USF Libraries
USF Digital Collections

Investigation into the hydrogen gas sensing mechanism of 3C-SiC resistive gas sensors

MISSING IMAGE

Material Information

Title:
Investigation into the hydrogen gas sensing mechanism of 3C-SiC resistive gas sensors
Physical Description:
Book
Language:
English
Creator:
Fawcett, Timothy J
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Cubic silicon carbide
Thermal detection
Thermal conductivity
Heat transfer
Silicon-on-insulator
Dissertations, Academic -- Chemical Engineering -- Doctoral -- USF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: The hydrogen (H2) gas sensing mechanism driving 3C-SiC resistive gas sensors is investigated in this work in which two hypotheses are proposed. One hypothesis involves the surface adsorption of H2 on the sensor surface with the adsorbed molecules influencing the flow of current in a resistive gas sensor, termed the surface adsorption detection mechanism. The second hypothesis includes the transfer of heat from the sensor to the gas, producing a change in the temperature of the device when the heat transfer characteristics of the gas change, termed the thermal detection mechanism. The heat transfer characteristics of the gas are dependent on the thermal conductivity of the gas, a property which is a strong function of gas composition. Thus, the thermal detection mechanism mainly detects changes in the thermal conductivity of a gas or gas mixture.Initial experiments suggested the surface adsorption mechanism as the detection mechanism of resistive 3C-SiC gas sensors. However, these experiments were performed in the absence of device temperature measurements. Recent experiments in which the device temperature was measured with a resistance temperature detector (RTD) in thermal contact with the device strongly support the thermal detection mechanism as being responsible for hydrogen gas detection. Experimental observations show the temperature of the resistive 3C-SiC hydrogen gas sensors changes greatly with changing hydrogen gas composition. For example, a 3C-SiC/SOI resistive sensor biased at 10 Vdc displayed a change in temperature from ~400°C to ~216°C, correlating to a change in current from ~41 mA to ~6mA, upon the introduction of 100% H2. The this 3C-SiC/SOI resistive sensor, this large decrease in temperature caused a large increase in resistance which is detected as a decrease in current. Several different experiments have also been performed to confirm the thermal detection mechanism hypothesis.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Timothy J. Fawcett.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 69 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001794030
oclc - 145741135
usfldc doi - E14-SFE0001537
usfldc handle - e14.1537
System ID:
SFS0025855:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Investigation into the Hydrogen Gas Sens ing Mechanism of 3C-SiC Resistive Gas Sensors by Timothy J. Fawcett A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Chemical Engineering College of Engineering University of South Florida Major Professor: John T. Wolan, Ph.D. Stephen E. Saddow, Ph.D. Andrew M. Hoff, Ph.D. Norma A. Alcantar, Ph.D. David A. Rabson, Ph.D. Anita Lloyd Spetz, Ph.D. Date of Approval: March 20, 2006 Keywords: cubic silicon carbide, thermal de tection, thermal conductivity, heat transfer, silicon-on-insulator Copyright 2006, Timothy J. Fawcett

PAGE 2

Dedication I would like to dedicate this dissertation to my family, especially my wife for supporting me for these past few years. Wit hout all of you, I would not be in the position I am today. Thank you.

PAGE 3

Acknowledgements I would like to acknowledge my mentor, Dr. John T. Wolan, for his guidance and support throughout my doctoral work. I would also like to thank Dr. Stephen E. Saddow for providing the epitaxial material used in th is work and continual advise as well as Dr. Andrew M. Hoff for his help in device fabr ication. Prof. Anita Lloyd Spetz has also provided a great deal of advise and suppor t for this work including providing the gas sensor packaging. I would al so like to thank Dr. Norma A. Alcantar and Dr. David A. Rabson for their helpful ideas. I would like to thank Dr. Rachael L. Meyers and Meralys Reyes for growing the 3CSiC epitaxial films described in the work, as well as Dr. Shailaja P. Rao for fabricating the 3C-SiC/SOI devices while I was away. I would also like to thank Jeremy Walker, Benjamin Grayson and Alaa Kababji for thei r help and support thr oughout this work. I would like to acknowledge Dr. Carlos Smith for his help in tuning the software PID controller. This work was supported by the Defense University Research Initiative on Nanotechnology (DURINT) program administer ed by the Office of Naval Research under Grant N00014-0110715 monitored Dr. Colin Wood. Dr. Woods support is gratefully acknowledged.

PAGE 4

Table of Contents List of Tables ....................................................................................................................iv List of Figures .....................................................................................................................v Abstract ....................................................................................................................ix Chapter 1: Introduction .................................................................................................1 1.1 Gas sensor overview ...............................................................................................1 1.1.1 Semiconductor based gas sensor device structures.....................................1 1.1.1.1 Catalytic metal gate devices ............................................................1 1.1.1.2 Resistive gas sensors .......................................................................2 1.1.1.3 Thermal conductivity sensors .........................................................3 1.2 Detection mechanism of wide bandga p resistive hydrogen gas sensors .................3 1.2.1 Surface adsorption detection mechanism ....................................................3 1.2.1.1 Adsorption fundamentals ................................................................4 1.2.1.2 H2 induced surface metallization on 3C-SiC ..................................4 1.2.2 Thermal detection mechanism ....................................................................5 1.2.2.1 Heat transfer basics.........................................................................6 1.2.2.2 Thermal conductivity of gas mixtures ............................................6 1.2.2.3 Temperature dependence of semiconductor resistivity ...................7 Chapter 2: Resistive gas sensor fabrication ................................................................11 2.1 First generation devices ........................................................................................11 2.1.1 3C-SiC/Si epitaxial growth .......................................................................11 2.1.2 Scheme 1 Kapton tape mask ...............................................................12 2.1.3 Scheme 2 Photolithography mask ..........................................................12 2.2 Second generation devices ....................................................................................13 2.2.1 Resistive sensor fa brication process .........................................................14 2.2.2 3C-SiC/SOI epitaxial growth ....................................................................14 2.2.3 Active area definition ................................................................................15 2.2.4 Metal layer 1 .............................................................................................16 2.2.5 Metal layer 2 .............................................................................................16 2.2.6 Contact pad metal .....................................................................................16 2.2.7 Packaging ..................................................................................................16 2.3 Problems with second generation fabrication process ..........................................17 Chapter 3: Gas sensor test beds ..................................................................................18 3.1 First generation test bed ........................................................................................18 3.2 Second generation test bed....................................................................................18 i

PAGE 5

3.2.1 Electronic circuits .....................................................................................19 3.2.1.1 Gas sensor and ceramic heater circuits .........................................19 3.2.1.2 RTD circuit ...................................................................................20 Chapter 4: Results .......................................................................................................21 4.1 First generation 3C-SiC sensors ............................................................................21 4.2 Second generation 3C-SiC sensors.......................................................................21 Chapter 5: Hydrogen gas sensors using 3C-SiC/Si epitaxial layers ...........................24 5.1 Introduction ...........................................................................................................24 5.2 Epitaxial growth of 3C-SiC on Si(001)................................................................24 5.3 3C-SiC/Si gas sensor fa brication and testing ........................................................25 5.4 Results and discussion ..........................................................................................26 5.5 Conclusions ...........................................................................................................27 5.6 Acknowledgements ...............................................................................................27 Chapter 6: Wide-range (0.33 100%) 3C-SiC re sistive hydrogen gas sensor development ..............................................................................................28 6.1 Introduction ...........................................................................................................28 6.2 Device fabrication .................................................................................................29 6.3 Results and discussion ..........................................................................................29 6.4 Summary ...............................................................................................................31 6.5 Acknowledgements ...............................................................................................32 Chapter 7: Thermal detection mechanism of SiC based resistive hydrogen gas sensors .......................................................................................................33 7.1 Introduction ...........................................................................................................33 7.2 Experimental .........................................................................................................34 7.3 Results and discussion ..........................................................................................35 7.4 Summary ...............................................................................................................38 7.5 Acknowledgements ...............................................................................................38 Chapter 8: Thermal detection mechanism of 3C-SiC resistive gas sensors ...............39 8.1 Experimental .........................................................................................................39 8.1.1 Device fabrication .....................................................................................39 8.1.2 Gas sensor test bed ....................................................................................40 8.2 Results and discussion ..........................................................................................41 8.2.1 Experimental results ..................................................................................41 8.2.2 Finite element method simulations ...........................................................46 Chapter 9: Conclusions and future work ....................................................................48 9.1 Conclusions ...........................................................................................................48 9.2 Future work ...........................................................................................................50 9.2.1 Further investigation into the thermal detection mechanism for resistive sensors ........................................................................................50 9.2.2 Improved gas sensor performance ............................................................50 ii

PAGE 6

9.2.3 Test resistive sensor sensitivity to other gases ..........................................50 9.2.4 On device temperature measurement ........................................................50 References ...................................................................................................................51 Appendices ...................................................................................................................58 Appendix A. Derivation of heat transfer equations ............................................59 Appendix B. Derivation of temperatur e dependence of carrier concentrations ...............................................................................63 Appendix C. Details of finite element method simulations ...............................66 C.1 Geometry ...................................................................................................66 C.2 Governing equations .................................................................................66 C.3 Subdomain properties ...............................................................................66 C.4 Boundary conditions .................................................................................67 C.5 Simulation procedure ................................................................................68 About the Author...................................................................................................End Page iii

PAGE 7

List of Tables Table 1.1. Thermal properties of several gases at various temperatures .....................5 Table 5.1. Empirical relationships between 3C-SiC/Si sensor response to H2 concentration at various temperatures ......................................................27 Table A.1. Notation for derivation of heat transfer equations....................................59 Table B.1. Notation for derivation of temp erature dependence of carrier concentrations ...........................................................................................63 iv

PAGE 8

List of Figures Figure 1.1. Catalytic metal gate gas sensing device structures .....................................2 Figure 1.2. Shift in I-V characteristic s in a Schottky diode sensor ...............................2 Figure 1.3. Schematic of general resistive sensor device structure ...............................3 Figure 1.4. Calculated band diagram from H2 induced surface metallization of 3C-SiC .........................................................................................................5 Figure 1.5. Thermal conductivity of a He-N2 mixture as a function of He mole fraction at room temperature .......................................................................7 Figure 1.6. Temperature dependence of majori ty carrier concentration for ntype Si .........................................................................................................8 Figure 1.7. Resistivity of n-type (a) 3C-SiC and (b) 6H-S iC as a function of temperature and doping level ....................................................................10 Figure 2.1. Schematic drawing of 3C-SiC/Si gas sensor ............................................11 Figure 2.2. Current versus voltage (I-V) ch aracteristics of a typical first generation 3C-SiC resistive hydrogen gas sensor .....................................12 Figure 2.3. Photolithography mask used for first generation devices .........................13 Figure 2.4. Example second generation device layout ................................................14 Figure 2.5. 3C-SiC/SOI resistive device schematic. ...................................................15 Figure 2.6. Schematic of fabrication problems using second generation process .......17 Figure 3.1. First generation gas sensor test bed. .........................................................18 Figure 3.2. Second generation gas sensor test bed. .....................................................19 Figure 3.3. Second generation test bed circuit ............................................................19 Figure 4.1. Gas sensor response at 50C to H2 in varying concentrations (10%100% with 10% increments) with Ar purge between H2 events ...............21 v

PAGE 9

Figure 4.2. Current (left axis) and percent change in current (right axis) flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in resp onse to various concentrations (10% to 100% in 10% increments) of H2 in N2 at an initial device temperature of 400C and a bias of 10 V DC ...........................................22 Figure 4.3. Temperature of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in resp onse to various concentrations (10% to 100% in 10% increments) of H2 in N2 at an initial device temperature of 400C and a bias of 10 V DC ...........................................22 Figure 4.4. Current flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% H2 (in N2) in 10% increments with device temperature controlled at 400C and a bias of 10 V DC ...................................................................23 Figure 4.5. Current flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% He (in Ar) in 10% increments with device temperature controlled at 100C and a bias of 10 V DC ...................................................................23 Figure 5.1. Schematic drawing of 3C-SiC/Si gas sensor ............................................25 Figure 5.2. Gas sensor test bed ....................................................................................25 Figure 5.3. Gas sensor response at 50C to various concentrations of H2 (10% to 100% with 10% steps)..........................................................................26 Figure 5.4. 3C-SiC/Si gas sens or response vs. H2 concentration at various temperatures ..............................................................................................26 Figure 5.5. n-type Si(001) gas sensor response vs. H2 concentration at various temperatures ..............................................................................................27 Figure 5.6. Gas sensor response at low H2 concentrations at 50C .............................27 Figure 6.1. Gas sensor response at 50C to H2 in varying concentrations (10%100% with 10% increments) with Ar purge between H2 events ...............30 Figure 6.2. Gas sensor response at 50C to H2 in varying concentrations (10%100% with 10% increments) w ithout Ar purge between H2 events ..........30 Figure 6.3. Gas sensor response at 50C to H2 in varying concentrations (10%100% with 10% increments) with Ar purge between H2 events ...............31 Figure 7.1. 3C-SiC/SOI resistive device schematic ....................................................34 vi

PAGE 10

Figure 7.2. Measured current (left axis) and percent change in current (right axis) flowing through a 3C-SiC/SOI resistive gas sensor, with 0.5 mm x 0.25 mm contacts and 0.25 mm ga p, in response to 10% to 100% H2 in N2 in 10% increments at an initial device temperature of 673 K ....................................................................................................35 Figure 7.3. Measured temperature of a 3C-SiC /SOI resistive gas sensor, with 0.5 mm x 0.25 mm contacts and 0.2 5 mm gap, in response to 10% to 100% H2 in N2 in 10% increments at an initial device temperature of 673 K ................................................................................36 Figure 7.4. Calculated resistance (left axis) and measured temperature (right axis) vs. H2 concentration (N2 ambient) for a 3C-SiC/SOI sensor, with 0.5 mm x 2 mm contacts and 0.25 mm gap, at an initial device temperature of 673 K for bias voltages of 2.5, 5, and 10 V DC...............37 Figure 8.1. Current versus voltage (I-V) charac teristics of typical 3C-SiC/SOI resistive gas sensor with 0.5 mm x 1 mm contacts and a 1 mm gap. .......40 Figure 8.2. (a) Cross sectional diagram of 3C-S iC/SOI resistive gas sensor and (b) photograph of packaged sensor ...........................................................40 Figure 8.3. Schematic of the experimental gas sensor test bed ...................................41 Figure 8.4. Current (left axis) and percent change in current (right axis) flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in resp onse to various concentrations (10% to 100% in 10% increments) of H2 in N2 at an initial device temperature of 400C and a bias of 10 V DC ...........................................42 Figure 8.5. Temperature of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in resp onse to various concentrations (10% to 100% in 10% increments) of H2 in N2 at an initial device temperature of 400C and a bias of 10 V DC ...........................................42 Figure 8.6. Resistance of several 3C-SiC/SOI sensors, with 0.5 mm x 1 mm contacts and various gap lengths as a function of temperature ................42 Figure 8.7. Current as a function of H2 concentration (in N2) of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperatures with a bias of 10 V DC ..................................43 Figure 8.8. Percent change in current of a 3C-SiC/SOI sensor with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperatures with a bias of 10 V DC ........................................................43 vii

PAGE 11

Figure 8.9. Resistance as a function of H2 concentration (in N2) of a 3CSiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperat ures with a bias of 10 V DC ......................44 Figure 8.10. Percent change in resistance of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperatures with a bias of 10 V DC ........................................................44 Figure 8.11. Resistance as a function of H2 concentration (in N2) of a 3CSiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at an initial device temperature of 400C with sensor biased at several voltages .........................................................................................44 Figure 8.12. Current flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% H2 (in N2) in 10% increments with device temperature controlled at 400C and a bias of 10 V DC ...................................................................45 Figure 8.13. Current flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% He (in Ar) in 10% increments with device temperature controlled at 100C and a bias of 10 V DC ...................................................................46 Figure 8.14. Device temperature as a function of H2 concentration (in N2) at several initial device temperatures ............................................................47 Figure 8.15. Resistivity as a function of H2 concentration (in N2) at several initial device temperatures ........................................................................47 Figure 8.16. Percent change in resistivity as a function of H2 concentration (in N2) at several initial device temperatures .................................................47 Figure C.1. Device simulation geometry .....................................................................66 Figure C.2. Resistivity versus temperature of a 3C-SiC/SOI resistive gas sensor with 0.5 mm x 1 mm contacts with a 1 mm gap .......................................67 Figure C.3. Device temperature under pure N2 at 373 K .............................................69 Figure C.4. Device temperature as a function of heater heat flux ...............................69 viii

PAGE 12

ix Investigation into the Hydrogen Gas Sensing Mechanism of 3C-SiC Resistive Gas Sensors Timothy J. Fawcett Abstract ABSTRACT The hydrogen (H2) gas sensing mechanism driving 3C -SiC resistive gas sensors is investigated in this work in which two hypotheses are proposed. One hypothesis involves the surface adsorption of H2 on the sensor surface with the adsorbed molecules influencing the flow of current in a resis tive gas sensor, termed the surface adsorption detection mechanism. The second hypothesis includes the transfer of heat from the sensor to the gas, pr oducing a change in the temperatur e of the device when the heat transfer characteristics of the gas change, te rmed the thermal detection mechanism. The heat transfer characteristics of the gas ar e dependent on the ther mal conductivity of the gas, a property which is a strong function of gas composition. Thus, the thermal detection mechanism mainly detects changes in the thermal conductiv ity of a gas or gas mixture. Initial experiments suggested the surface adsorption mechanism as the detection mechanism of resistive 3C-SiC gas sensors. However, these experiments were performed in the absence of device temperature measur ements. Recent experiments in which the device temperature was measured with a resistance temperature detector (RTD) in thermal contact with the device strongly s upport the thermal detection mechanism as being responsible for hydrogen gas detecti on. Experimental observations show the temperature of the resistive 3C-SiC hydroge n gas sensors changes greatly with changing hydrogen gas composition. For example, a 3C-SiC/SOI resistive sensor biased at 10 Vdc displayed a change in temperature from ~400 C to ~216C, correla ting to a change in current from ~41 mA to ~6mA, upon the introduction of 100% H2. The this 3C-SiC/SOI resistive sensor, this large decrease in temp erature caused a large in crease in resistance which is detected as a decrease in current. Several different experi ments have also been performed to confirm the thermal detection mechanism hypothesis.

PAGE 13

Chapter 1: Introduction 1.1 Gas sensor overview Advancing technology has driven the development of gas se nsors able to operate at high temperatures and in harsh environments. Gas sensors are needed to monitor fuel cell [1,2] and automotive exhaust emissions [3 ], air pollution [4,5], chemical processes [6], and flue gas compositions [7] for safety and process control. Many material systems have been investigated for use as gas sensors with the most successful being semiconductors [8-10] and semiconducting metal oxides [11,12]. For operation at high temperatures (> 200C), wide bandgap semi conductor materials are generally required [13] with silicon carbide (SiC) being one of the most studied material systems for high temperature harsh environm ent gas sensing [3,6-9,14-25]. Many different device structur es have been studied for gas sensing applications. These device structures range from surf ace acoustic wave (SAW) devices built on piezoelectric materials [26] to semic onductor based devices including metalsemiconductor (MS) [14,19] and meta l-oxide-semiconductor (MOS) devices [3,10,23,27]. Thermal conductivity sensors fabricat ed from a wide variety of materials, ranging from Si to Pt have also been devel oped [28-33]. This study focuses on the use of resistive MS devices for the detection of hydrogen (H 2 ). Semiconductor and thermal conductivity based gas sensing devices are described below. 1.1.1 Semiconductor based gas sensor device structures Several different semiconductor device struct ures have been employed as gas sensors including Schottky barrier diode [9,19-21] MOS capacitor [34], MOS field-effect transistor (MOSFET) [3,7,25,27,35], and resistive [14,36] structures. All of these device structures, except resistive devices, employ a catal ytic metal gate to aid in gas detection. The metal contacts for resistive structures do no t necessarily need catalytic properties. A brief overview of the structure and operation of these device structures is given below. 1.1.1.1 Catalytic metal gate devices Several gas sensing device structures employ th e use of a catalytic metal gate to assist in the detection of specific gases. Schottky barrier di ode, MOS capacitors and MOSFETs have all been used in conjunction with differe nt catalytic metal gates, usually Pt, Pd, Ru, or Ir, to detect specific gase s [23]. Schematics of these devi ce structures are presented in Figure 1.1 The catalytic metal gate, sometimes re ferred to as top-side contact, reacts with the detection gas, producing a response that can be measured electrically. For example, the detection of hydrogen containi ng species is routinely performed using catalytic metal gate devices. The hydrogen co ntaining species dissociates on the catalytic metal surface forming atomic hydrogen (H) [37]. These H atoms then diffuse to the metal-insulator interface and form a dipole layer which changes the electrical properties 1

PAGE 14

of the device [10,23,27]. For example, this di pole layer shifts the current vs. voltage (IV) characteristics for a Schott ky diode sensor, as shown in Figure 1.2 [24,38]. (a) Schottky diode (b) MOS capacitor (c) MOSFET Figure 1.1. Catalytic metal gate ga s sensing device st ructures [23]. Figure 1.2. Shift in I-V characteristics in a Schottky diode sensor [24]. 1.1.1.2 Resistive gas sensors Resistive gas sensors can be categorized by the substrate material from which they are built. Semiconducting metal-oxides [11,12,39] were the first resistive gas sensors while wide bandgap materials have recently at tracted much interest [14,40]. Although the device structure for both material systems is similar, as shown in Figure 1.3 their detection mechanisms are very different. Semiconducting metal-oxide resistive sensors detect gaseous species by induced changes in the stoichiometry of the metal-oxide surface upon reaction with the target species. Th ese changes in stoichiometry give rise to a change in the resistance of the device, wh ich is interpreted as the sensor response 2

PAGE 15

[36,41]. The detection mechanism behind wi de bandgap metal-semiconductor resistive sensors is the topic of this dissertation, and will be explored below. Figure 1.3. Schematic of general resistive sensor device structure. 1.1.1.3 Thermal conductivity sensors As described in section 1.2.2 thermal conductivity sensors are based on the transfer of heat from a device to a gas. The amount of heat transferred, and thus the temperature of the device, is related to the thermal conductivity of the gas. In the thermal conductivity sensors presented in [28-33], a resistive h eater of some sort was used to heat the gas with the temperature change being m easured directly by a resistance temperature detector (RTD) or thermocouple. The resi stive heaters used in previous thermal conductivity sensors range from air-bridge, us ually silicon or polysilicon [31-33], to standard, usually Pt [28-30], microheaters. The sensors described in this work measure the temperature change caused by thermal conductivity changes of the gas mixture, through the gas composition, directly wit hout the use of an additional RTD. 1.2 Detection mechanism of wide bandgap resistive hydrogen gas sensors The aim of this report is to identify the detection mechanism(s) driving 3C-SiC resistive hydrogen gas sensor s. Although this detection mechanism is not yet fully understood, several hypotheses are proposed in th is report. The first proposed detection mechanism involves the adsorption of H 2 on the sensor surface that, in turn, causes a change in the device resistance. The second proposed detection mechanism is based on heat being absorbed from the device by the gas, causing a characteristic change in temperature of the device thus affecting th e device resistance. Background information behind both of these detection mechanisms will be presented below. 1.2.1 Surface adsorption detection mechanism The adsorption detection mechanism is based on the first reports of contact potential changes with changing ambient environments in Ge in 1953 [42]. The dangling bonds of a semiconductor surface can act as adso rption sites for gases. Thus H 2 may adsorb on the semiconductor surface, most likely dissociating in the process. The presence of H atoms on the surface can influence the transport of electrons in the semiconductor, producing a measurable change in the a resist ive 3C-SiC sensors resistance. 3

PAGE 16

1.2.1.1 Adsorption fundamentals The adsorption of gaseous species on the surface of a solid is driven by either physical (van der Waals forces) or chemical bonding forces, termed physisorption and chemisorption respectively. The difference between these types of adsorption lie in the nature of the interaction. P hysisorption is usually driven by physical forces such as Coulombic attraction whereas chemisorption is the actual sharing of electrons, forming a weak covalent bond. This difference is qua ntified by the surface binding energy, with bond energies greater than 10 kcal/mol (or 0.43 eV/molecule) attributed to chemisorption and bond energies less than 10 k cal/mol to physisorption [43]. The quantity of gas that adsorbs on a solid su rface is a function of temperature, total pressure, and composition of the adsorbing gas n ear the surface. A plot of the amount of adsorbed gas, given in a variety of units from fractional surface coverage to volume of gas adsorbed per weight of sold (cc/g), vers us partial pressure of that gas at a fixed temperature is termed an adsorption isotherm. Different adsorption isotherms are utilized to describe different adsorption mechanisms. For example, the Langmuir adsorption isotherm is used for non-interacting gases a nd with solids whose ad sorption sites are all equivalent while the Tempkin adsorption isot herm is used for in teracting gases [43]. The adsorption of gases on solid surfaces is not only a function of temperature, total pressure and composition as described above but is also a function of the surface structure of the solid [43]. The surface struct ure of a solid is defined by the crystal plane and surface reconstruction of that surface. With tungsten (W), for example, N 2 readily adsorbs to the W(100), W(310), W(411), W(210) and W(311) crystal planes whereas very little adsorption of N 2 is found on W(110), W(211), W( 111), and W(321) [43]. The adsorption of gases on a solid surface can al so either induce a surface reconstruction or change a surfaces existing reconstruction. Fo r example, upon the adsorption of CO onto a (2x1)Pt(110) surface, the surface atoms move to form a (1x1) surface reconstruction, and remains in that surface reconstruction ev en after CO desorbs from the surface [43]. 1.2.1.2 H 2 induced surface metallization on 3C-SiC In the 3C-SiC material system, it has been shown both experimentally and theoretically that th e adsorption of hydrogen on the surf ace of 3C-SiC (001) induces a 3x2 surface reconstruction in ultrahigh vacuum (UHV). The band structure of 3C-SiC changes at the surface due to this induced su rface reconstruction, be nding the bands such that the conduction band crosses the Fermi level, as shown in Figure 1.4 With the Fermi level in the conduction band (at least at the surface) electrons no longer have an energy gap to cross and are free to move without re striction. Thus, the surface of 3C-SiC under these conditions becomes metallic in nature [44-46]. Under these conditions, an increase in electron flow (decreased resi stance), at least at the surface, should be observable. The surface adsorption mechanism is one hypothesis to explain the detection mechanism driving 3C-SiC resistive hydrogen gas sens ors. The second hypothesis, a thermal detection mechanism, is described below. 4

PAGE 17

Figure 1.4. Calculated band diagram from H 2 induced surface metallization of 3C-SiC [46]. The Fermi level (E F ) is that of the surface, and the shaded areas represent the predicted bulk band structure. 1.2.2 Thermal detection mechanism The thermal detection mechanism is based on recent evidence showing the temperature of the resistive gas sensor changi ng while subjected to different gases [15]. With the absence of any other thermal phenomena (i.e. the latent heat associated with a phase change), the only manner for the temperature of the gas sensor to change is by heat transfer (from the device to the gas). The heat transfer properties of different gases can vary dramatically, as shown in Table 1.1 [47]. Since a semic onductors resistivity is strongly temperature dependent, even small changes in temper ature can be de tected as a change in resistivity. Table 1.1. Thermal properties of several gases at various temperatures [47]. k (W/mK) C p (J/molK) Gas 300 K 600 K 300 K 600 K N 2 0.0222 0.0441 28.47 29.73 Ar 0.0152 0.0301 20.81 20.81 He 0.135 0.247 20.81 20.81 H 2 0.156 0.291 28.73 29.75 CH 4 0.0277 0.0948 36.8 51.25 CO 2 0.0129 0.0403 37.64 47.9 O 2 0.0225 0.0412 26.2 33.08 5

PAGE 18

1.2.2.1 Heat transfer basics The transfer of heat from the device to the gas can be analyzed by first writing a differential energy balance on a differential vo lume unit of the device, shown in equation 1.1 [48]. A glossary of terms and the full derivation of heat transfer equations used can be found in Appendix A TkV t T Cp 2 2 (1.1) This energy balance describes the transport of energy inside the device not its interaction with the gas. Boundary conditions are needed to completely define the heat transfer problem and define the interaction between th e device and the gas. The first boundary condition defines the initial state of the syst em by defining the temperature of the device at t = 0, shown in equation 1.2 The second boundary condition, equation 1.3 defines the heat flux at the heater/dev ice interface (see section 2.2.7 for details of the gas sensor packaging), which is constant for each experime nt not utilizing temper ature control. The final boundary condition defines the heat flux at all surfaces aside fr om the heater/device interface, shown in equation 1.4 The amount of heat transfe rred to the gas is described by Newtons law of cooling, equation 1.4 and contains a heat transfer coefficient ( h ) that describes the heat transfer ch aracteristics of the gas [48]. oTT t = 0, everywhere (1.2) device heater heater device heaterTkq q/ / t > 0, heater/device interface (1.3) surface surface surface gas surfaceTkqTThq t > 0, all surfaces except heater/device interface (1.4) Heat transfer, as defined by equations 1.1 through 1.4 will occur at all times a gas sensor experiment is in progress. During a gas sensor experiment, the concentration of the test gas (H 2 in this case) is changed in relation to a dilutant gas (usually N 2 Ar or air) while keeping the total flow over the sens or constant. With each change in the composition of the gas, the heat transfer char acteristics of that gas change. The thermal properties, namely specific heat ( C p ) and thermal conductivity ( k ), are a function of the gas composition and will always be between those of the pure components. Thus, the thermal properties of the test gas mixture will always be different from those of the pure dilutant gas. In this manner, the therma l detection mechanism is based on changes in thermal properties in relation to the dilutant or the reference gas. These changes in thermal properties, combined into the heat transfer coefficient, are measured as a temperature change due to the transfer of heat from the device to the gas. 1.2.2.2 Thermal conductivity of gas mixtures The thermal conductivity of a gas mixt ure can be expressed by the empirical relationship proposed by Wassiljewa, shown in equation 1.5 where k mix is the thermal 6

PAGE 19

conductivity of the mixture, n is the number of gases in the mixture, y i is the mole fraction of gas i k i is the thermal conductivity of pure gas i and A ij is a function of the pure gas properties [49]. n i n j ijj ii mixAy ky k1 1 (1.5) The functional form of A ij can be expressed using Mason and Saxenas proposed expression in equation 1.6 where M i is the molar mass of pure gas i [49]. 21 2 41 2118 1605.1ji jiji ijMM MMkk A (1.6) The thermal conductivity of a He-N 2 mixture as a function of He concentration at room temperature is presented in Figure 1.5 Figure 1.5. Thermal conductivity of a He-N 2 mixture as a function of He mole fraction at room temperature [29]. 1.2.2.3 Temperature dependence of semiconductor resistivity The temperature dependence of the resistivity of a semiconductor is related to both the concentration of electrons and holes ( n and p =[cm -3 ]) as well as their respective mobilities ( n and p =[cm 2 /Vs]), as shown in equation 1.7 [50]. pnqpn 11 (1.7) 7

PAGE 20

Assuming the semiconductor is n-type, i. e. doped with donor atoms such that n >> p, the expression for resistivity simplifies to the expression found in equation 1.8 The same arguments presented below also stands for p-type semiconductors. nqn 1 (1.8) Both the concentration of electrons and el ectron mobility are temperature dependent. A glossary of terms and the full derivation of the temperature dependence of carrier concentrations (electrons and holes) can be found in Appendix B The temperature dependence of the concentrat ion of electrons stems from the thermal energy of the system, k BT The concentration of el ectrons is given by equation which contains two temperatur e dependent terms [51]. B 1.9 TkEE c TkEE BeBcF BcFeN e h Tkm n 2 322 2 (1.9) The first term is the effective density of states for electrons which describes the number of electrons per unit volume that exist in the semiconductor. The second temperature dependent term describes the fr action of those electrons that will have enough energy to jump from the valence band to the conduction band. The combination of these two temperature dependent terms given in equation 1.9 yields three different regions, termed the freeze out, extrinsic, and intrinsic regions, of carrier concentration dependence on temperature, shown in Figure 1.6 for n-type Si. Figure 1.6. Temperature depende nce of majority carrier concentration for n-type Si [51]. 8

PAGE 21

In the first region, for T < ~ 100 K, ve ry few donor atoms are ionized, and any electrons that can move have insufficient thermal energy to excite them from the valence band to the conduction band. This region of insufficient thermal energy is termed the freeze out region. The second region is termed the extrinsic region and applies to temperatures between 100 K and 450 K. Th e thermal energy within this temperature range causes essentia lly approximately all donors to be ionized, in Si, but not to possess enough energy to cross the bandgap. This gives a relatively constant carrier concentration over this span of temperatures. If the temperature is increased even further, T > ~ 450 K, electrons will have enough ther mal energy to cross the bandgap with the carrier concentration approach ing the intrinsic carrier con centration, thus termed the intrinsic region [51]. The other term that factors in the temperat ure dependence of resistivity of an n-type semiconductor is the electron mobility. The el ectron mobility is a measure of the ease of electron motion in a semiconductor. Severa l different scattering effects govern the magnitude of this mobility such as phonon and ionized impurity scattering. Phonon scattering refers to the coll ision between electrons and la ttice vibrations while ionized impurity scattering refers to the Coulombic attraction or repulsion between electrons and the ionized donors. Other scattering mech anisms exist, but phonon and ionized impurity scattering are the dominant sc attering mechanisms [51]. The first order approximation of the m obility caused by phonon scatter yields the inverse relationship shown in equation 1.10 As temperature increases, the lattice vibrations (or phonons) also increase, increasing the scattering of electrons and decreasing the electron mobility, as demonstrated in equation 1.10 However, increasing temperature increases the velocity at which electrons move. This increased velocity decreases the time in which their movement is influenced by ionized impurities. Thus, the mobility associated with ionized impurit y scattering is directly proportional to temperature, as shown in equation 1.11 However, increasing temperature increases the number of ionized impurities, yielding th e inversely proportional term in equation 1.11 [51]. 2 3 Tphonon (1.10) AD ionizedNNT2 3 (1.11) Once the mobilities caused by the different sca ttering mechanisms have been determined, they are combined to form an overall mobility, shown in equation 1.12 [51]. nionized nphonon n ,111 (1.12) In general, the temperature dependence of resistivity described in this section is rather complex. The actual nature of this temper ature dependence is dependent on material, material quality, and doping leve l and not easily described or quantified. For example, the resistivity of 3C-SiC and 6H-SiC, shown in Figure 1.7 initially decreases with increasing temperature at lower dopings due to the creation of more carriers and less 9

PAGE 22

ionized impurity scattering, but then increases with te mperature due to the phonon scattering as previous ly described [51-53]. (a) (b) Figure 1.7. Resistivity of n-type (a) 3C-SiC and (b ) 6H-SiC as a func tion of temperature and doping level [52,53]. Upon examination of the resistivity versus temperature plots in Figure 1.7 a doping level for a resistive sensor ba sed on the thermal detection mechanism can be selected. For example, lower dopings, such as the unintentionally doped line for 3C-SiC in Figure 1.7 a, have a higher temperature coefficient of re sistivity at temperatures greater than 0C, however, this coefficient is not constant al ong all temperatures greater than 0C. Therefore, even with the highest change in re sistivity for a given change in temperature, that change will not be constant for all temp eratures. This gives rise to a non-linear sensor response. However, the degenerately doped line for 3C-SiC in Figure 1.7 a has a relatively constant temperature coefficient of resistivity but does not posses as high a magnitude as the unintentionally doped line. Thus, there is a trade off between sensitivity, magnitude of a change in resis tivity due to a change in temperature, and linearity. 10

PAGE 23

Chapter 2: Resistive gas sensor fabrication 2.1 First generation devices The fabrication process of first generation devices (devices fabricated prior to April 2004) is reported in detail in [54]. The general device st ructure, consisting of a 4 m thick 3C-SiC epitaxial layer grown on n-type Si (001) with two plan ar contacts on the surface is shown in Figure 2.1 The 3C-SiC epitaxial laye r was grown by low-pressure chemical vapor deposition (LPCVD), described in section 2.1.1 After the 3C-SiC epitaxial layer was grown, the samples were cleaned using a standard RCA process described elsewhere [55]. The contacts on th ese devices were fabricated using one of two schemes. The first scheme uses Kapton tape to mask the active area of the sensor while the contacts were be ing deposited, described in section 2.1.2 The second scheme uses one photolithography mask, described in section 2.1.3 Both of these schemes are described in more detail below. Figure 2.1. Schematic drawing of 3C-SiC/Si gas sensor. The 3CSiC film is 4 m thick. 2.1.1 3C-SiC/Si epitaxial growth The 3C-SiC epi layer was gr own on Si(001) in a horizonta l LPCVD reactor [56]. The process involves three main st eps. First, a native oxide H 2 etch of the silicon wafer was performed. This was followed by a carbonizatio n step, which converts the Si surface to SiC. The final step is the growth of the 3C -SiC epi layer on the car bonized surface. Prior to growth, the Si wafer was RCA cleaned a nd placed on a graphite susceptor, which was inserted into the reaction tube Purified hydrogen, with a fl ow of 9 SLM, was introduced at a pressure of 150 Torr. The susceptor and Si wafer were then heated by radio frequency (RF) induction to 1000C and hydrogen etched in this manner for 10 min. The RF generator was then switched off for 5 minutes. Propane was then added to the hydrogen stream at a partial pressure of 0.03 Torr. After 60 sec, the RF generator was switched back on until the growth temperat ure of 1380C was reached. After 90 sec SiH 4 and C 3 H 8 were introduced at partial pressures of 0.035 and 0.035 Torr respectively. 3CSiC growth was then conducted for 120 min after which the RF generator was turned off and the sample was cooled under an Ar flow to ambient temperature. A growth rate of ~ 11

PAGE 24

2m/hour is typical for this process thus re sulting in a 4 m film for subsequent gas sensor fabrication [14,40]. 2.1.2 Scheme 1 Kapton tape mask Kapton is a polyimide film, developed by DuPont, that is able to withstand temperatures up to 400C [57]. Kapton tape was folded over on itself so there was no exposed adhesive, cut to the desired thickne ss, and then placed on the sample surface. Both sides of the folded over Kapton tape were then taped to a support wafer to hold the sample and Kapton tape mask in place. The samples were then placed in the electron-beam evaporator for the deposition of ~ 2000 of NiCr. After the deposition, the Kapton tape was removed, the sample wa s rinsed in acetone and DI water then dried in nitrogen (N 2 ). The sample was then annealed in a rapid thermal processing (RTP) unit at 950C for 2 minutes, in order to form ohmic NiCr contacts with linear current versus voltage (I-V ) characteristics shown in Figure 2.2 These devices where then loaded onto a ceramic sample holder, cont acted to with Ni probes, and placed in the first generation gas sensor te st bed described in section 3.1 Figure 2.2. Current versus voltage (I-V) characteristics of a typical first generation 3CSiC resistive hydrogen gas sensor. 2.1.3 Scheme 2 Photolithography mask In this scheme, the photolithography mask shown in Figure 2.3 was used in conjunction with negative photoresist (NR1 -PY 3000 from Futurrex Inc.) to form the NiCr contacts. This negative photoresist was spun on the sample surface at 3000 RPM for 40 seconds followed by a soft bake for 60 seconds at 155C. After the device cooled, the mask shown in Figure 2.3 was aligned to the sample a nd exposed using UV radiation, using a Quintel mask aligner, for 17 seconds followed by a post-exposure bake for 60 seconds at 130C. After the device coole d, the photoresist was developed in RD6 for 20 seconds, from Futurrex Inc., after which the device was rinsed in de-ionized (DI) water then dried in nitrogen (N 2 ). The samples were taped to a Si wafer (for support) and placed in the electron-beam evaporator for th e deposition of ~ 2000 of NiCr. After the deposition, the unwanted NiCr was removed us ing a standard lift-off process by rinsing 12

PAGE 25

in acetone to dissolve the remaining photoresist. After all the unwanted metal was removed, the samples were rinsed in me thanol and DI water then dried in N 2 Figure 2.3. Photolithography mask used for first generation devices. After the contacts were deposited on the sa mple surface, the contacts were annealed in a RTP at 950C for 2 minutes to produce ohmic NiCr contacts. After the NiCr contacts were made ohmic, negative photoresist was again spun on the sample surface. The same mask shown in Figure 2.3 was aligned to the existing NiCr contacts, and exposed then developed using the process de scribed above. The samples were again taped to a Si support wafer and placed in an electron-beam evaporator for the deposition of a Ti (100 ) / Au (1800 ) stack. The T i/Au stack is required for Au wire bonding to the device [55]. The unwanted metal was removed using the lift -off process described above. The samples where then mounted in a ceramic package with Au leads that were wire bonded to the contacts on the device. Thes e leads were contacted to with Ni wire and stainless steel screws to allow for elec trical connections, and placed in the first generation gas sensor tested bed described in section 3.1 2.2 Second generation devices For the second generation devices, a four -layer photomask se t was designed and fabricated. The evolution to a full photolithogra phy fabrication process allows for the control of device dimensions to within a fe w micrometers. This four-layer mask set contains resistive, Schottky diode and/or capacitor, van der Pauw, and transmission line method (TLM) structures. The TLM and van der Pauw structures can be used for contact and material characterization whereas the resistive, Schottky diode and capacitor structures allow for gas sensing device fabrication. The TLM structures have a c onstant contact size (50 x 100 m), but with increasing distances between them (10 to 50 m in 5 m steps). The van der Pauw structures have four square contacts, measuring 1 mm on a si de, but with several gap sizes (0.25, 0.5, and 1 mm). Resistive devices on the mask were made from the 64 combination of 3 parameters, contact width (x), contact height (y), and gap length (L), with 4 different dimensions, 0.25, 0.5, 1, and 2 mm. These resistive devices were then grouped in several ways, i.e. devices with a constant exposed area were grouped as well as devices with the same gap length and contact width, and pl aced on the mask. The Schottky diode and capacitor devices consist of dots w ith diameters of 250, 500, 750, and 1000 m. The various device dimensions for the mask set previously described were chosen to aid in the investigation of the mechanism driv ing 3C-SiC resistive gas sensors. Devices with the same contact dimensions but incr easing gap lengths, i.e. 0.25 mm, 0.5 mm and 1 13

PAGE 26

mm, are used to test the surface adsorption detection mechanism presented in section 1.2.1 With constant contact dimensions, any ga s detection caused by the contacts should be constant for all devices. Thus, any differe nces in response can be attributed to the increasing exposed surface areas through the surface adsorption detection mechanism. However, increasing the ga p length increases the device resistance (R = L/A), which decreases the amount of Joule heating for a fixed voltage (P = V 2 /R). This will factor into the thermal detection mechanism presented in section 1.2.2 In this way, devices with constant exposed surface areas but differe nt contact areas can be used to assess the influence of contact dimensions on device performance. The four mask layers consis t of 1) Active area definition, 2) Metal layer 1, 3) Metal layer 2, and 4) Contact pad metal. Mask la yer 1, the active area definition, is performed for all devices, whereas the application of the other three mask layers is dependent on the experiment. An example device layout for th e four-layer mask previously described, utilizing the Active area definition, Metal la yer 1, and Contact pad me tal layers, is shown in Figure 2.4 The full fabrication scheme is described below. Active area definition Metal layer 1 Contact pad metal Figure 2.4. Example second generation device layout. 2.2.1 Resistive sensor fabrication process A schematic of the device structure used fo r second generation devices is presented in Figure 2.5 The fabrication process for these s econd generation devices starts with the epitaxial growth of a 1.4 m thick 3C-SiC layer on a thin Si layer bonded to a polycrystalline SiC wafer by a process described in section 2.2.2 After epitaxial growth, these samples were RCA cleaned, which rids the sample surface of most organic and metallic atoms/molecules [55]. Next, the su rface is covered with ~ 5000 of silicon dioxide (SiO 2 ) via plasma enhanced chemical vapor deposition (PECVD). The rest of the fabrication process is described below. 2.2.2 3C-SiC/SOI epitaxial growth A thin 3C-SiC film was deposited on a thin Si (001) layer (~ 150 ) miscut 4 towards the [211] direction, bonde d to a polycrystalline SiC substrate, which is referred to as silicon-on-insulator (SOI), using a wafer bonding process described elsewhere [56]. The SOI substrate was RCA cleaned prior to deposition to remove trace organic and 14

PAGE 27

metal species from the surface [55]. The 3C -SiC deposition was carried out in a lowpressure, hot-wall, horizontal chemical vapor depositio n (CVD) reactor using the traditional dual precursor gas chemistry, propane (C 3 H 8 ) and silane (SiH 4 ) [58]. Hydrogen chloride (HCl) was also incorporated into the gas steam as a growth promoter. During the CVD process a gas stream containing C 3 H 8 and H 2 was established in the chamber while the sample temperature was ramped from room temperature to the growth temperature of 1360C. Once at the growth temperature, the process pressure was lowered to 200 Torr after which SiH 4 and HCl were added to th e gas stream for growth. The Si/C and Si/Cl ratios during growth were set to 0.7 and 3.3, respectively. The 3CSiC epitaxial film was determined to be ~ 1.4 m thick measured via FTIR [59]. Figure 2.5. 3C-SiC/SOI resistive device schematic. 2.2.3 Active area definition The active area definition mask is us ed to etch windows in the SiO 2 layer to define the active area of the gas sensor After the deposition of the SiO 2 layer, the sample is placed in a 3:1 mixture of H 2 SO 4 and H 2 O 2 Piranha etch, for 1 minute. The sample is then rinsed in DI water and dried in N 2 A dehydration bake was then performed on the sample by placing it on a hot plate at a temperature ~ 150C for 5 minutes to remove any remaining water from the surface. After the samp le cools, it is then placed on a spinner. Hexamethyldisilazane (HMDS) is then placed on the sample surface and allowed to stand for ~ 1 minute, after which it is spun at 3000 RPM for 30 seconds. Next, Shipley 1813 positive photoresist is spun on the sample surface at 3000 RPM for 30 seconds followed by a softbake at 110C for 60 seconds. Next the active area mask is aligned to the sample and exposed for 21.5 seconds on the Quintel mask aligner. The photoresist is then developed in MF-319 for approximately 2 minutes, then rinsed in DI water and dried in N 2 The photoresist is then hardbaked at 110C for 60 seconds. After the sample cools, it is placed in a 10:1 buffered oxide etch (BOE) for ~ 15 minutes to etch the windows in the 5000 thick SiO 2 layer to define the active area of the sensors. After the BOE etch is finished, the samples are rinsed in DI water, but not dried. They are then placed in a 3:1 Piranha solution (descr ibed above) for 60 seconds to remove the remaining photoresist. After the photoresist is removed, the samples are again rinsed in DI water and dried in N 2 The samples are then ready for the next photolithography step. 15

PAGE 28

2.2.4 Metal layer 1 After the sensor active areas have been defined, metal must be deposited, in controlled regions, to form the ohmic contacts for the sensor. If the samples were not processed immediately following the active area definition step, the samples were dipped in a 3:1 Piranha solution for 60 seconds, th en rinsed in DI water and dried in N 2 Next, another 5 minute dehydration bake was performed. After coo ling, the sample is placed on the spinner and Futurrex Inc. negative photoresist NR9-1500PY is spun on the surface at 3000 RPM for 40 seconds, fo llowed by a softbake at 155C for 60 seconds. The metal layer 1 mask is then aligned to th e alignment marks etched in the SiO 2 layer formed with active area definition mask (described in section 2.2.3 ) and exposed for 7.5 seconds, followed by a post-exposure bake at 110C fo r 60 seconds. The photoresist is then developed in RD6 for 20 seconds, rinsed in DI water and blown dry with N 2 After the photoresist is developed on the sa mple surface, the metal layer 1 is ready to be formed by depositing metal on the sample. This deposition was performed either by electron-beam evaporation or by sputtering. For all resistiv e samples, ~ 2000 of NiCr was deposited, usually by electron-beam evapor ation. After the metal layer is deposited, the unwanted metal is removed by a lift-off proc ess. This lift-off process included either the submersion of the sample in resist remove r, RD4 from Futurrex Inc. at ~ 110C, or by sonicating the sample in acetone for several minutes. After all of the unwanted metal is removed from the sample surface, the sample is rinsed several times in acetone and methanol, then in DI water and dried in N 2 For resistive sensors, the contacts are then annealed in a RTP at 950C for 2 minutes. 2.2.5 Metal layer 2 The fabrication process for metal layer 2 is the same as for metal layer 1 in all respects. This metal layer allows the deposition of two different metals on the same sample, thus giving the possibility of testing bot h of their responses to the detection gas. For example, a resistive sensor with NiCr contacts could be tested alongside Schottky diode sensors with a Pd gate on the same sample. This would allow the de tection of low concentrations of H 2 using the Schottky diode sensor [19,38] as well as high concentrations of H 2 using the resistive sensor [14]. 2.2.6 Contact pad metal The contact pad metal layer was fabricat ed using one of two different masks depending on the application. Th e contact pads consist of a Ti (200 ) / Au (4000 ) stack. The Ti aids in adhesion, and Au a llows for wire bonding. Using the contact pad metal mask, the Ti/Au stack is deposited on the metal 1 layer that overlaps the SiO 2 layer, as seen in Figure 2.4 The second scheme uses the metal layer 1 mask to completely cover the contacts with the Ti/Au bonding layer. For this contact pad metal layer, the same negative photores ist process outlined in section 2.2.4 is used but with the Ti/Au stack being the metal deposited and with no annealing afterwards. 2.2.7 Packaging After fabrication, sensor samples were di ced into pieces measuring ~ 3 mm x 6 mm and mounted in a 16-pin gold package. Each sample was mounted to a resistance temperature detector (RTD) which is mounted to a ceramic heater. The contacts of the 16

PAGE 29

sensors, fabricated using the second generati on fabrication process described above, are wire bonded to the package pins using a ball bonder. This new packaging scheme was employed to measure the temperature of th e device, via the RTD, and control its temperature with the ceramic heater. The p ackaging of these devices was performed by Evald Mildh at Linkping University in Sweden. 2.3 Problems with second generation fabrication process The only problem with this second generation fabrication process lie s in the potential for the metal on the semiconductor surface not being in contact with the metal on the SiO 2 surface. The metal in metal layers 1 and 2 overlap the active area definition by 200 m on three sides of the contacts. This was done so the contact pad metal could be deposited on metal layers 1 and 2 without de positing on the metal that is in contact with the semiconductor surface (the active area). The problem lies in the fact that the SiO 2 layer is ~ 5000 thick, and the metal layers area usually 2000 thick. Then the metal layer is deposited, there is no guarantee that the metal on the substrate will contact the metal on the SiO 2 as shown in Figure 2.6 Figure 2.6. Schematic of fabrication problems using second generation process. To alleviate this problem, the second scheme outli ned in section 2.2.6 was developed and utilized by using the metal 1 layer mask to deposit the bonding layer over the entire contact area to generate the contact pads The metal 1 and metal 2 could also be deposited via sputtered instead of electron-beam evaporation. Sputtering the metal layer would coat the side-walls of the field oxide layer, giving good contact between the metal on the field oxide and the meta l on the semiconductor surface. 17

PAGE 30

Chapter 3: Gas sensor test beds 3.1 First generation test bed The first generation test bed is described in de tail in [54]. This test bed consists of a tube furnace, three mass flow controllers (M FCs), three solenoid va lves, a power supply, and a digital multimeter (DMM), shown in Figure 3.1 The sensors are placed in a 1 diameter quartz tube which resides in a tube furnace. The furnace controls the temperature of both the gas and of the sensor The mass flow controllers are controlled by an analog output card, through a MFC rea dout, and the solenoid valves are controlled by a digital input/output card, through an elec tromechanical relay module. Both the analog output and digital input/output cards are installed in a computer. The power supply and digital multimeter are also cont rolled by a computer via GPIB and serial communication respectively. The power suppl y biases the gas sensor while the DMM measures the voltage across the sensor, the current flowing through the sensor, and the temperature of the gas around the sensor. Due to the nature of the measurements, different measurement modes and channels in the DMM, data was collected at a sampling frequency of 0.5 Hz. All of these com ponents are controlled by a computer through Labview from National Instruments. Figure 3.1. First generation gas sensor test bed. 3.2 Second generation test bed The second generation test be d was developed to accommodate the sensor packaging described in section 2.2.7 The pins of the 16-pin gold package can only be contacted to using a printed circuit board (PCB) which can not be heated above 125C [60]. With the maximum operating temperature of the PCB being 125C, the PCB cannot be placed in the quartz tube used in the fi rst generation test bed describe above. Thus, an aluminum housing was fabricated to support the sensor package and allow gas to flow over the sensor. This aluminum housing provides onl y an ~ 7 mm (0.157) space for gas to flow over the sensor, greatly decrea sed from the first generation test bed described in section 3.1 This test bed was expanded to accommodate 6 MFCs, all of which are computer controlled, shown in Figure 3.2 In place of the digital multi meter, an 18-bit 16 channel 18

PAGE 31

analog input card was added to the computer which also drastically increased the sampling rate from 0.5 Hz to 10 Hz with 500 kHz sampling rates possible. Figure 3.2. Second genera tion gas sensor test bed. 3.2.1 Electronic circuits 3.2.1.1 Gas sensor and ceramic heater circuits A simple operational amplifier follower circuit, shown in Figure 3.3 is used to supply the bias to the gas sensor in place of the power supply. The OPA541 operational amplifier (op amp), available from Texas Inst ruments (TI), is capable of up to 10 A of output current. The current flowing through th e sensor (RV) is determined by the voltage drop across a precision resist or (RC). The precision resistor was either 1.02% or 100.005% The resistor RB ensures the voltage at the non-inverting input of the op amp to be grounded if V in is disconnected. The ceramic heater is powered by the same circuit that drives the gas sensor, shown in Figure 3.3 with the heater as RV and a 0.01 resistor as RC. Figure 3.3. Second generation test bed circuit. 19

PAGE 32

3.2.1.2 RTD circuit The RTD measurements were performed by the combination of a XTR105, 4-20 mA current transmitter with sensor excitation, and RCV420, precision current loop receiver, integrated circuits (ICs) from TI. The XTR105 passe s a known current (nominally 800 A) through the RTD and a matched current th rough a reference resistor. The voltage drop across the reference resistor is subtra cted from the voltage drop across the RTD, then multiplied by a gain and linearized to produce a 4-20 mA signal that is proportional to an initial temperature, set by the reference resistor, and a temperature span, set by the gain. This 4-20 mA signal is then passed to the RCV420 wh ich converts it to a 0-5V signal that is measured by the PCI-62 81 analog card previously described. 20

PAGE 33

Chapter 4: Results 4.1 First generation 3C-SiC sensors Initial hydrogen gas sensing experime nts performed on 3C-SiC/Si sensors demonstrated the potential of 3C-SiC as a hydr ogen gas sensing material. The fabrication of these device is described in section 2.1 while the gas sensor test bed is described in section 3.1 For example, Figure 4.1 shows the response, reported as the change in current relative to 0% H 2 of two 3C-SiC/Si sensors with constant contact areas but different gap lengths, 0.5mm a nd 1 mm, to high concentrat ions (10% to 100%) of H 2 in Ar at 50C. Initial experiments aimed at determining the detection mechanism suggested the surface adsorption mechanism discussed in section 1.2.1 The sensor responses shown in Figure 4.1 approximately scaled with expos ed SiC area (ratio between areas = 2:1 whereas the ration between sensor response s ~ 1.5:1). The increas ed surface area can give an increased opportunity for H 2 to influence the surface conductivity, a result depicted in Figure 4.1 These results are discussed in more detail in Chapters 5 and 6 Figure 4.1. Gas sensor response at 50C to H 2 in varying concentrations (10%-100% with 10% increments) with Ar purge between H 2 events. Dashed line L = 0.5 mm, solid line L = 1 mm. 4.2 Second generation 3C-SiC sensors Several experimental improvements were ma de for the second generation sensors in an effort to determine the gas sensing mechanism. First, a thin film (1.4 m) of 3C-SiC deposited on an insulating substrate, described in section 2.2.2 was used in these studies. The use of a thin film on an insulating substrate allows for better control of where current flows in the device, i.e. onl y through the 3C-SiC epi-laye r and not also through the Si substrate. Also, fabricated devices had an improved structure with the addition of a field oxide isolation layer with a four-layer phot olithography mask set, discussed in section 21

PAGE 34

2.2 The packaging of second generation sensors, discussed in section 2.2.7 also allowed for device temperature measurement and control. Figure 4.2 shows the current flowing through a 3C-SiC/SOI sensor and Figure 4.3 shows the device temperature while exposed to high concentrations (10% to 100%) of H 2 in N 2 at 400C. Figure 4.3 shows drastic decreases in device temperature upon the introduction of H 2 to the system. These decreases in device temperature cause changes in the resistivity of the 3C-SiC epi-layer which in turn cause a change in the current passing through the device at a fixe d voltage, as seen in Figure 4.2 With these changes in device temperature, the thermal detection mechanism outlined in section 1.2.2 seems to drive the gas sensing of these 3C-SiC sensors. Figure 4.2. Current (lef t axis) and percent change in current (r ight axis) flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to various con centrations (10% to 100% in 10% increments) of H 2 in N 2 at an initial device temperature of 400C and a bias of 10 V DC. Figure 4.3. Temperature of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to various concentrations (10% to 100% in 10% increments) of H 2 in N 2 at an initial device temperature of 400C and a bias of 10 V DC. When device temperature is controlled, the current flowing through a 3C-SiC/SOI sensor increases with the introduction of H 2 to the system, as shown in Figure 4.4 This increase in current is opposite of that when device temperature is not controlled as shown in Figure 4.2 This opposite behavior initially s uggests another detection mechanism. However, the current also increases when increasing concentrations of He are introduced in a Ar atmosphere as shown in Figure 4.5 Since He and Ar are moth monatomic noble gases, they should have similar interac tions with the surface, allowing any surface adsorption effects to be ignored. Since changes in current are still observed in Figure 4.5 and the thermal properties, mainly thermal conductivities are so different, the thermal detection mechanism still explains the detect ion. These results are discussed in more detail in Chapters 7 and 8 22

PAGE 35

Figure 4.4. Current flowing through a 3CSiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% H 2 (in N 2 ) in 10% increments with device temperature controlled at 400C and a bias of 10 V DC. Dotted line = H 2 on, dotted line with circle = H 2 off. Figure 4.5. Current flowing through a 3CSiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% He (in Ar) in 10% increments with device temperature controlled at 100C and a bias of 10 V DC. 23

PAGE 36

Chapter 5: Hydrogen gas sensors using 3C-SiC/Si epitaxial layers T.J. Fawcett 1 J.T. Wolan *1 R.L. Myers 2 J. Walker 2 and S.E. Saddow 2 1 Chemical Engineering Department, University of South Florida, Tampa, FL 33620, USA 2 Electrical Engineering Department, University of South Florida, Tampa, FL 33620, USA Adapted from: T. J. Fawcett, J. T. Wolan, R. L. Myers, J. Walker, and S. E. Saddow, Hydrogen gas sensors using 3C-SiC/Si epitaxial layers Materials Science Forum 457460, 1499-1502 (2004). Poster presentation at the International C onference on Silicon Carbide and Related Materials (ICSCRM) 2003 in Lyon, France. Abstract. A hydrogen gas sensor consisting of plan ar electrical NiCr contacts formed on the surface of a 3C-SiC epitaxial layer grown on Si(001) has been fabricated and tested. The n-type, 4 m thick 3C-SiC epi layer wa s grown under low-pressu re conditions with an approximate doping density of 10 18 cm -3 This sensor demonstrates a two-fold repeatable improvement in stability and sensi tivity in comparison to an n-type Si sensor of the same type also fabricated and test ed under the same conditions. Both the 3CSiC/Si and Si sensors operated up to 250 C; however, the 3C-SiC/Si sensor was able to detect hydrogen at concentrations far exceedin g that of the Si sensor. The 3C-SiC/Si device detected hydrogen at concentrations ra nging from 0.333% to 100% in Ar while the Si sensor could only detect hydrogen at concen trations ranging from 2% to 100% in Ar. Based on this preliminary data, it has been s hown that 3C-SiC/Si hydr ogen sensors of this type have a larger dynamic range and higher se nsitivity to hydrogen than Si sensors, thus allowing for harsh environmental applications. 5.1 Introduction High temperature extreme environment solid-state gas sensing is of great interest to the automotive, aerospace and chemical pro cessing industries [6,17]. The wide bandgap, high thermal conductivity and thermal and chemi cal stability make silicon carbide (SiC) an ideal solid-state gas sensing material for operation under harsh conditions [22,61]. According to literature, a silicon (Si) gas sensor reportedly failed between 200 C and 250 C because of its relatively small bandgap of 1.1 eV [8,19,61]. This gives a 3C-SiC gas sensor (bandgap = 2.36 eV), the potential to operate at far greater temperatures. The 3C-SiC polytype was chosen because of its high electron mobility (~10 3 cm 2 /V-sec) and ability to be epitaxially grow n on Si, thus greatly decreasing the cost of wafers in comparison to bulk 3C-, 4Hor 6H-SiC substrates [20]. 5.2 Epitaxial growth of 3C-SiC on Si(001) In this work, a 3C-SiC epi layer was grow n on Si(001) in a low-pressure horizontal cold-wall CVD reactor [56]. The process involv es three main steps. First, a native oxide 24

PAGE 37

H 2 etch of the silicon wafer was performed. This was followed by a carbonization step, which converts the Si surface to SiC. The final step is the growth of the 3C-SiC epi layer on the carbonized surface. Prior to growt h, the Si wafer was RCA cleaned and placed on a graphite susceptor, which was inserted into the reaction tube. Pu rified hydrogen, with a flow of 9 SLM, was introduced at a pressure of 150 Torr. The susceptor and Si wafer were then heated by radio frequency (RF) induction to 1000C and hydrogen etched in this manner for 10 min. The RF generator wa s then switched off for 5 minutes. Propane was then added to the hydrogen stream at a partial pressure of 0.03 Torr. After 60 sec, the RF generator was switched back on un til the growth temperature of 1380C was reached. After 90 sec SiH 4 and C 3 H 8 were introduced at part ial pressures of 0.035 and 0.035 Torr respectively. 3C-SiC growth was th en conducted for 120 min after which the RF generator was turned off and the sample was cooled under an Ar flow to ambient temperature. A growth rate of ~ 2m/hour is ty pical for this process thus resulting in a 4 m film for subsequent gas sensor fabrication. 5.3 3C-SiC/Si gas sensor fabrication and testing The gas sensing device was fabricated by evaporating NiCr ( 2000 ) equipotential contacts on the 3C-SiC face of the structure. The device was then annealed at 950 C for 2 minutes in an Ar atmosphere to produce ohmic contacts. 5 shows a schematic drawing of the 3C-SiC/Si gas sensor. An n-type Si(001) sensor was also fabricated in the same manner and simultaneously tested for comparis on. Both sensors were approximately 1cm x 1cm, with 2.5 mm wide contacts with an 5 mm wide gap between the contacts. The initial resistances of th e 3C-SiC/Si and Si devices were approximately 13 and 16.6 respectively. The two sensors were placed in a gas sensor test bed, shown in Figure 5.2 Both sensors were biased with 2V using a Hewlett-Packard model 6626A quad programmable power supply. The se nsors were exposed to UHP H 2 to determine sensor response at varying concentra tions (0.333% to 100%) of H 2 in Ar with pure Ar flows in between each H 2 concentration to allow for complete desorption. The temperature of the test bed was varied from 50 C to 250 C in 50 C increments. The current flowing through each sensor was measured by a Keithle y model 2700 Digital Multimeter (DMM) equipped with a Keithley m odel 7700 20-Ch. Multiplexer. Figure 5.1. Schematic drawing of 3C-SiC/Si gas sensor. The 3CSiC film is 4 m thick. Figure 5.2. Gas sensor test bed. 25

PAGE 38

5.4 Results and discussion The 3C-SiC/Si sensor was observed to be a great deal more sensitive to H 2 concentration than the Si sensor throughout th e temperature range tested in this work. The gas sensor response in figures 2.3-2.6 ( I) is the change in current of each device relative to the current in 100% Ar (0% H 2 ). Figure 5.3 illustrates the sensor response to various H 2 concentrations (10% to 100% in 10% increments) at 50 C with an Ar purge between H 2 events to allow for desorption. It can be clearly seen that at 10% H 2 the 3CSiC/Si sensor gave an approximately two-fold greater response than the Si sensor; while at 100% H 2 the 3C-SiC/Si sensor showed approximately 3.5 times the response of the Si sensor. The Si sensor also saturated at about 40% H 2 in Ar, thus not allowing for reliable detection beyond this concentration. Figure 5.4 and Figure 5.5 show the 3C-SiC/Si and n-type Si(001) gas sensor response respect ively to various concentrations of H 2 at various temperatures. Both sensors show a decrease in response, indicating a loss of sensitivity, as temperature increases. An empirical least squares fit to the relation between 3C-SiC/Si gas sensor response and H 2 concentration shown in Figure 5.4 is shown in Table 5.1 Figure 5.3. Gas sensor response at 50C to various concentrations of H 2 (10% to 100% with 10% steps). Solid line denotes 3CSiC/Si response, dashed line Si response. Figure 5.4. 3C-SiC/Si gas sensor response vs. H 2 concentration at various temperatures. The 3C-SiC/Si sensor showed the ability to detect extremely low concentrations of H 2 in Ar, while the Si sensor was limited in its concentration limits. The sensor response at low concentrations (0.333%, 0.5%, 1%, 2%, 3%, 4%, 4.7%, 5%, 7.5% and 10% H 2 in Ar) is shown in Figure 5.6 The total gas flow for these low concentrations was not kept constant (due to mass flow controller limita tions). Thus there is no correlation between gas sensor response amplitude and H 2 concentration. This leads to the conclusion that the adsorption process is mass transfer limited. The 3C-SiC/Si sensor was able to detect H 2 down to 0.333% H 2 in Ar, while the Si sensor was only able to detect down to 2% H 2 in Ar. 26

PAGE 39

Figure 5.5. n-type Si(001) gas sensor response vs. H 2 concentration at various temperatures. Figure 5.6. Gas sensor response at low H 2 concentrations at 50C. Table 5.1. Empirical relationships betw een 3C-SiC/Si sensor response to H 2 concentration at various temperatures. Temperature Empirical fit R 2 Correlation Factor 50C 2 2017.0 0089.02 H HC C I 0.9842 100C 2 20068.0 0028.02 H HC C I 0.9937 150C 2 20027.0 0008.02 H HC C I 0.9990 Note: I [mA] and C H2 [% H 2 in Ar] 5.5 Conclusions The 3C-SiC/Si gas sensor showed to be more sensitive to H 2 concentration than the ntype Si(001) sensor. The empi rical relationships shown in Table 5.1 allow for the determination of H 2 concentration by measuring the current relative to 0% H 2 The 3CSiC/Si gas sensor also showed no H 2 saturation effects allowing for the practical application in real-time H 2 concentration measurements. 5.6 Acknowledgements The work at USF was supported by the DURIN T program administered by the Office of Naval Research (Dr. C. Wood) under Grant N00014-0110715. 27

PAGE 40

Chapter 6: Wide-range (0.33 100%) 3C-SiC resistive hydrogen gas sensor development Timothy J. Fawcett and John T. Wolan Chemical Engineering Department, University of South Florida, Tampa, Florida 33620 Rachael L. Myers, Jeremy Walker, and Stephen E. Saddow Electrical Engineering Department, University of South Florida, Tampa, Florida, 33620 Adapted from: T. J. Fawcett, J. T. Wolan, R. L. Myers, J. Walker, and S. E. Saddow, Wide-range (0.33%-100%) 3C-SiC resistiv e hydrogen gas sensor development, Applied Physics Letters 85 (3), 416-418 (2004). Abstract. Silicon carbide (SiC) resistive hydrogen gas sensors have been fabricated and tested. NiCr planar ohmic c ontacts were deposited on both a 4 m 3C-SiC epitaxial film grown on n-type Si(001) and directly on Si to form the resistive sensor structures. Detection at concentrations as lo w as 0.33% and as high as 100% (H 2 in Ar) was observed with the 3C-SiC sensor while the Si sensor saturated at 40%. The 3C-SiC sensors show a remarkable range of sensitiv ity without any saturation effects typically seen in other solid-state hydroge n gas sensors. Under a cons tant 2 V bias, these sensors demonstrated an increase in current up to 17 mA upon exposure to pure H 2 Preliminary experiments aimed at determining the gas sens ing mechanism of these devices have been conducted and are also reported. 6.1 Introduction Hydrogen gas sensing is of great interest to the fuel cell, automotive, and aerospace industries, mainly for leak detection [1,18]. Th e aerospace and auto motive industries require hydrogen sensors that can operate at high temperatures and under potentially corrosive conditions [6,18]. Silicon based sensors only operate up to 250C because of the relatively small bandgap of Si (1.1 eV) [7 ]. For operation at temperatures greater than 300C and in harsh environments, silicon carbide (SiC) sensors have been employed [38]. The wide bandgap (~2.2 eV for the 3C polytype), high thermal conductivity, and high chemical stability make SiC a gas sensing material of choice for extreme environments [18]. Resistive, or surface conductivity, gas sens ors have mainly found their place in semiconducting oxide-based gas sensors [12] The resistance of these semiconducting oxide materials is dependent on the adsorption of gases on its surface. The surface resistance is lowered by the adsorption of reducing gases and increased upon the adsorption of oxidizing gases in n-type se miconductor oxide sensors, allowing for the presence of reducing or oxidizi ng gases to be easily detected [36,62]. In th is report, the 28

PAGE 41

fabrication and testing of non-oxide-based resistive gas sensors fabricated from SiC will be reported. 6.2 Device fabrication 3C-SiC resistive gas sensors we re fabricated by growing a 4 m thick 3C-SiC thin film on n-type Si(001) by low pressure chemical vapor deposit ion (LPCVD). NiCr planar contacts were deposite d by electron beam evaporation and then annealed for two minutes at 950C in an Ar atmosphere pr oducing a device showing good linear current versus voltage characteristics with a resistance of approximately 13 Identical Si sensors with NiCr contacts were fabricated from the same n-type Si(001) substrate that the 3C-SiC films were grown on to serve as a control, also giving linear current versus voltage characteristics with a resistance of approximately 15 A schematic of the 3CSiC resistive sensor is shown in the insert of Figure 6.1 Both devices were mounted on ceramic sample holders with Ni probes cont acting to the devices. These devices were then simultaneously placed in a quartz t ube for testing under different gaseous atmospheres. A more detailed description of the epitaxial growth, device fabrication, and gas sensor testing apparatus can be found in Ref. [63]. 6.3 Results and discussion The change in resistance, of both the 3C-S iC and n-type Si(001) gas sensors, was monitored by recording the current flowi ng through each sensor wh ile under a constant 2V DC bias. Figure 6.1 shows the sensor response at 50 C (relative current measured with respect to 0% H 2 ) to varying concentrations of H 2 from 10%-100% in 10% increments. It can be clearly seen that the 3C-SiC sensor shows a two-fold increase in response compared to that of the n-type Si( 001) sensor. Both the 3C-SiC and Si sensors reach their respective maximum response within 15 minutes following a change in targetgas concentration. This response time is slow compared to other SiC base d gas sensors (on the order of milliseconds [6]) because the response time of these resistive gas sensors is controlled by the adsorption kinetics, not the device structure. The only parameters that affect the adsorption kinetics are the concentration of the adsorbing species and the temperature of operation. However, the sens ors reported here are capable of sensing hydrogen from below the combustion limit to 100%, whereas other SiC sensors are typically limited to concen trations below 10% [64]. Figure 6.2 shows the relative current versus ti me of the same sensors to various concentrations of H 2 also at 50C. The concentration of H 2 was varied from 10% to 100% in Ar (in 10% increment) without an Ar purge between events to allow the desorption of hydrogen from the surface. Both the 3C-SiC and Si sensors show a reversible response with no need to the allo w hydrogen to desorb from the semiconductor surface. This allows either of these sensor s to be used in an environment where the concentration of H 2 is continuously changing without losing the sensitivity due to any previous environment. Figure 6.2 also shows the Si sensor becoming saturated with H 2 at a concentration ~40% in Ar, rendering the Si sensor useless for detecting H 2 at concentrations above 40%. The 3C-SiC sens or does not show any saturation effects, making it useful for the detection of any H 2 concentration above 0.33% [63]. The specific gas sensing mechanism for this SiC resistive sensor system is still under investigation. One possible mechanism could be a reduction in the surface resistance due 29

PAGE 42

to the molecular adsorption of H 2 onto the semiconductor surface, similar to the mechanism for the detection of reducing gases for the semiconducting oxide based sensors [36,62]. Another possible mechanism could be the catalytic dissociation of H 2 on the surface of the metal contacts. The atom ic H then diffuses through the metal to the metal-semiconductor interface creating a dipole layer at the interface, changing the electrical properties of the device. This mechanism ha s been put forth to explain the detection of H 2 in Schottky barrier diode sensors [38] as well as MOS based sensors [65]. Figure 6.1. Gas sensor response at 50C to H 2 in varying concentrations (10%-100% with 10% increments) with Ar purge between H 2 events [63]. Dashed line represents the n-type Si(001) sensor, and the solid line represents the 3C-SiC sensor. Inset cross-section of 3C-SiC resistive sensor structure. Figure 6.2. Gas sensor response at 50C to H 2 in varying concentrations (10%-100% with 10% increments) without Ar purge between H 2 events. In order to determine whether the response of the SiC resistive sensors is caused by the adsorption of H 2 on the SiC surface or by the catalytic dissociation of H 2 on the metal surface, the following experiment was performe d. New 3C-SiC sensors with ohmic NiCr contacts were fabricated in the same manne r discussed previously. In addition to the NiCr ohmic contact, a 100 Ti buffer la yer followed by an 1800 Au layer was deposited by electron beam evaporation. Th e Ti buffer layer improves the adhesion between the NiCr and Au. It ha s been reported that gold (Au) is inert to the adsorption of H 2 [18,66]. Thus, the Au layer should preven t a change in electrical properties of the NiCr contact as well as at the NiCr-SiC interface upon adsorption of H 2 If this assumption is correct the 3C -SiC sensor response to H 2 will only be caused by the adsorption of H 2 on the SiC surface. In addition to using a Au layer, the area between the contacts was also varied while the contact ar ea was kept constant. If the adsorption of H 2 on the SiC surface is the basis of the gas sens ing mechanism, an incr ease in surface area would be expected to gene rate a larger response. Two 3C-SiC sensors were fabricated w ith 1 mm x 3 mm ohmic NiCr/Ti/Au contacts on the same 3C-SiC on Si sample. One sensor had a 0.5 mm gap (1.5 mm 2 exposed SiC surface) between the contacts whereas the other had a 1 mm gap (3 mm 2 exposed SiC surface). Both of these sensors showed good lin ear current versus voltage characteristics 30

PAGE 43

with resistances of approximately 6.7 (1.5 mm 2 sample) and 6.5 (3 mm 2 sample). As seen in Figure 6.3 the 3C-SiC sensor with the grea ter SiC surface area between the contacts (3 mm 2 ) shows a greater response than that of the sensor with the smaller area (1.5 mm 2 ). Figure 6.3. Gas sensor response at 50C to H 2 in varying concentrations (10%-100% with 10% increments) with Ar purge between H 2 events. Dashed line L = 0.5 mm, solid line L = 1 mm. The ratio of the response of the SiC sensor with a 3 mm 2 space between contacts to that of the sensor with a 1.5 mm 2 space between contacts is expected to be equal to the ratio of the exposed SiC surface area. Doubl ing the surface area fo r gas detection would theoretically double the number of adsorption sites available for H 2 A doubling of the possible adsorption sites, assuming the interaction of the adsorption of H 2 is the same everywhere, should double the response of the sensor. The ratio of the responses shown in Figure 6.3 is 1.49 0.05 which is close to the exp ected value of 2. While this appears to indicate that the sensing mechanism is direct hydrogen adsorp tion on the exposed 3CSiC surface, further experimentation is needed to determine the exact gas sensing mechanism of the 3C-SiC resistive hydrogen gas sensors. 6.4 Summary In summary, 3C-SiC resistive hydrogen gas sensors have been fabricated and shown to be capable of detecting hydrogen at concentrations as low at 0.33% in Ar and as high as 100% without any saturation e ffects seen in other types of gas sensors. These sensors displayed a repeatable response to a continuous change in hydrogen concentration thus indicating that their response is independent of the sensors previous environment. Preliminary experiments aimed at determining the gas sensing mechanism for these resistive sensors have been performed and seem to indicate the mechanism is direct hydrogen adsorption on the exposed SiC surface rath er than catalytic effects at the metal contacts. Continuing investig ation is underway to unlock the specific hydrogen sensing mechanism and to determine the possible impact of gas contaminants on sensor performance. 31

PAGE 44

6.5 Acknowledgements This work was supported by the DURINT pr ogram administered by the Office of Naval Research (Dr. C. Wood) under Grant N00014-0110715. 32

PAGE 45

Chapter 7: Thermal detection mechanism of SiC based resistive hydrogen gas sensors Timothy J. Fawcett and John T. Wolan Chemical Engineering Department, University of South Florida, Tampa, Florida 33620 Anita Lloyd Spetz S-SENCE and Division of Applied Physics, Linkping University, SE-581 83 Linkping, Sweden Meralys Reyes and Stephen E. Saddow Electrical Engineering Department, University of South Florida, Tampa, Florida, 33620 Adapted from: T. J. Fawcett, J. T. Wolan, M. Re yes, A. Lloyd Spetz, and S. E. Saddow, Thermal detection mechanism of 3C-SiC based hydrogen resistive gas sensors, to be submitted to Applied Physics Letters. Abstract: Silicon carbide (SiC) resistive gas sens ors have been fabricated and tested for sensitivity to hydrogen. Planar NiCr contacts were depos ited through SiO 2 windows on a 1.4 m thick 3C-SiC epitaxial film grown on a 15 nm thick Si wafer, which is bonded to a polycrystalline SiC substrate. At 673 K, a resistive gas sensor was shown to detect H 2 at concentrations ranging from 1% to 100% H 2 in an N 2 ambient with up to a 51.75 0.04% decrease in sensor output current from the zero H 2 flow baseline value while operating at a maximum power of < 5 mW. The detection mechanism is shown to be caused by a decrease in the device temp erature of up to 163.1 0.4 K for 100% H 2 This change in device temperature has been shown to be driven by the transfer of heat from the device to the target gas, giving ri se to a thermal detection mechanism. 7.1 Introduction Gas sensors that can operate at high temp eratures and in harsh environments are needed for fuel cell [1], aerospace [16], and automotive [3] applications. These applications often require se nsor operation at temperatures in excess of 923 K and involve corrosive chemical species such as NO x [3,16]. Wide bandgap semiconductor materials with excellent chemical and thermal stability, such as silicon carbide (SiC), are being investigated for use in harsh high temperature environments [3,6,7,19-21]. The wide bandgap (2.36 eV for 3C-SiC [67]) allows these materials to operate at elevated temperatures in comparison to silicon whose band gap is only 1.1 eV [13]. Devices based on SiC for gas sensing have included Schottky barrier diodes [9,19-21] and metal-oxide-semiconductor field-effect transistors (MOSFETs) [3,6,7,38], each of which employ a catalytic metal gate to driv e their detection mechanisms. Recently, resistive device structures have been investigated for their potential use as gas sensing devices [14,68]. Resistive gas sensors posse ss several advantages over other sensor device structures including a simplified device fa brication and no catalytic gate metal that 33

PAGE 46

can be poisoned during operation [38]. Also, resistive sensors have demonstrated the ability to detect a wide range of H 2 concentration, from 0.33% to 100% H 2 in Ar [14,40,68], a range not seen in catalytic gate Schottky diode or MOSFET sensors [1,6,7,20,21]. 7.2 Experimental Resistive gas sensors were fabricated fr om a thin 3C-SiC f ilm grown on a 15 nm thick Si wafer, which is bonded to a polycrystal line SiC substrate, referred to as siliconon-insulator (SOI) in this letter, by a wafer bonding process that is described elsewhere [56]. In this work, a hot-w all SiC chemical vapor deposi tion (CVD) reactor operating at low pressure was used to grow the 3C-SiC fi lm on the SOI substrate. Details of the 3CSiC growth process are described elsewhere [58]. The 3C-SiC epitaxial film was measured to have a thickness of 1.4 m, as determined by FTIR measurement. After CVD growth a 5000 thick SiO 2 layer was deposited on the 3C-SiC surface via plasma enhanced chemical vapor depositi on (PECVD). The sensor active areas were then realized by etching windows in the SiO 2 layer followed by the deposition of NiCr (~2000 ) contacts via e -beam evaporation. These contacts were then rapid thermally annealed for 120 seconds at 1223 K in an Ar ambient to prod uce contacts with ohmic-like behavior. A Ti (200 )/Au ( 4000 ) stack was then sputtered on the annealed NiCr to allow for Au wire bonding. Several devices were fabricated on the same die, with contact dimensions of either 0.5 mm x 0.25 m or 0.5 mm x 2 mm and a gap of 0.25 mm between the contacts. A schematic of the device structure is shown in Figure 7.1 Figure 7.1. 3C-SiC/SOI resistive device schematic. Devices were then mounted on a Pt-100 re sistance temperature detector (RTD) and placed on a ceramic resistive heater. The c ontacts were then wire bonded to a 16-pin gold package described elsewhere [3]. Th e packaged sensors were placed in an aluminum plenum housing with each sensor compartment measuring approximately 25.4 mm x 25.4 x 9.5 mm with ~ 4 mm of space for ga s to flow over each sensor. The Al housing was connected to a gas manifold syst em and the gas sensors were tested in various H 2 /N 2 atmospheres.. The composition and flow rates of the gases (maintained at a total flow of 100 sccm for all experiment s) were digitally controlled via computeroperated mass flow controllers. The resistive gas sensors were tested under a constant dc bias while the current flowing through the sensors was measured by the voltage drop across a precision resistor [69], rate d at either a 1 0.02% or 100 0.005% value, allowing real-time measurement of th e sensor resistance via Ohms law. 34

PAGE 47

7.3 Results and discussion Figure 7.2 shows the current flowing through a resistive gas sensor, with 0.5 mm x 0.25 mm contacts and 0.25 mm gap, in response to H 2 exposure from 10% to 100% in a N 2 ambient. The H 2 concentration was varied in 10% increments at an initial device temperature of 673 K while under a constant 10 V dc bias. In this device, the current decreased as H 2 was added to the N 2 ambient with a decrease of up to 51.75 0.04%, calculated from the steady-stat e portion of the response, fr om the baseline value when pure H 2 was introduced to the gas sensor. Th e baseline is defined as the sensor response to 0% H 2 (i.e., 100% N 2 ambient). The time constant for this device, estimated as a first order exponential decay, was ~ 16 22 sec, with the full response occurring at ~90-120 sec after the initial in troduction of the gas mixture a nd is strongly dependent on the H 2 concentration. This response time is a dramatic improvement from our previous reports, where 420-900 sec to full response wa s not uncommon [14,40,68] This device demonstrated measurable change s in current of ~1 mA, at H 2 concentrations in a N 2 ambient as low as 1% at 673 K. Figure 7.2. Measured cu rrent (left axis) and percent change in current (right axis) flowing through a 3C-SiC/SOI resi stive gas sensor, with 0.5 mm x 0.25 mm contacts and 0.25 mm gap, in response to 10% to 100% H 2 in N 2 in 10% increments at an initial device temperature of 673 K. In a previous letter [14], we discussed th e possibility of the resistive gas sensor response being driven by the adsorption of H 2 on the 3C-SiC surface and its influence on the resistance of the surface. However, new evidence has pointed to thermal effects driving the gas detection. Figure 7.3 shows the temperature of the same device, as measured by the aforementioned RTD in intimate thermal contact with the sensor, while detecting the H 2 pulses whose responses are shown in Figure 7.2 It is clearly seen in Figure 7.3 that the temperature of the device decr eases when the sensor is exposed to H 2 with a temperature decrease of up to 163.1 0.4 K when pure H 2 was introduced. It is well known that the resistivity of a semic onductor is strongly temperature dependent [70]. Thus, the temperature decrease, shown in Figure 7.3 causes a change in the 35

PAGE 48

resistivity of the semiconductor material, as seen in the decreasing current shown in Figure 7.2 Figure 7.3. Measured temperature of a 3CSiC/SOI resistive gas sens or, with 0.5 mm x 0.25 mm contacts and 0.25 mm gap, in response to 10% to 100% H 2 in N 2 in 10% increments at an initial device temperature of 673 K. The temperature of the device is governed by the amount of heat transferred from the device to the gas. The heat generated by Jo ule heating of the resistive gas sensor is transferred to the gas with a magnitude depende nt on the thermal propert ies of gas. This heat transfer can be described by a differe ntial energy balance for a differential unit volume of the device, shown in equation 7.1 [48]: TkVTC tp2 2 (7.1) where is the density, C p is the specific heat, is the electrical conductivity, k is the thermal conductivity of the device, and V and T are voltage and temperature, respectively. The boundary conditions for this partial differential equation are 1) for t = 0 at all boundaries and 2) for t > 0 at the device/heater interface, where q is heat flux. Newtons law of cooling defines the heat fl ux for t > 0 and for all boundaries except the device/heater interface is shown in equation oTT heaterqq 7.2 [48]: surface gas surfaceTThq (7.2) The energy balance in equation 7.1 does not account for the heat transfer characteristics of the gas, but rather only de scribes the flow of heat in th e device. It is the boundary condition in equation 7.2 that describes the magnitude of heat transferred from the device to the gas (and vice versa). The inlet gas temperature ( T gas ) is approximately room temperature (300 K) and is relatively consta nt ( 2 K) throughout these experiments. Thus, the amount of heat transferred to the gas (q surface ) is related to the temperature of the 36

PAGE 49

surface (T surface ) only through the heat transfer coefficient ( h). It is the heat transfer coefficient that changes w ith the composition of the ga s through the specific heat ( C p ), thermal conductivity ( k ), and viscosity ( ) of the gas mixture [48] that is responsible for the sensing of gases through a thermal detection mechanism. The transfer of heat from the device to the gas will occur regardless of which gas and/or gas mixture is flowing over the sensor. However, the magnitude of heat transfer will clearly depend on the thermal properties of the gas. Thus, these sensors actually detect thermal changes relative to a reference gas which is N 2 for these experiments. A decrease in temperature upon the introduction of H 2 as exhibited in Figure 7.3 suggests that H 2 removes more heat from the sensor than N 2 a fact that is well known [71], and supported by the thermal properties of H 2 ( k = 0.291 W/mK C p = 29.73 J/molK) relative to N 2 ( k = 0.0041 W/mK, C p = 29.75 J/molK) at 600 K [47]. Figure 7.4 shows the mean resistan ce (as calculated from the measured instantaneous sensor voltage and current) of a resistive gas sensor, with 0.5 mm x 2 mm contacts and 0.25 mm gap, as a function of H 2 concentration at 673 K for several gas sensor voltages (2.5 V, 5 V, and 10 V). Increa sing gas sensor voltage increa ses the Joule heating effects (i.e., the 2V term from equation 7.1 ) so that more heat is transferred to the gas. With more heat transferred to the gas, the more the de vice surface temperature ( T surface ) will decrease, also shown in Figure 7.3 with a constant heat transf er coefficient, as governed by the boundary condition in equation 7.2 This increased change in temperature results in an increased change in resistance, as demonstrated in Figure 7.4 Thus, the higher the Joule heating, the higher the gas sensor response (i.e., change in resistance). Figure 7.4. Calculated re sistance (left axis) and measured temperature (right axis) vs. H 2 concentration (N 2 ambient) for a 3C-SiC/SOI sensor, with 0.5 mm x 2 mm contacts and 0.25 mm gap, at an initial device temperature of 673 K for bias voltages of 2.5, 5, and 10 V DC. These 3C-SiC resistive gas sensors can be compared to thermistor devices, made primarily from metal oxides such as BaTiO 3 [72] and SrTiO 3 [73] whose highly temperature dependent resistance is used as a temperature transducer for a variety of applications. Thermistor devices have also be en used to indirectly measure heat transfer 37

PAGE 50

coefficients for convective heat transfer [74] similar to those used in the measurements presented in this letter. Th ermal conductivity sensors fabricated from materials such as Pt [29] and poly-Si [32] have also been us ed to measure the thermal conductivity of gas mixtures and for gas detection. 7.4 Summary In summary, a 3C-SiC/SOI resistive sensor has been fabricated and shown to be capable of detecting H 2 over a wide range of concentrations from 10% to 100% H 2 in N 2 and at a relatively high temperature of 673 K. Initial invest igations proposed the dissociation of H 2 on the exposed SiC surface, givi ng rise to a cha nge in surface conductivity, as the gas sens ing mechanism [14,40,68]. In contrast to this surface conductivity proposition, this le tter provides evidence that th e gas sensing mechanism is most likely driven by thermal effects. With the addition of a temp erature sensor to the sensor package, the temperature of the se nsor was examined while the sensor was exposed to H 2 A large change in temperature (up to ~163 K) was observed upon the introduction of pure H 2 relative to that in pure N 2 This large change in temperature is driven by the transfer of heat from the device (heat that is provided by both the heater and the Joule heating of the resistiv e sensor) to the gas. Since the resistivity of semiconductor materials is largely dependent on temperature, even small changes in temperature manifest themselves as a significant change in resistivity. This change can be measured electronically and correlated to the concentrat ion of the target gas( es) via the proposed thermal detection mechanism for the 3C-SiC/SOI resistive sensor presented in this study. 7.5 Acknowledgements This work was supported by the DURINT pr ogram administered by the Office of Naval Research (Dr. C. Wood) under Grant N00014-0110715. 38

PAGE 51

Chapter 8: Thermal detection mechanism of 3C-SiC resistive gas sensors (unpublished data to be submitted to the Journal of Applied Physics) 8.1 Experimental 8.1.1 Device fabrication In this study, a thin 3C-SiC film wa s deposited on a silicon-on-insulator (SOI) substrate, diced into a 1 cm x 1 cm die. The SOI substrate consisted of a thin Si (001) layer (~ 150 ) miscut 4 towards the [211] direction, bonded to a polycrystalline SiC substrate using a wafer bonding process describe d elsewhere [56]. The SOI substrate was RCA cleaned prior to depositi on to remove trace organic and metal species from the surface [55]. The 3C-SiC deposition was ca rried out in a low-pressure, hot-wall, horizontal chemical vapor deposition (CVD) reactor using the traditional dual precursor gas chemistry, propane (C 3 H 8 ) and silane (SiH 4 ) [58]. Hydrogen chloride (HCl) was also incorporated into the gas steam as a grow th promoter. During the CVD process a gas stream containing C 3 H 8 and H 2 was established in the chamber while the sample temperature was ramped from room temperature to the growth temperature of 1360C. Once at the growth temperature, the pro cess pressure was lowered to 200 Torr after which SiH 4 and HCl were added to the gas stream for growth. The Si/C and Si/Cl ratios during growth were set to 0.7 and 3.3, respectively. The 3C-SiC epitaxial film was determined to be ~ 1.4 m thick measured via FTIR. Following deposition of the 3C-SiC epitaxi al layer, the sample was RCA cleaned, after which ~ 5000 of silicon dioxide (SiO 2 ) was deposited on the surface by plasmaenhanced CVD (PECVD) using SiH 4 and nitrogen dioxide (NO 2 ) precursors. Shipley 1813 positive photoresist was used to pattern the SiO 2 layer, to define the active sensor areas, which was subsequently removed using a 10:1 buffered oxide et ch (BOE) solution. Next, Futurrex Inc. NR9-1500PY negative photore sist was used to pattern the sample to define the ohmic contact placement. The contacts, ~ 2000 of NiCr (80:20), were deposited via e -beam evaporation, after which the negative photoresist and undesired NiCr was removed using a standard lift-off pro cess. The sample was rinsed in de-ionized (DI) water, dried in nitrogen (N 2 ) and then placed in a rapid thermal processing (RTP) unit to anneal the contacts at 950C for 2 minutes in an Ar ambient to produce ohmic contacts. After annealing, th e NiCr contacts were verifi ed to be ohmic, producing approximately linear current vs voltage (I-V) characterist ics, as demonstrated in Figure 8.1 Next, another negative photores ist step was performed to pattern the same contact areas previously described; a Ti (200 ) / Au (4000 ) stack was sputtered on the existing ohmic contacts to allow for Au wire bonding. The remaining negative photoresist and unwanted Ti/Au was removed using a standard lift-off process. The 1 cm x 1 cm sample was then diced into sma ller pieces (~ 3mm x 6 mm) with each die 39

PAGE 52

mounted on a resistance temperature detector (RTD) and placed on a ceramic heater with the resistive gas sensor contacts wire bonde d to a 16-pin package described elsewhere [3]. A cross sectional drawing of a typical device is shown in Figure 8.2 a and a photograph of an actual packaged gas sensor with three devices is shown in Figure 8.2 b. Figure 8.1. Current versus voltage (I-V) characteristics of typical 3C-SiC/SOI resistive gas sensor with 0.5 mm x 1 mm contacts and a 1 mm gap. Figure 8.2. (a) Cross sectional diagram of 3C-SiC/SOI resistive gas sensor and (b) photograph of packaged sensor. 8.1.2 Gas sensor test bed A schematic drawing of the experimental gas sensor test bed is shown in Figure 8.3 After device fabrication and packaging, th e resistive gas sensors were placed in an aluminum (Al) housing connected to a gas manifold system and tested. The composition of the detection gas was cont rolled by the six mass flow controllers (MFCs) while the timing of the gas mixtures was controlled by pneumatic valves. The resistive gas sensors were tested under a constant dc bias while the current flowing through the sensors was measured by the voltage drop across a preci sion resistor, either 1 0.02% or 100 0.005% allowing the real-time measurement of sensor resistance via Ohms law. Ceramic heater and gas sensor bias voltages as well as gas sensor voltage, current and temperature measurements were controlled and collected by an 18-bit multifunction 40

PAGE 53

analog data acquisition card, PCI-6281 from National Instruments, at a sampling frequency of 10 Hz. Figure 8.3. Schematic of the experimental gas sensor test bed. 8.2 Results and discussion 8.2.1 Experimental results The current flowing through a 3C-SiC/SOI resistive gas sensor, with 0.5 mm x 1 mm contacts and a 1 mm gap, in response to 10% to 100% in 10% increments of H 2 in N 2 with an initial device temperature of 400C and a 10 Vdc bias is shown in Figure 8.4 The decreases in current shown in Figure 8.4 range from 11.1 0.3 mA or 26.8% (10% H 2 in N 2 ) to 34.9 0.2 mA or 84% (100% H 2 ) with time constants, estimated as a first order exponential decay, ranging from ~16-40 s econds. With an RTD in thermal contact with the device, as described in section 8.1.1 the temperature of the device was able to be observed while detecting H 2 As demonstrated in Figure 8.5 the temperature of the device decreases as H 2 is introduced with magnitudes ranging from 35.9 0.7C (10% H 2 in N 2 ) to 184.1 0.4 C (100% H 2 ). As described in section 1.2.2.3 a semiconductors resistivity is dependent on temperature. Thus, the temperature changes demonstrated in Figure 8.5 cause the devices resistivity to change. Since resist ance is proportional to resistivity with the proportionality constant being related to the device geometry (as shown in equation 8.1 ) any changes in resistivity will manifest as changes in resist ance for any device with set geometry with cross-sectional area ( A ) and gap length ( L ). A L R (8.1) 41

PAGE 54

Figure 8.4. Current (left axis) and percent change in current (r ight axis) flowing through a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to various con centrations (10% to 100% in 10% increments) of H 2 in N 2 at an initial device temperature of 400C and a bias of 10 V DC. Figure 8.5. Temperature of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to various concentrations (10% to 100% in 10% increments) of H 2 in N 2 at an initial device temperature of 400C and a bias of 10 V DC. The resistance of several 3C-SiC/SOI devices with 0.5 mm x 1 mm contacts, giving a cross-sectional area of ~ 1.4 m x 1 mm ~ 1.4 -9 m 2 and gap lengths of 0.25 mm, 0.5 mm, and 1 mm is shown in Figure 8.6 as a function of temperature. For all these 3CSiC/SOI devices, the resistance decr eases as temperature increases. Figure 8.6. Resistance of several 3C-SiC/SOI sensors, with 0.5 mm x 1 mm contacts and various gap lengths, as a function of temperature. Therefore, the temperature decreases shown in Figure 8.5 cause the device resistance to increase in accordance with Figure 8.6 These resistance increases are recorded as decreases in current at a fixed voltage, as demonstrated in Figure 8.4 42

PAGE 55

The mean current as a function of H 2 concentration for a 3C-SiC/SOI sensor with 0.5 mm x 1 mm contacts with a 1 mm gap for several initial device temperatures, 100C, 200C, 300C and 400C, with a 10 V DC bias is shown in Figure 8.7 It is clear from Figure 8.7 that this sensor operating at 400C gives the largest change in current, I = 34.9 0.2 mA at 100% H 2 as already presented in Figure 8.4 It is more appropriate to view the data presented in Figure 8.7 as the percent change in current relative to the baseline (0% H 2 or 100% N 2 ), allowing the response to H 2 at the four different temperatures to be compared on the same scale. The percent change in current in response to 10% to 100% in 10% increm ents of H 2 in N 2 of a 3C-SiC/SOI sensor with 0.5 mm x 1 mm contacts and a gap of 1 mm biased at 10 Vdc is shown in Figure 8.8 It is clear from Figure 8.8 that operating at 400C gives the largest percent cha nge in current with ~ 84 % as previously stated. At H 2 concentrations greater than ~ 43.25%, the percent change in current increases with increasing temperature. However, at H 2 concentrations less than 43.25% the percent change in cu rrent for 200C operation is greater than that for 300C operation. The reason for this anomaly is not known and varies as the gap length changes wi th constant contact dimensions. Figure 8.7. Current as a function of H 2 concentration (in N 2 ) of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperatures with a bias of 10 V DC. Figure 8.8. Percent change in current of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperatures with a bias of 10 V DC. The gas sensor resistance is related to the measured current flowing through the sensor at a fixed voltage by Ohms law, R = V/I. Figure 8.9 shows resistance as a function of H 2 concentration for a 3C-SiC/SOI sens or, with 0.5 mm x 1 mm contacts with a 1 mm gap, biased at 10 V DC at several te mperatures. These resistances correlate to the currents presented in Figure 8.7 with 400C having the greatest change in resistance, but lowest magnitude. The percent increase in resistance for the same responses shown in Figure 8.9 is shown in Figure 8.10 with up to an ~ 525% increase in resistance at 400C in response to pure H 2 43

PAGE 56

Figure 8.9. Resistance as a function of H 2 concentration (in N 2 ) of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperatures with a bias of 10 V DC. Figure 8.10. Percent change in resistance of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at several initial device temperatures with a bias of 10 V DC. Figure 8.11 shows the mean resist ance as a function of H 2 concentration of a 3CSiC/SOI sensor, with 0.5 mm x 1 mm contact s with a 1 mm gap, at an initial device temperature of 400C with the sensor biased at several voltages, 2.5V, 5V, and 10V. The change in resistance increases with incr easing bias voltage, as demonstrated in Figure 8.11 Increasing gas sensor voltage give increasing Joule heating effects ( 2V term from equation 1.1 ) and the ability for more heat to be transferred from the device to the gas. With increased heat transfer comes an increased temperature change for a set heat transfer coefficient, yielding an increasing gas sensor response (change in resistance) as the bias voltage increases. Figure 8.11. Resistance as a function of H 2 concentration (in N 2 ) of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, at an initial device te mperature of 400C with sensor biased at several voltages. 44

PAGE 57

To test the use of the thermal detecti on mechanism to explain the gas sensor performance described above, the temperatur e of the device was controlled using a software proportional-integral-derivative (PID) controller. The temperature was measured using the RTD and controlled us ing the heater, both described in section 2.2.7 With temperature held constant by the PID c ontroller, the resistan ce of the device should be constant giving a constant current at a fixed voltage. Figure 8.12 shows the response of a 3C-SiC/SOI sensor, with 0.5 mm x 1 mm contacts and a 1 mm gap, to 10% to 100% in 10% increments of H 2 in N 2 with the device temperatur e controlled at 400C. Note, the spikes in Figure 8.12 are caused by the PID controllers compensation for the sudden changes in temperature. Even with te mperature controlled at 400C, the current flowing through the 3C-SiC/SOI sensor is not constant. The current shown in Figure 8.12 increases when H 2 in introduced to the sensor, oppos ite of when temperature is not controlled. Figure 8.12. Current flowing through a 3CSiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% H 2 (in N 2 ) in 10% increments with device temperature controlled at 400C and a bias of 10 V DC. Dotted line = H 2 on, dotted line with circle = H 2 off. The response to H 2 even with the temperature of the device held constant (shown in Figure 8.12 ) suggests the thermal detection mech anism may not be the only phenomena driving these resistive gas se nsor. For example, the adsorption detection mechanism presented in section 1.2.1 may play some part in the overall detection mechanism. However, the sensor was also tested for its response to He diluted in Ar at 100C, shown in Figure 8.13 Both He and Ar are monatomic nobl e gases and should react with the gas sensor surface in the same manner. However, He and Ar posses drastically different thermal properties, k = 0.0223 W/mK and Cp = 4.97 cal/molK for Ar and k = 0.180 W/mK and Cp = 4.97 cal/molK for He at 400 K [47] With the thermal conductivity of He being more than 8 times that of Ar, He can tr ansfer heat at a much faster rate than Ar. Thus, when the concentration of He in a He/Ar mixture increases the heat transfer coefficient of that mixture increases. An in creasing heat transfer coefficient lowers the 45

PAGE 58

surface temperature for the same surface heat flux through Netwons law of cooling, surface gas surfaceTThq in equation 1.4 Thus the changes in current seen in Figure 8.12 and Figure 8.13 are still the result of heat transfer Heat is still transferred from the device to the gas, producing a s light temperature gradient in the device, which manifests as changes current. Experiments testing the sensor response to CO 2 and CH 4 in N 2 while the temperature was controlled were also performed. The device temperature (resistance) decreased (increased) with the exposure of CH 4 and increased (decreased) with the exposure of CO 2 These results are expected with regard to the thermal detection mechanism presented in section 1.2.2 The thermal conductivity of CH 4 (0.0277 W/mK) is greater than that of N 2 (0.0222 W/mK), allowing more heat to be transferred from the device to the gas when CH 4 is introduced, causing the device temperature to decrease and resistance to increase. Howeve r, the thermal conductivity of CO 2 (0.0129 W/mK) is less than that of N 2 allowing less heat transferred from the device to the gas, causing the device temperature to increase and resistance to decrease. To completely eliminate heat transfer, the gas must be preheated to exactly the same temperature as the device, a topic addressed in section 9.2.1 Figure 8.13. Current flowing through a 3CSiC/SOI sensor, with 0.5 mm x 1 mm contacts with a 1 mm gap, in response to 10% to 100% He (in Ar) in 10% increments with device temperature controlled at 100C and a bias of 10 V DC. 8.2.2 Finite element method simulations Finite element method (FEM) simulations using Femlab 3.2 from COMSOL [75], were performed to model the thermal and elect rical characteristics of a simplified 3C-SiC resistive gas sensor in response to changing H 2 concentrations in N 2 at various initial device temperatures. The details of these simulations are discussed in Appendix C Figure 8.14 shows the device temperature as a function of H 2 concentration (diluted in N 2 ) at several initial device temperatures with the sensor biased at 10V DC. As shown in Figure 8.14 the magnitude of the decrease in device temperature increases with increasing initial device temperature. 46

PAGE 59

Figure 8.14. Device temperature as a function of H 2 concentration (in N 2 ) at several initial device temperatures. Device biased at 10V DC. Figure 8.15 shows the simulated resistivity as a function of H 2 concentration (in N 2 ) at several initial device temper ature with the sensor biased at 10V DC. These resistivity calculations agree with the experimental resistance data presented in Figure 8.9 Figure 8.16 shows the percent change in resistivity as a function of H 2 concentration (in N 2 ) at several initial device temperatures. The percent change in re sistivity, shown in Figure 8.16 increases with increasing initial device te mperature which is in agreement with the experimental result presented in Figure 8.10 Figure 8.15. Resistivity as a function of H 2 concentration (in N 2 ) at several initial device temperatures. Device biased at 10V DC. Figure 8.16. Percent cha nge in resistivity as a function of H 2 concentration (in N 2 ) at several initial device temperatures. Device biased at 10V DC. 47

PAGE 60

Chapter 9: Conclusions and future work 9.1 Conclusions The detection mechanism driving 3C-S iC resistive hydrogen gas sensors was investigated in this work. Two hypotheses were develope d to explain the change in current at a fixed voltage of a resistive device built on a 3C-SiC upon the introduction of different gases to the sensor. The first hypothesis is based on the adsorption of gaseous species on the 3C-SiC surface and subsequent alteration of the semiconductors overall conductivity as described in section 1.2.1 Initial experimental data from 3C-SiC/Si resistive sensors, presented in Chapters 5 and 6 suggested the surface adsorption detection mechanism to be responsible for the response of the gas sensors to H 2 As the gap length was doubled, with c onstant contact dimensions, th e change in current due to various concentrations of H 2 was ~ 1.5 times greater for the device with twice the exposed 3C-SiC surface area. Doubling the surface area doubles the opportunity for H 2 to adsorb on the surface and influence the devi ces resistance [14]. However, the device temperature was not monitore d during these experiments. The second hypothesis is based on the transfer of heat from the sensor to the gas, causing the temperature of the device to change with changing gas composition as described in section 1.2.2 and termed the thermal detection mechanism. Since a semiconductors resistivity is highly temperature dependent, see section 1.2.2.3 any changes in device temperature will manifest a change in resistivity. To experimentally test this hypothesis, a different sensor packaging scheme was employed. The sensor was attached to an RTD (to measure device temper ature) and placed on a ceramic heater (to control the device temperature) then placed in a 16-pin Au package (as described in section 2.2.7 ). As demonstrated in Chapters 7 and 8 the temperature of the device dramatically changes upon the introduction of a different ga s mixture to the sensor. A temperature change of up to ~ 184C upon the introduction of 100% H 2 relative to 100% N 2 was demonstrated with a 3C -SiC/SOI device operating at an initial temperature of 400C. This large decrease in temperature caused the 3C-SiC resistivity to increase, resulting in an increase in the observed resistance. To further test the thermal detection mechanism, the device was operated at various voltages, 2.5 V, 5 V and 10 V, in order to explore the influence the amount of heat generated by the device has on the device perfor mance. Increasing the gas sensor voltage increases the magnitude of the Joule heati ng term in the energy balance in equation 1.1 allowing more heat to be transferred to the gas at a given initial de vice temperature. The transfer of more heat from the device to th e gas causes the device temperature to decrease due to Newtons law of cooling for a set heat transfer coefficient (i.e. constant gas composition). The experimental data for two different 3C-SiC/SOI sensors, presented in Figure 7.4 and Figure 8.11 show the change in resistance increasing with increasing bias voltage, providing more evidence to support the thermal detection mechanism hypothesis. 48

PAGE 61

If all temperature changes were eliminated while detecting gases with different heat transfer coefficients, the sens or response should be eliminat ed if the thermal detection mechanism was solely response for gas detectio n. However, changes in current at a fixed voltage still occur when H 2 is introduced even when the device temperature is held at a fixed value but in the opposite direction than when temperature is not controlled (shown in Figure 8.12 ). The evidence presented in Figure 8.12 possibly suggest another detection mechanism in addition to the ther mal detection mechanism being responsible for the changes in current. However, the same pattern is seen when the same sensor is exposed to He diluted in Ar (shown in Figure 8.13 ). Both He and Ar have similar chemical reaction characteristics and should influence the device resistance in the same manner if surface adsorption is taking place. Yet, the response to He in Ar ( Figure 8.13 ) follows the same pattern as the response to H 2 in N 2 ( Figure 8.12 ). Moreover, He and H 2 have similar heat transfer characteristics while Ar and N 2 also have similar heat transfer characteristics, see Table 1.1 Thus, the thermal effects of detecting He in Ar are approximately the same as those when detecting H 2 in N 2 Also, heat transfer still occurs even while the temperature of the device is held constant by a PID controller since the gas enters the sensor housing at approximately room temperature. This small amount of heat transfer is most likely responsible for the small change s in current found in Figure 8.12 and Figure 8.13 Thus, the evidence presented in Chapters 7 and 8 supports that gas sensing is being driven by the thermal detecti on mechanism. The large chan ges in temperature upon the introduction of a gas with different heat transfer characteristics than the reference gas cause corresponding change s in the device resistance. In effect, any gas mixture made up of gases with different thermal properties wi ll produce a response in 3C-SiC resistive gas sensors. Thus, instead of a H 2 specific gas sensor, these 3C -SiC sensors can detect any gases that have different therma l properties. For example, H 2 can be detected in any gas other than He since the thermal propertie s of He are very close to that of H 2 CO 2 can also be detected in N 2 or H 2 since the thermal properties of CO 2 are different than those of both N 2 and H 2 These changes in resistance are measured as changes in current at a fixed voltage. The influence of the gas sensor bias voltage on the magnitude of the change in resistance, due to the Joule heatings dependence on volta ge, also supports the thermal detection mechanism being responsible for the gas se nsing. Although small changes in current where demonstrated even with the temperatur e of the device held constant, the thermal detection mechanism is still the mostly likel y explanation for the detection mechanism of wide bandgap resistive gas sensors. Finite element method (FEM) simu lations, presented in section 8.2.2 have confirmed many of the experimental fi ndings presented in Chapters 7 and 8 The device temperature was shown to decrease with increasing H 2 concentration in N 2 The magnitude of the decrease in device temperature increased with increasing initial device temperature. Both of these simulated re sults agree with the experimental trends presented in Chapters 7 and 8 Thus, the experimental and theoretical evidence presented in this work support the thermal detection m echanism as being the main driving force for the detection of H 2 by 3C-SiC/SOI resi stive gas sensors. 49

PAGE 62

9.2 Future work 9.2.1 Further investigation into the thermal detection mechanism for resistive sensors Preheating the gas inlet to the same temper ature as the resistiv e gas sensor will eliminate any temperature gradients in the syst em and therefore eliminate any transfer of heat from the device to the gas. Thus, preh eating the gas should essentially eliminate the possibility of gas detection vi a thermal effects. If a gas sensor response still arises upon the introduction of a gas or gas mixture with different thermal properties, the thermal detection mechanism is not the only mechanism responsible for the response of these 3CSiC resistive gas sensors. Varying the total gas flow over the sensor can also give further proof to the thermal detection mechanism. Changing the total gas flow will also change the heat transfer coefficient, through the Reynolds nu mber as discussed in section C.4 For example, if adjusting the total flow of pure N 2 to obtain the same heat transfer coefficient as pure H 2 yields the same sensor response, the therma l detection mechanism is the main driving force for H 2 detection. 9.2.2 Improved gas sensor performance Optimization of the gas sensor device struct ure has yet to be ascertained. The goal of this work was to understand the detection mechanism using different device geometries. With the gas sensor detection mechanism mo st likely being thermal in nature, device structures can be optimized. For example, as stated in section 1.2.2.3 lightly doped SiC yields a much larger temperature coefficient of resistance compared to highly doped SiC. Thus, a more sensitive sensor can be realized with using a very lightly doped substrate. Factors such as doping and devi ce dimensions should be fully characterized to determine the optimum sensor parameters. 9.2.3 Test resistive sensor se nsitivity to other gases The sensitivity of 3C-SiC re sistive sensors to other gases must also be thoroughly explored. This report investigated the sens itivity of 3C-SiC resistive sensors to H 2 in Ar and N 2 in order to ascertain the detection mechanism. With the thermal detection mechanism driving the gas sensor response, other gas mixtures should be able to be detected as long as differences in thermal properties exists. 9.2.4 On device temperature measurement The device temperature for the second ge neration devices described in section 2.2 is measured by a RTD located direction beneath the device. The device and RTD are in intimate thermal contact with a high ther mal conductivity epoxy connecting them. This temperature measurement can be improved if the temperature measuring structure was either directly on the device, without the epoxy layer, or in the device. The deposition of a standard Pt film, with know n resistance and temperature coefficient of resistance, directly on the device surface w ill give a better indi cation of temperature. With the Pt RTD directly on the device the epoxy layer is eliminated, giving a better indication of temperature. 50

PAGE 63

References [1] S. Nakagomi, K. Okuda, and Y. Kokubun, Electrical properties dependent on H 2 gas for new structure diode of Pt-thin WO 3 -SiC Sensors and Actuators B 96, 364371 (2003). [2] S. C. Pyke and L. Sadwick, Gallium nitride integrated gas/temperature sensors for fuel cell system monitoring for hydrogen and carbon monoxide Proceedings of the 2001 ODE Hydrogen Program Review, NREL/CP-570-30535, Baltimore, MD, unpublished (2001). [3] H. Wingbrant, H. Svenningstorp, P. Salomonsson, P. Tengstrom, I. Lundstrom, and A. L. Spetz, Using a MISiCFET device as a cold start sensor Sensors and Actuators B 93, 295-303 (2003). [4] A. Mandelis and C. Christofides, Physics, chemistry and t echnology of solid state gas sensor devices (John Wiley & Sons, Inc., New York, 1993). [5] W. Qu, R. Green, and M. Austin, Development of multi-f unctional sensors in thick-film and thin-film technology Meas. Sci. Technol. 11, 1111-1118 (2000). [6] A. L. Spetz, P. Tobias, A. Baranzahi, P. Martensson, and I. Lundstrom, Current status of silicon carbide based high-temperature gas sensors IEEE Transactions on Electron Devices 46 (3), 561-566 (1999). [7] A. L. Spetz, P. Tobias, L. Uneus, H. Svenningstorp, L.-G. Ekedahl, and I. Lundstrom, High temperature catalytic metal fiel d effect transistors for industrial applications Sensors and Actuators B 70, 67-76 (2000). [8] A. L. Spetz, A. Baranzahi, P. Tobias, and I. Lundstrom, High temperature sensors based on metal-insulator-silicon carbide devices phys. stat. sol. (a) 162, 493-511 (1997). [9] J. Kim, B. P. Gila, C. R. Abernath y, G. Y. Chung, F. Ren, and S. J. Pearton, Comparison of Pt/GaN and Pt/4H-SiC gas sensors Solid State Electronics 47 (9), 1487-1490 (2003). [10] I. Lundstrom, S. Shivaram an, C. Svensson, and L. Lundkvist, A hydrogensensitive MOS field-effect transistor Applied Physics Letters 26 (2), 55-57 (1975). 51

PAGE 64

[11] A. Gurlo, N. Brsan, M. Ivanovskaya, U. Weimar, and W. Gpel, In 2 O 3 and MoO 3 In 2 O 3 thin film semiconductor sensors: interaction with NO 2 and O 3 Sensors and Actuators B 47 (1-3), 92-99 (1998). [12] J. Mizsei, H 2 -induced surface and interface po tentials on Pd-activated SnO 2 sensor film Sensors and Actuators B 28, 129-133 (1995). [13] P. G. Neudeck, R. S. Okojie, and L.-Y. Chen, High temperature electronics A role for wide bandgap semiconductors? Proceedings of the IEEE 90 (6), 1065176 (2002). [14] T. J. Fawcett, J. T. Wolan, R. L. Myers, J. Walker, and S. E. Saddow, Wide-range (0.33%-100%) 3C-SiC resistive hydrogen gas sensor development Applied Physics Letters 85 (3), 416-418 (2004). [15] T. J. Fawcett, J. T. Wolan, A. L. Spetz, M. Reyes, and S. E. Saddow, Thermal detection mechanism of SiC based hydrogen resistive gas sensors submitted to Applied Physics Letters. [16] G. W. Hunter, L.-Y. Chen, P. G. Ne udeck, D. Knight, C.-C. Liu, Q.-H. Wu, and H.-J. Zhou, Chemical gas sensors for aeronautic and space applications NASA Technical Memo 107444 (1997). [17] G. W. Hunter, P. G. Neudeck, L. Chen, D. Knight, C. C. Lui, and Q. H. Wu, Silicon carbide-based hydrogen and hydrocarbon gas detection NASA Tech. Memo. 107064 (1995). [18] G. W. Hunter, P. G. Neudeck, G. D. Jefferson, and G. C. Madzar, The development of hydrogen sensor technol ogy at NASA Lewis research center NASA Tech. Memo. 106141 (1992). [19] C. K. Kim, J. H. Lee, S. M. Choi, I. H. Noh, H. R. Kim, N. I. Choi, C. Hong, and G. E. Jang, Pdand Pt-SiC Schottky diodes for detection of H 2 and CH 4 at high temperatures Sensors and Actuators B 77, 455-462 (2001). [20] S. Roy, C. Jacob, and S. Basu, Studies on Pd/3C-SiC Schottky junction hydrogen sensors at high temperature Sensors and Actuators B 94, 298-303 (2003). [21] S. Roy, C. Jacob, C. Lang, and S. Basu, Studies on Ru/3C-SiC Schottky junctions for high temperature hydrogen sensors J. Electrochemical Society 150 (6), H135H139 (2003). [22] C. I. Muntele, D. Ila, E. K. Williams, I. C. Muntele, A. L. Evelyn, D. B. Poker, and D. K. Hensley, Improved sensitivity SiC hydrogen sensor Mat. Res. Soc. Symp. 622, 661-666 (2000). 52

PAGE 65

[23] A. L. Spetz, S. Nakagomi, and S. Savage, in Advances in silicon carbide processing and applications edited by S. E. Saddow and A. Agrawal (Artech House, Boston, 2004). [24] P. Tobias, A. Baranzahi, A. L. Spetz, O. Kordina, E. Janzen, and I. Lundstrom, Fast chemical sensing with metal-insulator silicon carbide structures IEEE Electron Device Letters 18 (6), 287-289 (1997). [25] H. Wingbrant, I. Lundstrom, and A. L. Spetz, The speed of response of MISiCFET Sensors and Actuators B 93, 286-294 (2003). [26] C. G. Fox and J. F. Adler, in Techniques and mechanisms in gas sensing edited by P. T. Mosley, J. O. W. Norris, and D. E. Williams (Adam Higler, New York, 1991), p. 325-346. [27] M. I. Lundstrom, M. S. Shivaraman, and C. M. Svensson, A hydrogen-sensitive Pd-gate MOS-transistor Journal of Applied P hysics 46 (9), 3876-3881 (1975). [28] G. Pollak-Diener and E. Obermeier, Heat-conduction microsensor based on silicon technology for the analysis of twoand three-component gas mixtures Sensors and Actuators B 13-14 (1-3), 345-347 (1993). [29] P. Tardy, J.-R. Coulon, C. Lucat, and F. Menil, Dynamic thermal conductivity sensor for gas detection Sensors and Actuators B 98 (1), 63-68 (2004). [30] I. Simon and M. Arndt, Thermal and gas-sensing prope rties of a micromachined thermal conductivity sensor for the de tection of hydrogen in automotive applications Sensors and Actuators A 97-98 (1), 104-108 (2002). [31] M. Kimura, J. Manaka, S. Satoh, S. Takano, N. Igarashi, and K. Nagai, Application of the air-bridge microheater to gas detection, Sensors and Actuators B 24-25 (1-3), 857-860 (1995). [32] R. P. Manginell, J. H. Smith, and A. J. Ricco, An overview of micromachined platforms for thermal sensing and gas detection SPIE Symposium on Smart Structures and Materials 3046, 273-284 (1997). [33] S. Sorge and T. Penchstein, Hully integrated thermal conductivity sensor for gas chromatography without dead volume, Sensors and Actuators A 63 (3), 191-195 (1997). [34] A. J. Crocker, in Techniques and mechanisms in gas sensing edited by P. T. Mosley, J. O. W. Norris, and D. E. Williams (Adam Higler, New York, 1991), p. 281-323. [35] B. P. Luther, S. D. Wolter, and S. E. Mohney, High temperature Pt Schottky diode gas sensors on n-type GaN Sensors and Actuators B 56 (1-2), 164-168 (1999). 53

PAGE 66

[36] D. E. Williams, Semiconducting oxides as gas-sensitive resistors Sensors and Actuators B 57, 1-16 (1999). [37] L. M. Lechuga, A. Calle D. Golmayo, and F. Briones, Hydrogen sensor based on a Pt/GaAs Schottky diodes Sensors and Actuators B 4, 515-518 (1991). [38] J. Schalwig, G. Muller, U. Karren, M. Eickhoff, O. Ambacher, M. Stutzmann, L. Gorgens, and G. Dollinger, Hydrogen response mechani sm of Pt-GaN Schottky diodes, Applied Physics Letters 80 (7), 1222-1224 (2002). [39] S. Pitcher, J. A. Thie le, H. Ren, and J. F. Vetelino, Current/voltage characteristics of a semiconductor metal oxide gas sensor Sensors and Actuators B 93, 454-462 (2003). [40] T. J. Fawcett, J. T. Wolan, R. L. Myers, J. Walker, and S. E. Saddow, Hydrogen gas sensors using 3C-SiC/Si epitaxial layers Materials Science Forum 457-460, 1499-1502 (2004). [41] P. T. Mosley and D. E. Williams, in Techniques and mechanisms in gas sensing edited by P. T. Mosley, J. O. W. Norris, and D. E. Williams (Adam Higler, New York, 1991), p. 46-60. [42] W. H. Brattain and J. Bardeen, Surface properties of germanium The Bell System Technical Journal 32, 1-41 (1953). [43] R. I. Masel, Principles of adsorption and reaction on solid surfaces (John Wiley & Sons, New York, 1996). [44] V. Derycke, P. Fonteneau, N. P. Pham, and P. Soukiassian, Molecular-hydrogen interaction with -SiC(100)3x2 and c(4x2) surfa ces and with Si atomic lines Physical Review B 63, 201305 (2001). [45] V. Derycke, P. G. Soukiassian, F. Am y, Y. J. Chabal, M. D. D'Angelo, H. B. Enriquez, and M. G. Silly, Nanochemistry at the atomic scale revealed in hydrogen-induced semiconductor surface metallization Nature Materials 2, 253258 (2003). [46] R. D. Felice, C. M. Bertoni, C. A. Pignedoli, and A. Catellani, Hydrogen-induced surface metallization of -SiC(100)-(3x2) revisited by density functional theory calculations Physics Review Letters 94, 116103 (2005). [47] Perry's chemical engineer's handbook ; 6th ed., edited by R. H. Perry, D. W. Green, and J. O. Maloney (McGraw-Hill, New York, 1984). [48] R. B. Bird, W. E. St ewart, and E. N. Lightfoot, Transport phenomena 2nd ed. (Wiley, New York, 2002). 54

PAGE 67

[49] The properties of gases and liquids; 5th ed., edited by B. E. Poling, J. M. Prausnitz, and J. P. O'Connell (Mc Graw Hill, New York, 2001). [50] S. M. Sze, Physics of semiconductor devices (John Wiley & Sons, New York, 1981). [51] R. F. Pierret, Advanced semiconductor fundamentals Vol. 6 (Addison-Wesley, Reading, Massachusetts, 1987). [52] J. S. Shor, L. Bemis, and A. D. Kurtz, Characterization of monolithic n-type 6HSiC piezoresistive sensing elements IEEE Trans. Electron Devices 41 (`5), 661665 (1994). [53] J. S. Shor, D. Goldstein, and A. D. Kurtz, Characterization of n-type -SiC as a piezoresistor, IEEE Trans. Electron Devices 40 (6), 1093-1099 (1993). [54] T. J. Fawcett, M.S.Ch.E. Thes is, University of South Florida, 2004. [55] R. C. Jaeger, Introduction to microelectronic fabrication Vol. 5 (Prentice Hall, Upper Saddle River, 2002). [56] R. L. Myers, S. E. Saddow, S. Rao, K. D. Hobart, M. Fatemi, and F. J. Kub, Development of 3C-SiC SOI structures using Si on polycrystalline SiC wafer bonded substrates, Materials Science Forum 457-460, 1511-1514 (2004). [57] http://www.dupont.com/kapton/ [58] R. L. Myers, Y. Shishkin, O. Kordina, and S. E. Saddow, High growth rates (>30 m/h) of 4HSiC epitaxial layers using a horizontal hot-wall CVD reactor, J. Cryst. Growth 285 (4), 486-490 (2005). [59] T. J. Fawcett, J. T. Wolan, A. L. Spetz, M. Reyes, and S. E. Saddow, Thermal detection mechanism of 3C-SiC based resistive gas sensors to be submitted to Journal of Applied Physics. [60] http://www.expresspcb.com/ExpressPCBHtm/SpecsProduction.htm [61] R. N. Ghosh, P. Tobias, S. G. Ejakov, and B. Golding, Interface states in high temperature SiC gas sensing Proc. IEEE Sensors 6.4, 1120-1125 (2002). [62] B. Chwieroth, B. R. Patton, and Y. Wang, Conduction and gas-surface reaction modeling in metal oxide gas sensors Journal of Electroce ramics 6 (1), 27-41 (2001). [63] T. J. Fawcett, J. T. Wolan, R. L. Myers, J. Walker, and S. E. Saddow, in Hydrogen gas sensors using 3C-SiC/Si epitaxial layers Lyon, France, 2003 (Trans Tech Publicati ons Ltd.), p. (in press). 55

PAGE 68

[64] S. Nakagomi, K. Muto, and M. Itoh, Hydrogen sensitive negative switching behavior in metal-oxide-semiconductor devices Sensors and Actuators B 72 (2), 108-114 (2001). [65] I. Lundstrom, Approaches and mechanisms to so lid state based sensing Sensors and Actuators B 35 (1-3), 11-19 (1996). [66] E. Souteynard, D. Nicolas, and J. R. Martin, Semiconductor/metal/gas behaviour through surface-potential change Sensors and Actuators B 24-25 (1-3), 871-875 (1995). [67] Y. Goldberg, M. E. Levins htein, and S. L. Rumyantsev, in Properties of advanced semiconductor materials: GaN, AlN, InN, BN, SiC, SiGe edited by M. E. Levinshtein, S. L. Rumy antsev, and M. S. Shur (Wiley, New York, 2001), p. 93148. [68] F. Yun, S. Chevtchenko, Y.-T. Moon, H. Morkoc, T. J. Fawcett, and J. T. Wolan, GaN resistive hydrogen gas sensors Applied Physics Letters 87, 073507 (2005). [69] Available from Precision Resistor Company, Inc. [70] N. W. Ashcroft and N. D. Mermin, Solid state physics (Brooks/Cole, London, 1976). [71] I. Langmuir and G. M. J. Mackay, The dissociation of hydrogen into atoms. Part I. Experimental J. Am. Chem. Soc. 36, 1708 (1914). [72] Y. Luo and X. Liu, High temperature NTC BaTiO 3 -based ceramic resistors Materials Letters 59, 3881-3884 (2005). [73] L. Li, J. Zhao, and Z. Gui, The thermal sensitivity and dielectric properties of SrTiO 3 -based ceramics Ceramics International 30, 1073-1078 (2004). [74] C. J. Kobus, Utilizing disk thermistors to indirectly measure convective heat transfer coefficients for forced, natural and combined (mixed) convection Experimental Thermal and Fluid Science 29, 659-669 (2005). [75] COMSOL multiphysics modeling guide (COMSOL AB, 2005). [76] J. M. Smith, H. C. V. Ness, and M. M. Abbot, Introduction to chemical engineering thermodynamics (Mc Graw Hill, Boston, 2001). [77] G. K. Wachutka, Rigorous thermodynamic trea tment of heat generation and conduction in semiconductor device modeling IEEE Trans. Computer-Aided Design 9 (11), 1141-1149 (1990). [78] J. R. Hook and H. E. Hall, Solid state physics 2nd ed. (Wiley, New York, 1974). 56

PAGE 69

[79] A. J. Chapman, Heat Transfer 3rd ed. (Macmillan, New York, 1974). [80] S. v. Herwaarden and G. C. M. Meijer, in Semiconductor sensors, edited by S. M. Sze (Wiley, New York, 1994). [81] http://webbook.nist.gov/chemistry/fluid/ 57

PAGE 70

Appendices 58

PAGE 71

Appendix A. Derivation of heat transfer equations Table A.1. Notation for derivation of heat transfer equations. Variable Description Units t Time sec density kg/m 3 v magnitude of velocity m/s H enthalpy J/kg sourceQ source of heat W/m 3 e combined energy flux vector W/m 2 v velocity vector m/s g gravity vector m/s 2 pC specific heat or heat capacity J/kgK T temperature K j current density vector A/m 2 E electric field vector V/m electrical conductivity -1 m -1 V voltage V viscous momentum-flux tensor N/m 2 q heat flux by molecular conduction W/m 2 k thermal conductivity W/mK oT Initial temperature K device heaterq/ heat flux at the heater/device interface W/m 2 heaterq heat flux generate by the heater W/m 2 surfaceq heat flux at all surfaces ex cept heater/device interface W/m 2 h heat transfer coefficient W/m 2 K gasT bulk temperature of the gas K surfaceT temperature at the surface K The general differential energy balance on a differential volume unit, including a source of heat, is shown in equation A.1 adapted from [48]. gve QHv tsource 2 2 1 (A.1) The term on the left-hand-side of the equation represents th e rate of increase of energy per unit volume and consists of a kinetic ener gy and an enthalpy term, reading from left to right. In the gas sensor problem at hand, th e device is not moving, thus velocity is zero 59

PAGE 72

Appendix A (Continued) everywhere and for all times, leaving only the enthalpy term The enthalpy can be related to temperature through the thermodynamic de finition of specific heat [76], given in equation A.2 p pT H C (A.2) The definition of specific heat given in equation A.2 must be manipulated to fit the form needed for the differential balance in equation A.1 This slight manipulation, taking the time derivative, is shown in equation A.3 dTC t Hd t dTCHdp p (A.3) After the manipulations given above, the left-hand-side of equation A.1 reduces to the expression found in equation A.4 dTC t Hv t dTC t H t v t Hv tp p 0 2 2 1 2 2 1 2 2 1 (A.4) The first term in the right-hand-side of the differential energy balance, equation A.1 is the heat source term. In the gas sensor pr oblem, the heat source is the Joule heating of the device due to current flowing through the device. The derivation of this term starts with the definition of Joule heati ng, shown in is shown in equation A.5 [77]. EjQsource (A.5) Ej (A.6) VE (A.7) The definition of Ohms Law (equation A.6 ) [70] and electric field (equation A.7 ) are substituted into equation A.5 to yield a useful Joule heating term shown in A.8 2 2 2V Q VEEEEjQsource source (A.8) 60

PAGE 73

Appendix A (Continued) The second term in the right hand side of equation A.1 contains the combined energy flux vector and can be expanded into three te rms, including 1) convective transport, 2) rate of molecular done, and 3) rate of conduction by molecular means, shown in equation A.9 [48]. qv vHve 2 2 1 (A.9) The first and second terms in equation A.9 the convective heat transfer term and molecular work term respectively, will both eq ual zero since the device is not moving and has a velocity of zero. The third term in equation A.9 is the heat conduction term, and is defined by Fouriers Law, shown in equation A.10 [48]. Tkq (A.10) Substituting equation A.10 into the second term of the right hand side of equation A.1 yields equation A.11 2 2 2 2 100 Tke TkTke Tkqv vHve (A.11) The last term on the righ t hand side of equation A.1 evaluates to zero since the device is not moving and has zero velo city, shown in equation A.12 0 gv (A.12) Substituting equations A.4 A.8 A.11 and A.12 into the differential energy balance in equation A.1 yields equation A.13 the final form of the differential energy balance. TkVTC tp 2 2 (A.13) To completely define the heat transfer problem, boundary conditions must be specified along with the differential energy balance in equation A.13 The initial condition, a Dirichlet condition, for this problem is specified in equation A.14 The second boundary condition, a Ne umann condition, specifies the heat flux from the heater as being equal to the heat flux through the heater/device interface, as specified in equation A.15 oTT t = 0, everywhere (A.14) 61

PAGE 74

Appendix A (Continued) device heater heater device heaterTkq q/ / t > 0, heater/device interface (A.15) The final boundary condition, also a Neumann condition, sp ecifies the heat flux from all surfaces of the device with the exception of the heater/d evice interface, specified in equation A.16 surface surface surface gas surfaceTkqTThq t > 0, all surfaces except heater/device interface (A.16) At all surfaces associated with this final b oundary condition, heat is transferred from the surface to the gas by convection with a magnitude described by Newtons law of cooling [48]. It is convection that provides the mean s of transferring energy from the device to the gas, with Newtons law being the general m odel. Thus, the heat transfer coefficient (h) is the property of the gas or gas mixture that is being detected when the thermal detection mechanism is responsible for the gas sensing mechanism. 62

PAGE 75

Appendix B. Derivation of temperature dependence of carrier concentrations Table B.1. Notation for derivation of temperature dependence of carrier concentrations. Variable Description Units n concentration of electronics in conduction band cm -3 p concentration of holes in valence band cm -3 E energy eV g (E ) density of states (general) cm -3 V volume cm -3 2 h Js H Planks constant = 6.6310 -34 Js Js M Mass kg f (E ) Fermi function ratio E F Fermi level eV k B B Boltzmanns constant = 8.61410 -5 eV/K eV/K T temperature K g c (E ) density of states in the conduction band cm -3 E c conduction band edge eV g v (E ) density of states in the valence band cm -3 E v valence band edge eV m e mass of electron kg m h mass of hole kg E g bandgap = vcEE eV N c effective density of states in the conduction band cm -3 N v effective density of states in the valence band cm -3 n i intrinsic carrier concentration cm -3 DN concentration of ionized donors cm -3 AN concentration of ionized acceptors cm -3 g D Donor degeneracy factor E D Donor energy level eV g A acceptor degeneracy factor E A acceptor energy level eV q magnitude of an electronic charge = 1.6010 -19 C The number of electrons ( n ) and holes ( p ) per cubic unit volume of semiconductor material having an energy between E and E + dE is given by the density of states, g ( E ), where the density of stat es is given in equation B.1 without derivation [78]. 2 1 2 32 232Em V Eg (B.1) 63

PAGE 76

Appendix B (Continued) The probability density function that gives the rati o of the filled energy states to the total allowed energy states of a particle at a given energy is the Fermi function, f (E ), given in equation B.2 also without derivation [51]. TkEEBFe Ef 1 1 (B.2) This, the total number of carriers in the condu ction (electrons) and valence (holes) bands of a semiconductor is found by integrating th e total number of filled states, given by multiplying the total number of possible states, the density of states g (E ), by the fraction of those states that a oc cupied at a given energy, f (E ). This integration is displayed in equation B.3 with density of states shown in equation B.4 [51]. cE cdEEfEg V n 1 vE vdEEfEg V p 1 1 (B.3) 2 1 2 32 232Em V Ege c 2 1 2 32 232Em V Egh v (B.4) The Fermi function in equation B.2 can be simplified under the assumption that the Fermi is in the band gap, E g and more than 3k BT from either band edge. Under this assumption, the Fermi function becomes expressed by equation [51,78]. B B.5 TkEEB FeEf (B.5) Substitution and manipulation of the conduction band density of states from equation B.4 and the modified Fermi function from equation B.5 into the electron concentration equation, B.3 yields the electron con centration given in equation B.6 A similar process is used to calculate the hole co ncentration, given in equation B.7 [51,78]. TkEE c TkEE Be E TkE TkE e E TkEE e E cBcF BcF c B BF c B F ceN e h Tkm dEeEe m dEeEm V V dEEfEg V n 2 3 2 1 2 3 2 1 2 32 32 322 2 2 2 2 2 1 1 (B.6) 64

PAGE 77

Appendix B (Continued) TkEE v TkEE BhBFv BFveN e h Tkm p 2 322 2 (B.7) For a semiconductor with no impurities, charge neutrality must be maintained, giving rise to the law of mass action definition of the intr insic carrier concentration given in equation B.8 [51]. TkE vc i TkE vc TkEE vc TkEE v TkEE c iBg Bg BcV BFv BcFeNNn eNN eNN eN eNnpn 2 (B.8) The electron and hole concentrations given by equations B.6 and B.7 only account for the carriers associated with the intrinsic semiconductor at oms and not for impurity atoms (donors or acceptors). For an impurity atom to be able to transport charge, and act as a carrier, it must be ionized, i.e. a donor mu st be willing to give up an electron. The fraction of donor and acceptor atoms that are ionized is given in equations B.9 and B.10 respectively [51]. TkEE D D DBDFegN N 1 1 (B.9) TkEE A A ABFAegN N 1 1 (B.10) A total charge balance within the semiconductor is shown in equation B.11 [51]. 0 1 1 0 TkEE A A TkEE D D kTEE c kTEE v ADBFA BDF cF Fveg N eg N eN eN NNnp (B.11) 65

PAGE 78

Appendix C. Details of finite element method simulations C.1 Geometry Finite element method (FEM) simulations were performed on a simplified 3C-SiC resistive device usi ng Femlab 3.2 from COMSOL. The simulation utilized the Conductive Media DC and Heat Transfer by Conduction modules [75] with a geometry consisted of a 1 cm x 1 cm x 300 m solid sample with 0.2 cm x 1 cm contacts, as shown in Figure C.1 Figure C.1. Device simulation geometry. C.2 Governing equations The governing equation for the Conductive Me dia DC module is a charge balance on a differential volume of the device which includes electron conduction ( V ), external current sources ( eJ ) and internal current sources (), shown in equation jQ C.1 [75]. j eQJV (C.1) Equation C.2 shows the general energy balance e quation on a differential volume of the device [75]. See 0 for a detailed derivation of this equation. QTk t T Cpts (C.2) C.3 Subdomain properties The subdomain properties are the properties of the substance(s) in the simulation. For the Heat Transfer by Conductio n module, the built-in thermal properties of SiC, thermal conductivity = 490 W/(mK), density = 3216 kg/m 3 and heat capacity = 690 J/(kgK) were used. For the Conductive Media DC module, the resistivity of 3C-SiC was estimated using the resistance versus temperature data from a 3C-SiC/SOI device, as described below. 66

PAGE 79

Appendix C (Continued) The resistance versus temperature data fr om a 3C-SiC/SOI device with 0.5 mm x 1 mm contacts with a 1 mm gap was used to estimate the resistivity of the 3C-SiC epitaxial layer using equation C.3 where R(T) is the device resistance, (T) is the resistivity, L is the gap length and A is the cross-sectional area (epi -layer thickness x contact width). L A TRT A L TTR (C.3) The resistance versus temperature charac teristics of this device are shown in Figure 8.6 while the resistivity versus temperature characteristics of this device are shown in Figure C.2 The exponential relationship between resistivity and temperature was then extracted using standard linear regr ession to produce the relati onships given in equation C.4 7491.00108.0ln T Te0108.04728.0 m ][ KT][ (C.4) Figure C.2. Resistivity versus temperature of a 3C-SiC/SOI resistive gas sensor with 0.5 mm x 1 mm contacts with a 1 mm gap. C.4 Boundary conditions There are eight boundaries that require boundary conditions in this simulation. There are three boundaries on th e top of the device, two contac ts and the exposed, or active, area, four sides of the device each give a boundary, and one boundary on the bottom of the device. Thus, eight boundary conditions are required for both the Conductive Media DC and Heat Transfer by Conduction modules. For the Conductive Media DC mo dule, a voltage is applied to one of the contacts while the other contact is held at ground while all othe r boundaries are electrically isolated. For the Heat Transfer by Conducti on module, the bottom of the device has a specified heat flux (termed the heat flux from the heater) while all heat flux from all other boundaries is specified by Newtons La w of Cooling, shown in equation C.5 where 67

PAGE 80

Appendix C (Continued) q surface is the heat flux from the surface, h is the heat transfer coefficient, T gas is the temperature o the gas and T surface is the temperature of the surface. surface gas surfaceTThq (C.5) The heat transfer coefficient can be estimated using an empirical relationship, shown in equation C.6 relating the Nusselt number (Nu) to the Prandtl number (Pr) and Reynolds number (Re) for convective heat transfer over flat plates [47,79,80]. (C.6) 2/13/1RePr664.0 Nu The definitions of Nu, Pr and Re, are presented in equation C.7 where h is the heat transfer coefficient, L is the characteristic length, k is the thermal conductivity, is the viscosity C p is the heat capacity, is the density and v gas is the velocity of the gas. Lv k C k hL Nugas p Re,Pr, (C.7) The gas properties were taken from the Na tional Institute of Standards and Technology (NIST) Chemistry WebBook, Standard Referen ce Database 69 [81]. For a specific gas, H 2 or N 2 in these simulations, the Pr and Re can be calculated. Thus, through equation C.6 Nu can be calculated and h extracted using the definition of Nu in equation C.7 C.5 Simulation procedure With a contact gas composition, the bias voltage and heat flux from the heater determine the device temperature. Figure C.3 shows a slice plot from a device biased at 10V DC with the heat flux from the heater being 117.8351 W/m 2 to give a device temperature of 373.15 K (or 100C) and pure N 2 making up the heat tr ansfer coefficient (1.4 W/m 2 K). Under pure N 2 conditions, the device temperature was determined at various values of the heat flux from the heat er (Q) to determine the heat flux needed to obtain initial device temperatures of 373, 473, 573, and 673 K, shown in Figure C.4 Thus, the heat flux from the heater is c hosen according to the desired initial device temperature using the relationship in Figure C.4 The heat transfer coefficient is then swept from its pure N 2 value of ~ 1.4 W/m 2 K to its pure H 2 value of ~ 1.6 W/m 2 K with the heat flux from the heat constant. Th e equations governing he at transfer and DC conduction equations, presented in section C.2 are then solved using the boundary conditions specified in section C.4 The results of these simulations are presented and discussed in section 8.2.2 68

PAGE 81

Appendix C (Continued) Figure C.3. Device temperature under pure N 2 at 373 K. Device biased at 10V DC. Figure C.4. Device temperature as a function of heater heat flux. Device biased at 10V DC. 69

PAGE 82

About the Author Timothy J. Fawcett received his Bachelor of Science and Master of Science degrees in Chemical Engineering from the University of South Florida in May 2004 where his research explored the potentia l application of SiC based resistive devices as hydrogen gas sensors. He has been pursuing his Doctor of Philosophy degree in Chemical Engineering since that time and has been active in the SiC community with his research focused on understanding the gas sensing mechanism driv ing SiC based resistive gas sensors. Timothy has published several peer-reviewed jour nal articles in addition to presenting his gas sensor research at several conferences After graduation, Dr. Fawcett plans to continue SiC based gas sensor research at the University of South Florida.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001794030
003 fts
005 20070627134239.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 070627s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001537
040
FHM
c FHM
035
(OCoLC)145741135
049
FHMM
090
TP145 (ONLINE)
1 100
Fawcett, Timothy J.
0 245
Investigation into the hydrogen gas sensing mechanism of 3C-SiC resistive gas sensors
h [electronic resource] /
by Timothy J. Fawcett.
260
[Tampa, Fla] :
b University of South Florida,
2006.
3 520
ABSTRACT: The hydrogen (H2) gas sensing mechanism driving 3C-SiC resistive gas sensors is investigated in this work in which two hypotheses are proposed. One hypothesis involves the surface adsorption of H2 on the sensor surface with the adsorbed molecules influencing the flow of current in a resistive gas sensor, termed the surface adsorption detection mechanism. The second hypothesis includes the transfer of heat from the sensor to the gas, producing a change in the temperature of the device when the heat transfer characteristics of the gas change, termed the thermal detection mechanism. The heat transfer characteristics of the gas are dependent on the thermal conductivity of the gas, a property which is a strong function of gas composition. Thus, the thermal detection mechanism mainly detects changes in the thermal conductivity of a gas or gas mixture.Initial experiments suggested the surface adsorption mechanism as the detection mechanism of resistive 3C-SiC gas sensors. However, these experiments were performed in the absence of device temperature measurements. Recent experiments in which the device temperature was measured with a resistance temperature detector (RTD) in thermal contact with the device strongly support the thermal detection mechanism as being responsible for hydrogen gas detection. Experimental observations show the temperature of the resistive 3C-SiC hydrogen gas sensors changes greatly with changing hydrogen gas composition. For example, a 3C-SiC/SOI resistive sensor biased at 10 Vdc displayed a change in temperature from ~400¨C to ~216¨C, correlating to a change in current from ~41 mA to ~6mA, upon the introduction of 100% H2. The this 3C-SiC/SOI resistive sensor, this large decrease in temperature caused a large increase in resistance which is detected as a decrease in current. Several different experiments have also been performed to confirm the thermal detection mechanism hypothesis.
502
Dissertation (Ph.D.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 69 pages.
Includes vita.
590
Adviser: John T. Wolan, Ph.D.
653
Cubic silicon carbide.
Thermal detection.
Thermal conductivity.
Heat transfer.
Silicon-on-insulator.
690
Dissertations, Academic
z USF
x Chemical Engineering
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1537