A Comparison of Two and Three Dimensional Wire Antennas for Biomedical Applications. Shruthi Soora

Transcription

1 ABSTRACT SOORA, SHRUTHI. A Comparison of Two and Three Dimensional Wire Antennas for Biomedical Applications. (Under the direction of Prof. Gianluca Lazzi.) Miniature antennas are necessary to reduce the size of communications and biomedical devices, specifically for a retinal prosthesis. A comparison of two dimensional and three dimensional antennas are presented as an attempt to miniaturize and reduce the antenna footprint size while enhancing antenna characteristics such as bandwidth, maximum gain and the radiation pattern. This work explores various two dimensional antennas and converts them into three dimensional implementations by folding and rotating the dipole arms. This changes the current vector alignment which can enhance the antenna characteristics. Further comparison between the 2D and 3D antennas are investigated in a transmitter/reciever system. The 2D and 3D antennas are tested and compared in air and inside eye phantoms to replicate the communications link of a retinal prosthesis system. To accurately model the vitreous humor of the eye, methods to mimic the electrical properties of a biological medium are also investigated. This work demonstrates the feasibility of three dimensional wire antennas for the application of a retinal prosthesis. Three dimensional wire antennas are ideal due to their smaller planar size, wider bandwidth and coupling characteristics in comparison to the two dimensional implementation.

2 A Comparison of Two and Three Dimensional Wire Antennas for Biomedical Applications by Shruthi Soora A thesis submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the Degree of Master of Science Electrical Engineering Raleigh, NC 2005 Approved By: Dr. Kevin Gard Dr. Doug Barlage Dr. Gianluca Lazzi Chair of Advisory Committee

3 ii Biography Shruthi Soora grew up in Cary, North Carolina. She graduated Summa Cum Laude from North Carolina State University in 2004 with a Bachelor of Science in Computer Engineering and Electrical Engineering. She is a member of IEEE, Eta Kappa Nu and Phi Kappa Phi and was the chair of the NCSU IEEE Student Branch in Since 2004, she has been working towards her Masters in Electrical Engineering at NC State University. Her research interests include miniature antenna design for biomedical applications.

4 iii Acknowledgements I would like to thank my advisor, Dr. Gianluca Lazzi whose undergraduate electromagnetics class sparked my interest in antenna design. I appreciate all of your guidance, advice and the opportunity to work in your research group. I also would like to thank Dr. Kevin Gard and Dr. Doug Barlage. Thanks for all of your advice and taking the time to be part of my committee. Thanks to Keyoor Gosalia who taught me so much about antenna design and so much more in this field. To Anand Konanur who helped me with experimental setup and always provided advice for my ideas and questions. To Ajit Rajagopalan who helped me with learning the milling machine. To Stefan Schmidt who has been a great officemate and all of your advice. To Kevin Taff who helped with the fractal antennas. Also thanks to everyone in Dr. Lazzi s research group for all of your support and made coming to the lab fun and enjoyable: Carlos Cela, Vinit Singh, Srinivas Jasti, Gaurav Gupta, Randy Barlow, Sundar Srinivas and Ben Gaddy. I would like to thank my family in the US and abroad for all of your encouragement in everything I have pursued. I would also like to thank my close friends, Erin, Sarah, Kate and Bradley for all of their support. Finally, I would like to thank my parents and sister who always told me to follow my dreams and have encouraged and supported me through all of my endeavors.

5 iv Contents List of Figures List of Tables vi viii 1 Introduction Background of Implantable Devices Approaches to Visual Prosthesis Subretinal Implants Epiretinal Implants Optical Nerve Implants Cortical Implants NCSU/USC Approach to Retinal Prosthesis Motivation for Research in Miniature Antennas Design and Simulation of Miniature Wire Antennas Fractal Antennas Fractal Antenna Simulations and Results Disadvantages of Fractal Geometries Matching Techniques Three Dimensional Wire Antennas Spiral Dipole Antenna S-Spiral Dipole Antenna Triangular Spiral Dipole Antenna Comparison of 2D and 3D Antennas Fabrication Methods and Results Fabrication Methods Fabrication Process Fabricated Results and Comparison for Individual 2D and 3D Antennas Simulation Method

6 v Fabricated Antenna Results Methods of Replicating Biological Mediums Procedure to Measure Experimental Vitreous Humor Experimental Results for Vitreous Humor Comparison of 2D and 3D Antennas in a Rx/Tx system Coupling Measurements in Air Air Coupling Testing Apparatus D Spiral Antenna Measurements in Air D Spiral Antenna Measurements in Air D Rotate Spiral Antenna Measurements in Air Comparison between 2D and 3D antennas in Air Coupling Measurements in Eye Phantom Eye Coupling Testing Apparatus Air-Filled Eye Phantom Measurements Vitreous Humor Filled Eye Phantom Measurements Comparison of 2D and 3D Antennas in the Presence of an Eye Model 74 5 Conclusion and Future Work Future Work Bibliography 79

7 vi List of Figures 1.1 Eye Diagram (Photo Courtesy of National Eye Institute and National Institute of Health) Image as seen by an Age Related Macular Degeneration Patient (Photo Courtesy of National Eye Institute and National Institute of Health) Image as seen by a Retinis Pigmentosa Patient (Photo Courtesy of National Eye Institute and National Institute of Health) First Five Iterations of the Hilbert Fractal First Five Iterations of the Sierpinski Arrowhead Fractal First Four Iterations of the Peano Fractal Simulation Results for the 4th Iteration Hilbert Fractal Simulation Results for the 5th Iteration Hilbert Fractal Simulation Results for the 2nd Iteration Peano Fractal Simulation Results for the 3rd Iteration Peano Fractal Simulation Results for the 4th Iteration Sierpinski Arrowhead Fractal Simulation Results for the 5th Iteration Sierpinski Arrowhead Fractal Simulation Results for the 2D Spiral Antenna Simulation Results for the 3D Spiral Antenna Simulation Results for the 3D Rotate1 Spiral Antenna Simulation Results for the 3D Rotate2 Spiral Antenna Simulation Results for the 2D S-Spiral Antenna Simulation Results for the 3D S-Spiral Antenna Simulation Results for the 3D Rotate1 S-Spiral Antenna Simulation Results for the 3D Rotate2 S-Spiral Antenna Simulation Results for the 2D Triangular Spiral Antenna Simulation Results for the 3D Triangular Spiral Antenna Simulation Results for the 3D Rotate1 Triangular Spiral Antenna Simulation Results for the 3D Rotate2 Triangular Spiral Antenna Simulation Results for the 3D Rotate3 Triangular Spiral Antenna Simulation Results for the 3D Rotate4 Triangular Spiral Antenna.. 41

8 vii 3.1 Fabrication Process Steps D Spiral Antenna and Results D Spiral Antenna and Results D Rotate2 Spiral Antenna and Results Open Ended Coaxial Probe Vitreous Humor Experimental Results Air Coupling Testing Structure Location of Measurement Points D Spiral Antenna Air Coupling Results at Different Points D Spiral Antenna Air Coupling Results between 0 to 40 Degrees D Spiral Antenna Air Coupling Results between 0 to -40 Degrees D Spiral Antenna Air Coupling Results at Different Points D Spiral Antenna Air Coupling Results between 0 to 40 Degrees D Spiral Antenna Air Coupling Results between 0 to -40 Degrees D Rotate Spiral Antenna Air Coupling Results at Different Points D Rotate Spiral Antenna Air Coupling Results between 0 to 40 Degrees D Rotate Spiral Antenna Air Coupling Results between 0 to -40 Degrees Eye Coupling Testing Structure Air Filled Eye Phantom Results D and 3D Fabricated Antennas and Patch Antennas Vitreous Humor Filled Eye Phantom Results

9 viii List of Tables 2.1 Fractal Simulation Results for Various Geometries and Iterations D and 3D Spiral Antenna Simulation Results D and 3D S-Spiral Antenna Simulation Results D and 3D Triangular Spiral Antenna Simulation Results Comparison of 3D Antenna Characteristics to 2D Antenna Characteristics for the Spiral Antennas Comparison of 3D Antenna Characteristics to 2D Antenna Characteristics for the S-Spiral Antennas Comparison of 3D Antenna Characteristics to 2D Antenna Characteristics for the Triangular Spiral Antennas D and 3D Air Coupling Results at a 25 mm Distance D and 3D Air Filled Eye Phantom Coupling Results at a 25 mm Distance D and 3D Vitreous Humor Filled Eye Phantom Coupling Results at a 25 mm Distance

10 1 Chapter 1 Introduction Today s environment relies on communication devices for all aspects of our daily routine. Advances in science and technology over the past few decades have greatly enhanced our quality of life. In order to continue improving devices, engineers strive for more compact components with enhanced performance. Improvements in circuit and VLSI design have created compact, efficient and low power processors with faster data rates. Developments in technology have allowed researchers in the medical field to understand how the body communicates, heals itself and functions. With technological components becoming smaller and the body s workings become less of a mystery, it is possible to create devices that mimic the actual workings of human organs or sensory components. Since most prosthetic devices will be implanted inside of the body, it is necessary to have wireless communication between the external and internal components in order to provide a less invasive mechanism. As an alternative to using wires, miniature sized antennas are crucial to build compact biomedical devices. The motivation of this research project is to develop miniature antennas for communications and biomedical applications. One application for miniaturized antennas

11 2 is for use in retinal prosthesis devices to communicate images from an extraocular component to an intraocular one thus allowing a blind person to see. Further details on various approaches to create a retinal prosthesis will be explained in the following sections. The challenge of developing a compact implantable antenna to sit inside the vitreous humor of the eye is the main focus of this research project. More specifically three-dimensional antenna designs are explored as an effort to increase the bandwidth and gain while minimizing the overall antenna footprint size. 1.1 Background of Implantable Devices With the technology available today, it is possible to realize various ideas that were impossible in years past. Since the body communicates with electrical signals and pulses, it is possible to mimic the body s function by stimulating the neurons or muscles with externally generated pulses and signals [1]. With the medical field merging with modern technology, surgical procedures have become more advanced and it is becoming easier to replace biological parts with mechanical devices. Research in this field has allowed people to use pacemakers to control irregular and slow heartbeats. Muscle stimulators allow people with strokes to recover some strength and functionality of their extremities [1]. A cochlear prosthesis allows deaf people to recover some of their hearing [2, 3]. In the recent years, various research groups have been working on building a retinal prosthesis system to restore partial vision to the blind by converting the natural process of receiving and processing light signals into an electrical device. The human eye receives light that travels through the cornea to the lens. The iris dictates how much light enters the pupil by changing its shape. The lens focuses the light and forms an inverted image on the retina, which contains around 130 million

12 3 Figure 1.1: Eye Diagram (Photo Courtesy of National Eye Institute and National Institute of Health) photoreceptor cells [4]. The photoreceptor cells are made up of rods and cones. Rods provide color in low light and are associated with peripheral vision while the cones give the vivid color for images. The most sensitive portion of the retina is the macula region, which is responsible for the central portion of vision. The photoreceptors convert the light into electrical pulses. The electrical signal passes through the various cell layers (horizontal, bipolar and amaerine) of the retina and finally reach the retinal ganglion cells whose axons compose the optic nerve which connects the eye to the brain [5]. Two common retinal degenerative diseases are Retinis Pigmentosa (RP) and Age Related Macular Degeneration (AMD). Retinis Pigmentosa causes the rods in the retina to lose function causing a loss of peripheral vision. RP affects approximately

13 4 Figure 1.2: Image as seen by an Age Related Macular Degeneration Patient (Photo Courtesy of National Eye Institute and National Institute of Health) 400,000 people in the United States [6]. Age Related Macular Degeneration causes a loss of the central vision. Ninety percent of AMD patients have Dry AMD where part of the macula s cells stop working. Wet AMD occurs when new blood vessels grow in the macular region which tend to leak liquid under the macula causing the cells to stop operating. AMD affects 7.6 percent of people over the age of 40 in the United States [7]. Although people with AMD and RP gradually lose their sight, the function of the photoreceptors are lost but the remainder of the inner retinal layers may still function. Since the ganglion and other layers are still functional and intact, it is possible to artificially stimulate the remaining cells through electrode stimulation [4,8]. This idea has been demonstrated to restore the sensation of light for the blind through clinical investigations where electrodes were placed on the retinal layer of human patients. With the implanted electrodes, patients have been able to detect a phosphene, which is the perception of light [9]. Patients were also able to perceive basic shapes and

14 5 Figure 1.3: Image as seen by a Retinis Pigmentosa Patient (Photo Courtesy of National Eye Institute and National Institute of Health) identify location and movement with 4x4 and 5x5 electrode arrays [10, 11]. Studies show that at least 300 electrodes are needed to recognize small words in the central vision area while 600 electrodes are necessary for reading a full page and gaining useful vision. In addition, additional training is needed for the patient to adjust and the brain to adapt to the new form of vision [12 14]. Although these trials have demonstrated the feasibility of visual prosthetic devices, a larger array of electrodes and more complex devices are required to restore a usable sight for the blind. 1.2 Approaches to Visual Prosthesis There are various ways to recreate vision for the blind by stimulating various locations in the retina (subretinal and epiretinal), optic nerve or the visual cortex. The details of each approach and its advantages and disadvantages are explained in

15 6 the sections below Subretinal Implants A subretinal implant is located between the outer layer of the retina and the pigment epithelium where the photoreceptors (rods and cones) are located. The implant mimics the functionality of the photoreceptors by using microphotodiodes, which convert the external light into an electrical current that is sent to a microelectrode. The small current stimulates the working portions of the retinal layers, which send the visual signal through the optical nerve to the brain. This method replaces the photoreceptors but depends on the other retinal layers and optic nerve to transmit the electrical pulse to the brain. An advantage of this approach is that the device only relies on light to stimulate the microphotodiodes, therefore it does not require a power device or an external device such as a camera to provide the image and computational power. Surgically implanting the electrodes and photodiodes is not complicated compared to typical ocular operations. The implant is secure between the retina and the pigment epithelium and has a lower risk of moving around. On the other hand, the subretinal approach does not have a processing unit to monitor the device. Experiments have also shown that a single electrode may not generate enough current to stimulate the neurons in the retina. Therefore an additional source may be needed [4, 8]. Using external coils to provide additional energy for the subretinal implant has been explored. This implementation uses an external camera to capture the image and transmits the information to the subretinal implant. This approach can power the photodiodes using an outside source rather than just relying on light as the only power source. With the additional energy there could be enough power to stimulate the retinal cells to send the visual information to the brain [15].

16 Epiretinal Implants An epiretinal device is more complicated than the subretinal implant since it requires a camera or sensor to receive light information. This method omits the photoreceptor stage by allowing a camera to process the various properties of color, brightness, contrast, etc. The visual data is transmitted to an implanted microelectrode device located at the ganglion cell layer on the inner layer of the retina [16,17]. The device converts the data into an electrical pulse, which stimulates the electrodes in order for the retinal neurons to understand and transmit data to the optical nerve and brain. This implementation requires computational resources to process and communicate the visual information [4, 8, 18]. Since the data has to be transmitted from an extraocular device to an intraocular one, various methods have been proposed from using a laser to transmit data [18] and a wireless network [19 21]. The advantage of this structure is that part of the system is extraocular which can result in less interference with biological tissue. Since the device will be located close to the vitreous humor, the jelly like liquid can help dissipate the heat away from the implanted device. It is difficult to secure an epiretinal device in the eye and is subject to eye movement compared to the subretinal approach. A larger current is needed to stimulate the neurons in the ganglion cell layer due to their distance from the electrode array [4, 8, 18] Optical Nerve Implants An optic nerve prosthetic device directly stimulates the optic nerve instead of stimulating the retina. This approach uses an electrode cuff to surround the optic nerve and stimulate different axons, which are sending visual information to the brain. This approach can have dangerous side effects and requires a complicated surgical procedure. Due to the concentrated number of axons in a small amount of

17 8 area, it is difficult to map a spatial location to a specific axon in the optic nerve. This method has been shown to stimulate the visual cortex through the optic nerve in humans [8, 22] Cortical Implants A cortical approach consists of a camera and processing unit, which will translate the visual data to electrical signals. The information is sent to the skull to an electrode array, which is implanted on the visual cortex of the brain. This approach can give people with various ocular diseases the ability to see since the information is sent straight to the brain. The epiretinal and subretinal approach only helps people suffering from RP and AMD. The side effects of this approach can be severe due to the complicated surgery. The visual mapping at the brain is more complex than at the retina, therefore a spatial pattern on the retina may not always map to the same pattern on the visual cortex [8, 23, 24]. A better understanding of the visual mapping of the brain is needed to enhance the cortical implant approach and accurately implant the electrodes in the correct locations. 1.3 NCSU/USC Approach to Retinal Prosthesis Engineers at North Carolina State University and medical doctors at the University of Southern California (formally with Johns Hopkins University) have been working together to create a working prototype of a full epiretinal prosthesis system to restore vision to RP and AMD patients. It has been shown that patients can perceive the sensation of light through stimulating electrodes on the retina [10, 11]. The effects of an epiretinal device was tested for a prolonged period of time in dogs and showed no critical side effects [25]. The results of these experiments demonstrate

18 9 the practicality and safety of implantable retinal devices. The proposed system consists of two separate units, an intraocular and extraocular unit. This implementation allows for part of the processing to occur outside the eye resulting in a lower power implementation and a smaller device to be implanted inside the eye. The two components can transfer power wirelessly through inductive coils and transfers data through an RF link. The extraocular unit contains a video camera, processing unit, protocol encoding chip, amplifier and a coil. The intraocular unit consists of a coil, antenna, rectifier and regulator, retinal stimulator, protocol decoding chip and an electrode array which connects to the ganglion cell layer of the retina. The camera receives an image from the outside environment and sends the visual data to the extraocular processing unit. The processor converts the visual data into a specific protocol and transmits it through the RF link to the intraocular unit. The intraocular unit decodes the data and converts the signal into an electrical pulse to stimulate the electrodes which sends the visual signal through the retinal layers to the optic nerve and finally reaches the brain [19 21]. 1.4 Motivation for Research in Miniature Antennas A wireless data and power link is advantageous in a two element prosthesis device to avoid connecting wires through the patient s body. This implementation is safer because it eliminates the risk of disconnecting components and damaging tissue through movement and would allow the patient to be less constrained. In earlier designs, it was proposed that the coils would provide the data and power transfer between the intraocular and extraocular device [19 21]. With the number of

19 10 electrodes increasing to provide a better picture, the bandwidth of the data link must be large to transmit the data for each electrode. Therefore, an inductive link operating in the 1-10 MHz range will not be sufficient to transmit a larger amount of information. A higher frequency range is needed to increase the bandwidth thus increasing the data rate. Unfortunately, an inductive coil s performance degrades at a higher frequency. A mutually exclusive power and data system was proposed by transferring data at microwave frequencies (1-3 GHz) and power transfer at lower frequencies (1-10 MHz) [26, 27]. A miniature antenna operating at microwave frequencies will have a higher bandwidth and can be used to receive the visual data. The design of implantable miniature antennas for the use of a retinal prosthesis is challenging due to the size constraint and the effects of a biological medium. A microstrip patch antenna was explored for this application and the showed promising results [28]. The antenna would be surrounded by the inductive power coil and placed on the outside of the implanted processing chip. Since the microstrip patch antenna contains a ground plane, it would reduce the coupling and power transfer between the power coils. Two Dimensional dipoles were explored as another option due to its lack of ground plane and ease of fabrication [27]. The motivation of this research is to explore various dipole designs for implanted biomedical devices, specifically for the application of a retinal prosthesis. Different two and three dimensional designs were explored in an effort to enhance the bandwidth and gain. Since size is a serious constraint, three dimensional antennas were explored to reduce the footprint of the antenna while slightly increasing the thickness resulting in an overall smaller design. This research work investigates various two and three dimensional antennas for the application of implanted biomedical devices.

20 11 Chapter 2 Design and Simulation of Miniature Wire Antennas There are many challenges in designing an efficient miniature antenna for implantable biomedical applications. For the specific application of a retinal prosthesis, the proposed configuration consists of the processing chip, RF receiving antenna and electrode stimulators to be enclosed on the same substrate. The inductive coil surrounds the circuitry and the antenna. Various antenna structures and designs were approached to find the optimal antenna type for the specific application of a retinal prosthesis [26 28]. Since the data and power links are mutually exclusive, it is necessary that the two independent structures do not interfere with each other to achieve maximum power transfer. A microstrip patch antenna was considered in [27] and showed promising results but the addition of the ground plane has several disadvantages. Adding a ground plane to the silicon substrate increases the complexity of the fabrication process. With the antenna in close proximity to the inductive power coil, a ground plane could add interference and coupling between the two structures resulting in a reduction in power

21 12 transfer. A wire antenna structure such as a dipole can provide similar performance without creating interference in the system. Dipoles do not require a ground plane resulting in a simpler fabrication process and are a good candidate for an intraocular antenna. The goal in designing miniature antennas is to create an efficient radiating structure that is physically small but behaves electrically large. The use of fractal geometries and meander line structures are common size reduction techniques to miniaturize a radiating structure. The focus of this research is to explore three dimensional versions of two dimensional antennas in an effort to reduce the overall footprint size. The following sections describe various simulation techniques, the background of fractal antennas and simulation results, simulations of 2D and 3D meander line dipoles and a comparison. 2.1 Fractal Antennas Fractals are self replicating geometries and are found throughout nature and science. The mathematics of fractal designs have been described in [29]. There are various types and forms of fractals, but the space filling form is most commonly used for antenna designs due to its ability to compress a large amount of wire into a small area. Space filling fractals are based on the principle that there is a geometric design or shape that is replaced for every segment. This process is recursive, creating designs that become more complex at each iteration. This approach has been used as a size compression technique for all types of antennas such as dipoles, loops, patches and so on [30 36]. For implantable device applications, fractal antenna geometries were explored in an attempt for size compression. Three space filling fractals designs were chosen

22 13 Figure 2.1: First Five Iterations of the Hilbert Fractal Figure 2.2: First Five Iterations of the Sierpinski Arrowhead Fractal to simulate, fabricate and analyze, the Hilbert (Figure 2.1), Peano (Figure 2.3) and Sierpinski Arrowhead fractals (Figure 2.2). The Hilbert and Peano fractals fill a rectangular space where the Sierpinski Arrowhead fractal fills up a triangular area. The Peano design is most efficient in increasing the amount of wire added at every iteration but it becomes a very complex geometry after a couple iterations. Figure 2.3: First Four Iterations of the Peano Fractal

23 Fractal Antenna Simulations and Results The fractal and wire structures were simulated using Numerical Electromagnetic Code (NEC). NEC is a free online electromagnetic simulation code, which uses the method of moments technique to calculate the fields [37]. Due to the complexity of the fractal and wire antenna geometries, Matlab [38] code was used to generate the endpoints of each wire segment. The coordinates started at one end of the dipole arm and ended at the termination of the other dipole arm. The Matlab code formatted the coordinate data into the NEC input file format. All structures were simulated using a 10 um wire radius and the segment sizes were made as small as possible for the highest accuracy. Since NEC limits the maximum number of segments to 1500, it was not possible to simulate higher iterations of the fractal geometries. One technique to match the fractal antennas is to move the source point towards the end of one of the dipole arms and has been demonstrated to work for Hilbert and Peano fractals [39, 40]. This achieves a match at the first resonance for any desired resistance. This technique was used to match the 4th and 5th iterations of the Hilbert Fractal, 2nd and 3rd iterations of the Peano Fractal and the 4th and 5th iterations of the Sierpinski Fractal to 50 Ohms. All antennas were designed to fit a 6cm by 6cm footprint size. The location of the feed point and other simulation results are shown in Figures 2.4, 2.5, 2.6, 2.7, 2.8 and 2.9.

24 15 (a) Current Distribution (b) Radiation Pattern (c) Structure and Feed Point Figure 2.4: Simulation Results for the 4th Iteration Hilbert Fractal

25 16 (a) Current Distribution (b) Radiation Pattern (c) Structure and Feed Point Figure 2.5: Simulation Results for the 5th Iteration Hilbert Fractal

26 17 (a) Current Distribution (b) Radiation Pattern (c) Structure and Feed Point Figure 2.6: Simulation Results for the 2nd Iteration Peano Fractal

27 18 (a) Current Distribution (b) Radiation Pattern (c) Structure and Feed Point Figure 2.7: Simulation Results for the 3rd Iteration Peano Fractal

28 19 (a) Current Distribution (b) Radiation Pattern (c) Structure and Feed Point Figure 2.8: Simulation Results for the 4th Iteration Sierpinski Arrowhead Fractal

29 20 (a) Current Distribution (b) Radiation Pattern (c) Structure and Feed Point Figure 2.9: Simulation Results for the 5th Iteration Sierpinski Arrowhead Fractal

30 21 Fractal Type Resonant Frequency Bandwidth Max Gain (Iteration) (MHz) (VSWR=3, khz) (dbi) Hilbert (4) Hilbert (5) Peano (2) Peano (3) Sierpinski Arrowhead (4) Sierpinski Arrowhead (5) Table 2.1: Fractal Simulation Results for Various Geometries and Iterations Table 2.1 summarizes the results of the simulated fractal antennas. All of the antennas maintain a 6cm by 6cm footprint size. (Note: The Sierpinski Arrowhead fractal triangular space filling fractal and only uses half the footprint area) While maintaining the same footprint size, the higher iteration fractals more effectively lower the resonant frequency due to the increased amount of wire. The higher iteration fractals have a larger wire density resulting in a smaller bandwidth compared to the lower iterations. 2.2 Disadvantages of Fractal Geometries Although fractal antennas reduce the resonant frequency, it is only due to the larger amount of wire compressed in a small area. The frequencies of operation are similar to a dipole of the same size. It has been shown that the current vector alignment and other aspects of the antenna affect the resonant frequency. Therefore, simple meander line dipoles can achieve similar or better performance as fractal geometries if the current vectors on wires are aligned in a certain way. The current

31 22 vectors in adjacent wires in fractal geometries tend to cancel each other out resulting in a higher resonant frequency and a less efficient design [41 44]. In addition, the complexities of fractal geometries require a precise fabrication method for miniature size applications. For the application of implantable devices, higher iteration fractal antenna geometries are difficult to fabricate at a miniature scale due to their complexity and small wire segments. Therefore simpler designs such as meander line dipoles would be easier to fabricate due longer line lengths and a lower density of wires in a given footprint. The antenna presented in [27] uses a spiral meandering line antenna geometry for miniature implantable applications. This design is simple and easier to fabricate compared to the fractal designs Matching Techniques Matching at the miniature scale is very difficult due to the size of the wires compared to the source. The location of the feed point is critical as it would be tricky to insert a feed in a middle location. The matching technique presented in [27] places the feed point on the outer edge of the antenna (which is located at the center of the dipole arms in this case) which allows for an easy connection to the source. Unlike the method presented in [39, 40] where the source is moved along one dipole arm till a match is achieved at the natural resonance. The technique in [27] maintains the source at the center while shortening one of the dipole arms. The match occurs at a induced resonance right after the natural resonance. At the induced resonance, the phase reverses at the source point resulting in a different current vector alignment and radiation pattern. This method is ideal for miniature antenna applications due to the fixed location of the source point.

32 Three Dimensional Wire Antennas For implantable applications, the antenna size is a significant constraint. Therefore, three dimensional antennas were explored as an effort to reduce the footprint size, increase bandwidth and gain. Fractal designs can be implemented in a three dimensional scale, such as the three dimensional Hilbert fractal [45]. In order to have an efficient design, a high iteration would be needed resulting in a very complex geometry. The fabrication of this design would require multiple metal layers and many vias to connect between the layers. Therefore, three dimensional meander line dipoles were explored due to their simplicity and ease of fabrication. By taking efficient two dimensional designs, the antennas can be folded and rotated to create a three dimensional structure resulting in a more compact design while enhancing different characteristics like the bandwidth and gain. The following sections compare various two dimensional designs with their three dimensional implementations. The antenna geometries were created in Matlab and exported into an NEC file format. The antennas were simulated using NEC in free space with a wire radius of 10um. The segment size was made as small as possible to provide the most accurate results. All of the antennas simulated have a 5.25mm by 5.25mm two-dimensional footprint. Due to the size of the antennas, all of the antennas were center fed and matched to 50 Ohms by shortening one of the dipole arms following the technique shown in [27] Spiral Dipole Antenna The first antenna set simulated was based on the antenna presented in [27]. The two dimensional implementation has a very small bandwidth and a high directivity shown in figure The current is strongly aligned in the center of the antenna

33 24 Antenna Type Resonant Frequency Bandwidth Max Gain (VSWR=3) (dbi) Spiral 2D 3.6 GHz 8.8 khz 4.9 Spiral 3D 3.5 GHz 1.3 MHz 1.4 Spiral 3D Rotate1 GHz MHz 1.3 Spiral 3D Rotate2 GHz khz 5.4 Table 2.2: 2D and 3D Spiral Antenna Simulation Results making the radiation pattern very focused. In an effort to increase the bandwidth, the antenna is folded to create a three dimensional implementation where the current vectors are equally aligned throughout the x-y plane resulting in a wider bandwidth and a more omnidirectional pattern. This design is shown in figure As one of the arms is rotated along the z-axis, the bandwidth and the gain change according to the current vector alignment. The 3D rotate1 antenna shown in figure 2.12 has a much wider bandwidth with a more omnidirectional pattern. The 3D rotate2 antenna s performance shown in figure 2.13 is slightly higher than the two dimensional case with a wider bandwidth and a higher gain. This is due to a strong current alignment close to the feed point of the antenna. Table 2.2 summarizes the simulation results for the 2D and 3D Spiral Antennas. As shown in the current distribution plots for each of the respective antennas, the induced resonant frequency is similar to a third resonance where there is a phase reversal at the center. This is due to the matching technique described earlier. Figures 2.10, 2.11, 2.12 and 2.13 show the antenna structure, current distribution, radiation pattern, current distribution and current vector alignment for each of the 2D and 3D Spiral antennas.

34 25 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.10: Simulation Results for the 2D Spiral Antenna

35 26 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.11: Simulation Results for the 3D Spiral Antenna

36 27 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.12: Simulation Results for the 3D Rotate1 Spiral Antenna

37 28 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.13: Simulation Results for the 3D Rotate2 Spiral Antenna

38 29 Antenna Type Resonant Frequency Bandwidth Max Gain (VSWR=3) (dbi) S-Spiral 2D 3.4 GHz 2.0 MHz 1.9 S-Spiral 3D 4.4 GHz 17.9 khz 3.1 S-Spiral 3D Rotate1 3.5 GHz 64.6 khz 3.2 S-Spiral 3D Rotate2 3.1 GHz 1.5 MHz 0.7 Table 2.3: 2D and 3D S-Spiral Antenna Simulation Results S-Spiral Dipole Antenna Since the Spiral antenna design has a very small bandwidth, an S shaped spiral geometry was explored to achieve a higher bandwidth. The S shaped spiral 2D geometry, shown in figure 2.14 has a larger bandwidth compared to the antenna presented in [27] due to the uniform current alignment. The maximum gain is lower and the radiation pattern has a more omnidirectional shape. As shown in figure 2.15, when the S shaped spiral antenna is folded, the current alignment is centralized but located across different planes. This causes a more focused radiation pattern with a slight tilt due to the current flowing on different vertical surfaces. In the 3D rotate1 antenna, the two dipole arms are offset by 90 degrees along the z-axis as shown in figure This creates a current alignment along the center of the antenna causing a higher gain and a smaller bandwidth range. In the rotate2 antenna shown in figure 2.17, the upper dipole arm is flipped compared to the 3D rotate1 antenna. This causes a less centralized current alignment creating a weaker radiation pattern and larger bandwidth. Table 2.3 summarizes the simulation results for the 2D and 3D S-Spiral Antenna implementations. Similar to the Spiral antennas, the induced resonance for the S-Spiral behaves

39 30 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.14: Simulation Results for the 2D S-Spiral Antenna

40 31 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.15: Simulation Results for the 3D S-Spiral Antenna

41 32 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.16: Simulation Results for the 3D Rotate1 S-Spiral Antenna

42 33 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.17: Simulation Results for the 3D Rotate2 S-Spiral Antenna

43 34 like a third resonance where there is a phase reversal at the center and is shown in the current distribution plots. Figures 2.14, 2.15, 2.16 and 2.17 show the antenna structure, current distribution, radiation pattern, current distribution and current vector alignment for each of the 2D and 3D Spiral antennas Triangular Spiral Dipole Antenna The last set of antennas explored had a more triangular shape in an effort to enhance the bandwidth. Triangular shapes have been known to increase bandwidth such as bowtie antennas [46]. The 2D Triangular Spiral antenna, shown in figure 2.18 has the largest bandwidth compared to the other two dimensional antennas simulated and a fairly omnidirectional radiation pattern. When the two dimensional implementation is folded along the center into a 3D version, the current vectors align vertically as shown in figure Due to the angular orientation of the different wire segments, it is difficult to completely align the currents on the x-y plane causing a degradation in both gain and bandwidth compared to the 2D version. In the Triangular rotate1 antenna shown in figure 2.20, there is a slight alignment in the current across the wire segments, but due to the angular structure the vectors do not completely add or subtract causing a degradation in bandwidth and gain compared to the two dimensional implementation. As the upper dipole arm is rotated along the z-axis as shown in figures 2.21, 2.22 and 2.23 there is not a large change or large improvement in the bandwidth and gain compared to the planar implementation. The radiation pattern also rotates as the angular orientation between the upper and lower dipole arm changes. Table 2.4 summarizes the simulated results for the 2D and 3D Triangular Spiral structures. As expected, the shorter dipole arm creates an induced series resonance as shown in the current distribution plots. Figures 2.18, 2.19, 2.20, 2.21, 2.22 and 2.23 show

44 35 Antenna Type Resonant Frequency Bandwidth Max Gain (VSWR=3) (dbi) Tri-Spiral 2D 3.8 GHz 3.2 MHz 2.1 Tri-Spiral 3D 3.5 GHz 2.2 MHz 1.4 Tri-Spiral 3D Rotate1 3.6 GHz 2.6 MHz 1.4 Tri-Spiral 3D Rotate2 3.6 GHz 2.9 MHz 0.9 Tri-Spiral 3D Rotate3 3.5 GHz 2.4 MHz 1.2 Tri-Spiral 3D Rotate4 3.7 GHz 3.1 MHz 1.2 Table 2.4: 2D and 3D Triangular Spiral Antenna Simulation Results the antenna structure, current distribution, radiation pattern, current distribution and current vector alignment for each of the 2D and 3D Spiral antennas.

45 36 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.18: Simulation Results for the 2D Triangular Spiral Antenna

46 37 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.19: Simulation Results for the 3D Triangular Spiral Antenna

47 38 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.20: Simulation Results for the 3D Rotate1 Triangular Spiral Antenna

48 39 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.21: Simulation Results for the 3D Rotate2 Triangular Spiral Antenna

49 40 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.22: Simulation Results for the 3D Rotate3 Triangular Spiral Antenna

50 41 (a) Radiation Pattern (b) Current Distribution (c) Current Magnitude and Phase Distribution across Segments (d) Current Vector Alignment (e) Antenna Structure Figure 2.23: Simulation Results for the 3D Rotate4 Triangular Spiral Antenna

51 42 Antenna Type Resonant Frequency Bandwidth Max Gain Spiral 2D Spiral 3D Spiral 3D Rotate Spiral 3D Rotate Table 2.5: Comparison of 3D Antenna Characteristics to 2D Antenna Characteristics for the Spiral Antennas Antenna Type Resonant Frequency Bandwidth Max Gain S-Spiral 2D S-Spiral 3D S-Spiral 3D Rotate S-Spiral 3D Rotate Table 2.6: Comparison of 3D Antenna Characteristics to 2D Antenna Characteristics for the S-Spiral Antennas 2.4 Comparison of 2D and 3D Antennas The motivation behind converting the two dimensional antennas into three dimensional implementations was to enhance the bandwidth or gain while reducing the planar size of the antenna. Three dimensional variations of the Spiral antenna and the S-Spiral antenna showed an improvement in either bandwidth or gain or both. The 3D modifications on the Triangular Spiral antenna did not show an increase in either performance metric. Due to the angular orientation of the wire segments between the upper and lower dipole arms, it is difficult to entirely add current vectors to fully enhance the antenna. Tables 2.5, 2.6 and 2.7 compare the maximum gain and bandwidth characteristics between the two dimensional implementation with the three dimensional implementation.

52 43 Antenna Type Resonant Frequency Bandwidth Max Gain Tri-Spiral 2D Tri-Spiral 3D Tri-Spiral 3D Rotate Tri-Spiral 3D Rotate Tri-Spiral 3D Rotate Tri-Spiral 3D Rotate Table 2.7: Comparison of 3D Antenna Characteristics to 2D Antenna Characteristics for the Triangular Spiral Antennas It was found that when one section of the antenna is strongly aligned in one direction compared to the rest of the antenna, the antenna performs with small bandwidth and high directivity characteristics. When the currents on the antenna are equally aligned with each other where the currents are more or less symmetrical across the antenna, the structure has a wider bandwidth and a more omnidirectional radiation pattern. The Spiral and S-Spiral antenna both have perpendicular wire segments making it easier to align current in a specified location to create a focused radiation pattern. Since the three dimensional implementations of the Spiral Antenna provide improvement to both maximum gain and bandwidth, comparison between fabricated versions of the two dimensional spiral antenna with the three dimensional version and the 3D rotate2 version are explored in the following chapters.

53 44 Chapter 3 Fabrication Methods and Results As described in Chapter 2, the various three dimensional implementations of the Spiral antenna demonstrated enhancement in both bandwidth and maximum gain. The 2D Spiral structure had a very focused radiation pattern and a small bandwidth. The folded 3D implementation provided a wider bandwidth with an omnidirectional radiation pattern. The 3D Rotate2 antenna geometry provided a more focused radiation pattern and a slightly wider bandwidth compared to the 2D implementation. In order to completely compare the performance of the 3D Spiral antennas to their two dimensional implementations, the antennas were fabricated and tested to accurately compare the characteristics. Although the performance can be measured independently in air, a similar setup to the retinal prosthesis device needed to be created to test and compare the effects in a biological medium. The following sections describe the fabrication methods used to make the antennas, the fabricated results of the individual antennas tested in air and the methods used to create a solution to mimic the vitreous humor of the eye.

54 Fabrication Methods Due to the small scale of the miniature antennas, fabricating prototype designs is very difficult due to the size of the smallest feature of the antenna geometry without using complex cleanroom equipment. The footprint of the antennas simulated in Chapter 2 were 5.25mm by 5.25mm with wire widths in the micron range. Various fabrication methods were explored to find the optimal process. Milling machines have been used to fabricate antenna designs. This process uses drill and milling bits to mill out the design on a copper clad board. This option is ideal for larger antennas but becomes difficult to create fine wire widths for miniature antennas due to the limits of the milling bit sizes. Since milling out the copper was not an option, etching copper was explored as an alternative. There are multiple methods in transferring antenna structures onto a copper board. One method uses iron on transfer paper to which transfers the design onto the copper board. This method did not provide the fine resolution required for the miniature scale antennas. Since light waves can provide a fine and uniform resolution, photolithography methods were ideal to create a proper mask to replicate the design on a copper board Fabrication Process The 2D and 3D antennas were fabricated using a prototyping process similar to making PCB boards. The antenna coordinates were generated and plotted in Matlab [38] to create a mask. The figures were transferred to Adobe Photoshop [47] to create high resolution images and the masks were printed onto transparency paper (1200 dpi) using a laser printer. Imperfections in the mask were touched up using a black permanent marker.

55 46 The mask was placed in a dark room on top of a MG Chemical 1/32 inch copper clad board [48] with positive photoresist on top of the copper surface. A clear acrylic weight was placed on top of the mask to prevent movement as shown in figure 3.1(a). A fluorescent light was placed above the board and mask and turned on for exactly five minutes in the dark room to expose the photoresist. After 5 minutes, the exposure process was complete and the light was turned off. The exposure time dictates the resolution of the wire widths. If the board is placed under the light for too long, the widths decrease in size creating discontinuities in the antenna. The board was developed in the dark room using a solution of 10 parts water and 1 part positive photoresist developer [48]. The board was agitated in the developing solution as shown in figures 3.1(b) and 3.1(c) until all of the unwanted photoresist is removed. After the developing stage, the board was cleaned to prevent fingerprints. The oils from the skin act as a photoresist and can create unwanted patterns and spots on the antenna geometry. The board is then agitated in a copper etching solution of 1 part Ammonium peroxydisulfate powder [49] and 5 parts deionized water. The board is shaken in the copper etchant solution for minutes till the all of the unwanted copper is removed as shown in figure 3.1(d) and 3.1(e). Ammonium peroxydisulfate was chosen over other copper etchants due to its clear nature allowing constant monitoring throughout the etching process. Since it is a powder, the concentration of the etchant can be varied to provide the optimal etch solution to create the fine resolution needed to make miniature antennas. Finally, the developer solution was brushed onto the board to remove the photoresist leaving the desired antenna geometry. The process was loosely based on the procedure outlined in [48] which describes a PCB prototyping process. The procedure times and chemical concentrations were varied to provide optimal results for fabricating antennas at the miniature scale.

56 47 (a) Exposure (b) Early Developing (c) Later Developing (d) Early Etching (e) Later Etching Figure 3.1: Fabrication Process Steps

57 3.2 Fabricated Results and Comparison for Individual 2D and 3D Antennas Simulation Method The 2D and 3D antenna designs described in Chapter 2 were simulating using NEC (Numerical Electromagnetic Code) in free space using circular wires. In order to more accurately simulate the antenna structure, the antennas were simulated in ADS (Advanced Design System) [50] which accounts for the substrate, connectors and wire surface widths. A rectangular copper pad was added at the source connection points to solder to the antenna since the wire widths were around 200 um. The antennas were simulated using the properties of the copper clad board which has a substrate dielectric constant of 4.8. The 2D Spiral Antenna was simulated using a differential port and the 3D versions were simulated using a single port and a ground reference. ADS does not allow differential ports on different metal layers, therefore the ground reference was used. The use of a single port and a ground reference caused the 3D Antenna simulations to have a very small bandwidth compared to the 2D simulation. Also, the ADS simulations show the ideal 3D antenna where the two arms are located directly on top of each other. There is an offset between the two dipole arms in the fabricated antennas which results in a shift in the resonant frequency compared to the simulated results.

58 Fabricated Antenna Results The fabricated 2D, 3D and 3D Rotate2 versions of the Spiral Antenna were produced using the method described earlier in the chapter. The antennas were fabricated on a copper clad board with a substrate dielectric constant of 4.8 and attached to SMA connectors. All antennas were matched to a reference impedance of 50 ohms. The performance of the antennas were tested using a network analyzer. The fabricated 2D Spiral antenna and its fabricated results are shown in figure 3.2. When the arms are of equal length, only the natural resonance exists. The results show an induced resonance at a slightly higher frequency than the natural resonance when one of the dipole arms is shortened. The induced resonant frequency occurs at 2.1 GHz with a bandwidth of 50 MHz. Due to the addition of the connector, a capacitance is added to the overall structure. Also, the dielectric substrate changes the characteristics of the antenna making it harder to create a phase reversal. Therefore, a larger length of the dipole arm has to be removed to induce and match a new resonance. Figure 3.3 shows the fabricated 3D Spiral antenna and its performance. The graph shows that when the antenna s arms are at equal length, the natural resonance occurs. When one of the dipole arms are shortened, an induced resonance occurs. As the arm becomes shorter, the induced resonance moves to a higher frequency. The wider bandwidth structures are more difficult to match at the induced resonance due to the higher frequency and the addition of the dielectric. Therefore, a larger amount of the dipole arm has to be removed to create a phase reversal and appropriate match at the induced resonance. The induced resonant frequency occurs at 2.4 GHz with a bandwidth of 90 MHz. The fabricated 3D Rotate2 Spiral antenna and its performance are shown in figure 3.4. With equal arm lengths, the natural resonance is matched and there is an induced

59 50 (a) Fabricated Antenna (b) Scattering Parameters (db) vs. Frequency (GHz) (Simulated Results, Fabricated Antenna with Equal Arms, Fabricated Antenna with Unequal Arms) Figure 3.2: 2D Spiral Antenna and Results

60 51 (a) Fabricated Antenna (Side 1) (b) Fabricated Antenna (Side 1) (c) Scattering Parameters (db) vs. Frequency (GHz) (Simulated Results, Fabricated Antenna with Equal Arms, Fabricated Antenna with Unequal Arms) Figure 3.3: 3D Spiral Antenna and Results

61 52 resonance at a higher frequency. This induced resonance occurs from the coupling between the two planes of the antenna and is not matched. As the dipole arm is shortened, the induced resonance moves higher in frequency and is matched. A larger amount of the dipole arm was removed to create a 50 ohm match due to the effects of the dielectric and the wide bandwidth. For this implementation, the induced resonance occurred at 2.55 GHz with a bandwidth of 80 MHz. The relative performance of the antennas are similar to the results found in Chapter 2 where the bandwidth of the 3D Spiral antenna was the widest followed by the 3D Rotate1 Spiral antenna and the 2D Spiral antenna. The addition of the dielectric and large connectors caused capacitive behavior and decreased the performance of the antennas in comparison to the simulated results. Despite this, the relative performance between the antennas shows an enhancement in bandwidth and a smaller footprint for the 3D implementations making them ideal for implantable applications. 3.3 Methods of Replicating Biological Mediums Further investigation of the effects of a biological medium, spherical shape of the eyeball and the addition of a receiver are explored in the following chapter. In order to accurately model the eye, a liquid or gel-like solution was needed to replicate the electrical characteristics at the antenna s frequencies of operation. This section explores different techniques and methods to correctly measure and create a biological solution. Various research has been done to model the electrical properties of different biological tissues and organs through various frequency ranges [51 54]. This data can be used to study the effects of implanted electrical devices through simulations and or create replica biological solutions to test devices in body phantoms.

62 53 (a) Fabricated Antenna (b) Scattering Parameters (db) vs. Frequency (GHz) (Simulated Results, Fabricated Antenna with Equal Arms, Fabricated Antenna with Unequal Arms) Figure 3.4: 3D Rotate2 Spiral Antenna and Results

63 54 Creating a liquid or gel-like solution to mimic a biological tissue is challenging due to the variation of the dielectric constant and conductivity throughout a large frequency range. Different researchers have explored numerous mixtures of chemicals and normal household items to accurately model the biological medium [55 57]. Most solutions use water as the base liquid due to its accessibility and known electrical parameters. At lower frequencies (below 100 MHz), the electrical properties of the biological medium are usually much higher than water. This creates difficulty in creating solutions that have high dielectric constants. Therefore, materials with dielectric constants above water such as glycine, urea or formamide are added to water to increase the electrical properties to the desired amount. At higher frequencies (above 100 MHz), the electrical properties of the biological tissue are usually lower than water. Materials such as sucrose, ethanol and glycerol have been used to lower the dielectric constant. Salt is usually added to modify the conductivity parameter. There are various methods in measuring a dielectric medium from using resonators [58], coplanar lines [59], coaxial lines [60, 61] and other techniques [62]. Each of these methods are useful in certain frequency ranges. Since the vitreous humor solution is a liquid, the coaxial line is an ideal testing device due to its ability to be immersed in liquid and its frequency range of operation includes the 1-3 GHz range. The open ended coaxial probe shown in 3.5 was constructed using a cut SMA connector cable so the connector is on one side and a open faced on the other. The coaxial line was teflon filled to prevent liquid from entering the cable during testing. The open face was filed down to a very slight conical shape to further expose the inner conductor to prevent an indirect connection during the experiments.

64 55 Figure 3.5: Open Ended Coaxial Probe Procedure to Measure Experimental Vitreous Humor The procedure to measure the dielectric solution was based on [63]. The method described uses an open ended coaxial probe to test a dielectric solution. The procedure requires two different calibrations. The first is the normal network analyzer calibration which only calibrates the system up to the end of the cables. The second calibration step consists of connecting the coaxial probe to an open, short and known material to account for the addition of the probe. The procedure described in [63] was slightly modified to use less chemicals. The original method suggested using liquid mercury to create a short for calibration. A metal cap was constructed to short the outer and inner conductors of the coaxial probe instead. Also, acetone was used as the known material in the paper. Acetone and distilled water was used as known materials for calibration instead. The method from [63] uses the data from the open, short, known liquid and unknown liquid and several equations to solve for the unknown dielectric constant. The equations account for fringing effects on the open end of the coaxial cable ɛ open + 23 Z open = Z open ɛ open

65 ɛ short + 23 Z short = Z short ɛ short + 12 ɛ known + 23 Z known = Z known ɛ known From this, 12 and 23 can be found. The unknown electrical properties (ɛ unknown ) can be calculated using the following equations ɛ unknown + 23 Z meas = ɛ unknown = Zmeas 12 +Z meas 23 ɛ unknown = ɛ ɛ = ɛ σ ωɛ o Zmeas ɛ unknown Experimental Results for Vitreous Humor The vitreous humor solution was created by combining different concentrations of water, salt and sugar. Salt increases the conductivity of the solution while sugar decreases the dielectric constant. Various combinations were tested and the closest solution of 50 ml water, 1/16 tsp salt and 15/16 tsp sugar is shown in figure 3.6. This combination is close to the actual results in the frequency range of GHz where the fabricated antennas operate. The following chapter summarizes the antenna performance in a transmitter/reciever setup in air and inside an eye phantom. The results are compared between the two and three dimensional implementations.

66 Figure 3.6: Vitreous Humor Experimental Results 57

67 58 Chapter 4 Comparison of 2D and 3D Antennas in a Rx/Tx system The following chapter summarizes and compares the test results of the 2D and 3D miniature antennas in a transmitter/receiver system in air and the presence of the eye model. 4.1 Coupling Measurements in Air Air Coupling Testing Apparatus To completely test the performance of the miniature antennas, a full transmitter and receiver system was developed. Basic rectangular patch antennas were constructed to operate at each of the miniature antenna s induced resonant frequencies. The patch antennas were fabricated on copper clad boards with a substrate dielectric constant of 9.2. The patch antennas represent the extraocular antenna, therefore the high dielectric constant was used to reduce the size to around the area of an eyeglass lens.

68 59 Figure 4.1: Air Coupling Testing Structure A wooden testing apparatus shown in figure 4.1 was made to properly secure the patch and miniature antenna during measurements. The patch antenna was fixed to a slot in the base while the miniature antenna was attached to a small wooden slider. The slider could move and rotate along the base for various distance and angular measurements. The robustness of the antenna at various points was tested. The location of the points is shown in figure 4.2. By measuring the S21 parameters between the patch and miniature antenna in different locations, a slight realization of the radiation pattern can be deduced to verify the radiation pattern simulations done in Chapter 2. Points 1-4 are located on the x-y plane surrounding the antenna while point 5 is located directly above the antenna. For the application of the retinal prosthesis device, the extraocular antenna will be placed approximately 25mm from the intraocular antenna. Therefore, specific measurements were taken at 25mm to measure the performance at that distance. Since the antenna will be placed inside the eye, it has to be robust to angular eye movements. Measurements to check the performance from -40 to 40 degrees were

69 60 Figure 4.2: Location of Measurement Points taken to check the uniformity of the signal across that angular range D Spiral Antenna Measurements in Air The radiation pattern for the 2D Spiral antenna was very focused in a slightly oval shape above and below the antenna. The S21 measurements over distance show a stronger performance at point 5 (above the antenna). The S21 values are stronger at points 1 and 3 compared to points 2 and 4 as shown in figure 4.3. This matches the radiation pattern shape where the majority of the signal is focused above the antenna along the projection of the x axis. At 25mm distance, the S21 values range from -34 db to -42 db between the different points. For the application of creating a retinal prosthesis, the antenna has to perform uniformly across various angles due to the rotation of the eye during movement. Mea-

70 61 (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.3: 2D Spiral Antenna Air Coupling Results at Different Points (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.4: 2D Spiral Antenna Air Coupling Results between 0 to 40 Degrees

71 62 (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.5: 2D Spiral Antenna Air Coupling Results between 0 to -40 Degrees surements across an angular range of -40 to 40 degrees were taken and the performance is shown in figures 4.4 and 4.5. At a 25mm distance, the performance variation is not significant and therefore the antenna is robust to eye movement. As the distance between the extraocular and intraocular antenna increases, there is more variation in the S21 parameters between the different angles D Spiral Antenna Measurements in Air For the 3D Spiral antenna, the radiation pattern was more omnidirectional across the x-y plane. The results of the S21 measurements across different points show similar performance for points 1-3 and a drastic drop in performance for point 5 as shown in figure 4.6. This demonstrates that the fabricated antenna does hold an omnidirectional pattern as simulated earlier. It was not possible to measure at point 4 due to the orientation of the antenna and the cable location. At 25mm, the S21

72 63 (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.6: 3D Spiral Antenna Air Coupling Results at Different Points values range from -30 db to -32 db for points 1-3 and point 5 is much lower at -42 db. To check for variation due to angular eye movement, measurements for the 3D Spiral antenna were taken from -40 to 40 degrees as shown in figures 4.7 and 4.8. The variation in the 3D antenna is slightly higher compared to 2D implementation. The variation between the different angles is approximately 5 db. Similar to the 2D implementation, as the distance increases the variation between the different angles increases as well. At a 25mm distance, the performance variation across the angles are still fairly close but slightly larger making it less robust than the other designs for eye movement.

73 64 (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.7: 3D Spiral Antenna Air Coupling Results between 0 to 40 Degrees (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.8: 3D Spiral Antenna Air Coupling Results between 0 to -40 Degrees

74 65 (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.9: 3D Rotate Spiral Antenna Air Coupling Results at Different Points D Rotate Spiral Antenna Measurements in Air The radiation pattern for the 3D Rotate Spiral antenna is more focused than the 2D antenna and slightly tilted to one direction. Through taking measurements at different points, the same pattern is realized in the fabricated implementation. As shown in figure 4.9, the strongest S21 value occurs at point 5. Points 1-3 have similar performance and point 4 has lower values compared to the other locations. These values show a strong focus above the antenna and that the pattern tilts towards points 1-3. At 25mm, the S21 values range from -29 db for point 5, -35 db to -37 db for points 1-3 and -46 db for point 4. For angular movement, the 3D Rotate Spiral antenna is more robust. The variation between -40 to 40 degrees is fairly close and uniform across all angles as shown in figure 4.10 and At 25mm distance, the performance does not vary much and is robust to eye movements. Similar to the other antenna, as the distance increases

75 66 (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.10: 3D Rotate Spiral Antenna Air Coupling Results between 0 to 40 Degrees (a) S21 at Varied Distances (b) S-Parameters at a 25 mm Distance vs. Frequency Figure 4.11: 3D Rotate Spiral Antenna Air Coupling Results between 0 to -40 Degrees

76 67 Antenna Type 2D Spiral 3D Spiral 3D Rotate Spiral Max S db -31 db db Table 4.1: 2D and 3D Air Coupling Results at a 25 mm Distance the variation across the angles increases as well Comparison between 2D and 3D antennas in Air Table 4.1 compares the performance of the 2D and 3D antennas in air. The coupling performance of the 3D implementations are stronger than the 2D antenna. The 3D Rotate antenna has the best S21 value and is also robust when subjected to eye movement. 4.2 Coupling Measurements in Eye Phantom In order to fully understand the effects of the miniature antennas, a full transmitter/reciever system inside an eye phantom was needed to characterize the antennas. In the previous section, measurements were taken in air without any scatterers or interference. This is the ideal case, but the introduction of a testing apparatus will introduce error. Therefore, measurements were first taken with an air-filled eye phantom to see the error introduced by the testing apparatus and then compared with the vitreous humor filled eye phantom results.

77 68 (a) Entire Structure (b) Close View of the Eye Phantom Figure 4.12: Eye Coupling Testing Structure Eye Coupling Testing Apparatus To test the performance of the miniature antennas for the application of a retinal prosthesis, and eye phantom testing apparatus was built as shown in figure The base is a metal rod and all of the testing components are connected and moved vertically along the base. The eye phantom is placed inside a foam base and is suspended from a wooden plank from above. The foam base is fixed by wooded rods to prevent movement. To mimic the properties of the eye, the eye phantom is filled with the vitreous humor solution described in Chapter 3. The miniature intraocular antenna is encapsulated in a plastic covering to prevent it from making contact with the vitreous humor solution. The patch antenna is connected to an adjustable horizontal rod and is aligned to be directly below the miniature antenna. The metal base and rods are covered with foam to prevent interference. A ruler is attached to take distance measurements.

78 69 Antenna Type Bandwidth Max S21 2D Spiral 79.7 MHz -41 db 3D Spiral 87.6 MHz db 3D Rotate Spiral MHz db Table 4.2: 2D and 3D Air Filled Eye Phantom Coupling Results at a 25 mm Distance Air-Filled Eye Phantom Measurements As shown in figure 4.13, the S21 values drop in comparison to the tests done in air. This is due to different error elements introduced such as the eye testing structure and the plastic covering on the antenna. Over distance, the S21 values start off at a lower value and rapidly decline with distance. Compared to the air measurements, the S21 measurement for the 2D Spiral antenna drops from db to -41 db at a 25mm distance. The 3D Spiral antenna values drop from -31 db to db and the 3D Rotate Spiral antenna decreases from db to db at 25mm. Therefore, a significant amount of error is introduced with the testing structure as well as the addition of the plastic covering surrounding the antenna. Although the testing structure is covered with foam to reduce some interference, it is not completely removed. In addition, the plastic covering introduces a discontinuity in the propagation and may cause a smaller amount of power to be completely transmitted. Table 4.2 summarizes the results for the air-filled eye phantom Vitreous Humor Filled Eye Phantom Measurements With the addition of the vitreous humor solution (described in Chapter 3) to the eye phantom, the S21 and bandwidth characteristics are greatly enhanced compared to the air filled eye phantom. Due to the added dielectric, the resonant frequency is lowered but all of the antennas remained well matched. New patch antennas were

79 70 (a) S-Parameters of the 2D Spiral at 25mm vs. Frequency (b) S-Parameters of the 3D Spiral at 25mm vs. Frequency (c) S-Parameters of the 3D Rotate Spiral at 25mm (d) S21 Comparison of 2D and 3D Antennas over vs. Frequency Distance Figure 4.13: Air Filled Eye Phantom Results

80 71 (a) 2D Spiral Antenna and Corresponding (b) 3D Spiral Antenna and Corresponding Patch Antenna Patch Antenna (c) 3D Rotate Spiral Antenna and Corresponding Patch Antenna Figure 4.14: 2D and 3D Fabricated Antennas and Patch Antennas fabricated using a copper clad board with a substrate dielectric constant of 9.2 to maintain a size similar to an eyeglass lens. The miniature antennas with their respective patch antennas are shown in figure The bandwidth for both the 2D and 3D antennas increased by a significant amount compared to the air filled eye experiment. The 2D Spiral antenna s bandwidth increased from 79.7 MHz to MHz. The bandwidth of the 3D Spiral antenna