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 001478811
007 cr mnu|||uuuuu
008 040811s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000458
Chaudhari, Amol V.
Development of surface acoustic wave sensors using nanostructured palladium for hydrogen detection
h [electronic resource] /
by Amol Chaudhari.
[Tampa, Fla.] :
University of South Florida,
Thesis (M.S.Ch.E.)--University of South Florida, 2004.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 80 pages.
ABSTRACT: This thesis addresses the development of new gas sensor using surface acoustic wave (SAW) technology. SAW sensors detect the change in mass, modulus, and conductivity of a sensing layer material via absorption or adsorption of an analyte. The advantage of SAW sensor includes low cost, small size, high sensitivity. We investigated the use of nano-crystalline palladium film for sensing hydrogen gas. We also investigated SAW fabrication for radio frequency (RF) range operation where high signal-to-noise ratios can be achieved. A test-bed consisting of a gas dilution system, a temperature-controlled test cell, a network analyzer, and computer-based measurement system was used for evaluating the performance of SAW gas sensors at very low concentrations. Both single and dual delay line SAW devices were fabricated by means of photolithography on a lithium niobate substrate. Tests are carried to determine response speed, resolution, reproducibility, and linear characteristics, over a range of analyte concentrations.
Adviser: Venkat Bhethanabotla.
x Chemical Engineering
t USF Electronic Theses and Dissertations.
Development o f Surface Acoustic Wave Sensors Using Nanostructured Palladium f or Hydrogen Detection by Amol Chaudhari A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Chemical Engineer ing Department of Chemical Engineering College of Engineering University of South Florida Major Professor: Venkat Bhethanabotla, Ph.D. Thomas Weller, Ph.D. Shekhar Bhansali, Ph.D. Date of Approval July 8, 2004 Keywords: e beam lithography net work analyzer Copyright 2004, Amol Chaudhari
Acknowledgements I wish to express sincere appreciation to my major professor Dr. Venkat Bhethanabotla for his valuable guidance and support for this thesis work. Thanks to Dr. Shekhar Bhansali and Dr. Thomas Weller, for agreeing to serve on my supervisory committee. I would like to thank Dr. Deepak Srinivasdasgupta and Stefan Cular for helping me at every step. I also appreciate Dr. Senthil Sambandam for his efforts in making sensing films. Special thanks are extended to Dr. Bert Lagel and Nhan Nyugyen for their help in E beam lithography operations. I also appreciate Balaji Lakhminarayan, Kevin and all of my friends for their support and encour a gement throughout this research.
i Table of Contents List of Tables iii List of Figures iv Abstract vi Chapter 1 Introduction 1 1.1 Sensors 1 1.2 Need and I mpact of S ensors 2 1.3 Hydroge n S ensors 4 1.4 Acoustic W ave S ensors 8 1.5 Thesis Organization 9 Chapter 2 Theory and Design 11 2.1 Piezoelectricity 11 2.2 SAW D evice F undamentals 11 2.3 SAW D evice D esign 14 2.4 SAW P erturbation M echanisms 19 2.4.1 Mass Loading Effect 21 2.4.2 SAW Acousto electric Re sponse 23 Chapter 3 Fabrication 26 3.1 Introduction 2 6 3.2 Optical L ithography 26 3.2.1 SAW F abrication D etails 29 3.3 E beam L ithography 32 3.3.1 Process Flow 36 3.4 Sensing L ayer F abrication 40 Chapter 4 Measurements 43 4.1 SAW Device M easurements 43 4.2 Design and Fabrication of T est C ell 45 4.3 Problems E ncountered 48 4.4 SAW S ensor T est b ed 50 Chapter 5 Results and Discussions 52 5.1 Fabrication of D evices 52 5.2 SAW Device C haracterization 54 5.3 SAW S ensor M easurements 58 5.4 Temperature E ffects on M easurements 60
ii 5.5 Repeatability 64 5.6 Effects on Palladium F ilm 6 4 Chapter 6 Conclusion and future work 66 References 69 Appendi c es 71 Appendix A 72
iii List of Tables Table 2.1 Different piezoelectric materials and their properties 12 Tab le 2 2 Independent p arameters for SAW sensor d esign 18 Table 5.1 Response times of 200 MHz SAW hydrogen sensor 59 T able 5.2 Average frequency shifts for a 200 MHz SAW hydrogen sensor 59 Table A.1 The various s pin o n photoresists used in fabrication 72
iv List of Figures Figure 2 .1 Schematic of SAW delay line 12 Figure 2.2 Simulated frequency r esponse o f a t ypical IDT 17 Figure 2.3 Frequency magnitude r esponse o f a t ypical SAW 19 Figure 2.4 Equivalent c ircuit model for SAW a cous to electric r esponse 23 Figure 3.1 Mask for SAW devices with frequency 50 MHz to 100 MHz 27 Figure 3.2 Mask for SAW devices with frequency 200 MHz 28 Figure 3.3 Process flow for etching 30 Figure 3.4 Pr ocess flow for lift off 31 Figure 3. 5 Typical EBL system 33 Figure 3.6 Pattern written by EBL on lithium niobate with 10nm Au Pd 37 Figure 3.7 Pattern written by EBL on lithium niobate with 3nm Au Pd 38 Figure 3.8 SEM sample holder w ith faraday cup, gold standard and lithium niobate with bonding pads 39 Figure 3.9 Pattern written by EBL with different area doses 4 0 Figure 3.10 XRD plot of nano crystalline Pd film 41 Figure 4.1 Schematic of SAW device measur ement set up 45 Figure 4.2 SAW device test cell schematic 47 Figure 4.3 Fabricated test cell 48 Figure 4.4 Scratched SAW device IDTs 49 Figure 4.5 Test cell with improved lid 49
v Figure 4.6 SAW device measurement set up with dilut io n system and data acquisition 50 Figure 5.1 Fabricated SAW devices, (a) IDT of a 100 MHz SAW, (b) IDT of a 200 MHz SAW, (c) 100MHz and 80 MHz SAW single and dual delay line devices. (d) 200 MHz SA W dual delay line de vices. 52 Figure 5.2 (a) 900 MHz IDTs written by EBL, (b) Misaligned IDT of a 900 MHz SAW device 53 Figure 5.3 (a) S21 transmission curve, (b) S11 reflection c urve for a 70 MHz device 5 5 Figure 5.4 ( a) S21 transmission curve, (b) S11 reflection curve for a 100 MHz device 56 Figure 5.5 (a) S21 transmission curve, (b) S11 reflection curve for a 200 MHz device 57 Figure 5.6 Response of a 200 MHz SAW sensor to hydrogen exposures 58 Figure 5. 7 Temperature effects on SAW device frequency 6 0 Figure 5. 8 Response of a 200 MHz SAW sensor at a temperature of 10 0 C 62 Figure 5.9 Response of a 200 MHz SAW sensor at a temperature of 65 0 C 63 Figure 5.1 0 Response of a 200 MHz SAW sensor for 1% Hydrogen cycling 64 Figure 5.1 1 Nano crystalline palladium film before and after hydrogen cycling 65 Figure 5.1 2 Nano crystalline palladium film peeling an d cracking 65
vi DEVELOPMENT OF SURFACE ACOUSTIC WAVE USING NANOSTRUCTURED PALLADIUM FOR HYDROGEN DETECTION Amol Chaudhari ABSTRACT This thesis addresses the development of new gas sensor using surface acoustic wave (SAW) technology. SAW sensors detect the change in mass, modulus, and co nductivity of a sensing layer material via absorption or adsorption of an analyte. The advantage of SAW sensor includes low cost small size, high sensitivity We investigated the use of nano crystalline p alladium film for sensing hydrogen gas. We also inve stigated SAW fabrication for radio frequency (RF) range operation where high signal to noise ratios can be achieved. A test bed consisting of a gas dilution system, a temperature controlled test cell, a network analyzer, and computer based measurement syst em was used for evaluating the performance of SAW gas sensors at very low concentrations. Both single and dual delay line SAW devices were fabricated by means of photolithography on a lithium niobate substrate. Tests are carried to determine response speed resolution, reproducibility, and linear characteristics, over a range of analyte concentrations
1 Chapter 1 Introduction 1.1 S ensors Smart sensors are of great interest in many fields of industry, control systems, biomedical applications, etc. Rapid advances in IC technologies have brought new challenges in the physical design of integrated se nsors and Micro Electrical Mechanical Systems (MEMS). MEMS offer new ways of combining sensing, signal processing and actuation on a microscopic scale and allow both traditional and new sensors to be realized for a wide range of applications and operationa l environments A good working definition of a sensor is a device that allows the transduction of chemical and physical properties at an interface into usable information. An ideal chemical sensor is an inexpensive, portable, foolproof device that respo nds with perfect and instantaneous selectivity to a particular chemical substance (1). A chemical sensor is thus a device that generates an electrical signal which is a function of chemical identity and/or concentration. Chemical sensors have the potential to be employed in both fundamental studies of interfaces and the detection and measurements of chemical species. Unfortunately, due to the wide range of concentrations (many orders of magnitude) and substances encountered in real life, it is not p ossible to achieve this using a single technology. Out of necessity, one must adapt the sensing systems to the sensing
2 needs. The techniques used for chemical sensing can be divided into the broad categories like; 1. Thermal sensors (that utilize the energy g enerated through direct contact with an analyte) 2. Mass sensors (that detect changes in mass loading caused by the analyte), 3. Electrochemical sensors (that detect changes in electrical properties) 4. Optical sensors (that detect changes in spectra of light) (2 ). Sensors are characterized in many ways. Their sensitivity is a measure of the magnitude of output signal produced in response to an input quantity, their resolution is a measure of the minimum change of input quantity which they can detect an d their selectivity is a measure of degree to which they can distinguish one input quantity from another. Besides, the following list gives both constraints and requirements for any sensor Reversibility, Fast and highly sensitive, Durability, Small size an d simple operation, Easy to fabricate, Temperature stability, etc. 1.2 Need and I mpact of S ensors The trend towards the small began with the miniaturization of macro techniques, which led to the now well established field of micro techno logy. Electronic, optical, and mechanical micro technologies have all profited from the smaller, smarter, and less costly sensors that resulted from work with ICs, fiber optics, other micro optics, and MEMS. As we continue to work with these minuscule buil ding blocks, there will be a convergence of nanotechnology, biotechnology, and information technology, among others, with benefits for each discipline. Substantially smaller size, lower weight, more
3 modest power requirements, greater sensitivity, and bette r specificity are just a few of the improvements well see in sensor design. We need smart sensors now with increased accuracy, reliability and speed. Today intelligent sensors are extremely necessary for industrial, business and defense applications as electronic noses, smart vision systems, personnel (human body) detection, authentication systems, building monitoring system, etc. This purpose is most effectively achieved by a combination of technological and structural algorithmic methods. It allow s to achieve the same performances (or even better), but is significant smaller money and human expenses and much faster. Microsensors have applications in many industries, among them transportation, communications, building and facilities, medicine, safety, and national security, including both homeland defense and military operations. Consider nanowire sensors that detect chemicals and biological material (3). Many start up companies are already at work developing these devices in an effort to get in at the beginning. Funding for nanotechnology increased by more than a factor of 5 between 1997 and 2003, and is still on the rise (4). So this is a good time to examine the possibilities and the limitations of this small new world. While presenti ng a significant challenge, integration of technologies could lead to tiny, low power, smart sensors that could be manufactured inexpensively in large numbers. Their service areas could include in situ sensing of structural materials, sensor redundancy in systems, and in size and weight constrained structures such as satellites and space platforms (5).
4 Two functions often separated in many sensors, especially those for chemicals and biological substances, are recognition of the molecule or other obje ct of interest and transduction of that recognition event into a useful signal. Nanotechnology will enable us to design sensors that are much smaller, less power hungry, and more sensitive than current micro or macro sensors. To foresee some of the possible devices and applications sensors for physical properties were the focus of some early development efforts, but nanotechnology will contribute most heavily to realizing the potential of chemical and biosensors for safety, medical, and other purpo ses. Nanotechnology is certain to improve existing sensors and be a strong force in developing new ones. The field is progressing, but considerable work must be done before we see its full impact. Among the obvious challenges are reducing the cost o f materials and devices, improving reliability, and packaging the devices into useful products. Nevertheless, we are beginning to see nano scale materials and devices being integrated into real world systems, and the future looks very bright indeed for tec hnology on a tiny scale. 1.3 Hydrogen S ensors Today, hydrogen is an important industrial chemical. Hydrogen is used as a feedstock in several well established industries such as chemical, food, metallurgical, electronic, and others, and therefore a l imited production, distribution and usage network already exists. Furthermore, prototype hydrogen fueled automobiles are already on the road (6).
5 Hydrogen Now 1. ~500 billion Nm 3 produced annually worldwide 2. Nearly half (48%) is produced via steam ref orming of natural gas 3. Used to upgrade crude oil and in the chemical industry (methanol, ammonia, fertilizer, etc) 3. NASA and other space agencies use hydrogen as fuel Hydrogen in the Future 1. As a chemical 2. As a fuel for space exploration 3. As a transportation fuel for fuel cell vehicles 4. As an energy storage medium for intermittent renewable sources (wind and PV) 5. As an energy carrier for supplying power, heat, and fuel for residential buildings Hydrogen in the Meantime 1. Hydrogen produc ed from natural gas will continue to provide cost effective hydrogen to industry 2. Transport of hydrogen via liquid tanker trucks or dedicated pipelines will provide fuel to emerging transportation applications 3. Infrastructure development will continue, focusing on small scale, on site production of hydrogen
6 However, the increasing use of hydrogen gas should not be considered as one without disadvantages. In fact, a number of problems arise involving the storage of this gas. A hydrogen leak in lar ge quantities should be avoided because hydrogen, when mixed with air in the ratio of 4.65 93.9 vol % is explosive (7). Hence, it is important to develop highly sensitive hydrogen detectors to prevent accidents due to its leakage, thus, saving lives and in frastructure. Such detectors should allow continuous monitoring of the concentration of the gas in the environment in a quantitative and selective way. Different types of hydrogen sensors (8) 1. Pyroelectric sensors: Pyroelectric materials exhibit the prop erty that the polarization vector is a function of temperature. As the temperature varies, the surface charge also varies due to dimensional changes in the pyroelectric. This property results in a potential difference between the two opposite surfaces of t he material, generally known as pyroelectricity. There are AC and DC pyroelectric hydrogen sensors. The response arises due to the thermal energy transfer as a result of the adsorption and dissociation of hydrogen molecules on the Pd surface. 2. Piezoelectric sensors: These sensors are b ased on piezoelectric crystals, which allow transduction between electrical and acoustic energies. There are mainly two hydrogen sensors in these category quartz crystal microbalance (QCM ) and surface acoustic wave sensor (SAW) The response arises due to the changes wave properties upon adsorption of hydrogen on palladium surface. 3. Fiber optic sensors: Fiber optic sensor is a means whereby light guided within an optical fiber can be modified in response to an external physical o r chemical
7 influence. When hydrogen gas molecules get absorbed in palladium coated fiber part, they change the optical properties of Pd. The response can be measured in terms of absorbance, reflectance, luminescence, or scattering. 4. Electrochemical sensors : Any electrochemical cell is made up of an electrical conductive path involving two electrodes immersed in an ionic electrolyte. Each electrode exhibits its own characteristic potential. In recent years, solid state electrochemical cells have been develop ed as convenient devices for continuous measurements of flowing hydrogen. Hydrogen provides the proton path, equivalent to electrolytic ionic conduction. The electron flux from the ionized protons manifests itself as a current in the external circuit, due to charge exchanges at the working electrode. Thus the measured current is a monitor of the hydrogen/proton concentration in the cell. 5. Semiconductor Sensors: The semiconductor chemical sensor is based on the metal oxide semiconductor (MOS) junction princip le. MOS sensors are of two types, MOS capacitors and MOS transistors. Hydrogen molecules when adsorbed on a metal surface act as dipoles and they give rise to macroscopic measurable voltage drop. The dipole layer shifts the energy levels at the metal insu lator interface. The voltage is recorded as a function of time for various hydrogen concentrations. 6. Resistive sensors: Resistive sensor is based on the change in the resistance of the metal film upon the adsorption or absorption of hydrogen. These are very simple sensors employing various films like Pd, tin oxide, etc. The change in resistance is related to the hydrogen concentration. They often operate at very high temperatures.
8 1.4 Acoustic W ave S ensors Acoustic wave sensors fall under the category of mass sensors and use changes in the acoustic wave transmission through a substance to detect changes in mass loading. I t has been shown quite some time ago that chemical vapor sensing could be accomplished using a device originally designed as a delay l ine (9). Currently available surface acoustic wave (SAW) sensors utilize frequencies of the order of 100 MHz. The most commercially developed of the acoustic wave sensors is the thickness shear mode (TSM) device, also known as the quartz crystal microbalan ce (QCM). Acoustic wave devices have been in commercial use for more than 60 years (10). SAW devices are widely used in the telecommunications industry as components in filters, with estimated annual sales of about $3 billion. It is anticipated that th eir use as sensors may eventually equal the demand of the telecommunications market. Potential applications include automotive applications (torque and tire pressure sensors), medical applications (chemical sensors), and industrial and commercial applicati ons (vapor, humidity, temperature and mass sensors). Acoustic wave sensors are economical, rugged, sensitive and reliable. Capability of being passively and wirelessly interrogated is developing rapidly, with such devices having the potential for use in di stributed sensing applications (11, 12). SAW devices are favored for use in chemical sensing applications because of their small size, low cost, high sensitivity and reliability. Most SAW chemical sensors monitor changes of the SAW phase velocity and a ttenuation as the vapor interacts with the sensing layer. The shift in the phase velocity and/or attenuation is measured by recording the frequency and insertion loss of the SAW device, respectively. Various
9 effects, including mass loading, viscoelastic lo ading, and acousto electric coupling, contribute to SAW sensor response. Typical chemical sensors take advantage of one or more of these mechanisms. One of the advantages of SAW sensors is the ability to accommodate very high sensitivities without res orting to pre concentration. This allows for considerations of sensors relevant to environmental monitoring also, in addition to detection for leaks and health related applications. While ppb levels of detection are indeed possible with SAW devices, most a pplications of this sensor so far have been for organics using polymer sensing layers with higher levels of gas phase concentrations. With the use of high frequency SAWs, novel sensing layers such as those described in this proposal, the use of sensing lay ers comprising of two or more layers that take advantage of the mass, modulus and/or electro acoustic perturbation mechanisms of the SAW sensor (13), and novel response interpretation techniques (14) it is now possible to realize relevant SAW sensing techn ologies for environmental gas monitoring at the required low ppb levels. Using higher frequencies can also increase sensitivity using modern fabrication techniques. 1.5 Thesis O rganization The work presented here deals with issues involved in the develo pment of surface acoustic wave sensors and its application as a hydrogen gas sensor. Efforts were made to fabricate the SAW sensors of different frequencies and to carry out their testing. The gas dilution system with the test bed to measure the sensor cha racteristics was designed and built.
10 The Chapter 2 discusses the theory of the SAW sensor, how it works and the detailed design of our SAW devices. It also gives an insight to the working principles of SAW sensor. It is followed by fabrication details of these devices with the sensing film in Chapter 3. It includes the processes which were selected for fabrication and the optimization of different required parameters. Chapter 4 deals with the measurements of the devices and the required test bed assembly. The results for experiments done are discussed and analyzed in Chapter 5. Conclusions and the directions for the future work are presented in Chapter 6.
11 Chapter 2 Theory and Design 2.1 Piezoelectricity In 1880, Jacques and Pierre Curie discovered an unusual characteristic of certain crystalline materials which lack the centre of inversion symmetry in their crystalline structure (15). When these materials are subjected to a mechanical force, the crystals become electrica lly polarized. Tension and compression generates voltages of opposite polarity, and in proportion to the applied force. Subsequently, the converse of this relationship was confirmed: if one of these voltage generating crystals was exposed to an electric fi eld it lengthened or shortened according to the polarity of the field, and in proportion to the strength of the field. These behaviors were labeled the piezoelectric effect and the inverse piezoelectric effect, respectively, from the Greek word piezein, me aning to press or squeeze. 2.2 SAW D evice F undamentals Acoustic waves can be generated in piezoelectric materials. The piezoelectric material may be either a polished substrate such as quartz, lithium niobate, lithium tantalate or a thin film such as zin c oxide. Quartz is the most commonly used substrate because of its temperature stability for certain crystal orientations. However, for high acousto electric coupling applications, lithium niobate is preferred over quartz. Thin films of zinc oxide deposi ted on a non piezoelectric substrate such as silicon are used
12 when the device needs to be integrated with microelectronics. The crystal orientation, the thickness of the piezoelectric material and the geometry of the metal transducers determine the type an d mode of the acoustic waves. The most commonly used piezoelectric materials and their properties like electromechanical coefficient ( K 2 ), capacitance per unit length and sound velocity through the material are listed in T able 1, Table 2.1 Different piezo electric materials and their properties Substrate K 2 C 0 (pF/cm) V(m/s) St quartz 0.0011 0.5 3158 Y Z LiNbO3 0.048 4.6 3488 Y X LiTaO3 0.016 4.4 3379 GaAs 0.0007 1.2 2864 In a delay line SAW device, surface acoustic waves are ge nerated using interdigital metal transducers (IDT) patterned on a substrate (see Figure 1). One set of the IDT is used as a transmitter that converts the applied voltage variation into acoustic waves and the other IDT receives these acoustic waves and con verts them back to an output voltage. Amplitude of particle displacement is on the order of a wavelength. Figure 2.1 Schematic of SAW delay line
13 As the acoustic wave propagates through or on the surface of the material, any changes to the characteristi cs of the path affect the velocity and/or the amplitude of the wave. Changes in the velocity can be monitored by measuring the frequency change and then be correlated with the physical quantity that is being measured (16). The generation of the SAW w aves is achieved by application of a voltage to metal film interdigital transducers deposited on the surface of a piezoelectric substrate. The electrodes that comprise these arrays are connected with alternate polarities so that an RF signal voltage of the proper frequency applied across them causes the surface of the crystal to expand and contract. This generates the Rayleigh wave, or surface wave, as it is more commonly called. Two IDTs are required in the basic SAW device configuration sketched in Figure 1.1. One of these acts as the device input and converts signal voltage variations to mechanical acoustic waves. The other IDT is employed as an output receiver to convert mechanical SAW vibrations back into output voltages. This kind of the arran gement is also referred as a delay line (16). If we use two delay lines on the same substrate, then the system will be a dual delay line system. Here, one delay line is used for measurement of the parameters and the other can be used as the reference. T he basic SAW transducer is a bidirectional radiator. That is, half of the power is directed toward the output transducer while the other half is radiated toward the end of the crystal and is lost. By reciprocity, only half of the intercepted acoustic energ y at the output is reconverted to electrical energy; hence, the inherent 6 dB loss associated with this structure. Numerous second order effects, such as coupling efficiency, resistive losses, and impedance mismatch, raise the insertion loss of practical f ilters to 15 30 dB
14 Since the surface wave or acoustic velocity is much slower than the speed of light, an acoustic wavelength is much smaller than its electromagnetic counterpart. This results in the SAWs unique ability to incorporate an incredible amou nt of signal processing or delay in a very small volume. As a result of this relationship, physical limitations exist at higher frequencies when the electrodes become too narrow to fabricate with standard photolithographic techniques and at lower frequenci es when the devices become impractically large. Hence, at this time, SAW devices are most typically used from 10 MHz to about 3 GHz. 2.3 SAW D evice D esign SAW device operation cannot be truly understood without some mention of the actual design procedures involved in their creation. With the possible exception of resonator type devices, SAW components utilize finite impulse response design techniques very similar to those used for digital filters. Hence, the principal design tool is the Fo urier transform, which is used to relate the time and frequency responses of the transducers and resultant filter. In general, the designer derives two impulse responses for the two transducers whose transforms can be added together (in dB) to produce the desired total frequency response characteristics. These two impulse responses are then etched onto the surface of a metalized piezoelectric substrate. The design and simulation of a surface acoustic wave (SAW) sensor follows from the eq uations of piezo electric effect, the model of formation of the surface Raleigh waves, and the electro acoustic effect.
15 The stress free boundary imposed by the surface of the crystal gives rise to a unique acoustic mode whose propagati on is confined to the surface and is therefore known as a surface acoustic wave (SAW). Lord Rayleigh discovered this mode of propagation in which acoustic energy is confined very near the surface of an isotropic solid. This mode is also known as Rayleigh w ave. These waves can be excited by surface electrodes in piezoelectric materials. This wave is extremely sensitive to surface perturbations. To satisfy the stress free boundary condition, coupled compressional and shear waves propagate together i n a SAW such that surface traction forces are zero. The generalized surface acoustic wave, propagating in the z direction, has a displacement profile u(y) that varies with depth y into the crystal as ( ) ( ) ( ) ( ) [ ] z t e z e y u y e y u x e y u t z y x u j j z j y j x g w f f f + + = , 3 2 1 (1) Where w is the angular frequency ( ) f p 2 g is the complex propagation factor x u y u and z u are displaceme nt components in the x y and z directions, respectively i f are the phases of the components with respect to z u The displacement compone nts ( ) y u i vary approximately as l p / 2 y e
16 Where l is the SAW wavelength along the surface of the material and y is the distance into the substrate. Thus, the amplitude decays rapi dly with distance into the bulk of the crystal. To design a SAW device or filter with a given resonant frequency 0 f and fractional bandwidth B (measured null to null on either side of the resonant fr equency), we make use of the following equations: The acoustic wavelength with reference to the above nomenclature is 0 0 v f l = (2) The width of each finger that results in this synchronous frequency is /4 l and th e inter digital spacing measured center to center is /2 l The number of finger pairs needed to achieve this fractional bandwidth specification is: 2 p N B = (3) For the best response characteristics, the impedance of the IDT ( Z ) should be matched with the impedance of the measurement system (typically 50 ohm). The IDT behaves as a capacitive system with the total capacitance determined by the number of finger pairs, their spacing, as well as the degree of overlap. The total capacitance required is: 0 1 2 t C fZ p = (4)
17 The aperture (overlap between fingers, in length units) is then given below: 0 t p C W CN = (5) The frequency magnitude response of the IDT appro ximated as the incoherent addition of contributions from individual fingers is: 1 sin () () X f X f = (6) where ( ) 0 0 p N ff X f p = Table 2 lists the typical parameters involved in the design of our SAW sensor. A typical response is show n in Figure 2.2. Figure 2.2 Simulated f requency m agnitude r esponse of a t ypical IDT
18 Table 2.2 Independent parameters for SAW sensor d esign Parameter Value Piezoelectric Crystal Y Z Lithium Niobate Surface Acoustic Wave Velocity ( 0 v ) 3488 m/s Center Frequency ( 0 f ) 200 MHz Design Impedance ( Z ) 50 ohm No. of Finger Pairs ( p N ) on a single IDT 50 Delay Path Length ( L ) (in terms of wavelength l ) 300 l Capacitance/finger pair length ( C 0 ) 4.6 pF/cm Normalized surface particle velocity in x direction ( 2 0 x v P w ) 0 cm.g Normalized surface particle velocity in y direction ( 2 0 y v P w ) -6 0.83 x 10 cm.g Normalized surface particle velocity in z direction ( 2 0 z v P w ) -6 0.56 x 1 0 cm.g Electromechanical coupling coefficient (squared) ( K 2 ) 4.8 % When two IDT are in series spaced apart by a delay path of length L as in a SAW device, the path delay affects only the phase delay and not the magnitude response, as the attenuation losses are negligible at low frequencies. The cumulative frequency magnitude resp onse for the SAW is simply the dot product: 2 1 12 ( ) ( ) .() S f ff ff (7)
19 Figure 2.3 shows the frequency magnitude response of a SAW as computed using Equation (7). Figure 2.3 Frequency m agnitude response of a t ypical SAW 2.4 SAW P erturbation M echan isms All acoustic wave sensors are sensitive, to varying degrees, to perturbations from many different physical parameters. The range of phenomena that can be detected by acoustic wave devices can be greatly expanded by coating the devices with materia ls that undergo changes in their mass, elasticity, or conductivity upon exposure to some physical or chemical stimulus. These sensors become pressure, torque, shock, and force detectors under an applied stress that changes the dynamics of the propagating m edium. They become mass, or gravimetric, sensors when particles are allowed to contact the
20 propagation medium, changing the stress on it. They become vapor sensors when a coating is applied that absorbs only specific chemical vapors. These devices work by effectively measuring the mass of the absorbed vapor. If the coating absorbs specific biological chemicals in liquids, the detector becomes a biosensor. As previously noted, a wireless temperature sensor can be created by selecting the correct orientation of propagation. The propagating medium changes with temperature, affecting the output. The change in the acoustic velocity in a SAW device depends on many factors like mass, stiffness, conductivity etc. + D + D + D + D + D + D = D ... .......... 1 0 0 p p T T c c m m n n s s n e e n n n n n n (8) where m mass c stiffness e dielectric constant s surface conductivity T temperature p pressure The measured response of the SAW arise from the perturbations in a wave propagation characteristics, specifical ly wave velocity & attenuation, resulting from interactions between the surface acoustic wave and a surface layer. As SAWs propagating in a piezoelectric medium generate both mechanical deformation and an
21 electrical potential, there are both mechanical and electrical couplings between SAWs and sensing film. 2.4.1 Mass L oading E ffect It is the most utilized interaction in SAW sensor applications. If the mass layer on the surface is sufficiently thin or rigid, then 1) it moves synchronously with the wave 2) Average kinetic density k U increases The change in the average kinetic energy per unit area of surface is ( ) 2 0 2 0 2 0 4 z y x s k U n n n r + + = D (9) where 0 x n 0 y n and 0 z n are the SAW particle velocities at the surface and s r is the surface mass density. The particle velocities are related to particle displacement u by i i u j w n = This increase in the kinetic energy results in the decrease in the wave velocity. The power density P(power/area) carried by a wave can be related to wave energy U (energy/volume) stored in a lossless medium as n U P = (10). As in the lossless medium P constant implicitly differentiating above equation yields 0 0 U U D = D n n (11)
22 Combining equations (9) and (11) + + = D P P P z y x s w n w n w n r wn n n 2 0 2 0 2 0 0 0 4 (12) All the wave energy is concentrated near the surface hence as operating frequency increases, surface particle velocities increase in proportion to ( ) 2 w P Thus, the quantity in the parenthesis P i w n 2 0 being inde pendent of wave amplitude and depending only on the substrate material, remains constant. Taking all the substrate dependent constants together, the mass induced change in SAW propagation velocity can be written as; s m f c r n n 0 0 = D (13) where m C is the mass sensitivity factor and is given as + + = P P P C z y x m w n w n w n pn 2 0 2 0 2 0 0 2 (14) The incremental center frequency change due to mass loading effect is; 2 22 0 0 00 00 4 y m m s xz v f v v vv f v PPP wr www DD = = ++ (15)
23 2.4.2 SAW Acousto electric Re sponse SAW propa gating in the piezoelectric medium generates a layer of bound charge on the surface which accompanies the mechanical wave. When a conductive film is deposited on to the acoustic path, the charge carriers in the film redistribute to compensate for the layer of bound charge. The effect of the coupling of wave and charge carriers can be illustrated by the an equivalent circuit model shown below Figure 2.4 Equivalent c ircuit m odel f or SAW acousto electric response The current genera ted per unit area of surface, 0 I is given as ( ) P k K I s e e w + = 0 2 2 2 0 2 (16) i 1 i 2 i 3 I 0 e jwt k e 0 k e 1 AIR SOLID F k 2 s s
24 where 2 K Electromechanical coupling co efficient e 0 and e s are air and substrate dielectric permitivities k is acoustic wave number P is power density Displacement currents generated in the substrate and air arise from capacitances of s k e and 0 e k respe ctively. Shunt conductance s k s 2 accounts for the conduction currents in the film overlay To derive the changes in the velocity and attenuation because of acousto electric coupling we will study two different cases 1. Without conductive fil m In the absence of conductive film the energy generated by the wave gets stored in the evanescent electric field. The power flow in this case is given as, ( ) s T k j I P e e w + = 0 2 0 1 2 2. With the conductive film the power flow b ecomes [ ] ) ( 2 0 2 2 0 2 s s T k j k I P e e w s + + =
25 The total acousto electric response is nothing but the difference between the two power flows, which entirely depend on the film conductivity ( ) + = = s s s s T T T kc j k kc j k I P P P w s w s 2 2 2 0 1 2 2 (17) where s s c e e + = 0 Change in Complex propagation f actor ? is given as P k P v v j k T 0 0 0 2 = D D = D a g (18) which gives the general relationship between power transferred from the wave T P and the resul ting changes in the wave propagation characteristics. Substituting the above value of total power transferred into equation (17) and equating the real and imaginary parts gives ( ) + = D 2 0 2 2 2 0 2 s s s c v K v v s s (19) (20) The above equation shows that the SAW velocity decreases as the attenuation goes up. Also, the magnitude of the acous to electric response is proportional to 2 K and thus is substrate dependent. ( ) + = D 2 0 2 0 2 2 s s s s c v c v K k s s a
26 Chapter 3 Fabrication 3.1 Introduction SAW devices can be fabricated by using several different methods of lithography. L ithium niobate was selected as a substrate due to its high electromechanical coupling coefficient and to avail the benefits of strong acousto electric effect in sensor response. Low frequency devices up to 200 MHz were fabricated by using traditional optical lithography while the high fre quency devices (900 MHz) were fabricated by using electron beam lithography because of submicron line widths. 3.2 Optical L ithography Lithography is the transfer of a pattern to a photosensitive material (photo resist) by selective exposure to a radiat ion source. Optical lithography is nothing but the exposure of the photosensitive material to a radiation source with the wavelengths within the visible spectrum. The lithography process consists of several key steps ; 1. Design and Layout 2. Mask fabrication 3. Photoresist application 4. Mask alignment 5. Development
27 The design of a mask will vary according to ones specific needs. The first step is to design the device and identify the layers, which will comprise the final device followed by the layout in any CAD software. The next step is to make a photomask which can be used to transfer pattern on many wafers. Photomasks are fabricated on various types of fused silica. The most important properties of the mask include a high degree of optical transparency, a smal l thermal expansion coefficient and a flat highly polished surface that reduces light scattering. On one surface of this glass is a patterned layer, mostly of chromium. The first mask was designed to get SAW devices with the frequency ranging from 50 to 1 00 MHz as shown below Figure 3.1 Mask for SAW devices with frequency 50 MHz to 100 MHz Although the device yield was good, the dicing was too difficult and time consuming for the wafers processed by using above mask. So, while designing the mask
28 for 200 MHz devices, the problem of dicing was considered and the following mask was designed It had twelve 200 MHz dual delay line SAW devices. Figure 3.2 Mask for SAW devices with frequency 200 MHz The masks for sensing layer were fabricated by simple mill ing of an aluminum plate because the dimensions were large enough. The aluminum pate acted as a shadow mask. Photoresist which is typically an organic material or polymer can be applied to the surface of wafers by various techniques like spinning, s praying or laminating. Generally, spinning is the most used method because of its simplicity and excellent uniformity of layer. Photoresists can be divided according to the polarity as positive and negative
29 photoresists. Following exposure of the photoresi st to the light, the wafer is immersed in the developer solution. The positive photoresist responds to the light in such a way that the exposed region gets dissolved in the developer. In the case of negative photoresist, the unexposed region gets dissolved Positive resists tend to have better resolution and hence they are much popular in the fabrication industry. For the exposure of the photoresist expensive mask aligners are used. 3.2.1 SAW F abrication D etails For the SAW devices, we designed two mask s, one with the devices having their center frequency ranging from 50 MHz to 100 MHz and the second comprising devices of 200 MHz center frequency. The fabrication of the IDT is a single mask process. We used two different techniques, etching and lift off for fabrication of our devices. a) Etching An etching procedure utilizing the positive photoresist requires use of a lithographic mask that is opaque in the regions where metal is to be retained this is called a clear field mask. First the lithium nio bate wafer is coated with the aluminum layer with chromium as a seed layer to improve adhesion, which can be done by several ways, and simplest is by e beam evaporation in a high vacuum system. Then, the completely metalized substrate is over coated with a layer of photoresist by flooding the substrate surface with a solution of the resist, then spinning at a high speed ( 1000 4000 rpm ), producing a very uniform layer of the resist. The list of different photoresists used in this research and the required conditions are supplemented in an appendix. This layer
30 is heated to drive off solvents, and then exposed to UV light with the mask in direct contact or extremely close to the substrate. Exposure was provided by Karl Suss mask aligner, which can hold the s ubstrate and mask in close, uniform alignment. It is very important to align the IDT pattern so that the acoustic wave will propagate in the intended crystallographic direction. Round substrate wafers are typically provided with a flat oriented normal to the wave propagation direction for this purpose. Developing the exposed resist with a selected mixture of solvents removes the resist where it was exposed to light. The metal is then etched, in the regions unprotected by photoresist, by an appropriate etc hant. Stripping away the remaining photoresist by using acetone leaves the finished devices. Figure 3.3 Process flow for etching Cleaning Photoresist coating Cr Al metallization UV expos ure PR development Metal etching PR removal
31 The etching procedure using a negative photoresist is identical except that the lithographic ma sk must be opaque in the regions where metal is to be removed this is known as the dark field mask. Exposure of negative resist to UV light makes it insoluble in its developer. b) Lift off technique A lift off procedure is somewhat different. A layer of photoresist is deposited on the bare substrate, then exposed and developed prior to metallization. Figure 3.4 Process flow for lift off Cleaning Photoresist coating UV exposure PR development Cr Al metallization Lift off in acetone
32 The substrate wafers are cleaned using acetone, methanol. Then, the wafers are rinsed in De ionized water for about 5 minutes. After this, the substrate is heated at high temperature to remove any surface moisture. After cooling, the photoresist is spin coated, soft baked, and then the, wafer is exposed to UV light, so that the regions of the resist that are exposed become soluble to the developer. The wafer is developed until the sections that have been exposed to UV light are etched away. Then the thin patterned wafer can be coated with aluminum by using evaporation. The wafer is submerged in the solvent to facilitate liftoff. 3.3 E beam L ithography Electron beam lithography (EBL) is a specialized technique for creating extremely fine patterns (down to 10 nm ) on wafers. This technique consists of scanning of an electron beam across a surfa ce covered with a resist film sensitive to those electrons. The beam of electrons deposits the energy in the desired pattern in the resist. The exposed resist undergoes bond fission process so that after immersing in the developer solution it can get diss olved to form desired pattern of resist on the surface of resist. Despite having very high resolution capability and versatility to use with different materials, EBL has certain disadvantages such as cost, complexity and very slow speed.
33 Typical EB L system consists of a scanning electron microscope with beam blanker and computer. Shown below is a block diagram of sun an EBL system. Figure 3.5 Typical EBL system Scanning electron microscopes consists of distinguished s ections of 1. A column which generates a beam of electrons. 2. A specimen chamber where the electron beam interacts with the sample. It also has the facility to load and unload the sample. Associated with the chamber is a vacuum system needed to maintain an ap propriate vacuum level throughout the machine and also during the load and unload cycles. Stage H.V.Power Supply Computer Pattern generator Blanking amplifiers Pattern data storage Stag e controller Lens power supplies Vacuum system Chamber Column Electron gun
34 3. Detectors to monitor the different signals that result from the electron beam/sample interaction. A set of control electronics supplies power and signals to various parts of the machine. 4. A viewing system that builds an image from the detector signal. The SEM has a five axis motorized stage controller that can move the sample in one of five ways: Horizontal (X axis) Vertical (Y axis) Up/down (Z axis) Tilt Rotation T he SEM uses a beam of electrons to scan the surface of a sample to build a three dimensional image of the specimen. The electron gun is housed on the top of the column and generates the beam of electrons that rushes towards the sample housed in the specim en chamber. Electrons are very small and easily deflected by gas molecules in the air. Therefore, to allow the electrons to reach the sample, the column is under a vacuum. The vacuum is maintained by two vacuum pumps: a rotary pump and oil diffusion pump w hich are housed inside the SEM and are water cooled. Thus, the SEM needs a filtered water cooling line to cool the oil diffusion pump. Within the electron gun is the filament which is the source of the beam of electrons. The filament is made of tungsten a nd is heated to generate a fine beam of electrons
35 Current must flow through the filament for the electron beam to be generated. As voltage is slowly applied across the filament, electrons begin to flow from the filament to form the electron beam which wil l scan the sample. Most tungsten filaments emit electrons initially at one side of the filament loop as the filament begins to heat up. This causes a first peak of saturation of the current of the electron beam. As more voltage is applied slowly to the fil ament, the filament current increases and the electron beam moves towards the tip of the wire loop of the filament. As this happens, the current of the electron beam falls slightly. As the filament heats further when more current is applied, the electron b eam will flow steadily from the very tip of the filament's wire loop. This causes a second peak of saturation of the current of the electron beam. This is the point of filament saturation at which a maximum number of electrons gets emitted from the filamen t Beyond this point, if the filament current is increased further, the beam current no longer increases. Increasing the filament current beyond the point of saturation is called over saturation which greatly decreases the life of the filament and could ac tually melt the filament in a few minutes. The accelerating voltage is the speed of the electrons in the electron beams that stream towards the specimen. They are measured in kilovolts (KV). A high accelerating voltage allows the electrons to penetrate de eply into the sample, and a low accelerating voltage provides information from the very surface of the sample. For EBL, high accelerating voltage is preferable to get a good resolution. The spot size refers to the width of the electron beam that comes fr om the filament. These electrons will be scanning the surface of the specimen. It is like scanning
36 a structure in the distance with a spot light, or a finely focused laser beam. What influences our choice? Smaller the spot size, greater the resolution. The re's a catch though. Although it gives the better resolution, the small spot size produces fewer electrons that are more difficult to detect. Images appear very grainy. Before doing the actual pattern writing the beam should be focused perfectly with opti mized spot size. 3.3.1 Process F low 1. A SAW device was designed for center frequency of 900MHz and drew the IDTs in DesignCad software which is very well compatible with the Nano Pattern Generation System (NPGS), which is the system installed on top of the JEOL 840 SEM for EBL. After that, a pattern file was created in NPGS menu followed by a run file. While creating the run file we have to supply information such as area dose and beam current. The values of those were obtained by optimizing the parameters and after drawing several different test patterns. 2. The bond pads with the alignment marks for IDTs were fabricated on the lithium niobate wafer by using simple optical lithography because it is not feasible to write the large, 4mm X 5mm bonding pads by us ing EBL. The mask was made on an emulsion coated glass plate by using projection photolithography. For the gold bonding pads, simple lift off was carried out using PR 3000 PYi. 3. The wafer having bond pads for two devices was coated with the photoresist PMM A. PMMA, poly(methyl methacrylate), with a molecular weight of 950,000 is the most common resist used for fine lithography because of its high resolution. PMMA resist was spun on to the wafer and baked for 2 hours at 160C. The recipe uses 3% by weight of
37 950,000 MW (molecular weight) PMMA in chlorobenzene spun at 500 rpm for 3 seconds then spun at 4000 rpm for 20 seconds. The achieved thickness of resist was 195 nm. 4. The NPGS equipped JEOL 840 SEM can handle maximum sample size of 1 inch X 1 inch. Hence wa fer was diced in to two devices of 1cm X 2 cm. 5. For SEM lithography, the sample should be conductive enough to avoid charging of the sample at some spots. In our case, lithium niobate is non conductive so to avoid charging issues we decided to put a very t hin layer of Au Pd layer. 6. Au Pd layer was put by sputtering. D ifferent patterns with different thicknesses of Au Pd were studied to get the optimized layer thickness. Figure 3.6 Pattern written by EBL on lithium niobate with 10nm Au Pd Fi rst, a sample of lithium niobate was coated with photoresist PMMA 950K molecular weight. Then, it was coated with 10 nm of Au Pd by sputtering. After writing a pattern, the sample was developed. As we can see from the image in Figure 3.6, the Au Pd layer w as too thick to be removed and also the resolution of the pattern is not good. Then, few pieces of lithium niobate wafer were coated by using the same procedure but different thicknesses of Au Pd such as 3, 5, 7nm. Out of which the pattern
38 with 3nm coatin g yielded the best results. As we can see in Figure 3.7 below the sample was clean and the resolution was also very good. Figure 3.7 Pattern written by EBL on lithium niobate with 3nm Au Pd 7. After putting 3nm of Au Pd alloy the sample was mo unted on a sample holder. As shown in Figure 3.8, the sample holder has a Faraday cup and a gold resolution standard. While mounting the sample, several precautions must be taken such as putting the sample right below the Faraday cup with the perfectly squ ared upper right hand corner. A reference corner in the upper right is convenient when the stage coordinates increase when the area viewed in the SEM moves down and to the left. Mounting the sample firmly and perfectly flat by using spring clips and copper tape is important. 8. After characterizing the Au Pd film thickness, a sample of lithium niobate wafer was coated with 3 nm of Au Pd alloy and wrote a practice pattern so as to optimize the values of area dose and beam current.
39 Figure 3.8 SEM sample hold er with Faraday cup, gold standard and lithium niobate with bonding pads Figure 3.9 shows an array of written patterns. Here, each pattern is written at different area dose. We selected the value of area dose which yielded pattern with the best resolution. The measured value of area dose was found to be 420 C/cm 2 9. Load actual sample in SEM chamber and get a good gold resolution standard image. 10. Then, align the actual sample and set the beam blanker on an external control. Start writing the pattern by usin g NPGS menu.
40 Figure 3.9 Pattern written by EBL with different area doses 11. After writing pattern, take the sample out of the chamber. To develop the sample, use a solution of 3:1isopropyl alcohol to methyl isobutyl ketone for 20 seconds, followed by 20 s econds in isopropyl alcohol and 30 seconds in water. 12. After writing, put gold by evaporation followed by simple lift off to get the desired structures. 3.4 Sensing L ayer F abrication After the fabrication of SAW delay line by above procedures, we put the sensing layer. The sensing layer is deposited on the acoustic path of the devices. The sensing layer consists of a bilayer of metal free phthalocyanine and nanocrystalline palladium.
41 To deposit the sensing layer a shadow mask made up of aluminum which pro tects the IDTs was used We aligned the wafer and mask manually and clamped them together so that only the acoustic path of delay line was exposed. A thin layer of metal free phthalocyanine was coated by using sublimation. The process was carried out in a chamber under a vacuum of around 5 X 10 5 Torr. The powdered metal free phthalocyanine was contained in a baffled tantalum box covered with sieved lid. This specialized boat was required to direct the vapors straight and to get material deposited on wafer which was mounted at a large distance from the source. The thickness of the phthalocyanine layer was 115 nm. Nano crystalline thin films are known to have much larger grain boundary volume in the material, leading to increased rates of diffu sion of gases. Nano crystalline thin films of several materials such as Cu, Fe, Mo and Si have been deposited using sputtering. 20 30 40 50 60 0 20 30 40 50 60 0 20 30 40 50 60 0 PdSi (200) Pd (111) RTA 350 o C RTA 250 o C as deposited Pd on Silicon substrate Intensity (a.u.) 2 q Figure 3.10 XRD plot of nano crystalline Pd film
42 Our preliminary work on Pd films was carried out by sputtering using a mul ti gun sputterer at pressures of 1.7x10 2 Torr to 5x10 2 T orr. These films were subjected to rapid thermal annealing at 250 and 350 C. X ray diffraction (XRD) measurements on these films revealed that Pd crystallizes as deposited. XRD profiles are shown in Figure 3.10. For deposited films, these showed reflecti ons of (111) and (200) peaks, both broadened, indicating nanocrystalline nature of the films. Films subjected to anneal also showed broader peaks, however, a small peak of PdSi emerged due to the reaction between the substrate and the film. From these pro files, a crystallite size of 30 nm was obtained. The thickness of the film used was 200 nm.
43 Chapter 4 Measurements 4.1 SAW Device Measurements The response a SAW device to an external perturbation for chemical sensing can be better underst ood if the device frequency response is known in advance. Measurement of the frequency response is also important if the most stable and accurate measurement system is to be designed for a particular device. Our SAW devices are measured by using the Agilen t 8753ES vector network analyzer. It is very useful to monitor velocity and attenuation both simultaneously. Most of the acoustic sensor research relies solely on the measurement of velocity perturbations. Such an approach is adequate if it is known in ad vance that all perturbations will affect wave velocity alone; such perturbations include mass effects and changes in the elastic and mechanical properties. The complete characterization of an acoustic wave device is always obtained from a complete freque ncy response spectrum including all the scattering parameters. Scattering parameters (S parameters) are the set of parameters describing the scattering and reflection of traveling waves when a network is inserted into a transmission line. S parameters are normally used to characterize high frequency networks, where simple models valid at lower frequencies cannot be applied. S parameters are normally measured as a function of frequency.
44 For each port, the incident (applied) and reflected wave properties are measured. When the incident wave travels through the network, its value is multiplied (i.e. its gain and phase are changed) by scattering, thus giving the resulting output value. S parameters can be considered as the gain of the network, and the subscript s denote the port numbers. The ratio of the output of port 2 to the incident wave on port 1 is designated S 21 Likewise, for reflected waves, the signal comes in and out of the same port, hence the S parameter for the input reflection is designated S 11 F or a two port network, (assuming use of matched loads and characteristic impedence of 50ohms), S 11 is the reflection coefficient of the input, S 22 is the refleciton coefficient of the output, S 21 is the forward transmission gain, and S 12 is the reverse tra nsmission gain From output to input, S parameters are used to help characterizing a network. They have a specific importance for high frequency applications. S parameters as seen above are voltage ratio, so they have no units. In practical usage, S parame ters are expressed in dB. S parameters are complex values, with magnitude and phase, so we cannot just use an oscilloscope or signal analyzer which will give us only the amplitude information. This is why we used a Vector Network Analyzer, VNA, which is ca pable of measuring both the magnitude and the phase of a signal simultaneously. To determine the scattering parameters of any two port network, the measurement technique must be capable of measuring incident, reflected and transmitted waves in amplitude an d phase and establishing the relative values. Such facilities are provided on the network analyzer which allows signals to be routed to the input or output ports which are typically coaxial of the APC 7 type. These ports are used as the reference planes fo r the device under test (which may require
45 mounting in standard or specially designed test fixture to provide the interfacing to the measurement ports). Figure 4.1 Schematic of SAW device m easurement set up System accuracy is enhanced by the use of error correction techniques to remove systematic errors such as source and load mismatches, directivity, coupling variations with frequency etc. The error terms are derived by utilizing known stan dards such as fixed open and short circuits, fixed loads, sliding loads and through calibration. To determine the S parameters of the device mounted in the test fixture, the measurement reference planes must be referred through the test fixture to the devi ce reference terminals. 4.2 Design and F abrication of T est C ell SAW devices are sensitive to a large number of physical and chemical measurands. These include parameters such as temperature, pressure, stress, viscoelast ic parameters and electrical conductivity. In most of the cases, one or two of these sensitivities are studied and rest other responses become undesirable interferences. Thus, it is essential that the sensor environment be carefully controlled to eliminate the effects Vector network analyzer Port 1 Port 2
46 of sensor cross sensitivities. Again, frequency of the SAW device gets affected due to the perturbations from external ambient noise, i.e. the external RF signals which are mostly radio station signals or signals from wireless phones. Hence, S AW devices must be properly shielded in a metal RF cage. So, we designed and fabricated a test cell. The schematic of the test cell is shown in Figure 4.2. The test cell is comprised of a base which has gas inlet and outlet; it also has a groove to hold a SAW device. The base is fabricated by using 316L stainless steel. The lid of the cell was also made of stainless steel and has pogo pins inserted in ceramic sleeves to avoid short circuiting. The pogo pins have spring loaded tips to touch the bonding pads of the SAW device to avoid scratching and are gold plated to enhance the conductivity. The pogo pins were soldered to the SMB connectors mounted on the sidewalls of lid with the help of copper wires.
47 Figure 4.2 Saw device test cell schematic Aluminum Cradle Mount Lid Allen Screws Pogo pin insulated with ceramic sleeve Heating Water in Heating Water Out Flange C a s e 3 Pogo pins grounded to case with conductive epoxy Cradle Casing with Cradle and Mount O-ring Mount SAW-1 SAW-2 Dual Delay Line Configuration + + + + Input-1 Input-2 Output-1 Output-2 Gas In Gas Out
48 The cell can be placed on top of an aluminum cradle through which hot water can be circulated to control the temperature of the system. Then, the whole assembly sits on a wooden cradle to avoid heat losses. To avoid the gas leakage from the cell, an O ring was used in between the lid and the base. The O ring was fixed in a groove made up on the top surface of the base. Figure 4.3 Fabricated test cell 4.3 Problems E ncountered Although the test cell assembly looked very robust, there was a design flaw. The pogo pins which were soldered to the SMB connector had a little movement (play). While removing or fixing the lid, that little movement led to scratching of the devices. The scratched devices are shown in Figure 4.4.
49 Figure 4.4 Scratched SAW device IDTs As the IDTs got damaged, it was impossible to get desirable and stable response from the devices. Hence, we decided to change the design of the lid. Figure 4.5 Test cell with impro ved lid In the new lid design, t he upper aluminum housing was removed It was replaced by a FR4 board having pogo pins inserted in the same manner as in the previous lid. The FR4 board was fabricated with transmission lines which connected pogo pins with R F connectors. Gold plated SMA edge connectors were used for connection between network analyzer ports and test cell.
50 4.4 SAW S ensor T est bed A fully instrumented and automated gas dilution radio frequency test bed consisting of arrays of mass flow contro llers to achieve hydrogen concentrations of a few volume percent to ppm range has been designed and fabricated. An array of five mass flow controllers (MKS 1479A series) is used to mix hydrogen with nitrogen to get the desired concentration. Solenoid valve s are used to select specific gas flow lines. Both gases go through the mixing chamber to the temperature controlled test cell, where the SAW device is placed. Mixing Chamber Agilent 8753ES Network Analyzer Computer Input Port Output Port Hydrogen Mass Flow Controllers RF DPDT Switch Nitrogen To exhaust Solenoid Valve Solenoid Valve Temperature Control System Test Cell NI PCI-6033E Data Acquisition Card Cell Temperature Switching Signal Figure 4.6 SAW device measurement set up with dilution system and data acquisition The two po rt radio frequency measurements are done by an S parameter VNA For the dual delay line configuration, an RF switch is used to switch between the two delay lines. An IBM PowerPC functions as the control station using PCI bus data acquisition and GPIB cards to communicate with the instrumentation. The set up permits
51 us to analyze high speed and long term response characteristics of different gas sensors at various concentrations, sub/supra ambient temperatures, and pressures, with and without contaminants.
52 Chapter 5 Results and Discussions 5.1 Fabrication of Devices SAW devices of center frequency ranging from 50 MHz to 200 MHz were fabricated successfully by optical lithography by using two different masks as discussed earlier. Optical photo and micro g raphs of these fabricated devices are shown in Figure 5.1. (a) (b) (c) (d) Figure 5.1 Fabri cated SAW devices, (a) IDT of a 100 MHz SAW, (b) IDT of a 200 MHz SAW, (c) 100MHz and 80 MHz SAW single and dual delay line devices. (d) 200 MHz SAW dual delay line devices coated with sensing layer.
53 Fabrication of 900 MHz devices was attempted using E bea m lithography. As we can see in Figure 5.2, the electrodes were overexposed because lithium niobate is almost non conductive. Although the substrate was coated with optimized layers (PMMA 950K and 3nm Au Pd alloy), overcharging of the substrate happened wh ich resulted in overexposing of electrodes. Figure 5.2 (a) 900 MHz IDTs written by EBL (b) Misaligned IDT of a 900 MHz SAW device
54 Another problem we faced while writing the pattern was of alignment. After focusing the beam and measuring the beam curre nt with the help of gold resolution standard and Faraday cup we tried to move the sample so that the pattern can be written exactly between the bonding pads. But, at very high voltage and low current it was difficult to locate the exact position. In Figure 5.2, we can see that the written pattern is misaligned, where we can also see the alignment marks and part of bonding pads 5.2 SAW D evice C haracterization All the SAW devise were characterized by Agilent 8753ES Network Analyzer. Network analyzer with all the test leads and terminations were calibrated by using calibration kit 85033E. The maximum center frequency of every device fluctuates between +/ 1000 Hz depending on the external noise, vibrations and test fixture stability. The measurements were 1601 points in each sweep. Before measurements, devices were checked by using multimeter for any shorts. A simple continuity test was carried out between each pair of bonding pads. The S 21 and S 11 curves measured by network analysis were compared with the char acteristic curves of a SAW device. In some sweeps, averaging and smoothing options were used to get a stable curve. Some S parameter curves for SAW devices with center frequency 70, 100 and 200 MHz, respectively, are shown in Figures 5.3 through 5.5.
55 (a ) (b) Figure 5.3 (a) S 21 transmission curve, (b) S 11 reflection curve for a 70 MHz device Figure 5.3 shows the S 21 transmission curve with the peak frequency near designed frequency of 70 MHz and S 11 reflection curve showing a dip near resonance frequenc y with minimum insertion loss.
56 (a) (b) Figure 5.4 (a) S 21 transmission curve, (b) S 11 reflection curve for a 100 MHz device
57 (a) (b) Figure 5.5 (a) S 21 transmission curve, (b) S 11 reflection curve for a 200 MHz device
58 5.3 SAW S ensor M easureme nts The responses of a 200 MHz SAW sensor coated with 115 nm metal free phthalocyanine a nd 200 nm nano crystalline film is shown in figure 5.6 Figure 5.6 Response of a 200 MHz SAW sensor to hydrogen exposures The test was carried out at room temperatu re inside the test cell. The hydrogen was cycled when the maximum center frequency of the device repeated its value for a couple of minutes. As shown in the figures, different concentrations of hydrogen ranging from 1% to 6% were tried. Desired concentrati on of hydrogen was achieved by mixing it with nitrogen. The flow rate was kept at a total of 1000 SCCM.
59 Following are the response times of this sensor at different concentrations of hydrogen. This time includes the time it takes fro hydrogen to reach th e test cell. The transport lag was calculated, and it is roughly 8 30 seconds, d epending on the concentration. Table 5.1 Response times of the 200 MHz SAW hydrogen sensor Following are the average frequency shifts at different hydrogen percentages. Ta ble 5.2 Average frequency shifts for a 200 MHz SAW hydrogen sensor
60 The observed frequency shifts are much larger and are in easily detectable ranges by simple and inexpensive electronics. The best values of exact frequency changes can be established by experimenting with different thicknesses of Pd and using several runs. 5.4 Temperature E ffects on M easurements First of all the sensor was tested for its temperature stability. A 200 MHz SAW device was put inside the cell. Only temperature was the var iable and the rest of the system was kept undisturbed. The temperature was varied from 10 0 C to 50 0 C. As we can see from the graph in Figure 5.7 the frequency of the SAW device decreases linearly with increasing temperature. The measured temperature coeffi cient was ~113 ppm/ 0 C. Figure 5. 7 Temperature effects on SAW device frequency
61 This graph shows that temperature has a direct effect on the acoustic wave device performance. Changes in temperature produce change in the density of the substrate material which in turn changes the velocity of acoustic wave. In this case, it becomes difficult to operate the sensor in an environment where temperature fluctuates. The heat of adsorption and absorption of the analyte on to and into the sensing film, respectivel y, can change the device temperature in most cases.. There are three main methods by which the effects of temperatures can be minimized: 1. Use of low temperature coefficient material such as Quartz. 2. Incorporate a temperature sensor and compensation circuit ry 3. In situ control of temperature The stringent need of temperature control can be taken care of by using a dual delay line SAW device. In dual delay line configuration, one delay line is coated with the sensing layer while other is left uncoated. If we m easure the difference between the two device parameters, all the external effects including temperature will be nullified, except for the very local effects from the heats of adsorption and absorption. One more advantage of this particular set up is that t he resultant frequency shifts due to the analyte absorption will be in KHz range and can be measured very easily and economically. Figure 5.8 shows the response of 200 MHz SAW sensor at a temperature of 10 0 C. The same device which yielded the hydrogen res ponse at room temperature as shown in figure 5.6 was used.
62 The device was tested for a pulse of 1% hydrogen. As we can see from the graphs, the sensor is having longer response time. Also desorption of hydrogen from palladium film is a very slow process. Figure 5.8 Response of a 200 MHz SAW sensor at a temperature of 10 0 C Figure 5.9 is the response of a 200MHz SAW sensor at elevated temperature. The test cell is heated from room temperature to 65 0 C. Once the cell reached the desired temperature, a pulse of 1% hydrogen was given. Although the frequency of the device increased by 30 KHz in a shorter period of time than at room temperature, power loss takes longer response time and is not stable mainly because of very small fluctuation in temperature.
63 As w e can see in the plot of frequency vs time, there is a drop in the frequency of the device marked by a small curve on left side. This is mainly because of the temperature effect on SAW device frequency as we have discussed this earlier in Figure 5.8. Fig ure 5.9 Response of a 200 MHz SAW sensor at a temperature of 65 0 C Also, if we compare frequency plot vs temperature plot we can see that there is a significant time lag. The temperature of the cell is monitored by using a thermocouple and the substrate ta kes longer time to reach an equilibrium temperature than the cell itself. Hence, the pulse of the hydrogen was given right after cell and sensor reached the same temperatures.
64 5.5 Repeatability Repeatability of the sensor was also studied. A pulse of 1% hydrogen was cycled at room temperature. As shown in Figure 5.1 0 the sensor shows very good repeatability with a frequency shift of around 40 KHz every time when there are no temperature fluctuations. The second peak is widened because hydrogen pulse dur ation was longer than the previous one. Figure 5.10 Response of a 200 MHz SAW sensor for 1% Hydrogen cycling 5.6 Effect on P alladium F ilm Palladium film before and after the test runs was studied. After several cycles of absorption and desorption of hy drogen, the palladium film started peeling and cracking.
65 Figure 5.1 1 Nano crystalline palladium film before and after hydrogen cycling When hydrogen diffuses into palladium a hydride bond develops that crinkles the pure metal film, as can be seen eas ily in Figure 5.1 1 When hydrogen gets absorbed in a palladium film, it expands. This expansion of palladium with an underlying metal free phthalocyanine layer leads to internal stresses between both the films and eventually the film gets cracked. We can s ee in Figure 5.1 2 that the palladium film is cracked badly and the brick colored metal free phthalocyanine is exposed. 5.1 2 Nano crystalline palladium film peeling and cracking
66 Chapter 6 Conclusions and Future Work In this thesis, the m odeling and design of SAW sensors of different center frequencies have been described The designs of SAW parameters are done so as to achieve impedance matched input and output IDT sets and to avoid power losses. Synthesis of nano crystalline palladium i s achieved by optimizing the parameters of sputtering process. SAW devices of different center frequencies have been fabricated by using optical lithography. A fully instrumented gas dilution system and RF test bed to test sensors has been constructed. A SAW sensor of 200 MHz center frequency is tested for different hydrogen concentrations. The nano crystalline palladium reduced the response time to order of a few seconds; a significant improvement over existing palladium based hydrogen sensors, where the response times are on the order of hundreds of seconds. Temperature stability of the sensors has been studied. The temperature coefficient of lithium niobate was found to be 113 ppm/ 0 C. So, to operate the sensor with maximum efficiency and linearity, the temperature should be kept constant. The sensor response at different temperatures has been studied. At lower temperatures, sensor response is slow and the regeneration times are also longer. The sensor was also tested at higher temperatures of around 65 0 C. The response is faster than
67 that at room temperature but power loss takes longer response time and is not stable mainly because of very small fluctuation in temperature. The optimum temperature has to be found by testing the sensor at different temperat ures and comparing their results for maximum response and recovery time. Also, limit of detection with respect to temperature has to be studied. The sensor was also studied for its repeatability. At room temperature the sensor has shown the same frequency shifts. The sensing film thickness has to be studied for the response and the detection limit. Especially, very thin films will have very low regeneration times compared to the film we have used. Again, different nanostructures of palladium has to be put or grown on the acoustic path of SAW device and should be tested at different hydrogen concentrations. In this case, the area exposed to the hydrogen will be many times higher than the surface area of a film, which will surely improve the adsorption and de sorption of hydrogen and so the sensor response. With the 200MHz SAW sensor we have detected 0.5% (5000ppm) of hydrogen, which can further go down but a more stable measuring system such as oscillatory circuits to drive dual delay line devices which will c ompensate for temperature and other substrate dependent effects. Again, high frequency SAW devices have to be fabricated and tested, which promise an even faster response to lower concentrations of hydrogen. The fabrication of 900 MHz SAW device was att empted More tests and optimization of E beam lithography for SAW devices are needed which involve faster pattern writing and
68 alignment of SAW devices. New process flow has to be designed so that lithium niobate will be slightly conductive and then overcha rging issues will be avoided.
69 References 1) NRC (National Research Council) Expanding the Vision of Sensor Materials, National Academy Press, Washington D.C., 1995 2) Janata. J., Principles of Chemical Sensors, Plenum Pres s, New York, 1989. 3) Cui, Y., et al., Nanowire Nanosensors for Highly Sensitive and Selective Detection of Biological and Chemical Species, Science Vol. 293, pp. 1289 1292, Aug. 17, 2001. 4) Moore, S.K., U.S. Nanotech Funding Heads for $1 Billion Horizon, IEEE Spectrum Online 5) W. Wallraff and On behalf of the AMS Collaboration. The Si tracker for the AMS 02 experiment on the Space Station, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associate d Equipment, Volume 511, Issues 1 2, 21, Pages 76 81, September 2003. 6) US department of Energys Hydrogen program, Catherine Gregoir Padro, National Renewable Energy Lab. 7) A. Katsuki, K. Fukui, H2 selective gas sensor based on SnO2, Sens. Actuators B 52, 30 37 1998 8) C. Christofides, A. Mandelis, Solid state sensors for trace hydrogen gas detection, J. Appl. Phys. 68 (6), R1 R30, (1990). 9) Wohltjen, H. and R. Dessy, Anal. Chem., 51, 1465,1979. 10) Drafts, B., Sensors, 17 October, 2000. 11) Pohl, A., IEEE Trans. Ultrasonics, Ferroelectrics, and Freq. Control, 47 317 2000 12) Dong, Y., Cheng, W., Wang, S., Li, Y., and Feng, G., Sensors and Actuators B, 76 130, 2000. 13) Jakubik, W.P., Urbaczyk, M.W., Kochowski, S., and Bodzenta, J., Sensors and Actuators B, 82 26 5, 2002
70 14) Mller, C., Nirmaier, T., Rgemer, A., and Schickfus, M.V., Sensors and Actuators B 68 69, 2000. 15) Shaul Katzir, Measuring constants of nature: confirmation and determination in piezoelectricity, Studies In History and Philosophy of Science Part B: Studies In History and Philosophy of Modern Physics, Volume 34,Pages579 606, Issue 4, December2003. 16) Ballentine, D.S., Jr., White, R.M., Martin, S.J., Ricco, A.J., Frye, G.C., Zellers, E.T., and Wohltjen, H., Acoustic Wave Sensors: Theory, Design and Physico Chemical Applications Academic Press, New York, NY, 1997.
72 Appendix A Table A.1 The various s pin o n photoresists used in fabrication Name Spin speed (rpm) Spin time (se c) Soft bake ( 0 C) Time (sec) Exposure (sec) Developer Solution name(sec) Hard bake ( 0 C) Time (sec) Shipley 1813 3000 30 90 60 8 MIF319(60) 120 60 Shipley 1827 2000 30 100 80 18 MIF319(100) 120 90 AZ 4620 4500 40 100 80 30 AZ726(1800) 120 90 3000 PY 4000 50 155 60 13 RD 6(25) 110 60 Note: The exposure time varies with the intensity of the lamp used in the mask aligner. The above values of exposure time are characterized for Karl Suss mask aligner. In the case of photoresist 3000 PY, the hard bake is done prior to the development.