0 ) + ρ0

(6.5)

where a is the temperature coefficient, typically a positive constant for a specific metal and or an alloy. Although any metal can be theoretically functioned as the material for a RTD, there are strict limitations on the selection of a metal. For example, the metal selected needs to be corrosion resistant and it will have a high melting temperature. In order to ensure a good linearity of temperature dependence, high purity of the metal is preferred. Platinum is one the metals that can satisfy these requirements and therefore it has been selected as the metal of choice for RTD in this study. The temperature dependence of a platinum sensing element could be described by the Callendar-Van Dusen equation:

132

Rt = R0 + R0 * α [t − δ (0.01t − 1) *0.01t − β (0.01t − 1) *(0.01t )3 ]

(6.6)

where α, β, and δ are constants usually determined at selected temperatures (100°C, 182.96°C, and 444.7°C, respectively). For a winding of a pure, strain free, annealed platinum, typical values of these constants are α=0.003925, β=0.11 for negative t and β=0 for positive t, and δ=1.49. Equation 6.6 can be further simplified to equation 6.7 if the temperature is between 0 and 100°C: Rt = aR0 * t + R0

(6.7)

Platinum RTDs can be fabricated either as a wire wound or as a thin film structure. The thin film element can be manufactured by depositing a thin layer of platinum in a specifically designed pattern. One of the advantages of the thin film type structure is that it will allow for a greater resistance, consequently an increased resolution. 6.2.2 Principle of the limiting current amperometric measurements Amperometric oxygen gas sensors are based on the measurements of oxygen reduction currents at the working electrode at a pre-determined reductive potential. With an addition of diffusion barrier, the current measurement can be carried out at the limiting current region, where the diffusion of oxygen is the rate-limiting electrochemical process. In this case, the current is determined by the concentration gradient across the double layer on the surface of the working electrode. This diffusion-limited current is determined by a combination of Fick’s and Faraday’s Laws as shown in eqaution 6.8 [17]:

iL = −nAFDO

CO*

(6.8)

δ

133

where iL is the limiting current, n is the number of electrons transferred, A is the surface area of the working electrode, F is the Faraday constant, DO is the diffusion coefficient, CO* is the bulk concentration, and δ is the thickness of the diffusion barrier layer. In

accordance to the standard electrochemical conventions, the reductive cathodic current is taken to be negative. 6.2.3 Principle of the potentiometric measurements When a potentiometric measurement is carried out, the electromotive force (EMF) can be used to correlate to the analyte concentration, in this case, the oxygen concentration. An electrochemical cell as shown in figure 6-1 can be used as an example to describe the potentiometric measurement:

POII2

Oxygen ionic conducting electrolyte

POI2

Figure 6-1 Schematic of an electrochemical cell

In our study, the EMF induced by the oxygen concentration gradient across the oxygen ionic conducting electrolyte at steady state can be represented by the Nernstian equation 6.9 [18] similar to that of pH sensors: POII2 RT EMF ∝ ln( I ) nF PO2

(6.9)

134

where POI2 and POII2 are the oxygen partial pressures on the two sides of the oxygen ionic conducting electrolyte respectively, R is the gas constant, T is the temperature, n is the number of electrons transferred, and F is the Faraday constant. When an internal solid state metal-metal oxide reference was used, equation 6.9 can be further simplified to demonstrate that a typical logarithmic relationship between the EMF and the oxygen partial pressure to be determined can be expressed in equation 6.10: EMF = a ln( POII2 ) + b

(6.10)

Equation 6.10 shows that the EMF can be expressed as in linear dependence on the logarithmic of the oxygen potential/partial pressure. 6.2.4 Mechanism of oxygen transport While yttria-stabilized-zerconia YSZ is a pure oxygen ionic conductor, other materials such as perovskite materials are mixed conductors, meaning they can conduct both oxygen ions and electrons. A mixed conductor nature can be attractive for selected perovskite materials to various applications such as ferroelectrics, catalysts, sensors and superconductors, solid oxide fuel cells, and oxygen separation membranes [19]. Perovskite is a relatively rare mineral on the Earth's crust discovered in the Ural mountains of Russia by Gustav Rose in 1839 and named for Russian mineralogist, L. A. Perovski

[20]

. Perovskite is also the collective name of a more general group of crystals

which take the same crystal structure. The basic chemical formula follows the pattern ABO3 stoichiometry, where A and B are cations of different sizes [19, 20]. Typically, the A-

135

site cation is large (such as a rare earth element) with a 12-coordinate and the B-site is smaller (frequently a transition metal element) with a 6-coordinate. The general crystal structure of perovskite is a primitive cube, with the A-cation in the middle of the cube, the B-cation in the corner and the anion, commonly oxygen, in the centre of the face edges. The structure is stabilized by the 6 coordination of the Bcation (octahedron) and 12 of the A cation. The packing of the ions can be thought of the A and O ions together forming a cubic close packed array, where the B ions occupy a quarter of the octahedral holes [20]. Although the primitive cube is the idealized structure, differences in radius between the A and B cations can alter the structure to a number of different distortions, of which tilting is the most common one. Complex perovskite structures contain two different A-site cations and/or two different B-site cations. The ordered and disordered variants can thereby be generated. A schematic of the ABO3 perovskite structure is shown in figure 6-2 [20]:

Figure 6-2 Schematic of perovskite structure. The red spheres are oxygen atoms, the deep blue are smaller metal cations and the green/blue are the larger metal cations [20]

136

The oxygen ionic conductivity of a perovskite material is induced when the material loses its lattice oxygen forming oxygen vacancies. This also applies to YSZ. The oxygen vacancy generation process typically occurs at an elevated temperature depending on the specific A-site and B-site cations

[20, 21]

. An appropriate partial substitution of the

A-site and/or B-site cations with divalent metal cations and/or trivalent metal cations can promote the generation of the oxygen vacancies, and hence, the increase in the oxygen ionic conductivity

[21]

. The general reaction at the surface of the mixed conducting

perovskite materials can be rerpresented by equation 6.11 [1]: 1 O2 + VO ⇔ OO2− + 2h + 2

(6.11)

where VO is an oxygen vacancy, OO2− is an oxygen ion, and h is an electron hole. The oxygen ion will then diffuse through the diffusion barrier to the other side of the perovskite layer driven by the externally added voltage gradient, where the opposite reaction as shown by equation 6.12 takes place [7]: 1 OO2− ⇔ VO + O2 + 2e− 2

(6.12)

For amperometric measurements, the diffusion rate of the oxygen through the perovskite diffusion barrier will determine the value of the steady state diffusion-limited oxygen reduction current. Therefore, a linear correlation between the current and the gaseous oxygen concentration can be established, according to equation 6.8. In this study, perovskite material La0.8Sr0.2MnO3-δ was selected to be the diffusion barrier material

[5]

.

Meanwhile, Nernst equation can also be applicable even with the dense perovskite diffusion barrier as long as there is no externally added potential gradient. Therefore, the 137

amperometric sensing element in this study can in principle, operate at potentiometric mode as well. This potentiometric mode of operation of the oxygen sensor has been conducted in this study and the results will be presented and discussed in the following sections of this chapter. 6.3 Experimental The RTD elements of the high temperature oxygen gas sensor are shown in figure 6-3 as those designs surrounding the serpentine-style heaters in the middle. These RTD elements were thin layers of platinum covered by an aluminum oxide insulation layer. The widths of these RTD were 0.2mm and 0.05mm for the large oxygen sensor design (referred to as big sensor hereafter) and the small design (referred to as small sensor hereafter), respectively. The thickness of the RTD element was 3000Å platinum thin film. The RTD element on the sensor chip was first bonded with platinum wires. Then the sensor chip was placed in a VULCAN 3-1750 thermal furnace manufactured by NEY Global (Yucaipa, CA). The resistances of the RTD element at selected temperatures were measured with an OMEGA 9303 digital multimeter. During the RTD element calibration process, it was discovered that once the furnace temperature exceeded 550°C, the LSM diffusion barrier might break in some instances. Therefore, the subsequent oxygen concentration tests were carried out at temperatures less than 550°C. After the RTD element was calibrated, the sensor chip was bonded and packed as discussed in chapter 2. The contact pads of the heater were then connected to a HP E3611A DC power supply. At the same time, the RTD element was connected to a digital multimeter and the resistance at each heating voltage was recorded. The heating voltages

138

across the heater were correlated with the temperatures according to the results from the RTD calibration process. When a stable temperature had been reached, gas with controlled oxygen concentration flew through the opening surface of the sensor and electrochemical measurements (cyclic voltammetry measurements, amperometric i-t measurements, and open circuit potential measurements) were then carried out. The assembly of the gas flow system was identical to that described in chapter 5. The big sensor each had a sensing area of 0.132cm2, while the small sensor each had a sensing area of 0.015cm2. The test results of the two-electrode-configuration cyclic voltammetry measurements were inconclusive and unstable with large hysterisis and the test results of the amperometric i-t measurements were noisy and unreliable. Therefore, only the open circuit potential measurements were taken and reported in this study. The cause of the failure of the cyclic voltammetry and amperometric i-t measurements was not clear and due to the time limitation, detailed analysis was not performed in this study. It might be attributed to the possible processing defects of any step of this eight-layer thin film process as described in detail in chapter 2. Due to the thin film nature of these layers (each layer had a thickness of only 3000Å), any defect in any layer could have affected the overall performance of the sensor, especially when the current measurement was involved. On the other hand, the open circuit potential/electromotive force measurements were probably less sensitive to the possible existence of defects and therefore could yield useful results.

139

Figure 6-3 Schematic of RTD elements of the oxygen gaseous sensors

6.4 Results and Discussion Figure 6-4 and figure 6-5 show the resistance of the RTD element of the big sensors and the small sensors, respectively. These figures indicate that the resistance of the RTD element yielded a good linear relationship with the temperature. Comparing these results with equation 6.7, the calculated values of R0 and α for the big sensors were 51.85Ω and 0.142%, respectively, and 374.52Ω and 0.136% for the small sensors, respectively. The two α values are comparable to each other, while the R0 values follow the physical principle of the resistance of resistors. These results indicate that the RTD elements employed in this study were well-defined.

140

In order to measure the resistance of a platinum RTD element with a digital multimeter, it is unavoidable to pass a current through the RTD. Such a process resulted in an additional current and generated additional heat in the RTD element. This was often termed “Self-heating effect”, or the “Joule heating effect”. Therefore the temperature indicated by the sensor would be slightly higher than the actual temperature. The usage of a small measuring current could minimize such effects. This phenomenon, however, was not a focus of this study and was therefore not further explored. Our major assessment in this phase of investigation focused on the relationship of the RTD element resistance and the temperature of the sensor operation.

100 90 80 resistance (ohm

70 60 50 40

R = 0.0736T + 51.847 R2 = 0.9995

30 20 10 0 0

100

200

300 temp. (C)

400

500

600

Figure 6-4 RTD element resistance calibration curve for big sensors: resistance is in linear relationship with temperature

141

700 600

Resistance (ohm)

500 400

R = 0.5094T + 374.52 2 R = 0.9988

300 200 100 0 0

100

200

300

400

500

600

Temperature (C)

Figure 6-5 RTD element resistance vs. temperature calibration curve for small sensors: resistance is in linear relationship with temperature

The relationship between the voltage across the heater and the resulted RTD resistance are shown in figure 6-6 and figure 6-7 for the big sensors and the small sensors, respectively. Table 6-2 summarizes selected heater voltage and temperature results.

142

120

detector resistance (oh

100 80 60 40 20 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 heating voltage (v)

Figure 6-6 Heater voltage and corresponding RTD response for the big sensors: linear dependence

750

resistance (ohm)

700

650

600

550

500 0

1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 21

heater heating voltage (v)

Figure 6-7 Heater voltage and corresponding RTD response for the small sensors: linear dependence

143

Table 6-2 Summary of heater voltage and corresponding temperature Big sensors Heater

Small sensors

0

0.8

1.0

1.2

1.5

0

5

8

10

12

15

56.70

76.83

82.56

88.30

96.89

571.0

596.4

613.4

625.4

638.4

658.8

20

330

410

480

560

20

440

460

490

520

570

voltage (v) RTD resistance (Ω) Temperature (°C)

Figure 6-6 and figure 6-7 show a good linear dependence of the resistance of RTD elements on the heating voltage imposed on the heating elements for both the big sensors and the small sensors. Meanwhile, the linear relationship between the resistance of the RTD elements and the temperature has been established by figure 6-4 and figure 6-5. Therefore, the heating voltage can be used to directly indicate temperature as summarized in table 6-2. This can greatly simplify the data recording process. A typical open circuit potential measurement result (taken in air) is shown in figure 6-8. Both the amperometric sensing element and the potentiometric sensing element exhibited the similar open circuit potential (OCP) response for the big sensor and small sensor. A first derivative of the open circuit potential measurement is shown in figure 6-9. Unlike the fast response of the thick film dissolved oxygen sensor,

144

unfortunately, figure 6-9 indicates that the kinetics of the open circuit potential measurement of the thin film oxygen gas sensor was rather slow, it took about 10 minutes for the open circuit potential to reach a quasi-equilibrium value.

Increasing the

temperature when it was in a relatively low temperature range (20°C to 300°C) improved the sensor response. When it was in a high temperature range (400°C to 550°C) an increase in temperature did not noticeably improve the sense response. This phenomenon can be better illustrated in figure 6-10.

Figure 6-8 Typical OCP measurement result: OCP decreases with time

145

Figure 6-9 High temperature oxygen sensor OCP measurement: slow kinetics

146

Figure 6-10 Temperature effect on sensor response: (a) 0.2v heating; (b) 0.4v heating; (c) 0.8v heating; (d) 1.5v heating: OCP decay kinetics increases with the increase in heater voltage (temperature)

Figure 6-10 shows that the rate of change (represented by the first derivative) in open circuit potential increased with the increase in temperature (represented by the heating voltage). When the temperature is over 400°C, the open circuit potential reaches quasi-equilibrium in about 10 minutes. Further increase in temperature does not noticeably improve the sensor performance. On the contrary, when the temperature is

147

below 300°C, it will take at least 15 minutes for the open circuit potential to reach quasiequilibrium. Figure 6-11 demonstrates the reproducibility of the oxygen sensors developed in this study. The open circuit potential measurements at a given oxygen concentration showed acceptable reproducibility for both the amperometric element and the potentiometric element of the big sensor and the small sensor. The variation in the equilibrium open circuit potential measurements is typically within a few mv (about 0.3% of the equilibrium value) for a given testing condition. The temperature effects on the OCP measurements in air of the amperometric elements of big sensors are illustrated in figure 6-12 (a). It shows that the absolute value of the equilibrium OCP increases with temperature. This appears to be more obvious for the small sensors as shown in figure 6-12 (b).

Figure 6-11 (a) Big sensor amperometric elements OCP at 330°C: good reproducibility of sensor measurements

148

Figure 6-11 (b) Big sensor amperometric elements OCP at 410°C: good reproducibility

Figure 6-11 (c) Small sensor potentiometric elements OCP at 520°C: good reproducibility

149

Figure 6-12 (a) Temperature effects on OCP measurements of big sensor amperometric unit in air: 330°C (red), 410°C (blue), 480°C (green), and 560°C (grey): OCP becomes more negative as temperature increases

Figure 6-12 (b) Temperature effects on OCP measurements of small sensor amperometric elements in air: 570°C (red), 520°C (blue), and 490°C (grey): OCP becomes more negative as temperature increases

150

Figure 6-11 and figure 6-12 show that the open circuit potential measurements demonstrate acceptable reproducibility and agree well with equation 6.9. The open circuit potential becomes more negative as temperature increases. By increasing the temperature, the sensitivity of the oxygen sensor measurements (the slope of the OCP vs. ln(PO2) plot) will be increased. Typical OCP measurements for the big sensor and the small sensor (on both amperometric element and potentiometric element) were carried out at various temperature conditions. The results are shown in figure 6-13 and the results indicated that the equilibrium open circuit potential became more negative as the gaseous oxygen concentration increased. This was in good agreement with the Nernstian equation, previously shown as equation 6.9.

Figure 6-13 (a) Typical OCP measurements for the amperometric elements of the big sensors at 480°C: OCP becomes more negative as oxygen concentration increases (red: 21%, purple: 0%)

151

Figure 6-13 (b) Typical OCP measurements for the amperometric elements of the big sensors at 560°C: OCP becomes more negative as oxygen concentration increases (red: 21%, purple: 0%)

Figure 6-13 (c) Typical OCP measurements for the potentiometric elements of the big sensors at 480°C: OCP becomes more negative as oxygen concentration increases (red: 21%, green: 0%)

152

Figure 6-13 (d) Typical OCP measurements for the amperometric elements of the small sensors at 490°C: OCP becomes more negative as oxygen concentration increases (purple: 21%, red: 0%)

Figure 6-13 (e) Typical OCP measurements for the amperometric elements of the small sensors at 520°C: OCP becomes more negative as oxygen concentration increases (red: 21%, green: 0%)

153

Figure 6-13 (f) Typical OCP measurements for the potentiometric elements of the small sensors at 570°C: OCP becomes more negative as oxygen concentration increases (red: 21%, purple: 0%)

Figure 6-13 shows that the open circuit potential becomes more negative as time increases. It also demonstrates that the open circuit potential becomes more negative as the oxygen concentration increases, which is in good agreement with equation 6.9. A long waiting time will be needed for the open circuit potential to reach an equilibrium value as the kinetics of the OCP measurements is relatively slow. However, after 15 minutes the differences among the open circuit potentials corresponding to different oxygen concentrations remain stable. Therefore, in order to make practical use of these oxygen sensors, the OCP measurements can be recorded at a fixed time (15minutes). The values of the OCP at this fixed time can be used to calibrate the oxygen sensors. The equilibrium open circuit potential values of the OCP measurements were used to establish the calibration curves for both the amperometric and the potentiometric

154

measurements of the big sensor and the small sensor, respectively. Using this approach, the sensor calibration curves were obtained and are shown in figure 6-14. The open circuit potential of 100% N2 measurements were used as the reference points to calculate the formalized open circuit potential values in figure 6-14. Figure 6-14 demonstrates good linear dependence of the formalized open circuit potential on the logarithmic oxygen concentration. The linearity of the calibration curve increases as temperature increases for both amperometric and potentiometric units. Potentiometric units demonstrated better open circuit potential measurement performance compared with the amperometric ones. Also figure 6-14 shows that the performance of the big sensor was relatively better than the small sensor. Overall the potentiometric unit of the big sensor demonstrated best performance in open circuit potential measurements at elevated temperature (around 500°C).

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0 -0.02

V = -0.1434*log(C) - 0.2166 R2 = 0.9374

-0.04

OCP

-0.06

-0.08

-0.1

-0.12

-0.14

-0.16

log(oxygen concentration)

Figure 6-14 (a) calibration curve for amperometric element of the big sensor 480°C

155

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

-0.05

OCP (v

-0.1

-0.15

-0.2

V= -0.2394*log(C) - 0.3346 2

R = 0.964 -0.25

log(oxygen concentration)

Figure 6-14 (b) Calibration curve for amperometric element of the big sensor 560°C

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0 -0.02

V = -0.1631*log(C) - 0.2453 2 R = 0.9914

-0.04

OCP

-0.06

-0.08

-0.1

-0.12

-0.14

-0.16

log(oxygen concentration)

Figure 6-14 (c) Calibration curve for potentiometric element of the big sensor 480°C

156

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0 -0.1 -0.2 -0.3

OCP

V = -0.3775*log(C) - 0.9691 2 R = 0.9914

-0.4 -0.5 -0.6 -0.7 -0.8

log(oxygen concentration)

Figure 6-14 (d) Calibration curve for potentiometric element of the big sensor 560°C

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0 -0.05

OCP

-0.1

-0.15

-0.2

V = -0.257*log(C) - 0.3647 R2 = 0.9128

-0.25

log(oxygen concentration)

Figure 6-14 (e) Calibration curve for amperometric element of the small sensor 490°C

157

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0 -0.02 -0.04

V = -0.1549*log(C) - 0.2332 R2 = 0.9816

OCP

-0.06 -0.08 -0.1 -0.12 -0.14 -0.16

log(oxygen concentration)

Figure 6-14 (f) Calibration curve for amperometric element of the small sensor 520°C

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

-0.05

-0.1

OCP

V = -0.1776*log(C) - 0.2975 R2 = 0.9867

-0.15

-0.2

-0.25

log(oxygen concentration)

Figure 6-14 (g) Calibration curve for potentiometric element of the small sensor 490°C

158

0 -1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0 -0.05

V = -0.334*log(C) - 0.5081

-0.1

2

R = 0.9776

OCP

-0.15

-0.2

-0.25

-0.3

-0.35

log(oxygen concentration)

Figure 6-14 (h) Calibration curve for potentiometric element of the small sensor 520°C

159

6.5 Summary In this part of the study, solid state high temperature gaseous oxygen sensor based on YSZ electrolyte were fabricated and tested. One big sensor and one small sensor with different sensing areas were designed, fabricated, and characterized (The big sensor each had sensing area of 0.132cm2, while the small sensor each had sensing area of 0.015cm2). Each sensor had one amperometric unit and one potentiometric unit with a shared platinum heater and two platinum RTD. The calculated values of R0 and α for the RTD of the big sensor were 51.85Ω and 0.142%, respectively, and 374.52Ω and 0.136% for the small sensor, indicating that the thin platinum RTD used in this study were well defined. A good linear dependence of the resistance of the RTD with the temperature had been established, indicating that the RTDs were capable of sensing temperature. The amperometric measurements were not conclusive, probably due to some defects of the layered structure. The potentiometric measurements on both the amperometric unit and the potentiometric unit yielded reasonably good correlation between the OCP and the gaseous oxygen concentration. Experimental results indicated that the potentiometric unit of the big sensor exhibited the best Nernstian response on the oxygen concentration at around 500°C.

160

References: [1] R. Ramamoorthy, P. Dutta, and S. Akbar, “Oxygen sensors: materials, methods, designs and applications”, J. of Materials Science 38 (2003) 4271-4282 [2] J. Ramirex-Salgado and P. Fabry, “Feasibility of potentiometric oxygen gas sensor based on perovskite and sodium titanate measuring electrode”, Sensors and Actuators B 82 (2002) 34-39 [3] L. Clark, “Monitoring and control of blood and tissue oxygen tensions” Trans. Am. Soc. Art. Internal Organs 2 (1956) 41-49 [4] L. Nei and R. Compton, “An improved Clark-type galvanic sensor for dissolved oxygen” Sensors and Actuators B 30 (1996) 83-87 [5] F. Garzon, I. Raistrick, E. Brosha, R. Houlton, and B. Chung, “Dense diffusion barrier limiting current oxygen sensors” Sensors and Actuators B 50 (1998) 125-130 [6] J. Hendrikse, W. Olthuis, and P. Bergveld, “The MOSFET as an oxygen sensor: constant current potentiommetry” Sensors and Actuators B 59 (1999) 35-41 [7] J. Lee, H. Kim, and B. Kim, “Oxygen sensing characteristics of limiting current-type sensors with microstructural and structural variations in diffusion barrier” Materials Letters 26 (1996) 27-33 [8] B. Elyassi, N. Rajabbeigi, A. Khodadadi, S. Mohajerzadeh, and M. Sahimi, “An yttria-doped ceriabased oxygen sensor with solid state reference”, Sensors and Actuators B 103 (2004) 178-183 [9] H. Kaneko, T. Okamura, H. Taimatsu, Y. Matsuki, and H. Nishida, “Performance of a miniature zirconia oxygen sensor with a Pd-PdO internal reference” Sensors and Actuators B 108 (2005) 331-334 [10] N. Rajabbbeigi, B. Elyassi, S. Mohajerzadeh, and M. Sahimi, “A novel miniaturized oxygen sensor with solid-state ceria-zirconia reference” Sensors and Actuators B 100 (2004) 139-142 [11] S. Yu, Q. Wu, and C. Liu, “Development of a silicon-based yttria-stabilized-zirconia (YSZ) amperometric oxygen sensor” Sensors and Actuators B 85 (2002) 212-218

161

[12] G. Meijer, “Thermal sensors”, Institute of Physics Publishing 1994 [13] D. Garvey, “So, what is a RTD?” [14] http://www.sensorsmag.com/articles/0900/17/main.shtml [15] W. Gopel, “Sensors”, Vol. 4 Thermal sensors, VCH, 1990 [16] J. Li, “Development of a microfabricated sensor array for oil evaluation” Thesis dissertation 2005 [17] C. Vallieres, J. Gray, S. Poncin, and M. Matlosz, “Potentiometric detection of oxygen based on the mixed potential of zinc” Electroanalysis 10 (3) (1998) 191-197 [18] W. Maskell, “Inorganic solid state chemically sensitive devices: electrochemical oxygen gas sensors” J. Phys. E. Sci. Instrum. 20 (1987) 1156-1168 [19] S. Skinner and J. Kilner, “Oxygen ion conductors”, Materialstoday, Mar. 2003, 30-37 [20] http://en.wikipedia.org/wiki/Perovskite [21] V. Kharton, A. Viskup, E. Naumovich, and V. Tikhonovich, “Oxygen permeability of LaFe1-xNixO3-δ Solid solutions” Materials Research Bulletin, Vol. 34 (8) (1999) 1311-1317 [22] http://www.answers.com/topic/matthiessen-apos-s-rule-solid-state-physics?cat=technology

162

Chapter 7 Summary of This Study and Recommendations for Future Work

7.1 Summary of This Study Microfabriated electrochemical sensors including a thin film conductivity sensor, a thick film palladium based solution pH sensor, and a gold based thick film solution oxygen sensor were designed and fabricated at the Electronics Design Center of Case Western Reserve University and used for corrosion evaluation. A thin film solid state YSZ based high-temperature gaseous oxygen sensor with one amperometric sensing element and one potentiometric sensing element was also fabricated and characterized in this study. Electrochemical tests including AC impedance, cyclic voltammetry, amperometric i-t, and open circuit potential were carried out for solution conductivity, solution pH value, solution oxygen concentration, oxygen diffusion coefficient, and gaseous oxygen concentration measurements. Environmental factors could influence the progress of corrosion process, and these factors included the hydrogen-ion concentration (pH) in the solution and the dissolved oxygen concentration. In addition, the conductivity of the solution played a key role in determining the rate of progress of the corrosion process. In order to be able to evaluate the long-term progress of a corrosion process, it was essential to be able to monitor these environmental factors. Corrosion would be a primary determinant of waste package performance at the proposed Yucca Mountain repository and would control the delay time for radionuclide transport from the waste package. It is the most likely degradation process that will determine when packages will be penetrated and also the shape, size, and distribution of those penetrations. The sensor development and findings reported here are important complements to a multi-investigator program that strives for

163

increased scientific understanding, enhanced process models, and advanced technologies for long-term corrosion performance. In this study, microfabricated electrochemical sensors for the solution conductivity measurement, the dissolved oxygen concentration monitoring, and the pH value evaluation were developed and characterized. These sensors would be potentially adapted to meet the needs of measuring environmental parameters for corrosion monitoring applicable to the long-term stability modeling of the Yucca Mountain Repository project. The sensors could be used individually and the sensors were capable of being used in a multi-sensor array. Individual sensors for the measurements of solution conductivity, pH, and oxygen concentrations were developed and demonstrated for use with simulated or real corrosion cells in this study. The operational principles of these micro sensors were electrochemical and the processes of the fabrication of these micro sensors employed silicon based microfabrication techniques. This included both thin and thick film metallization processes. Photolithographic reduction, photoresist patterning and chemical etching as well as screen printing techniques were used. In this research, silicon-based solid state resistivity sensors were designed, constructed, and tested. The design and usage of such conductivity sensors provided a simple yet effective means of measuring resistance of solution or particulate layer saturated with solution. The electrode material was platinum and the process used for sensor fabrication was thin-film metallization technique. Experimental results indicated

164

that AC impedance technique was an effective method to measure the particulate layer/solution conductivity over a wide thickness and electrolyte concentration range. The measurement of the solution pH value was conducted with a three-electrode configuration (two Pd/PdO electrodes and one Ag/AgCl electrode) pH sensor using open circuit potential decay method. The fabrication of this sensor employed the thick film printing metallization. In order to measure the solution pH value, in solution pre-cleaning of the working electrode surface was first performed using cyclic voltammetry technique, followed by open circuit potential (OCP) measurements. Good reproducible linear Nernstian dependence of the OCP on the solution pH value was established with close to theoretical sensitivity. Test results indicated that the microfabricated pH sensor was capable of reliable pH monitoring over a wide pH value range (alkaline region). A solid state solution oxygen sensor was also fabricated on alumina substrate. The fabrication of this sensor was carried out using thick film printing metallization process. This was a three-electrode configuration sensor with a resistance temperature detector (RTD) on the side. The working and counter electrodes were made of gold while the reference electrode was a standard thick film silver film which was subsequently electrochemically chloridized after the sensor had been fabricated. Steady state oxygen reduction currents at -0.55v (determined by cyclic voltammetry measurements) were used to correlate with the solution oxygen concentration and to establish sensor calibration curve. Test results exhibited a good linear dependence of the oxygen reduction current on the oxygen concentration. Transient current time response of the sensor to the change in oxygen concentration was utilized to extract the effective oxygen diffusion coefficient.

165

The second part (not related to corrosion) of this dissertation involved the design, the fabrication, and the testing of a high temperature solid state gaseous oxygen sensor. Two major types of oxygen sensing principles were incorporated: amperometric and potentiometric. In this dissertation, thin film microfabrication of a solid state oxygen sensor was discussed based on the silicon-based MEMS technology. Each sensor contained two separate units, one intended to operate in an amperometric mode while the other in a potentiometric mode. Through such design, the gaseous oxygen sensors developed in this study were capable of measuring both oxygen lean (ppm) environment (using the potentiometric element) and oxygen rich (percent) environment (using the amperometric element). Two different sizes of sensing area were designed as well to explore the influence of area on the sensitivity of these oxygen sensors. Experimental results indicated that the amperometric measurements of the amperometric element of the sensor were inconclusive, while the potentiometric measurements suggested that the potentiometric element of the big oxygen sensor exhibited a good Nernstian dependence of the OCP on the concentration of the gaseous oxygen. 7.2 Recommended Future Work

The three microfabricated electrochemical sensors for the solution conductivity measurement, the pH value determination, and the dissolved oxygen concentration monitoring were fabricated and characterized individually in this study. In order for these sensors to be effective and efficient for the corrosion process evaluation application, future work would be recommended towards the integration of the three individual sensing units into one single sensor array and the careful determination of the possible interference among sensor measurement operations.

166

Despite of the fact that the amperometric element of the high temperature solid state gaseous oxygen sensor did not perform well in this study, such a concept of integrating the amperometric and the potentiometric oxygen sensing units into one single sensor was still highly recommended. This concept could enable the use of one single sensor to accurately measure both the lean oxygen and the rich oxygen environment. The future work to improve this sensor would be recommended towards the defect proofing of each individual layer of the eight layers to make sure that no defect would exist to interfere with the amperometric measurements. Also a different sensor design, possibly a parallel structure rather than the vertical layered structure, to reduce the fabrication difficulty (therefore to increase the sensor quality) was recommended.

167

BIBLIOGRAPHY A. bard and L. Faulkner, “Electrochemical methods: fundamentals and applications”, 2nd edition, 2001 John Wiley & Sons, Inc. A. Fog and R.P. Buck, “Electronic semiconducting oxides as pH sensors”, Sensors and actuators, 5, 137146, 1984 A. Safavi and H. Abdollahi, “Optical sensor for high pH values”, Analytica Chimica Acta 367 (1998) 167173 ABM Inc., “Operation manual for ABM mask alignment & exposure system” B. Elyassi, N. Rajabbeigi, A. Khodadadi, S. Mohajerzadeh, and M. Sahimi, “An yttria-doped ceria-based oxygen sensor with solid state reference”, Sensors and Actuators B 103 (2004) 178-183 B. Sohn and C. Kim, “A new pH-ISFET based dissolved oxygen sensor by employing electrolysis of oxygen” Sensors and Actuators B 34 (1996) 435-550 C. Herard, P. Bowen, J. Lemaitre, and J. Dutta, “Chemical synthesis and characterization of nanocrystalline palladium oxide”, Nanostructured Materials 6 (1995) 313-316 C. Liu, “Development of chemical sensors using microfabrication and micromachining techniques” Materials Chemistry and Physics 42 (1995) 87-90 C. Liu, D. Bocchicchio, P. Overmyer, and M. Neuman, “A palladium-palladium oxide miniature pH electrode”, Science, vol. 207, 11 Jan. 1980 C. Pan, J. Chou, T. Sun, and S. Hsiung, “Development of the tin oxide pH electrode by the sputtering method”, Sensors and Actuators B 108 (2005) 863-869 C. Vallieres, J. Gray, S. Poncin, and M. matlosz, “Potentiometric detection of oxygen based on the mixed potential of zinc” Electroanalysis 10 (3) (1998) 191-197 D. Garvey, “So, what is a RTD?”

168

F. Garzon, I. Raistrick, E. Brosha, R. Houlton, and B. Chung, “Dense diffusion barrier limiting current oxygen sensors” Sensors and Actuators B 50 (1998) 125-130 FY06-67: Research summaries fiscal years 2006-2007, office of the Chief Scientist, Science, Technology, and Management, US Department of Energy, Office of Civilian Radioactive Waste Management, Washington, DC, DOE/RW-0594 Also see: http://www.ocrwm.doe.gov/science/targeted_thrusts/matperf_targetedthrusts.shtml G. John, I. Klimant, C. Wittmann, and E. Heinzle, “Integrated optical sensing of dissolved oxygen in microtiter plates: a novel tool for microbial cultivation” Biotechnology and Bioengineering, 81 (7) (2003) 829-835 G. Meijer, “Thermal sensors”, Institute of Physics Publishing 1994 Gamry Inc., “Electrochemical impedance spectroscopy theory: a primer”, 2005 H. Kaneko, T. Okamura, H. Taimatsu, Y. Matsuki, and H. Nishida, “Performance of a miniature zirconia oxygen sensor with a Pd-PdO internal reference” Sensors and Actuators B 108 (2005) 331-334 H. Norton, Sensor and analyzer handbook, Prentice, Hall Inc., 1982 H. Ogino and K. Asakura, “Development of a highly sensitive galvanic cell oxygen sensor” Talanta 42 (2) (1995) 305-310 H. Suzuki, “Advances in the microfabrication of electrochemical sensors and systems” Electroanalysis 12 (9) (2000) 703-715 H. Suzuki, “Microfabrication of chemical sensors and biosensors for environmental monitoring” Materials Science and Engineering C 12 (2000) 55-61 http://en.wikipedia.org http://en.wikipedia.org/wiki/chemical_vapor_deposition.html http://en.wikipedia.org/wiki/Cyclic_voltammetry

169

http://en.wikipedia.org/wiki/image:angular velocity.svg http://en.wikipedia.org/wiki/Perovskite http://en.wikipedia.org/wiki/physical_vapor_deposition.html http://snf.stanford.edu/process/lithography/lift-off.html http://ts.nist.gov/MeasurementServices/ReferenceMaterials/upload/260-142-2ndVersion.pdf http://www.answers.com/topic/matthiessen-apos-s-rule-solid-state-physics?cat=technology http://www.ciks.cbt.nist.gov/garbocz/papers3/nodes3.html http://www.corrosion-doctors.org/principles/theory/html http://www.fuelcell-magazine.com/eprings/free/agilentfeb03.pdf http://www.gamry.com/app_notes/EIS_primer/EIS_primer.html http://www.kayelaby.npl.co.uk/chemistry/3_9/3_9_1.html http://www.ocrwm.doe.gov/science/targeted_thrusts/matperf_targetedthrusts.shtml http://www.physclips.unsw.edu.au/jw/ac.html#impedance http://www.radiometer-analytical.com/en_meterlab_molar.asp http://www.sensorsmag.com/articles/0900/17/main.shtml http://www.yorogawa.com/tm/ty/tm-tr0605_01.html J. Currie, A. Essalik, and J. Marusic, “Micromachined thin film solid state electrochemical gas sensors” Sensors and Actuators B 59 (1999) 235-241 J. Hale and M. Hitchman, “Some considerations of the steady state and transient behavior of membrane covered dissolved oxygen detectors” J. Electroanal. Chem. 107 (1980) 281-294

170

J. Helbert, “Handbook of VLSI microlithography”, 2nd ed., Noyes Publications, 2001 J. Hendrikse, W. Olthuis, and P. Bergveld, “The MOSFET as an oxygen sensor: constant current potentiommetry” Sensors and Actuators B 59 (1999) 35-41 J. Lee, H. Kim, and B. Kim, “Oxygen sensing characteristics of limiting current-type sensors with microstructural and structural variations in diffusion barrier” Materials Letters 26 (1996) 27-33 J. Li, “Development of a microfabricated sensor array for oil evaluation” Thesis dissertation 2005 J. Ramirex-Salgado and P. Fabry, “Feasibility of potentiometric oxygen gas sensor based on perovskite and sodium titanate measuring electrode”, Sensors and Actuators B 82 (2002) 34-39 J. Shen, L. Dudik, and C. Liu, “An iridium nanoparticles dispersed carbon based thick film electrochemical biosensor and its application for a single use disposable glucose biosensor”, Sensors and Actuators B 125 (2007) 106-113 K. Kreider, M. Tarlov, and J. Cline,“Sputtered thin-film pH electrodes of platinum, palladium, ruthenium, and iridium oxides”, Sensors and Actuators B 28 (1995) 167-172 K. Mancy, D. Okun, and C. Reiley, J. Electroanal. Chem. 4 (1962) 65-92 L. Clark, “Monitoring and control of blood and tissue oxygen tensions” Trans. Am. Soc. Art. Internal Organs 2 (1956) 41-49 L. Nei and R. Compton, “An improved Clark-type galvanic sensor for dissolved oxygen” Sensors and Actuators B 30 (1996) 83-87 L. Nei and R. Compton, “Novel approaches to galvanic oxygen analysis” Analytical Communications 33 (1996) 319-321 M. Albareda, A. Merkoçi, and S. Alegret, “Configurations used in the design of screen-printed enzymatic biosensors” A Review, Sensors and Actuators B-chem., 69 (2000) 153-163 M. Jeong, CH Pyun, and IH Yeo, J. Electrochem. Soc. 140 (1993) 1986

171

M. Madou, “Fundamentals of microfabrication: the science of miniaturization”, 2nd ed., CRC Press, 2002 M. Shao, X. Xing, and C. Liu, “pH measurements based on a palladium electrode”, Electroanalysis, 6, 245249, 1994 M. Wang, S. Yao, and M. Madou, “A long-term stable iridium oxide pH electrode”, Sensors and Actuators B 81 (2002) 313-315 N. Rajabbbeigi, B. Elyassi, S. Mohajerzadeh, and M. Sahimi, “A novel miniaturized oxygen sensor with solid-state ceria-zirconia reference” Sensors and Actuators B 100 (2004) 139-142 N. Sheppard, M. Lesho, P. McNally, and A> Francomacaro, “Microfabricated conductimetric pH sensor”, Sensors and Actuators B 28 (1995) 95-102 P. Kissinger and W. Heineman, “Laboratory techniques in electroanalytical chemistry”, Marcel Dekker Inc., 1984 P. Shuk, KV Ramanujachary, and M. Greenblatt, “New metal-oxide-type pH sensors”, Solid State Ionics 86-88 (1996) 1115-1120 Q. Wu, K. Lee, and C. Liu, “Development of chemical sensors using microfabrication and micromachining techniques”, Sensors and Actuators B 1-6 (1993) 13-14 R. Buck, S. Rondinini, AK Covington, FGK Baucke, CMA Brett, and MF Camoes, “Measurement of pH: definition, standards, and procedures”, Pure Appl. Chem. 74 (2002) 2169-2000 R. Jameel, NIST Special Publication 260-142. 2000, National Institute of Standards and Technology R. Kocache, J. Swan, and D. Holman, “Applications of oxygen-ion-conducting solid electrolytes” in “Solid State Gas Sensors”, Philadelphia, 1987 R. Meruva and M. Meyerhoff, “Potentiometric oxygen sensor based on mixed potential of cobalt wire electrode” Analytica Chimica Acta 341 (1997) 187-194

172

R. Ramamoorthy, P. Dutta, and S. Akbar, “Oxygen sensors: materials, methods, designs and applications”, J. of Materials Science 38 (2003) 4271-4282 R. Robinson and R. Stokes, “Electrolyte solutions”, 2nd ed., Butter Worths, 1959 S. Arrhenius, Z Phys. Chem. 4 (1889) 226 S. Badwal, M. Bannister, F. Ciacchi, and G. Hooshmand, “Response rate techniques for zirconia-based Nernstian oxygen sensors”, J. Appl. Electrochem., 18 (1988) 608-613 S. Glab, A. Hulanicki, G. Edwall, and F. Ingman, “Metal-metal oxide and metal oxide electrodes as pH sensors”, Crit. Rev. Anal. Chem. 21 (1989) 29-47 S. Refaey and G. Schwitzgebel, “Electrochemical impedance spectroscopic investigation of dissolution, passibation and pitting corrosion of tin in Na2CO3 solution and the effect of Cl- and I- ions”, Applied Surface Science, 135 (1998) 243-253 S. Skinner and J. Kilner, “Oxygen ion conductors”, Materialstoday, Mar. 2003, 30-37 S. Sorenson, “Enzyme studies II: the measurement and meaning of hydrogen ion concentration in enzymatic processes”, Biochem. Z 21 (1909) 131-200 S. Sotiropoulos and K. Wallgren, “Solid-state microelectrode oxygen sensors”, Analytica Chimica Acta, 388 (1999) 51-62 S. Wolf and R. Tauber, “Silicon processing for the VLSI era”, Vol. 1 2nd ed., Lattice Press, 2000 S. Yu, Q. Wu, M. Tabib-Azar, and C. Liu, “Development of a silicon-based yttria-stabilized-zirconia (YSZ) amperometric oxygen sensor”, Sensors and Actuators B, 85 (2002) 212-218 T. Seiyama, “Current state and future outlook” in “Chemical Sensor Technology”, Vol. 1 (1988) Kodansha Ltd. T. Yang and S. Pyun, “Hydrogen adsorption and diffusion into and in palladium”, Electrochimica Acta. 41 (6) (1996) 843-848

173

V. Karagonis, C. Liu, M. Neuman, L. Romankiw, P. Leary, and J. Cuomo, “A Pd-Rd film potentiometric pH sensor” IEEE Trans. Biomed. Eng. BME-33 (1986) 113-116 V. Kharton, A. Viskup, E. Naumovich, and V. Tikhonovich, “Oxygen permeability of LaFe1-xNixO3-δ Solid solutions” Materials Research Bulletin, Vol. 34 (8) (1999) 1311-1317 W. Cao, O. Tan, W. Zhu, B. Jiang, and C. Reddy, “An amorphous-like solid solution system for low temperature resistive-type oxygen sensing” Sensors and Actuators B 77 (2001) 421-426 W. Gopel, “Sensors”, Vol. 4 Thermal sensors, VCH, 1990 W. Grubb, “Palladium-palladium oxide pH electrodes”, Anal. Chem. 5 (1980) 273-276 W. Maskell, “Inorganic solid state chemically sensitive devices: electrochemical oxygen gas sensors” J. Phys. E. Sci. Instrum. 20 (1987) 1156-1168 W. Vonau and U. Guth, “pH monitoring: a review”, J. Solid State Electrochem., 10 (2006) 746-752 X. Wu, X. Wu, M. Tian, S. Zhang, and Y. Wang, “Research on n+p type oxygen sensor”, Solid State Electronics 46 (2002) 97-101 Y. Miao, J. Chen, and K. Fang, “New technology for the detection of pH”, J. Biochem. Biophys. Methods 63 (2005) 1-9

174

Suggest Documents