Resonant Plasmonic Photomixers

MISSING IMAGE

Material Information

Title:
Resonant Plasmonic Photomixers
Physical Description:
1 online resource (106 p.)
Language:
english
Creator:
O'brien, Kathryn E
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Materials Science and Engineering
Committee Chair:
Davidson, Mark R
Committee Co-Chair:
Holloway, Paul H
Committee Members:
Norton, David P
Pearton, Stephen J
Tanner, David B

Subjects

Subjects / Keywords:
lithography -- mems -- plasmon -- plasmonic -- terahertz -- thz
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre:
Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Photomixing over nano/micro scale plasmonic structures is a novel concept for generating narrow band radiation, specifically in theterahertz (THz) range. The plasmonic structures can serve as antennas for absorbing incoming photons and conversely emit radiation of a lower frequency if it is generated from mixing. In this experiment, antenna structures are excited by two tuned laser diodes to output THz radiation. Various designs for these antennas are have been considered, with presented results focusing on 2dimensional arrays of elliptical antennas fabricated from Ag using electron-beam lithography and lift-off. The plasmonic antenna arrays exhibit polarization dependent absorption when excited by visible light in agreement with results from simulations. The effect of varying antenna size on absorption will be discussed, as well as results from photomixing experiments. The effects of absorption and transmission on the arrays have also been simulated 2dimensionally using COMSOL and MEEP. Array structures were excited by individual and multiple sources in the visible and near-IR range. The effects of varying antenna size, position, and incoming frequencies are examined and compared to some experimental results. Photomixing was observed in a multitude of simulations, and was particularly strong in arrays of shorter antennas with longer antennas placed at intervals of the desired output wavelength.Experimental tests on such an array of 780 nm and 2 µm length antennas showed polarization dependent behavior in the IR regime.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Kathryn E O'brien.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Davidson, Mark R.
Local:
Co-adviser: Holloway, Paul H.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2014-02-28

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045259:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 RESONANT PLASMONIC PHOTOMIXERS By A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

PAGE 2

2

PAGE 3

3 To Mom and Dad

PAGE 4

4 ACKNOWLEDGMENTS First off, none of this would have been possible without the help, support, and guidance of my advisors, Dr. Mark Davidson and Dr. Paul Holloway. Dr. Davidson has been a wealth of knowledge on every subject and has helped me every step of the way. Every da y in his group was a learning experience, and I could not have chosen a better advisor. Dr. Holloway has been a much needed voice of reason throughout this project and without his go ahead attitude and interventions; I may not have been able to complete t his project. His countless accolades and contributions to the field materials science have made him an honor to work with. I also enjoyed the office gator pranks and AVS sugar rush. I would also like to thank my committee members: Dr. David Norton, Dr. Ste phen Pearton, and Dr. David Tanner for their time and participation. T hank you to Chuck Rowland for all the help with building and machining, lab upkeep and safety, the common sense, and, of course, the amazing puppy. Without him I never would have surely overcomplicated every step of my project. Ludie Harmon has been a huge help throughout my journey. From planning my travel, to ordering necessary parts, and making sure I had necessary forms on file. Even when her position changed, she did not stop helping me with my administrative needs. She brought sunshine to some of my most stressful days. To Jon Marburger and Jean Tokarz at Advanced Plasmonics, Inc. for all their help with the lithography. Because of their knowledge, assistance, and access to equipment I was able to meet countless deadlines.

PAGE 5

5 Po Yuan Wang, for the help with the COMSOL simulations and fun times in the lab. To Wesley Hamiln, for all his work on the optical set up and the constant soldering. Their contributions were vital to this dissertat ion. Next, I need to thank my lab mate, Nick Ptschelinz ew for all his help in the lab (especially at the last minute), providing another opinion, keeping me company in the office, and for being a great friend. To my trivia team: Mario Mendoza, Chris Bohling, Ember Patterson, Allison Bodnar, Sarah Onofrio, Jesus Acosta, and Ida Berglund for giving me a bi weekly source of sanity and for naming the year. Thank you for being such great friends. Lastly to my family : Mom, Dad, Rachel, and Molly, for all their love and support through this chapter of my life.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 2 LITERATURE REVIEW ................................ ................................ .......................... 19 Uses and Advantages of THz Radiation ................................ ................................ 19 Summary of THz Radiators ................................ ................................ ..................... 22 Photomixing ................................ ................................ ................................ ............ 22 Plasmonics ................................ ................................ ................................ ............. 25 Plasmonic Photomixers ................................ ................................ .......................... 29 Concept ................................ ................................ ................................ ............ 29 Materials Selection ................................ ................................ ........................... 31 3 EXPERIMENTAL PROCEDURE ................................ ................................ ............ 34 Device Design ................................ ................................ ................................ ......... 34 Method of Selectio n ................................ ................................ .......................... 34 Arrays of Ellipses ................................ ................................ .............................. 34 Arrays of Alternating Ellipses ................................ ................................ ............ 35 ................................ ................................ ................ 36 ................................ ................................ ...................... 36 Final Sele ction ................................ ................................ ................................ .. 37 Simulations ................................ ................................ ................................ ............. 38 COMSOL ................................ ................................ ................................ .......... 38 MEEP ................................ ................................ ................................ ............... 40 Electron Beam Lithography ................................ ................................ ..................... 44 Step 1: Spincoat PMMA ................................ ................................ ................... 46 Step 2: Electron Beam Writing ................................ ................................ ......... 47 Step 3: Development ................................ ................................ ........................ 51 Step 4: Elec tron Beam Deposition ................................ ................................ .... 52 Step 5: Lift off ................................ ................................ ................................ ... 54 Optical Absorption Testing ................................ ................................ ...................... 55 Photomixing ................................ ................................ ................................ ............ 57

PAGE 7

7 Custom Set Up ................................ ................................ ................................ 57 Procedure ................................ ................................ ................................ ......... 61 4 CHARACTERIZATION AND TESTING ................................ ................................ .. 63 Fabrication ................................ ................................ ................................ .............. 63 Simulation Results and Discussion ................................ ................................ ......... 64 COMSOL ................................ ................................ ................................ .......... 64 MEEP ................................ ................................ ................................ ............... 72 Optical Absorption Testing ................................ ................................ ...................... 81 Photomixing ................................ ................................ ................................ ............ 86 5 CONCLUSIONS ................................ ................................ ................................ ..... 93 Fabrication ................................ ................................ ................................ .............. 93 Simulation ................................ ................................ ................................ ............... 94 Photomixing and Testing ................................ ................................ ........................ 95 6 FUTURE WORK ................................ ................................ ................................ ..... 98 APPENDIX: MODELING OF A 3D PLASMONIC SPHERE ................................ ........ 100 REFERENCES ................................ ................................ ................................ ............ 101 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 106

PAGE 8

8 LIST OF TABLES Table page 2 1 A summary of common THz radiators and their capabilities and limitations. ...... 22 2 2 A selection of excitation wavelengths predicted to achieve target THz frequencies. ................................ ................................ ................................ ........ 30 2 3 A summary of possible materials for the plasmonic photomixer and their properties. [29, 58 61] ................................ ................................ ......................... 32 3 1 A summary of potential plasmonic photomixer device designs and their features. ................................ ................................ ................................ ............. 37 3 2 Geometric parameters that were altered and tested using MEEP. ..................... 44 3 3 Typical spin coater parameters for a dilution of 6% PMMA in anisole. ............... 47 4 1 Expected output wavelength and frequency by the plasmonic photomixer for varied tuned wavele ngths from 780 nm and 850 nm lasers. .............................. 86

PAGE 9

9 LIST OF FIGURES Figure page 2 1 An example of the detection of a concealed weapon using THz radiation. [9] ..... 20 2 2 Images of a cancerous human liver. ................................ ................................ ... 21 2 3 A rough diagram of a typical semiconductor photomixer on a glass substrate. .. 23 2 4 A diagram showing two metal nanoparticles undergoing localized surface plasmon resonance (LSPR). [43] ................................ ................................ ......... 27 2 5 A diagram showing incoming light with a wavelength that is greater than the width of the nanowire structure it is exciting. The light will couple to the sur face and excite free electrons, creating a propagating surface plasmon (PSP) which can be contained in the plasmon waveguide. [43] .......................... 28 2 6 Rough diagram of the plasmonic photomixing device. ................................ ....... 30 3 1 An array of ellipse antennas. ................................ ................................ .............. 34 3 2 An example of alternating short and long antenna spaced at wavelength intervals. ................................ ................................ ................................ ............. 35 3 3 ................................ ..... 36 3 4 Proposed desig antennas. ................................ ................................ ................................ ............ 37 3 5 The transmitted mode of the light on the periodic antenna structures. ............... 39 3 6 The reflected mode of the light on the periodic antenna structures. ................... 40 3 7 An example of a 2D array simulated in MEEP. ................................ ................... 42 3 8 Visual representa tions of the parameters changed ................................ ............ 43 3 9 A schematic of the electron beam lithography process. ................................ ..... 46 3 10 PMMA thickness vs. spin speed for 2 7% dilutions in anisole from the manufacturer. [82] ................................ ................................ ............................... 47 3 11 tear ............. 49 3 12 A diagram of the e beam evaporator system used for these experiments. ......... 53 3 13 An example of a silver elliptical antenna array on a silver substrate fabricated for these experiments as per t he procedure in Chapter 3. Some minor

PAGE 10

10 artifacts are visible on the surface as well as residual PMMA from the lift off process (red circled areas). ................................ ................................ ................ 54 3 14 The custom optical absorption testing set up used for the reflection and refraction experiments. ................................ ................................ ....................... 56 3 15 An overhead view of the light path taken through the custom optical absorption set up. ................................ ................................ ............................... 56 3 16 A photograph of the first custom photomixing set up that was designed and used for this dissertation. ................................ ................................ .................... 57 3 17 A close up view of the two laser diodes in TECs for the first photomixing set up. The chopper is in front of one diode. ................................ ............................ 58 3 18 The detectivities of the far IR detector at various wavelengths. [84] .................... 58 3 19 The altered custom set up for photomixing utilizing the pyroelectric detector. ... 60 3 20 The relativ e responsivity vs. wavelength plot for the pyroelectric detector (70124) from the manufacturer. [87] ................................ ................................ .... 60 4 1 The 2D periodic area of the arrays simulated using COMSOL.. ......................... 65 4 2 A time slice image of three 500 nm diameter periodic antennas excited by 444 nm waveleng th light and visually showing efficient plasmonic coupling from one to the next. ................................ ................................ ........................... 65 4 3 Varying the antenna length while holding incoming radiation constant. Wavelength of 450 nm is incoming at a 53 angle. Antenna dimensions are 500 nm in length and 150 nm in height.. ................................ ............................. 67 4 4 Varying the antenna spacing while keeping antenna size constant. Wavelength of 450 nm is incoming at a 53 angle. Antenna dimensions are 500 nm in length and 150 nm in height ................................ ............................. 68 4 5 Varying the antenna spacing while keeping antenna size constant. Wavelength of 675 nm is incoming at a 53 angle. Antenna dimensions are 500 nm in length and 150 nm in height ................................ .............................. 69 4 6 Simulated excitation from a 500 nm antenna.. ................................ ................... 70 4 7 Reflectance vs. wavelength simulation for three arrays. ................................ ..... 71 4 8 Time slice images of an array of 500 nm antennas excited by light of a wavelength of 1 m ................................ ................................ .......................... 74

PAGE 11

11 4 9 Three time slices of a simulated antenna array showing plasmonic excitation, but not plasmon resonance. ................................ ................................ ............... 76 4 10 The transmission vs. wavelength for the antenna structures in Figure 4 9.. ....... 77 4 11 A simulation with the longer antennas spaced at the mixing wavele ngth of 3880 nm at a time of 157 s. Stronger modes are visible on the longer structures than the short ones. ................................ ................................ ........... 78 4 12 The transmission vs. wavelength for the antenna structures in Figure 4 11 ... 78 4 13 A 99 s time slice image of an array with longer ante nna spaced every four short antenna. Incoming wavelengths are 780 nm and 635 nm. ........................ 79 4 14 The transmission spectra for the array in F i gure 4 13 ................................ ....... 80 4 15 Intensity vs. wavelength for a set of simulations in which the length of the large antenna was varied by hundredth of a micron (antenna lengths given in legend). The 2.1 m length antennas show the strongest intensity at a wavelength of 6 m. ................................ ................................ ........................... 81 4 16 The 500 nm length antennas tested for this experime nt. The imperfect shape and size are artifacts of the beam stigmation and focusing. ............................... 82 4 17 Experimental reflectance vs. simulated reflectance measurements for a 500 nm antenna array taken at polarizations varying 0 90 ................................ ... 82 4 18 Integrated reflectance vs. polari zation angle for two absorption dips from the reflectance vs. wavelength graph in Figure 4 17A ................................ .............. 83 4 19 An SEM micrograph of the 780 nm antenna array. Many antennas were lifted off in the fabrication process. ................................ ................................ .............. 84 4 20 Experimental reflectance vs. simulated reflecta nce for a 780 nm antenna array.. ................................ ................................ ................................ ................. 84 4 21 Integrated reflectance vs. polarization angle for two absorption dips from the reflectance vs. wavelength graph in Figure 4 21A ................................ .............. 85 4 22 A silver on silver antenna array sample with alternating antenna sizes of 780 nm and 85 0 nm to phase match the lasers. ................................ ........................ 88 4 23 An array of 780 nm antenna with 2 m antenna placed at intervals of every four antenna. ................................ ................................ ................................ ...... 89 4 24 An array of alternating 1 m and 2 m antennas at a wider spacing than the previous samples. ................................ ................................ ............................... 91

PAGE 12

12 A 1 Transverse electric (TE) simulation of the excited plasmonic silver sphere in air. ................................ ................................ ................................ .................... 100

PAGE 13

13 LIST OF ABBREVIATION S BWO Backward Wave Oscillator CW Continuous Wave E B Electron Beam (lithography) EO Electro optic FESEM Field Emission Scanning Electron Microscopy FDTD Finite Difference Time D omain FIB Focused Ion Beam FTIR Fourier Transform Infrared Spectroscopy HEMT High Electron Mobility Transistor IPA Isopropyl A lcohol, C 3 H 8 O IR Infrared LSPR Localized Surface Plasmon Resonance MEEP MIT Electromagnetic Equation Propagation MIBK Methyl isobutyl ketone, (CH 3 ) 2 CHCH 2 C(O)CH 3 MRI Magnetic Resonance Imaging PML Perfectly Matched Layers PMMA Poly(methyl methacrylate), (C 5 O 2 H 8 ) n PMT Photomultiplier Tube PSP Propagating Surface Plasmon PVD Physical Vapor Deposition QCL Quantum Cascade Laser QCM Quartz Crystal Monitor SEM Scanning Electron Microscopy SERS Surface Enhanced Raman Spectroscopy

PAGE 14

14 SPR Surface Plasmon Resonance TDS Time Domain Spectroscopy TE Transverse Electric TEC Ther moelectric Cooler TEM Transmission Electron Microscopy TH Z Terahertz TM Transverse Magnetic UV Ultraviolet

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy RESONANT PLASMONIC PHOTOMIXERS By Kathryn E August 2013 Chair: Mark Davidson Coc hair: Paul Holloway Major: Materials Science and Engineering Photomixing over nano/m icro scale plasmonic structures is a novel concept for generating narrow band radiation, specifically in the terahertz (THz) range. The plasmonic structures can serve as antennas for absorbing incoming photons and conversely emit radiation of a lower frequency if it is generated from mixing In this experiment, antenna structures are excited by two tuned laser diodes to output THz radiation. Various designs for these antennas are have been considered, with presented results focusing on 2 dimensional arrays of elliptical antennas fabricated f rom Ag using electron beam lithography and lift off. The plasmonic antenna arrays exhibit polarization depende nt absorption when excited by visible light in agreement with results from simulations. The effect of varying antenna size on absorption will be discussed, as well as results from photomixing experiments. The effects of absorption and transmission on the a rrays have also been simulated 2 dimensionally using COMSOL and MEEP. Array structures were excited by individual and multiple sources in the visible and near IR range. The effects of varying antenna size, position, and incoming frequencies are examined an d compared to some experimental results. Photomixing was observed in a multitude of simulations, and was particularly strong in arrays of shorter antennas with

PAGE 16

16 longer antennas placed at intervals of the desired output wavelength. Experimental tests on such an array of 780 nm and 2 m length antennas showed polarization dependent behavior in the IR regime.

PAGE 17

17 CHAPTER 1 INTRODUCTION Radiation detectors and emitters have been used increasingly in the past century for a number of applications. X ray radiation is commonly used for detection of broken measure ionizing radiation to detect radioactivity. Infrared (IR) radiation is frequently used in materials science for Fourier Transform Infrared Spectroscopy (FTIR) to analyze various compounds. For certain applications, a different type of radiation det ector would be more desirable. For example, x ray radiation has become commonplace, but its ionizing radiation is dangerous and it is incapable of imaging tissues. Magnetic resonance imaging (MRI ) is a non ionizing technique with a great contrast in detect ing various tissues. [1] However, the MRI is not at all portable and requires a great deal of power to operate A portable, non ionizing source of radiation for medical imaging could revolutionize the field and improve care in a number of scenarios. A THz device could be the perfect solution for this problem. In another example, FTIR is excellent for a identifying a great deal of materials, but it cannot be used for free radicals or weakly bonded compounds. In such cases, a terahertz (THz) emitter can be an ideal replacement for these devices. THz radiation is a form of black body radiation and is emitted weakly fr om all things with a temperature greater than 10 K. [2] THz radiation is safe and non ionizing, so there are no concerns from exposure. [3] Additionally, THz radiation can penetrate plastics, fibers, and fog, but is strongly absorbed by metal and water.

PAGE 18

18 A number of THz emitters have been developed but all have major drawbacks. Some of these radiators will be discussed in further detail in Chapter 2. This dissertation wil l discuss a novel technique for emitting THz radiation: the plasmonic photomixer. This device meets all the crite ria for a viable source and has the potential to be backwards compatible as a detector. While the THz sources are the primary motivation for th is work, it should be noted that plasmonic photomixing is a novel method for enhancement of photomixing and this work is thus focused upon potential designs and predicted performance of a variety of potential photomixer devices. A literature review of THz radiation, plasmonics, and photomixers is given in Chapter 2. Experimental procedures and device design are covered in Chapter 3. Characterization and testing results of these novel devices, both by simulated and experimentally, are contained in Chapter 4. Conclusions of these studies are summarized in Chapter 5. Finally, future studies recommended for this project and area of research are given in Chapter 6.

PAGE 19

19 CHAPTER 2 LITERATURE REVIEW New and more portable means of generating narrowband radiatio n is of great interest, especially in the terahertz range. Until recent years, despite being weakly emitted as black body radiation by almost all things, THz radiation was almost i mpossible to efficiently produce and detect. [2] Thus, the range of 300 GHz to 30 THz became named challenging and problems with portability and efficiency remain. [4] Thus, there are no portable low cost high power THz sources currently available. [5] Uses and Advantages of THz R adiation THz radiation offers potential for more powerful detection of concealed weapons and explosives, materials spectroscopy, and medical imaging. [6] In terms of weapons detection, THz radiation can penetrate plastics and fabrics in order to uncover concealed weapons (Figure 2 1). [7 9] New full body scanners in airport secu rity have already begun utilizing what can be considered sub terahertz radiation to scan for weapons through bulky clothing. Concealed weapons detection from distances of up to has already been demonstrated by the New York Police Department. [10] Further the THz absorption spectroscopy can be used to detect specific explosive materials from distances of te ns of meters away, which would be specifically useful in military and bomb squad applications. [11]

PAGE 20

20 Figure 2 1 An example of the detection of a concealed weapon using THz radiation. [9] Materials spectroscopy using THz radiation known as THz time domain spectroscopy (TDS), is similar to Fourier transform infrared (FTIR) spectroscopy, but is capable of detect ing materials having strong vibrational modes in the THz range. Thi s is important in the study of reactive species such as free radicals, ions, high temperature superconductors, and weakly bound compounds. [12 16] In addition, many excitonic transitions of semiconductors occur in the THz range. [17 19] THz TDS is already in use in many research laboratories, but is currently limited by low power sources and slow acquisition times (~1 ms) [16] Typical systems also use expensive femtosecond Ti:sapphire lasers as opposed to more desirable semiconductor diode lasers. [17] Therefore, further developments are required before THz TDS systems become more popularized. Lastly, THz radiation is very promising in medical imaging, as the radiation is non ionizing and a million tim es lower in energy than x rays. [3] Therefore, it is less harmful and a much safer alternative than commonly used x ray radiation. [20] In addition, THz imaging is capable of measuring the densi ty of water in tissue and bone,

PAGE 21

21 which is the primary factor i n the contrast in THz imaging. [21] Thus, THz imaging would aid in noninvasively identifyin g specific tissues in the image, much like the imaging contrast t hat can be achieved using an MRI. However, THz devices can be portable and a fraction of the cost of an MRI. In addition, t he imaging contrast achieved using THz is specifically useful in detection of tumors, as water retention in tumors is very different th an that in healthy tissue. For example, Figure 2 2A depicts a photograph of a cancerous human liver. Figure 2 2B shows the THz image of the same liver, with the cancerous tissue clearly identified in the dark areas. [22] The safe, non invasive detection of tumors is one of the most exciting uses of this technology. Again, current market models are limited by expensive sources and slow acquisition times. Overall, a continuous wave THz source has the potential to be a safe, portable, and cost effective form of medical imaging which would lead to lead to fast detection of diseased tissue. [23] Figure 2 2 Images of a cancerous human liver. A ) Photograph of a human liver containin g tu mors (metastasis) B ) THz image of the same liver where the dark areas represent the cancerous tissue [23]

PAGE 22

22 Summary of THz Radiators There are a number of devices on the market which are able to produce THz radiation. However, there is not one portab le, tunable, high power CW device. For example, the quantum cascade laser (QCL) is the most advanced and high power source available; but it is not tunable and it is both expensive and complicated to fabricate and use as it is only operable at cryogenic temperatures High electron mobility transistors (HEMTs) on the other hand, have been able to produce up to approximately 6 THz, but are limited by low power and are difficult to tune. [24, 25] A summary of the pros and cons of some of the most popular devices is shown in Table 2 1, and a the most relevant sources to this project will be discussed in further detail in the coming sections. Table 2 1. A summary of c ommon THz radiators and their capabilities and limitations Generator High Power Portable Tunable Auston Switch X Smith Purcell Limited Backward Wave Oscillators (BWO) Limited HEMTs X Quantum Cascade Lasers (QCL) X Synchrotron X Photomixers X X Plasmonic Photomixers X X X Photomixing Photo mixing is defined as the conversion of wavele ngth through a medium or a meta material. In this technique, two continuous wave (CW) co linear incoming laser frequencies are focused over the photomixing material and the nominal difference in their frequ encies is the output in the form of a beatnote (Equation 2 1 ) (Figure 2 3 ). Photomixers are the most commonly used and researched CW THz source due to their low cost, ability to be operated at r oom temperature, and tunability. [26, 27] For the

PAGE 23

23 output of THz radiation, the photomixing medium is typically a crystal, a semiconductor thin film, or a nano scale semiconductor structur e. This process can also be used in reverse to create a large area THz detector. [28] (2 1) Figure 2 3 A r ough diagram of a typical semiconductor photomixer on a glass substrate. Generally, a nonlinear optical material is utilized for photo mixing, such as III V and II VI semiconductors, which tend to have a large third order nonlinear optical susceptibility, (3) Equation ( 2 2 ) shows the relation of polarizati on density, P (t), to electric field, E (t), through the n th order susceptibilities, ( n ) ( 1 ) is the linear susceptibility, which describes most basic optical processes. ( 2 ) is the second order nonlinear susceptibility, which is dependent on crystal structure and controls processes such as second harmonic generation. Lastly, (3) as previously stated, is the third order nonlinear susceptibility It is related to third harmo nic generation and the phenomenon of intensity dependent refractive index. [29] The relation of the real and imaginary ( 3) to the nonlinear refractive index, n 2 is shown by e quation s ( 2 3 ) and ( 2 4 ) respectively. [30] In these equations, c denotes the speed of light in air 0 is the permittivity of free space, n 0 is the linear refractive index, is the wavelength, and is

PAGE 24

24 the nonlinear absorption coefficient This relationship is known as the Kerr effect or the quadratic electro optic (EO) effect. [31] (2 2 ) (2 3) (2 4) The majority of p ast photomixing experiments have used semiconductor structures and thin films, most commonly low temperature grown GaAs antenna structures (typically a single interdigitated electrode) on glass substrates as the photomixer i n order to output THz radiation. [21, 32, 33] GaAs is the material of choice for these devices as it is extremely photoconductive has a short carrier lifetime (0.2 ps) and has a h igh third order nonlinear susceptibility Additionally, the design of the antenna has a significant effect on the efficiency of the photomixing. [34] When the two incoming lasers of identical polarization illuminate the G aAs antenna structure, an applied bias collects the photogenerated carriers and the antenna emits THz radiation through the glass substrate which forms a lens to enhance outcoupling [17] The instantaneous optica l power incident on the photomixer in terms of the average powers, P 1 and P 2 and average frequencies, 1 and 2 of the lasers is given by Equation 2 4 Where 1 2 is the nominal difference from e quation ( 2 1 ) and 1 + 2 is the mixing product that is not co upled well to the photomixer. [21] The sizes of these devices are limited to that of the laser spot size on the substrate. Thus, the interaction area of traditional photomixers is very small. To date, photomixing experiments have only been able to output radiation from 300 GHz to 5 THz, and tend to have extremely

PAGE 25

25 low output powers typically on the order of W, due to the low thermal conductivity of GaAs. [21, 32, 35] (2 4 ) As previously stated, the design of the antenna has a significant effect on photomixing efficiency. Recent studies have shown that replacing the traditional interdigitated electrodes (IDEs) with arrays of nanoantennas can enhance the photomixed output of a 1 THz wave by more than two orders of magnitude. [26] The goal of this project is to reach much farther into the THz regime and to create higher power tunable devices, which will requi re a modification of this photo mixing technique. Plasmonics A way to extend the range o f radiation output of the photo mixing technique is to increase the interaction area. This can be done by employing plasmonics. A plasmon is a quantum of a plasma oscillation, or a single oscillation of a single plasma quasiparticle. This is a quasiparticle created from plasma oscillations and can be thought of in a s imilar manner to a phonon, which is a single vibration from a crystal. The plasmon particles are many orders of magnitude larger than photons; therefore their interaction area is much greater. [36] While plasmo nics have become more understood only in recent years, their properties have been observed and used for centuries. The most common example of this is stained glass. The colors in stained glass are caused by the metal nanoparticles in the glass matrix. The particles interact with incoming light to change the wavelength of the outgoing light. This is all due to the phenomenon known as plasmon resonance,

PAGE 26

26 where the plasmonic oscillations around the nanoparticles absorb and reflect different wavelengths. [37] When these metal nanoparticles are confined to a surface, which in this case is considered as the interface between a metal and any dielectric material, they may interact more strongly with light. The incoming e lectromagnetic light wave causes the surface plasmons to resonate and induces an oscillating electric field on the surface. [38] If the specific wavevector of the incoming light, k matches the wavevector of the surface plasmons, k sp then the entire system is in resonance and the electromagnetic field couples to the surface plasmons. This event creates a surface plasmon polariton, which is an electromagnetic excitation propagating at the metal/dielectric interface. [39] Equation ( 2 5 ) is the expression for the wavevector of the incident light where is the incident wavelength, n is the refractive index of the medium through which the light travels, and is the angle of incidence. Equation ( 2 6 ) shows the expression for the 1 is the dielectric permittivity constant of the metal, and 2 is the dielectric permittivity constant of the ex it medium. [40] Equation ( 2 7 ) shows the condition for this process. (2 5 ) ( 2 6 ) (2 7 ) The result of an incoming beam of light striking the surface at this precise angle is a highly attenuated light reflection (not to be confused with emission) This process by which an incident light excites surface plasmons and produces an attenuated reflected light is known as surface plasmon resonance (SPR) (Figure 2 4 ). [41] A common use of

PAGE 27

27 SPR is in surface enhanced Raman spectroscopy (SERS). The surface enhancement of this technique is caused by metal nanoparticles, which when in resonance with the light to b e detected, can greatly increase the field intensity. [42] Figure 2 4 A diagram showing two metal nanoparticles undergoing localized surface plasmon resonance ( L SPR) [43] Conversely, if light strikes a metal nanostructure, particularly one where the wavelength of the light is larger than the width of the structure, the nanostructure may serve as what is known as a plasmon waveguide. In this case, the light couples to the metal surface, which in turn excites its free electrons. These electrons then travel across the waveguide and the corresponding effect is known as a propagating surface plasmon (PSP) (Figure 2 5) [43] Plasmon waveguides can be fabricated as metal efficient coupling from one to the next. [39] Plasmon waveguides have become more prevalen t in electronic devices in recent years for their efficient propagation and have been an area of great interest of research [44 47] Of particular interes t, plasmon waveguides have been used to effectively transmit and confine THz radiation waves on over long distances, which could not otherwise be achieved using traditional wires or dielectric IR optical fibers [27, 46, 48 51] These waveguides have been able to effectively transmit up to approximately 4 THz.

PAGE 28

28 Figure 2 5. A diagram showing incoming light with a wavelength that is greater than the width of the nanowire structure it is exciting. The light will couple to the surface and excite free electrons, creating a propagating surface plasmon (PSP) which can be contained in the plasmon waveguide. [43] To make the se plasmonic effects, particularly SPR, useful for device physics the metal nanoparticles from Figure 2 4 can be replaced with arrays of resonant metal nanos tructures or nano scale antennas W ith the introduction of the antenna structures, a grazing incidence angle is no longer required. When light strikes one structure, it is able to couple to those near it and over a large interaction area. Devices utilizing su ch plasmonic nano scale antennas have been successfully fabricated at Advanced Plasmonics, Inc. (Gainesville, FL). [52] These devices have been shown to emit visible light when excited by an electron beam source. Thus, this s ource excites tuned resonant nano scale structures and output visible light. This process is similar to the techni que of Smith Purcell radiation in which an electron beam passes through a metal grating to output radiation The wavelength, of this radia tion is dependent on the grating period, l the spectral order, n the electron velocity relative to the speed of light, = / c and the angle of emission normal to the surface of the grating, (Equation 2 8 ) (2 8 )

PAGE 29

29 While the plasmonic devices designed by Advanced Plasmonics, Inc. are similar to Smith Purcell radiation in that they use metal structures excited by an electron beam, they differ in that the structures are tuned in multiple dimensions to become resonant with the system. These structures are designed to be excited at a specific plasmon frequency. Neighboring antenna in the array are closely coupled, thus, they resonate, causing a shift in their eigenfrequencies. If the entire system is designed su ch that the wavevectors of all of the modes match, then the entire system can become resonant, and becomes an efficient oscillator pumped by the energy of electron beam space charge field. Plasmonic Photomixers Concept This project comb ines the concepts of photo mixing and plasmonics, as it has been predicted that plasmonic devices are able to handle greater than 100 THz and can be tunable over a large range of frequencies. [47, 53, 54] T he concept is theref ore photo mixing two incoming wavelengths over an array of plasmonic antenna structure t o output a tunable range of frequencies which could include the THz regime In principle, the plasmonic structures allow for the wavevectors to be c ollinear and collocated in a small space to allow mixing just as was the case in the traditional photomixer. However, the main difference between the traditional photomix er and the plasmonic photomixer is that the light e nergy is converted into plasmon ene rgy in this new mixer, rather than exploiting the photons by using a photoconductive material in the traditional mixer. By making this conversion to plasmons, we can take advantage of their long range (micron scale) interactions. Thus, these devices can be much larger than prior photomixers (not limited to laser spot size) due to the coupling from one structure to the next. By forcing

PAGE 30

30 the wavevectors to travel along the plasmonic antenna array, the mixing is allowed to take place (Figure 2 6 ). Figure 2 6 Rough diagram of the plasmonic photomixing device. The wavelengths to be converted are those of two incoming solid state near infrared lasers. The primary excitation of these tes t prototypes will be two solid state semiconductor lasers, one held constant at 780 nm, while the secondary excitation from a is tuned and held constant at a slightly different frequency. Depending on the desired output, this laser could also be 780 nm or 850 nm. The exact tuning frequency can be determined by the desired output wa velength and the sizes of the antenna structures. For example, emission of 1 THz could be achieved by mixing the 780 nm wavelength laser with one tuned to 782 nm by equation ( 2 1 ). A summary of potential excitation wavelengths to achieve various THz freque ncies is contained in Table 2 2. To keep the lasers held constant and tuned, thermoelectric coolers (TEC) are utilized. These lasers will require much less power than an electron beam and could be battery powered, allowing the entire device to become porta ble. In addition, 780 nm and 850 nm laser diodes are particularly inexpensive and tunable, making them ideal for this application Table 2 2. A selection of excitation wavelengths predicted to achieve target THz frequencies. Target Frequency Primary Excita tion Secondary Excitation 0.1 THz 780 nm 780.2 nm 1.0 THz 780 nm 782 nm 10 THz 780 nm 800.9 nm

PAGE 31

31 In addition, the device also has the potential to be reversed into becoming a detector. As such, the detector would be constantly illuminated by one laser of a set wavelength, for example 780 nm. When the radiation to be detected illuminates the device, a detectable side band should be generated. This potential phenomenon was not tested for this project, but would be an interesting area to examine in the fut ure. Materials Selection In terms of the materials selection for the device, there are a number of considerations to be taken into account. As previously mentioned, photo mixing devices commonly use photoconductive semiconductor structures with a large (3 ) such as low temperature grown GaAs, as the mixing media. The GaAs structure is typically deposited on glass and emission is detected from the bottom of the substrate. For this project, glass substrates will not be utilized. Rather, the surface plasmons of these plasmonic structures are on the tops of the antenna structures on a substrate of the same material, where air acts as a dielectric (rather than a glass substrate). Further, to fully take advantage of the effects of surface plasmons, a very electri cally and thermally conductive material will be needed ; areas where GaAs falls short Lastly the material should have a wide range of plasmon frequencies and a high plasmon mean free path so that the device may be tunable over a large range of frequencies The three most electrically conductive metals, from highest to lowest conductivity, are silver, copper, and gold. [55] In addition, these three noble materials all display very different optical properties when in nanoparticle form, which is attributed to plasmon resonance. [56] As copper is less commonly used for patterning, silver a nd gold remain as the top candidates. Because silver has the highest conductivity of any metal, is relatively inexpensive, and has a huge range of plasmon frequencies, it will be

PAGE 32

32 the prime material for this project. [57] However, other metals and nonlinear optical materials, such as GaAs, can be considered for this project due to their use in traditional photomixing devices photoconductivity, and strong optical nonlinearity Work on such nonlinear optical materials will not be covered in this dissertation due to time constraints but would be an exciting area to research in future studies. A summary of the potential materials for this device and their properties is given in Table 2 3 Table 2 3 A summary of possibl e materials for the plasmonic photomixer and their properties [29, 58 61] Material Ag Au Cu GaAs Conductivity ( ) (S/m) at 20 C 6.3 x 10 7 4.1 x 10 7 5.96 x 10 7 5.0 x 10 3 Thermal Conductivity 1 1 ) 429 318 401 55 Nonlinear susceptibility ( (3) ) (esu) 2 x 10 11 5.4 x 10 11 1 x 10 10 Plasmon Energy ( E p ) (eV ) 25 24.8 19.3 16.5 Plasmon Mean Free Path ( p ) (nm) 125 120 100 Space Group Fm m Fm m Fm m F 3m A problem with the use of silver for this project is its high oxidation rate. [62, 63] For these experiments, the silver samples can be stored in a desiccated vacuum box. However, a more long term solution will need to be examined for the future. In this case, gold may be preferable, or a thin film coating will need to be applied to the material. This coating will certainly affect the efficiency of the device a nd reduce plasmonic excitation. Some research has been done into conductive polymer coatings on copper and silver to reduce corrosion. In the past, benzotriazole has been used as a coating for these metals. Specifically, benzotriazole is commonly used as a coating for plasmonic silver nanoparticles in SERS. [64] More recently, the ethoxy derivative of polyanaline has been

PAGE 33

33 found to have good adhesion to silver and copper and excellent corrosion resistance, superior to that of benzotriazole. [65 68] Another major challenge in this materials selection is that m etals are not traditionally thought of as nonlinear opti cal materials. H owever recent research has found nonlinear plasmonic excitations from silver and gold nanoparticle clusters and metastructures. [69 73] Even more promising is the use of plasmonic metal gratings in four wave mixing. In these experiments, a three fold increase in the (3) of gold was found when using gold metastructures as opposed to flat films. [72] In addition, nonlinear optical behavior of platinum and palladium nanoparticles has also been shown when excited by a femtosecond laser. [74] The excellent results from t hese studies confirm that metals such as silver and gold can be used for t he plasmonic photomixing device, as the plasmonic behavior is inherently nonlinear.

PAGE 34

34 CHAPTER 3 EXPERIMENTAL PROCEDU RE Device Design Method of Selection A variety of device designs wer e proposed for this project. These designs were of varying shapes and sizes and included arrays of r ods, ellipses and combinations of said structures. The motivation for and advantages of each of these designs will be discussed in detail in this section. Arrays of Ellipses As it is already known that plasmonic rod structures are excellent at emitting light and are known to strongly couple to one another it is natural to design similar structures for this application.[52] Such rod shapes were considered however the fabrication process used for this experiment would leave their corners rounded. As the nanorods tend to be rounded on the ends in the electron beam writing process, which will be discussed later in this chapter, arrays of similar elliptical structures were chosen as a candidate for this project (Figure 3 1) and modeling was based on elliptical antennas. The elliptical antenna designs were limited to those in which the antennas were in close proximity to one anoth er to ensure that they are closely coupled and form a resonant system Figure 3 1. An array of elliptical antennas.

PAGE 35

35 Arrays of Alternating Ellipses As previously stated, the ellipse array was identified as a strong candidate for this application due to it s demonstrated plasmonic resonance. However, since the system involves absorption of multiple wavelengths, design modifications were made to predict and optimize absorption and photomixing of both incoming laser wavelengths For example, p atterns with repe ated elements of the same elliptical shape but with alternating sizes were hypothesized to be a strong candidate to induce photomixing on the silver surface due to the multiple size scales which could then be tailored for the design wavelengths (Figure 3 2 ) The alternating sizes were designed to absorb the second laser wavelength or couple the predicted photomixed wavelength Variations on these designs were examined in simulations and t he results of these experiments will be discussed in detail in Chapter 4. Figure 3 2. An example of alternating short and long antenna spaced at wavelength intervals.

PAGE 36

36 Another potential antenna structure would be one of rows of interlocking resonant Figure 3 3 ). The interlocking of the structures w as investigated as a method for enhancing long range coupling from one structure to the next to form a resonant system. It was also thought that the could better couple by aligning to the wavevectors o f the incoming laser light. Lastly, the larger size of such antenna could enhance the long wave emission from the device. Figure 3 3 Another variation on the traditional bow tie structure of non plasmonic itself would have sharp corners, which are f avorable for plasmonic excitation due to the field concentration and when close to one another ar e excellent for coupling. Thu s, the not directly underneath each other from row to row, but shifted to preserve this important design factor. Adding arrays of ellipses such that the larger scale bowties couple w ith one another and absorb the light from the laser d iodes further increase the long scale interactions and create a resonant system

PAGE 37

37 (Figure 3 4 ). One drawback to this design is the difficulty of creating sharp corners via electron beam lithography. Figu re 3 4 antennas. Final Selection There were a number of potential designs considered. In order to provide the baseline information the ellipse arrays, both regular and alternating w ere chosen as the primary systems to be analyzed in simulations and experimentally for this project The results of this work will be presented in Chapter 4 H owever future work will focus on detailed analysis of the more complex systems For reference, a summary of the potential device designs for this project and their features and advantages is presented in Table 3 1. Table 3 1. A summary of potential plasmonic photomixer device designs and their features. Design Features Ellipse Array Ease of fabrication Demonstrated plasmon coupling and resonance Alternating long/short ellipses Ease of fabrication Increased nonlinearity for photomixing Long wave propagation Enhanced coupling Long scale symmetry Long wave propagation shapes with interspaced ellipses Increased nonlinearity for photomixing Sharp corners for plasmon excitation Long wave propagation

PAGE 38

38 Simulations The devices were simulated using t wo different programs: COMSOL Multiphysics (Stockholm, Sweden) and MEEP (MIT Electromagnetic Equation Propagation) The combination of the two simulation software packages gives a clear picture of the plasmonic interactions taking place on and around the arrays and helps in the design of the arrays to improve device quality. The s imulations also aid in explaining the phenomena observed in optical experiments. COMSOL COMSOL Multiphysics, a finite element analysis and engineering software, was used to create 2D and 3D simulations of the antenna excitation. COMSOL has superior gr aphics to MEEP and can be used to examine the variation of antenna size, incoming light angle, light frequency, and antenna shape on absorption. It is also much easier to develop 3D models in COMSOL due to its graphic interface and high powered platform H owever, a disadvantage is that it is very difficult to do time domain simulations with this software. The 2D simulations were of a n infinite periodic array of rectangular silver antenna on silver substrates in air in the xz plane The material properties for silver were obtained from the COMSOL materials library. The antennas were excited by a point source of light ranging from the visible to near IR. First, the frequency of the light source was varied over this range at a fixed angle of 45 and the effect s on the antenna were qualitatively observed in the simulated images. After finding a frequency that the ant enna strongly responded to, the strongest intensity response frequency was fixed and the incoming angle was varied. Once the most intense response f rom the varying

PAGE 39

39 incoming angle on the antenna was found, reflectance and refracted data was simulated over the visible range to near IR at that fixed angle. The simulated transmittance and reflectance data was later compared to experimental data for sample s of silver periodic antenna array to confirm validity of the results and draw further conclusions Data for both transmitted and reflected modes were analyzed using COMSOL. For the transmitted mode, the solution must satisfy equation ( 3 1 ) where m is an integer for the order of diffraction 0 is the wavelength of light in vacuum, n is the refractive index of the material, and m is the transmitted diffracted light of the m th order. Figure 3 5 shows a diagram for light striking a periodic antenna array in these conditions. [75] (3 1) Figure 3 5 The transmitted mode of the light on the periodic antenna structures. For the reflected mode, the simulation must follow the conditions of equation ( 3 6 ) where m is the reflected light of diffraction in the m th order. A diagram of a periodic antenna array under these conditions is shown in Figure 3 5. (3 2)

PAGE 40

40 Figure 3 6 The reflected mode of the light on the periodic antenna structures. Because the polarization of the incoming wave is important to these simulations, two modes must be examined. In the transverse electric (TE) case, the electric field component is in the y direction, which is the direction out of the 2D x z plane we are imag ing. In the transverse magnetic (TM) case, the electric field component is in the x z plane and perpendicular to the direction of propagation of the incoming wave. [75] MEEP MEEP, a Linux based software, was used to create 2D simulations of coupling and photomixing. It is programmed in Scheme, which is a subset of GNU Guile. The reactions of antenna arrays to Gaussian sources of near IR light were simulated. MEEP is an excellent tool for viewing these interactions over a long period of time, and movies can be made by compiling time slice images. The interactions between light and the structures can be examined qualitatively through these time slice images as well as quantitatively by measuring transmission and reflection through the structures. MEE P solv for electromagnetic materials and systems, and is particularly useful for analyzing re sonant modes, which are the basis for the

PAGE 41

41 plasmonic photomixing device. 4), (3 5), (3 6), and (3 7), where D denotes the dielectric displacement, E denotes electric field, H denotes magnetic field, B denotes ma gnetic field, J ext denotes current density and ext denotes the external charge density. [39] (3 4) (3 5) (3 6) (3 7) MEEP also recognizes third order nonlinear susceptibility, which will be employed to simulate the photomixing of the incoming sources. To avoid interactions perfectly matched layers (PML) is employed. The PML are added with a chosen thickness to absorb light. Like those simulations performed using COMSOL, the MEEP simulations were performed in 2D. However, rather than examining a single row pe riodic array, the xy plane was investigated ( Figure 3 7 ). Therefore, the interactions between and on top of antennas were able to be clearly visualized. The simulated structures were initially designed as perfect conductors, but were later given a high (3 ) value on the order of 10 2 to induce nonlinearity in the system In addition, by including the quantitative transmission and reflection measurements t hrough the structures, a representation of the device in the xy plane was able to be created. The quantitative measurements were subtractions can be made from the initial intensity of light versus the intensity of

PAGE 42

42 radiation measured by the port. The designs used in MEEP can easily be fabricated and tested physically; this will be discussed in greater detail in the following section. Therefore, by simulating the devices, only those devices showing strong simulated resonance are fabricated in order to keep costs down. In ad dition, the simulations may be able to explain interactions seen in the experimental results. Figure 3 7 An example of a 2D array simulated in MEEP. The parameters available to change and investigate using MEEP are seemingly infinite. A combination of t wo frequencies were used and varied to excite the antennas. However, due to time constraints, the frequency investigations were kept to a minimum (one source wavelength was always held constant at 780 nm to mimic experimental ly available equipment ) in orde r to further observe the effects of geometry. The change in coupling, resonance, and transmission are able to be observed quantitatively and qualitatively through even the smallest geometric alteration in the system One objective

PAGE 43

43 of this project is to fin d a resonant array using the tuned IR and near IR lasers available Visual representations of these changes are shown in Figure 3 8 A B C D E F Figure 3 8 Visual representations of the parameters changed. A) Antenna length B) Long antenna length C) Antenna spacing D) Antenna width E) Placement of long antenna originally F) Periodicity of long antenna rows varied

PAGE 44

44 Further, a summary of the geometric parameters changed in the MEEP simulations is contained in Table 3 2 A variety of frequencies were also tested over these arrays, with long antenna step sizes corresponding to the wavelength of the desired photomixed output. Table 3 2 Geometric parameters that were altered and tested using MEEP. Parameter changed x axis y axis Antenna length (all) 400 10 00 nm Long antenna length 800 2000 nm Antenna spacing 30 200 nm 20 400 nm Antenna width 30 100 nm Placement of long antenna Every 2 4 Electron Beam Lithography The devices used for this project were fabricated using electron beam (eB) lithography. This method is carried out on a SEM. This method is capable of fabricating structures down to approximately 10 nm in size, as opposed to ~750 nm for photolithography, or 50 100 nm for ultraviolet (UV) litho graphy. In addition, eB lithography can write overall patterns that are very large (>3 mm). [76] Unlike conventional photolithography or UV, there is no need for costly custom masks that take time to be produced. This is especially important at smaller feature sizes as masks for deep UV lithography can cost on the order of a million dollars. Another great advantage of eB is that the resist used for the process can be exposed to visible light, eliminating the need t o confine the experiments to a cleanroom. T he field emission SEM (FESEM) used for these experiments was located in the Major Analytical Instrumentation Center (MAIC) at UF is equipped with an eB writing system Nabity Pattern Generation System (NPGS) [77] Designing structures is easily (Upperspace Corporation) For

PAGE 45

45 the large arrays used for this project, a custom written macro was employed to quickly and accurately gen erate antenna array patterns. Thus, designing structures and patterning them can be done quickly by way of this process This system has been successfully used for many years now to wr ite sub 100 nm patterns over mm scale ar eas for other research programs, and was henceforth chosen as the primary means of fabrication for this process. [76, 78] If this device is to be commercialized in the future, it will require a faster means of processin g, such as nanoimprinting, an ultrahigh resolution process in which a template mold of a sample is formed for easy replication, the resonant plasmoni c photomixer device could potentially be very inexpensive to produce. [79, 80] The electron beam lithography process consists of five major steps: spincoating resist, electron beam writing, developing, metal deposition, and lift off (Figure 3 9 ) The procedures for each of these steps shall be detailed in this section. It should be noted that for cost reasons, rather than using a pure silver substrate, the substrate used in this process is (100) silicon coated with a 150 nm thin film of Ag. Because silver does not adhere well to silicon, a 5 10 nm layer of Ge was deposited onto the silicon substrate prior to silver deposition [81] The Ag thin film and adhesion layer of Ge are deposited by electron beam deposition on a (100) n type silicon wafer prior to th e start of this process discussed in this section The wafer is diced into small chips, which are stored in a desiccated vacuum box until they are need ed for lithography to help prevent the corrosion of the silver films T he details of the electron beam de position process used in this preparation step will be discussed later in this section in Step 4 in the

PAGE 46

46 lithography process as the same method is also used to deposit the silver antenna structures. Figure 3 9 A schematic of the electron beam lithograph y process. Step 1: Spincoat PMMA First, the thin film silver on silicon sample can be coated with resist. The chosen resist is a mixture of polymethyl methacrylate (PMMA) in chlorobenzene or anisole. It is a negative resist, which means that the areas expo sed in patterning will be those where the devices will be deposited. This resist is spin coated onto the sample using a Specialty Coating Systems Inc. Model P6708 spin coater The desired thickness of the resist is controlled by the concentration of PMMA in chlorobenzene o r anisole and the spin speed. Resist thicknesses may vary from 250 900 nm, depending on the desired structure thickness. A chart from the manufacturer is used to determine this spin speed (Figure 3 10 ) and typical parameters for a coating are given in Table 3 3 It is also

PAGE 47

47 important to note that, over time, the PMMA tends to cross link, causing the resist to thicken. This can be corrected for by increasing spin speed or adding more solvent to the dilution. After spin coating, the sample is baked in air on a hot plate at 180 C for 15 minutes to harden the resist. Table 3 3 Typical s pin coater parameters for a dilution of 6% PMMA in anisole Parameter Value Units RPM1 6 00 Revolutions per minute (rpm) Ramp1 0 Seconds (sec) Time1 15 Sec RPM2 25 00 RPM Ramp2 5 Sec Time2 30 Sec RPM3 6 00 RPM Ramp3 5 Sec Figure 3 10 PMMA thickness vs. spin speed for 2 7% dilutions in anisole from the manufacturer. [82] Step 2: Electron Beam Writing The microscope used for this project is a Phillips XL 40 FESEM which has nanometer scale resolution. As there is no load lock on this system, the whole chamber

PAGE 48

48 is evacuated using a turbomolecular high vacuum pump backed by a mechanical vacuum pump Operating chamber pressure is typically 1.25 x 10 5 mBa r. Writing and imaging are done at a set 10 mm working distance, an accelerating voltage of 30 kV, with probe spot size 1 3 depending on the desired pattern feature size. Using a smaller probe spot size (1) is ideal for very small features, but it significantly slows the write time. For some patterns, sweep s of the size 3 beam which is 1 nm in diameter at a 30 kV accelerating voltage, is a fast and easy way to create lines. Because of the varying beam sizes, a center to center distance and line spacing must be set by the user to ensure overlap of the beam spots for large pattern elements. When a pattern was opened in the DesignCad LT 2000 program, an NPGS add on, allowed for the maximum and preferred magnific ation settings on the microscope to be determined by the program itself. The program also ran through an it eration to check for elements t hat would not be recogn ized by NPGS, or may have problems during writing. 2 ) was also specified befor e writing. This is an important parameter, as it controls the energy of the beam striking the sample. The beam dosage is set by the user, but is also calculated from the given beam current which is found after initial focusing. The area dosage is determine d by equation (3 8 ), which correlates to the beam current, i B the exposure time, t e the center to center distance, d c and line spacing, x Likewise, the line dosage is determined by equation (3 9) This parameter is extremely important to the eB writing process. If the dosage is too low, the electrons will not penetrate deep enough into the PMMA, leaving a layer unexposed. When lift off occurs, the entire pattern will be removed, as none of the silver will be in contact with the substrate. If the dose is too high, the pattern will be overexposed and elements may

PAGE 49

49 blend together. Even a slight overdose can cause the pattern to become overexposed at its base layer because of the backscattered electrons (Figure 3 11 ). (3 8 ) (3 9 ) Figure 3 11 The dosage was typically 2 50 C/cm 2 for the PMMA thickness and pattern sizes used in these experiments Ho wever, this value is not accurate as it does not account for the electrons backscattered from the sample during the writing process (Figure 3 9 ) [78] This is known as the proximity effect and can be corrected for manually. [83] The appropriate dosage for a specific resist thickness was determined using a variety of test patterns. The test patterns, which are included in the software, consist of arrays of repeating patterns. The dosages for different elements in the patterns are varied, usually in increasing increments of 5 or 10 C/cm 2 for area dose elements. After

PAGE 50

50 development, the patterns are examined and th e element with the minimum number of artifacts that are acceptable is chosen. This element has had the correct dosage applied. To avoid unnecessary focusing on the sample itself and the subsequent exposing of unwanted areas on its surface a number of prec autions are taken. First, before the sample is placed in the FESEM, it is scored with a three scratch es to point towards the area where the pattern will be written and mark two other areas for focusing This helps prevent unnecessary focusing on the area t hat the pattern is to be written and aids in locating the position of the very small pattern after development. Then, the sample is placed on one of the five positions on the SEM stage. Second, initial focusing of the beam is done on gold nanoparticles. Th e gold nanoparticle sample is a standard that is on a separate sample holder and stage location from the sample to be patterned so they are at slightly different height s. However, correction s for this difference can be made later in the writing process O nce the gold particles are in focus, no further corrections are made to the focusing or stigmations, with the exception of the change in z height. Next, t he beam current is measured using a Faraday cup which has been placed at another stage position The Faraday cup is a 100 m transmission electron microscope ( TEM ) aperture that has been attached to a modified SEM sample holder. Probe current is measured from a picoammeter that has been added to the SEM. The current which is usually on the order of 103 p A, is logged for use in the NPGS program file. Next, the work on the sample can begin. First, the edge of the sample is slowly brought into focus and the score mark is located. Because the sample is not perfectly

PAGE 51

51 flat, a batch file can be run through the NPGS program to estimate the variations in z height across the surface of the sample and account for the variation between the sample height and gold nanoparticle reference height This is done by focusing at the e nds of two of the score marks on sides perpendicular to the score mark where the pattern will be written. The program records data at each focus point, then uses them to calculate the incline of the sample This focusing is carefully done on an artifact at the end of each score mark. This must be done quickly as the PMMA charges very rapidly ; especially at high magnifications. When the focus is properly set, the beam is blanked through Nabity and the SEM controls and beam blanker hat the NPGS program may now take over SEM controls. The run file is then executed. The beam moves to a location, which was set by the user, which is some distance from the focusing area. The actual writing can take from seconds to an hour depending on the size and complexity of the pattern or the chosen spot size Once writing has completed, a custom batch file is employed via the run file and it automatically blank s the beam and return s the stage to the Faraday cup position, to avoid any accidental exposu re of the sample. With this task accomplished, the run file sets the SEM back to manual control. With writing completed, the chamber is vented with dry nitrogen, and the sample is removed and replaced into its sample case and portable dessicated vacuum bo x (DESI VAC) to protect from moisture. This step is important for both the PMMA and silver in the sample as both are extremely sensitive to humidity Step 3: Development Following patterning, the sample is developed in a 3:1 mixture of isopropyl alcohol ( IPA) in methyl isobutyl ketone (MIBK), IPA, and distilled water, sequentially for 90 seconds each. To ensure agitation of the solution, the sample in solution was either

PAGE 52

52 ultrasonicated or manual waved in the solution for the 90 second duration. Lastly, a 2 minute descum is performed in ozone using an Ultraviolet ozone cleaning system (UVOCS) model T10X10/OES to remove residual resist in the deep trenches. This is an important step for patterns containing narrow elements as the solution may have not made it to the bottom of the trench. If too much silver is deposited, to the point where the silver in the trench makes contact with the silver on top of the PMMA layer, then the entire pattern will be lifted off. Step 4: Electron Beam Deposition As previously stated, the silver antennas are fabricated on top of silver thin films, which are both deposited using electron beam (e beam) deposition. This physical vapor deposition (PVD) technique is favorable for both initial deposition and lift off, as it is quick, is capable of fine resolution on nano scale features, and does not require substrate heating. The non substrate heating requirement is vital bec ause when lift off is performed, the PMMA cannot be heated or it will melt and ruin the pattern The e beam evaporator consists of one large chamber with no load lock (Figure 3 12 ) A carousel has slots for four sources with are typically hold crucibles for s ource materials, but occasionally a source material may be placed directly in the carousel The specific or lack thereof, with the given crucible material. Germanium and silver were both housed in alumina crucibles for this project. The system is pumped down first via a dry compre ssion roughing pump ( Leybold DryVac 50P ). Once the system reaches 500 mTorr, the high vacuum valve is opened to a cryopump (CTI Cryogenics Cryo Torr 8). The chamber is pumped down to approximately 8 x 10 6 T orr for metal deposition. At this time, the high voltage is turned on for the tungsten filament and the current is

PAGE 53

53 manually increased at a slow rate to avoid thermal shock of the filament The electron beam then begins to heat the metal source An INFICON XTC/3 thin film deposition controller connected t o a quartz crystal monitor (QCM) is used to measure the deposition speed. The deposition rate is determined from a combination of the QCM readout, and the manually input density and Z ratio for the given material. While the deposition rate is being stabili zed, a metal shield sits in place between the source and sample. Once the desired deposition rate has stabilized, the thin film deposition he program allows the shutter to be opened until the required film thickness has been reached. Once the desired thickness has been achieved, it is shuttered closed and returned to manual control. After deposition is completed, the high voltage is turned off and the filament is allowed to cool for some time to prevent thermal shock The n the system is vented with dry nitrogen the sample and sources are removed, and the system is pumped back down to its idle state Figure 3 12 A diagram of the e beam evaporator system used for these experiments

PAGE 54

54 Step 5: Lift off Next, lift off is performed by e beam. In this step, 100 500 nm of silver, depending on the desired height of the antennae, is deposited into the developed areas of the sample. Remaining silver and PMMA are removed by placing the sample in acetone. Due to the small feature size, the beaker containing the sample and acetone were sonicated to ensure that excess silver and PMMA were completely removed from the sample. An example of a final product of the eB process is shown in Figure 3 13 A variety of artifacts are visible in the SEM micrograph. For example, the bright white areas (circled in red) are residual PMMA. The malformed ellipse shapes are likely caused by the stigmation of the electron beam. Dirt and excess silver are also present on the surface of this particular sam ple. Figure 3 13 An example of a silver elliptical antenna array on a silver substrate fabricated for these experiments as per the procedure in Chapter 3. Some minor artifacts are visible on the surface as well as residual PMMA from the lift off proces s (red circled areas ).

PAGE 55

55 Optical Absorption Testing A custom built optical set up was designed to test whether the antennas were absorbing visible light. The arrays were fabricated according to the procedure described in Chapter 3 to create silver antennas on a silver substrate ( e.g. Figure 3 13 ). The arrays used for optical te sting were of same size antenna elements and were not fully optimized for resonance via calculations prior to fabrication and testing Rather, these experiments were performed to confi rm the plasmonic absorption on the antenna arrays and to confirm that any effects observed were indeed from the structures and not from the bare silver surface. As previously mentioned, a flat surface will have little to no response from polarized light. However, when light is polarized over the antenna surface, we will see a visible response in the absorption spectra. Therefore, we can confirm the plasmonic behavior of the system through the absorption spectra. For these optical experiments, incoming ligh t was generated from a Xe light source ( ORIEL 66902) passed through a monochrometer (ORIEL 77700) focused through a series of lenses, modulated by a mechanical chopper which was correlated with a lock in detector, and finally through a motorized polarize r, before it hit the sample (Figure 3 1 4 ). The modulated reflect ed light was detected with a photomultiplier tube (PMT) (ORIEL 77345) which was attached to the Merlin lock in detector (ORIEL Instruments) (Figure 3 15 ) Responses were measured using a custom software (Tra c q32) Polarization of the light beam was changed from 0 90 in 10 steps using a linear polarizer controlled by a Newport ESP300 3 axis motion controller. Measurements were taken at each polarization step both on the bare surface and t he sample area. Finally, the data was extracted into Excel for further analysis.

PAGE 56

56 Figure 3 1 4 The custom optical absorption testing set up used for the reflection and refraction experiments. (Photo courtesy of author) Figure 3 15 An overhead view of t he light path taken through the custom optical absorption set up. (Photo courtesy of author)

PAGE 57

57 Photomixing Custom Set Up Testing of the photomixing device required a custom built set up which was arranged in a number of ways (Figure 3 16 ). In all cases, th e f irst the two incoming lasers were mounted in thermoelectric coolers (TECs) (Thor Labs) and connected to a laser driver /controller and power source. One laser diode was held constant at the near IR 780 nm. The second laser diode was tuned to 850 nm which according to equation (2 1) should produce frequencies in the far IR or mid Terahertz (~30 THz) (Figure 3 17 ) Next, b ecause the sample area is difficult to see with the naked eye and the lasers are in the near IR region, we used an optical microsc ope to locate the antenna region and an IR camera attachment to ensure that the lasers were both focused onto the sample area. To aim the lasers onto the sample, two small flat mirrors were used Figure 3 1 6 A photograph of the first custom photomixing set up that was designed and used for this dissertation. (Photo courtesy of author)

PAGE 58

58 Figure 3 17 A close up view of the two laser diodes in TECs for the first photomixing set up The chopper is in front of one diode. (Photo courtesy of author) A few dete ctors were selected for these experiments to ensure proper detection. First, a liquid N2 cooled duel photodiode InSb and HgCdTe far IR detector (Kolmar Technologies) was used exclusively for THz detection. The sensitivities of this detector in the far IR regime are shown in a detectivity vs. wavelength plot from the manufacturer in Figure 3 18 The HgCdTe range was more commonly used in these experiments. Figure 3 1 8 The detectivities of the far IR detector at various wavelengths. [84]

PAGE 59

59 The change between the two far IR detector s is easily made by switching an input coaxial cable. The far IR detector was connected to a Merlin digital lock in radiometry system (ORIEL Instruments). [87] I n order for the lock in detector to confirm signal, eliminate background noise, and improve the overall signal to noise of the read out, a chopper was required The frequency of the chopper (ORIEL 75152) is set using the Merlin lock in and the chop is selected at a low frequency to eliminate background noise, but not THz waves. A number of chopping frequencies and detection rates were experimented with, b ut a 9 Hz chop and a 3 sec ond detection rate were most commonly used for the photomixing experiments. The chopper was placed in front of one of the laser diodes. After initial tests, a second detector with a larger sensitivity range was added to the optical set up i n an altered design (Figure 3 19 ) A lithium tantalate pyroelectric detector with a thallium bromoiodide window (ORIEL 70124) was used for its greater signal strength and larger range, which spanned 0.6 12 m (Figure 3 20 ) This detector was also connected to the Merlin lock in system, and was able to receive signal from the individual lasers. To eliminate this laser signal and ensure the observed effects were in the THz regime the black polyethylene and an optical bandpass filter with a 980 nm cut off were both used. The detector was u sed in conjunction with the IR spectrometer, which was able to detect wavelengths up to 1800 nm, to analyze the scattering from the lasers and the photomixed output. The entire set up was covered with a black velvet cloth to reduce ambient light and scattering effects.

PAGE 60

60 Figure 3 19 The altered custom set up for photomixing utilizing the pyroelectric detector. (Photo courtesy of author) Figure 3 20 The relative responsivity vs. wavelength plot for the pyroelectric detector (70124) from the manufacturer. [87]

PAGE 61

61 Lastly, as the pyroelectric detector exhibited a very strong noise level when filtered, one more detector was added to the system. A high performance thermoelectric cooled lead selenide photoconductor detector (Electro optical Systems PBSE 050 TE2 H) was employed. The range of this detector is limited to 1 but it was also filtered with black polyethylene to reduce near IR radiation. Procedure The sample was placed on the rotating stage of the optical microscope. The arra y area to be tested was located using the optical scope. The score mark made to identify the area in the electron beam lithography procedure is helpful in this step as it Next, the lasers were aligned using the mirro rs. A third mirror was placed near the sample to ensure the lasers general placement. T he trajectory of the lasers was also verified to be over the s ample area using a n IR spectrometer (Ocean Optics, Inc. ) attached to a fiber optic cable. The spectrometers were also useful in determining the tuned output via Overture a spectroscopy software designe d specifically for Ocean Optics instruments. Another check to verify the laser trajectory was performed using an IR camera (COHU 4915 2000) attached to the optic al microscope was used. The video output was displayed using computer software. The software clearly showed when both lasers were in the path of the camera. The IR camera showed the lasers exciting the resonant structures when they were properly aligned wi th the sample. Once alignment was finalized the entire optical set up was covered with black velvet cloth to keep out ambient light. As in the optical absorption testing, measurements were taken both on the array area and the bare substrate to ensure tha t the measured effects were indeed from the antenna structures. In both cases, one laser was kept at a constant wavelength of 780

PAGE 62

6 2 nm. The wavelength of the second laser was slowly increased or decreased and readouts from the detector were acquired from the Merlin lock in The read outs from the far IR detector s on the lock in are only somewhat quantitative. A voltage or power value is displayed; h owever, a s shown in Figures 3 18 and 3 20 the detectivity is fairly consistent over the IR range. Any responses elicited by the detector were logged. The sample was also rotat ed, as it should be polarization dependent, and a variation in intensity should be observed at various angles as the lasers themselves are highly linearly polarized.

PAGE 63

63 CHAPTER 4 CHARACTERIZATION AND TESTING The antenna arrays for the plasmonic plasmonic photomixer device were simulated using COMSOL Multiphysics and MEEP software using the procedures described in Chapter 3 These simulations were carried out before sample fabricati on to choose patterns that were promising for the device. They were also carried out after fabrication and testing of devices to further explain and confirm observed results. Samples of varying array patterns were fabricated by eB lithography, electron be am deposition, and lift off. Characterization of these samples will be discussed in this chapter. Further, the samples were optically tested on a custom built optical bench to confirm absorption on the antennas. Photomixing of the devices was performed on the custom built set up described in Chapter 3 and the results of these tests will be reported in this Chapter. Fabrication There were some challenges with the electron beam fabrication process but ultimately changes to the process were made to account fo r them. Firstly, very long and skinny structures, as well as very small structures were initially difficult to create, as they off process. To help prevent this, the descum procedure was added to remove the developed re sist from deep trenches. However, the descum process tended to harden the remaining PMMA, causing its removal in a beaker of acetone to become difficult. A few solutions were tested, including applying the acetone to the sample while it was spun on the spi ncoater. The best solution proved to be p rolonged times under agitation in the ultrasonicator, as well as some manual agitation.

PAGE 64

64 Another fabrication issue was the aforementioned crosslinking of PMMA over time. PMMA has a relatively long shelf life, but ten ds to thicken and become more difficult to spincoat. Higher spin speeds were required, and thickness measurements were periodically taken using a Filmetrics thickness measurement system. One final issue was the tendency for elements spaced closely together to develop into one another. This is undesirable for the periodic array, as small end to end spacings are desired to optimize coupling between antenna structures. To account for this, larger spacings were required, or a smaller beam spot size was chosen. After adjusting the procedure for these issues, the fabrication process was able to consistently produce elements as small as 400 nm x 150 nm and as large as 2 m by 330 nm. The arrays were drawn in DesignCad LT 2000 100 m 2 areas that could be stepped to form a total pattern area as large as 500 m 2 Spacings between patterns were generally 5 10 m. Larger overall pattern areas are certainly achievable, but were not tested for this dissertation. Simulation Results and Discussion COMSOL Periodic antenna arrays were first simulated using COMSOL Multiphysics. Initially, a 2D periodic antenna array was simulated using TE and TM modes Visible incoming light was varied both in wavelength and incoming angle. Using COMSOL Multiphysics, we examined 2D periodic a ntenna arrays (Figure 4 1 ) in the xz plane An extended view of the simulation time slice image (Figure 4 2 ) visually shows the plasmonic coupling from one antenna to the next. This coupling is essential to creating resonance and a large interaction area f or the plasmonic photomixer. The plasmonic

PAGE 65

65 coupling and resonance is also responsible for the light absorption observed from these structures. The absorbed light is converted into plasmon energy. A B Figure 4 1 The 2D periodic area of the arrays simulated using COMSOL. A) A schematic of the 2D simulation B) The 2D simulation excited by 444 nm light. Figure 4 2 A time slice image of t hree 500 nm diameter periodic antennas excited by 444 nm wavelength light and visually showing efficient plasmonic couplin g from one to the next.

PAGE 66

66 We varied the wavelength of the incoming light over the visible spectrum (to include the plasmon frequency of silver), the incident angle of light from 0 90, the feature size from 200 800 nm, and the feature spacing from 50 400 nm. We also obtained quantitative reflectance and refraction measurements from these simulations and compared them to our experimental reflectance data. The details of this comparison will be discussed in the optical absorption testing section of this chapter First, the length of the antenna was varied, while keeping the height, incoming wavelength, and incoming light angle constant. The antenna height is 150 nm, and the 450 nm light is entering at a 53 angle relative to the normal of the antenna, as in Figu re 4 6. The spacing between antennas was increased with decreasing antenna length, however. In all four simulations, with dimensions varying from 200 500 nm, a strong electromagnetic field is visually observed on the corners of the antennas (Figure 4 3 ). P shaped plasmonic structures are often seen in photomixers and in other research [88 90] In addi tion, most notably in Figure 4 3 A, a dispersion of the light in air is observed, which may either be caused by reflection from the surface of the structure or a plasmonic emission. Further quantitative absorption simulations would be required to corroborate either hypothesis. In any case, a stronger overall electric field on the surface is shown for the 200 nm length structures a nd the 500 nm length structures, and the electric field in air shows a stro ng interference from the coupled structure in Figure 4 3A as well.

PAGE 67

67 A B C D Figure 4 3 Varying the antenna length while holding incoming radiation constant. Wavelength of 450 nm is incoming at a 53 angle. Antenna dimensions are 500 nm in length and 15 0 nm in height. A) 200 nm length. B) 300 length. C) 400 nm length. D) 500 nm length. Next, as the antennas showed a strong surface interaction at 450 nm as well, this parameter was held constant and the spacing distance between individual 450 nm structures was varied. As in the prior simulations the incoming light was held constant at 450 nm and at an incoming angle 53 relative to the normal. Spacings, or the end to end distances between individual antennas, were va ried from 200 500 nm (Figure 4 4 ). Small er end to end spacings were found to show a much stronger surface interaction, with a 300 nm spacing showing the strongest sur face interactions (Figure 4 4B), which was expected.

PAGE 68

68 A B C D Figure 4 4 Varying the antenna spacing while keeping antenna size constant. Wavelength of 450 nm is incoming at a 53 angle. Antenna dimensions are 500 nm in length and 150 nm in height. A) 200 nm spacing. B) 300 nm spacing. C) 400 nm spacing. D) 500 nm spacing. The same experiment was repeated for light of 650 nm at a 5 3 angle relative to the normal. Again, antenna lengths were 450 nm and end to end spacings were va ried from 200 500 nm (Figure 4 5 ). Strong corner excitation is again visible in all simulation time slice images, however, none of the surface excitations ar e as strong as those excited by 450 nm light. This is likely due to the wavelength size matching in the latter case. Therefore in the design of the plasmonic photomixer, the antenna sizes should be close in size to the incoming laser wavelengths. Some chan ges must be made however to manipulate and resonate the output photomixed peak. This will be discussed in greater detail in the coming sections.

PAGE 69

69 A B C D Figure 4 5 Varying the antenna spacing while keeping antenna size constant. Wavelength of 675 nm is incoming at a 53 angle. Antenna dimensions are 500 nm in length and 150 nm in height. A) 200 nm spacing. B) 300 nm spacing. C) 400 nm spacing. D) 500 nm spacing. Using the results from the qualitative simulations, the antennas with parameters showi ng the strongest surface excitation were tested quantitatively for reflection and refraction over the visible spectrum. These experiments parallel the real life experiments done on a custom built optical bench with polarized light. First, because of its ea rlier strong plasmonic interactions, an antenna of length 500 nm, end to end spacing 300 nm was examined over the visible spect rum from 400 800 nm (Figure 4 6 ). The incoming angle of light was held at 53 relative to the normal in accordance with prior sim ulations. Strong dips in the reflectance spectra, where absorption is occurring, are observed close to 450 nm and 670 nm. The strong absorption peak at 450 nm corresponds to the plasmon frequency of silver, which is in resonance when the size of the antenn a sizes

PAGE 70

70 matches this wavelength. The strong absorption is also an indication of the transfer of light energy into plasmonic energy. A B Figure 4 6 Simulated excitation from a 500 nm antenna. A) A 500 nm antenna in 2D 445 nm light at 53 relative to the normal B) Reflectance and refractance vs. wavelength for this antenna These reflection and refraction experiments were repeated for antennas of increasing size. When antenna size was increased, a trend was observed in absorption intensity (Figure 4 7) The 500 nm antenna absorbed much stronger tha n the larger antenna which is shown in the m =1 reflected mode of Figure 4 7 B In addition, a slight shift in the absorption peak was observed with varying antenna size It sho uld be noted that a ll antennas have the same period spacing ratio with respect to antenn a size. All three simulated reflectance vs. wavelength plots for the antenna arrays showed the strongest absorption at 450 nm, with relative absorption increasing with increasing antenna size.

PAGE 71

71 A B C Figure 4 7 Reflectance vs. wavelength simulation for three arrays A) 350 nm antenna array and B) 500 nm antenna array C) 750 nm antenna array

PAGE 72

72 Overall, the 2D simulations confirmed plasmonic coupling in the silver antenna and showed a strong resonance when the size of the antenna correlated to t he plasmon frequency of silver near 450 nm. Changing the diameter of the antenna reduced the resonance in the system and resulted in a shift in the absorption peak. Th ese results serve as an excellent model for plasmonic absorption and coupling over the visible IR frequencies and will be compared to experimental data later in this chapter. In addition, q ualitative r esults from work in 3D simulations in COMSOL of a silve r plasmonic sphere in air are shown in Appendix A. MEEP Having confirmed the plasmonic absorption and coupling on the 2D periodic arrays in the xz plane using COMSOL, the a ntenna arrays were simulated in 2D in the xy plane using MEEP software. MEEP was use d to demonstrate conditions under which antennas not only absorb incoming light, but also reemit it. Arrays of conducting ellipse shaped antennas were created with an extended space to show the non conducting area beyond the array (Figure 4 1 ). Transmission monitoring sites are placed near the source and at a point far from the array so that the fraction of power that is transmitted or reflected across the array is monitored. It should be noted that t he color scale in the time slice images pre sented in this section represents the electric field intensity with red being positive and blue representing negative values. These time slice images were also compiled into gif videos to get a better sense of the surface interactions taking place. For the initial simulations, we wanted to expand upon the COMSOL simulations by examining the plasmonic antenna array in the xy plane, as opposed to the xz plane. For initial simulations, a basic array of identical ellipse shaped antenna was chosen for

PAGE 73

73 the factor s discussed in Chapter 3. Therefore, a single Gaussian source placed at the far left of the x axis of the array field was used to excite an array of 500 nm length identically sized elliptical antenna. The frequency of this source was varied and its effects on the antenna structures were qualitatively observed in a series of time slice images and corresponding video. In order to ensure the observation of ideal plasmonic behavior in these simulations, t he antennas were designated as perfect conductors and the area around them was designated as air. A number of interesting observations were made from these initial tests. First, w hen the wavelength of the source was much greater than the length of the antennas, lsed over. There was no response from the antennas in this case. However, i f the incoming light was chosen at a wavelength that was close in size or smaller than the length of the antennas, then the light would couple to the antennas. Then, the antennas wo uld couple to one another and the intensity of the electric field on their surface and in the spaces between the antennas would remain long after the source light had been turned off (Figure 4 9) This is most clearly shown in the videos compiled from this simulation, however, Figure 4 9D clearly shows the electric field intensity on the antennas long after the source has been switched off. Lastly, w hen the system was in resonance, the coupling was the strongest and most efficient, and the electric field on the surface of the antennas was the most intense. Plasmonic resonance of the photomixed peak is most desirable for the plasmonic photomixer device in order to obtain a strongly coupled output frequency

PAGE 74

74 A B C D Figure 4 8 Time slice images of an array of 500 nm antennas excited by light of a wavelength of 1 m. A) 80 s showing strong excitation from the source B) 122 s C) 146 s D) 171 s, long after the source has been turned off, there is still strong coupling and electromagnetic fields present on the antennas.

PAGE 75

75 With the plasmonic coupling of the antenna structures confirmed, a second source was added just below the first and the chal lenge of simulating the photomixing these two frequencies began. After first running these simulations on the antenna array shown in Figure 4 9, i t became clear that the system would need some irregularity to induce the mixing. However, the system would still require enough regularity to be come resonant. Also, because the photomixing proc ess is a nonlinear optical one, the perfect conductor condition for the antennas would have to be changed. A high nonlinear susceptibility value, (3) was applied. However, this value was not optimized, which may have been a cause for a decreased intensit y observed in the photomixed output. In the first attempts at inducing irregularity into the system smaller antennas, which were about the length of the incoming wavelengths were placed near the sources and longer antenna were added at the same spacing (F igure 4 9 ). Source frequencies were chosen at 780 nm, to mimic the actual experiments, and 62 8 nm in hopes of outputting a far IR/THz mixing peak at 3880 nm or 77 THz Long range coupling was observed in this system, but it was not in resonance. A detecto r port was place in the open space in the center of the upper right hand quadrant to calculate the transmission through the system. As expected, the strongest peaks occurred at the source wavelengths, 780 nm and 638 nm (Figure 4 9 A) Closer inspection of t he predicted location of the mixing peak at 3880 nm which was calculated from e quation ( 2 1) revealed a very low intensity peak (Figure 4 9 B). Again, t he extremely low intensity of the mixing peak is likely attributed to the system not being in resonance and a non optimized (3)

PAGE 76

76 A B C Figure 4 9 Three time slices of a simulated antenna array showing plasmonic excitation, but not plasmon resonance. A) 101 s B) 144 s C) 276 s (sources turned off).

PAGE 77

77 A B Figure 4 10 The transmission vs. wavelength for the antenna structures in Figure 4 9 A) The two strong peaks are from the sources. B) The very weak intensity mixing peak from this simulation. The peak is at the predicted position of 3880 nm. To create a system that was in resonance and increase the intensity of the mixing peak, the longer antennas were moved to a horizontal spacing that matched the wavelength of the desired photomixing peak, 3880 nm (77 THz) The source wavelengths remained at 7 80 nm and 635 nm for comparison. Visually, the intensity of the electric field on the longer antenna structures was much stronger than on the smaller antennas which is hypothesized to have been a result of the mixing (Figure 4 11 ). The transmission curves for this simulation showed the strong intensity at the expected peaks at the source wavelengths (Figure 4 12 A), but showed a third peak at a surprising location which likely corresponds to a higher order mode of the mixing peak (F igure 4 12 B). This peak is of much stronger intensity than the mixing peak shown in Figure 4 10 B, but i s not at its predicted locat ion of 3880 nm calculated from e quation ( 2 1 ). This shift may also be caused by a variation of refractive index in the materi al. The

PAGE 78

78 plasmon frequency is strongly dependent on the refractive index, so this change may shift the plasmon frequency. In addition, the higher order mode of the mixing peak may also be more favorable to the plasmonic structure resonances. Figure 4 11 A simulation with the longer antennas spaced at the mixing wavelength of 3880 nm at a time of 157 s Stronger modes are visible on th e longer structures than the short ones A B Figure 4 12 The transmission vs. wavelength for the antenna structures in F igure 4 11. A) Transmission intensity vs. wavelength for the simulation in Figure 4 11 showing the high intensity of the source peaks. B) The third peak in this simulation.

PAGE 79

79 The next step in inducing resonance in the system was to increase the amplitude of the incoming light sources The real life laser counterparts to these simulated sources are polarized. This resulted in an increase in intensity of the resonant peak, but the peak was still not at the predicted position for non p lasmonic photomixing. Therefore, the position of the long antenna rows was shifted to shift the peak position. As a strong resonance a t approximately 1 m was repeatedly observed in the simulations, the position s of the longer antenna were varied at differ ent intervals (e.g. Fig ure 3 6 E) to see if this would shift the resonance peak closer to the predicted mixing position (Figure 4 13) This shift was not observed, and the mixing peak remained at approximately 1 m (Figure 4 14A). H owever, the intensity of the 1 m peak did largely increase when the longer antenna were placed at intervals of every four antenna (Figure 4 14B) Figure 4 13 A 99 s time slice image of an array with longer antenna spaced every four short antenna. Incoming wavelengths are 780 nm and 635 nm.

PAGE 80

80 A B Figure 4 14 The transmission spec tra for the array in Figure 4 13 A) Spectra showing two strong peaks for source emission with the third peak just barely visible near 1000 nm B) The third mixing peak around 1 m that was present in this simulation. The length of the long antenna from Figure 4 11 was increased and d ecreased in small increments to see if a trend would arise in the i ntensity of the photomixed peak (Figure 4 15 ) Very small changes in this length, ~0. 01 m, had the ability to change the intensity of the photomixed peak. In that simulation set, a strong photomixed peak was shown near 6 m when the long antenna was 2.1 m in length. However, when the length of these antenna were changed by even 0.01 m, the intensity of this peak decreased, despite the overall increase in intensity for the simulation. While interesting, this is not ideal for the plasmonic photomixer, because this precision cannot be kept controlled using the eB process. An important note: the 2D simulations do not account for a variety of factors affecting the device performance including incoming angle of light in the z direction, feature height, (3) value of silver structures and substrate, or excitation light angles relative to the sam ple and one another. All of these factors will be

PAGE 81

81 present in the experimental testing of the device and should be considered when comparing simulated data to actual data. Figure 4 15. Intensity vs. w avelength for a set of simulations in which the length of the large antenna was varied by hundredth of a micron (antenna lengths given in legend) The 2.1 m length antennas show the strongest intensity at a wavelength of 6 m. Optical Absorption Testing Optical absorption testing was measured using the custom built optical table and procedures described in Chapter 3. Results from these experiments showed reflectance spectra with two to three strong absorption dips. For an array of antennas of 450 nm in length shown in Figure 4 16 the strongest absorption dip was close to 470 nm (Figure 4 17 A). This may correlate to the antenna length. When this plot is compared to the

PAGE 82

82 simulated refl ectance from COMSOL (Figure 4 17 B), it can be seen that the plots follow a very similar trend, which corroborates the data sets. Figure 4 16. The 500 nm length antennas tested for this experiment. The imperfect shape and size are artifacts of the beam stigmation and focusing. A Experimental Simulated B Figure 4 17 Experimental reflectance vs. simulated reflectance measurements for a 500 nm antenna array taken at polarizations varying 0 90 A) Experimental reflectance for varying polarization angles B) Simulated reflectance and refractance for a 500 nm antenna We observed a definite trend in the variation of the reflectance with respect to the polarization of light on the sample, whereas this trend was not observed on the bare

PAGE 83

83 substrate. To further illustrate this, we integrated the reflectance over the major ab sorption peaks and plotted it with respect to polarization angle (Figure 4 18 ). The same trend was observed for each peak, which verified the polarization dependent light absorption on the structures. Therefore, absorption from the plasmonic array was con firmed. Figure 4 18 Integrated reflectance vs. polarization angle for two absorption dips from the reflectance vs. wavelength graph in Figure 4 17 A The fabrication of an array of 780 nm length antennas posed some problems, as it was difficult to perform lift off on very long and narrow structures. This resulted in some of the antennas being lifted off w ith the excess PMMA (Figure 4 19 ). However, optical absorption tests were still performed on this sample. Results from an array of 780 nm length antennas do not show strong absorption for a number of reasons (Figure 4 20 A). First, the predicted absorption from these larger antenna sizes is much lower than f or smaller antennas (Figure 4 20 B). The aforementioned defects on the sample would also reduce the cou pling between antennas and therefore decrease the overall

PAGE 84

84 light absorption. Previous studies on gold nanoantennas have shown a significant reduction in plasmonic coupling when symmetries of antennas were reduced. [91 ] Figure 4 19 An SEM micrograph of the 780 nm antenna array. Many antennas were lifted off in the fabrication process. A B Figure 4 20 Experimental reflectance vs. simulated reflectance for a 780 nm antenna array. A) Experimenta l reflectance for varying polarization angles B) Simulated reflectance and refractance for a 780 nm antenna The integrated reflectance confirmed that there was a polarization dependant trend in absorption on the antenna structures, with a very strong absorption dip at a

PAGE 85

85 polarization angle of 60 (Figure 4 21). Therefore, the antennas were a prominent enough futu re to contribute to the absorption, and this absorption was most likely caused by the light coupling to the surface plasmons. Figure 4 21 Integrated reflectance vs. polarization angle for two absorption dips from the reflectance vs. wavelength graph in Figure 4 21 A Overall, with the comparison of the COMSOL simulations to the experimental absorption data, we confirm that the antennas are absorbing light and coupling between one another. Intensity was much stronger when smaller antennas were used in both methods Thereby, a t smaller sizes, we can increase the plasmonic interactions. If the arrays were in resonance, the absorption should be much greater than those for the samples tested in this section. However, the observed absorption still points towards the coupled plasmonic structures as a suitable platform for photomixing.

PAGE 86

86 Photomixing Testing for photomixing utilized the custom set up and procedures described in Chapter 3. The expected output frequency from the mixed lasers was calculated using equation ( 2 1 ) From Figure 3 15, the range of the InSb far IR detector is approximately 1 6 m, and the range of the HgCdTe detector is approximately 6 13 m. The range of the pyroelectric detector is 0.6 20 m. This encompasses the majority of wavelengths predic ted to be output by the tuned 780 nm and 850 nm lasers (Table 4 1) In addition, it was important to analyze over a range of frequencies, b ecause the MEEP simulations showed that the strongest photomixing peak was not a lways at its predicted location. This is due to the changing refractive index in the system, and henceforth changing plasmon frequency and higher order modes of the photomixed peak that may find a stronger resonance in the system. Table 4 1. Expected output wavelength and frequency by the pl asmonic photomixer for varied tuned wavelengths from 780 nm and 850 nm lasers. Wavelength 1 Wavelength 2 Expected Wavelength Expected Frequency 780 nm 830 nm 13.0 m 23 THz 780 nm 835 nm 11.8 m 25.3 THz 780 nm 840 nm 11.0 m 27.4 THz 780 nm 845 nm 10.1 m 29.6 THz 780 nm 850 nm 9.47 m 31.7 THz 780 nm 855 nm 8.89 m 33.7 THz 780 nm 860 nm 8.39 m 35.8 THz As a control, like in the optical absorption testing, the bare silver substrate was tested at a variety of mixed wavelengths prior to testing on the antenna structures. This was to verify that the observed phenomena were indeed from the coupled antenna structures. In addition, the effects of excitement by each laser on its own were tested to ensure the observed effects were a product of photomixing. As only one of the two laser

PAGE 87

87 diodes was chopped (tuned to 780 nm) the non chopped laser (tuned to 850 nm) did not elicit any response from the detectors on its own. The first detector utilized for these experiments was the HgCdTe far IR detector. However, the bulkiness of the HgCdTe detector made it difficult to place near the sample, and the incorporated dewar prevented the detector from being angled towards the sample in any way. In addition, the HgCdTe detector had a very strong noise level, which made it impossible to distinguish any signal from the noise. Therefore, it was replaced with the pyroelectric detector. The first sample that was tested was one of alternating antennas of 780 nm and 850 nm in length in an effort to phase match to the incoming lasers (Figure 4 22). The overall pattern area for this and all samples tested in this section was 300 m 2 The antenna elements were 330 n m in width and 50 nm in height polarization was in line with the sample, the pattern area was illuminated in the IR camera. As the sample was rotated, the surface excitation decreased. When the sample was rotated 90 the polarization did not match the sample well and the pattern was barely visible in the IR camera. This visual ef fect was helpful in aligning the sample to the chopped laser. The second non chopped laser was not polarization matched to the sample. When the 780 nm laser was switched on, the voltage readout from 1 to 30 V. On the bare substrate, the voltage rose sligh tly when turning on the chopped laser. To eliminate scattering effects, the black polyethylene and 980 nm bandpass filters were employed. However, when these filters were in place, on the strong noise level remained and it became difficult to register any signal from this noise. Importantly, the optical density of these filters should be such that a signal should be observed when the

PAGE 88

88 chopped laser is turned on. However, this rise was not seen, and the noise level remained extremely high. Therefore, it would be far less likely that we would be able to register a weak photomixed peak even if it is beyond the cut off range of the filters. Figure 4 22 A silver on silver antenna array sample with alternating antenna sizes of 780 nm and 850 nm to phase match th e lasers. Because the MEEP simulations showed the strongest photomixing from an antenna array with long antennas placed every four antenna, such a design was fabricated for photomixing testing. An array of 780 nm and 2 m antennas were fabricated for this test (Figure 4 2 3 ). The width of the antennas was 330 nm and they were 50 nm in height. The 2 m antennas propagate in the THz range, and were chosen to help induce resonance and promote the coupled long wave output When the 780 nm laser was switched on, the voltage rose from 1 to 30 V. When the 850 nm laser was added, the signal consistently increased to 40 V. Because, this laser was not chopped, it did not elicit a response from the detector on its own, but the measurement may indeed be from the mixed signal of the two lasers. Recall that the 850 nm laser is not polarized in line with the array and is not chopped, so it is not expected to cause the voltage to increase on its own. When the sample was rotated, there was a significant

PAGE 89

89 decrease in signal a nd photomixing was not observed, which successfully showed that the effect was polarization dependent, and further suggest that the voltage shown was the effect of the photomixing of the lasers rather than scattering Figure 4 23. An array of 780 nm antenna with 2 m antenna placed at intervals of every four antenna. However, when the filters were employed to assess if this was truly a photomixing peak as opposed to scattering the problems with detector signal to nois e were again presented. From Table 4 1, the predicted position of this peak is 9.47 m, which is in the range of the pyroelectric detector. As demonstrated in the MEEP simulations, a higher order mode of this peak may find resonance in the system as well. However, we also observed that the strength of signal of the photomixed peak in MEEP is only 1% the intensity of the incoming lasers. So if there is difficulty registering the laser signal in the detector, it could be possible that the photomixed peak beco me lost in

PAGE 90

90 the noise as well. As an additional check, the IR spectrometer and lead selenide photoconductor detector were used in conjunction with the pyroelectric detector. The IR how a mixed peak. The photoconductor peak. Therefore, further more sensitive measurements will be required to fully corroborate this data. After the findings on the prior sample of an array of short anten nas periodically interspaced with longer structures on e additional pattern design was tested. An array of wider ellipse shaped antenna spaced further apart with longer antenna spaced at intervals of every four antenna were also examined (Figure 4 24 ). Thi s pattern was specifically designed to optimize the eB fabrication process, as the larger width and spacing make the patterns much more repeatable and are easier to perform lift off on. All antennas were 500 nm in width and 50 nm in height. These smaller e lements were 1 m in length and the larger elements were 2 m in length. The 780 nm laser coupled well to the structures, producing a 35 V peak from the pyroelectric detector when the sample was perpendicular to the laser. Adding the second 850 nm laser d id not elicit an effect from th e detector. Rotating the sample showed a decreased intensity in the detected peak, confirming the coupling to the structures and the sample was shown to couple strongly to the laser in the IR camera P otential p hotomixing was not observed on this sample and the signal to noise issues from the filtered detectors were an additional issue This could be caused by the decreased sizes of the long antennas or the very large spacings which could have greatly reduced coupling. In add ition, the

PAGE 91

91 spacings may not have been small enough to force the two lasers to colineate and photomix. Figure 4 24 An array of alternating 1 m and 2 m antennas at a wider spacing than the previous samples. Overall, of the three samples tested, only repeatable potential photomixing was shown on the second sample with alternating 780 nm and 2 m antennas. The non filtered signal was shown to be polarization dependent, suggesting that it was an effect on the ar rays. This pattern design was modeled after the MEEP simulation that produced the highest intensity mixing peak (Figures 4 13 and 4 14). However, filtering the detectors resulted in an extremely strong noise making signals virtually impossible to detect. S till, t he possible correlation between the simulated and experimental results further suggests that the observed intensity increase from the detector is indeed from the photomixing peak. Signal from this structure could further be improved by increasing th e overall pattern size, increasing the thickness of the structures, and refining the pattern so that it is in resonance. Most importantly, more sensitive measurements with

PAGE 92

92 bolometer detectors, for example, should be taken to verify the potential photomixed output from these plasmonic structures.

PAGE 93

93 CHAPTER 5 CONCLUSIONS F abrication A number of conclusions were drawn from this work, but we will first focus on the results of the fabrication process. Device fabrication was successfully performed using eB lithogra phy and lift off. The center to center dimensions of the arrays was varied from 50 200 nm along the major and minor axis. The feature size was varied from to 150 nm to 2 m by 50 500 nm along the major and minor axis, respectively. Overall pattern sizes as large as 500 m 2 were created, and could effectively be made much larger. Artifacts remained on the surface of the samples after processing and fabrication was difficult for arrays that were both long and narrow, as they had the tendency to be lifted off. Once these size limits were established and a consistent PMMA thickness and electron dosage were obtained, however, the eB process became very reliable for device fabrication. Another challenge in the fa brication process was the cross linking of the PMMA r esist over time. The cross linking caused the resist to thicken, which resulted in the need for higher spin speeds upon application. In the resist removal process, the use of acetone became less effe ctive when the resist was cross linked which seemed to be accelerated by the descum process A great deal of agitation via prolonged times in ultrasonication were required to completely remove the resist in the lift off process. Once these processing issues were addressed, patterns were able to be consistently p roduced.

PAGE 94

94 Simulation The plasmonic photomixer device was simulated using COMSOL and MEEP softwares. COMSOL succeeded in showing that the periodic antenna array displayed plasmonic behavior when excited by visible light. The sizes, end to end spacings, excit ation wavelength, and incoming light angle were all varied and analyzed. The strongest absorption occurred on 450 nm antennas spaced 300 nm apart at an incident angle of 53 relative to the normal. Strong excitation was repeatedly shown on corners of the s tructures, which would be interesting to incorporate into pattern design. A few 3D simulation s were performed. Due to lack of computing time, the majority of simulations were made using various 2D models which provide a reasonable design insight. Simulated reflectance and refractance testing of these periodic arrays showed strong absorption dips near a wavelength of 450 nm The depth of this dip decreased with increasing antenna size and also showed a slight shift in wavelength. Thus smaller antenna sizes were shown to absorb much stronger than larger sized antennas. This data is in accordance with typical plasmonic behavior. The simulations in MEEP were performed primarily to confirm plasmonic coupling and photomixing from the device. Coupling was easily seen in time slice images of light sources exciting the periodic antenna array. Even after the source had been turned off, the electric fields remained on the surface of the antennas. When the system was in resonance, the intensity of the electric field on the surface of the antennas was much stronger Calculations of transmission through the array were also performed using a measurement port placed in the backfield. All quantitative results showed strong peaks from the incoming sources, and some showed sma ller peaks near the photomixing frequency.

PAGE 95

95 The MEEP simulations presented here are idealized 2D models and do not take into account a number of factors from the actual experiments that would improve plasmonic efficiency and resonance. Firstly, as seen in the COMSOL simulations, the antennas were most strongly excited and coupled when excited by light striking the antennas at a 53 angle, where the 2D MEEP simulations have the sources parallel to the plane of the antennas. In addition, within the xy plane, the two sources could not be angled in any way. Secondly, the antennas in MEEP are simulated as a perfect conductor with some nonlinearity added through a (3) value. Therefore, the actual material properties of silver and its inherent optical nonlinearit y were not accounted for. All of these factors could reduce the simulated plasmonic efficiency, and therefore should be used only as a guide for sample design. The simulations were utilized to determine several likely candidate designs. The primary candidate for this project was identified to be arrays of elliptical antenna alternated with longer antenna placed in periodic positions to correspond with the long wave output. These were designated for their ease of fabrication and known resonant properties. Arrays of alternating antenna corresponding to the source frequencies were also identified as promising for this project. Photomixing and Testing Optical testing proved the resonant silver antenna structures to be plasmonic and polarization dependent. Plasmonic behavior was demonstrated by the polarization dependent absorption of light by the antennas. The polarization dependence of this absorption was proven to be an effect of the antenna structures, as the bare s ilver substrate did not display any polarization dependent behavior. Further, smaller antenna size s elicited a much stronger absorption, which matched the reflection data simulated

PAGE 96

96 in COMSOL. In addition, the reflectance vs. wavelength spectra followed a s imilar trend to the simulated data, with both data sets showing a strongest absorption at a wavelength approximately 450 nm, thereby corroborating the experimental results. Later, photomixing on a number of patterns were tested using a custom set up that was altered in a number of ways in an effort to maximize the response from the structures. As expected, tests from the bare silver substrate did not elicit a THz signal, which served as a control for the experiments. When the chopped laser was aligned with the sample, it displayed a polarization dependent signal. On a sample of alternating 780 nm and 2 m antennas, which were modeled after the MEEP s imulation showing the strongest potential mixing peak, a mixing peak was observed when the sample was in line with the polarized 780 nm laser. The mixing peak was not observed when the sample was rotated, but was repeatable when returned to its original position. Filters were employed to verify that this was not purely a scattering effect However, the reduced si gnal to noise within the system when utilizing the full filter system prevented reliable detection of signal above the noise. Further experiments with a more robust system with better signal to noise will be required to unambiguously determine the source o f the detected signals. Photomixing could p otentially be improved by creating samples with a larger overall area. This would help with properly alignin g the lasers on the sample area, a nd possibly encourage a stronger signal from the larger scale coupling. Defining the resonance of these structures should also help improve the signal of the photomixed output. As a whole however, these initial photomixing experiments show that the device

PAGE 97

97 has the potential to be a viable CW source, but needs further experimen tal testing as part of future projects.

PAGE 98

98 CHAPTER 6 FUTURE WORK The initial results on the prototype of the plasmonic photomixer point towards it becoming an excellent CW tunable radiation source. However, f uture work is necessary in a number of areas First off, many more designs will need to be screened and potentially fabricated and analyzed in order to converge on an optimal array pattern that exhibits strong resonance and photomixing in the THz range. While this work has found some initial designs a nd provided a framework for screening, the breadth of design choices with endless geometrical possibilities is limited only by imagination and engineering. In addition larger overall pattern sizes would aid in aligning the lasers onto the arrays and would increase the interaction area These samples should also be optimized for resonance via further experimental work and simulations Finally t he confirmation of strong corner interactions by COMSOL would be interesting to study in relation to these pattern s. However, a different fabrication process may be necessary to mak e these sharp corners a reality, as the eB process tends to create elements with more rounded edges. In terms of simulations, it would be interesting to expand on the work done in COMSOL an d create a 3D simulation. This would be helpful in analyzing the system. In MEEP, a step function could be run over the frequency variable to optimize transmission through the array. It would also be of interest to optimize the (3) value for silver in ord er for the simulations to be more comparable to the actual data. If possible, a 3D simulation or angling of the sources would also be helpful. For very sensitive measurements in the THz range liquid helium cooled bolometers will be required Bolometers a re more advantageous for THz detection due

PAGE 99

99 to their high speed response and low noise. These measurements can assess how far the device can reach into the THz range. Efficiency and power outputs of the plasmonic photomixer device will also be necessary. Th e eB lithography process will be appropriate for the fabrication of the device through the prototype through early production stage. However, to be more commercialized, a faster and less expensive process must be chosen. Modern templating processes would b e a great option to explore. In addition, the stability of the device is in need of further research. The corrosion of silver could reduce device lifetime, so experiments should be done using coatings on the device, such as polyanaline; or a less corrosive metal, such as gold, should be chosen in order to improve the lifetime of the device. The potential for this device to be used as a radiation detector, as discussed in Chapter 2, is also an interesting area to be examined and tested in the future. The dev ice should not need to be cooled, making it potentially efficient and cost effective. Efficiencies will also have to be tested for this detector device so that it can be better compared to competing technologies Lastly, once a strong mixed signal has been achieved, the entire plasmonic photomixer device could be miniaturized and refined into a more stable platform. This platform would be more user friendly and less fragile and would help promote the device as an excellent CW tunable THz source.

PAGE 100

100 APPENDIX MODELING OF A 3D PLA SMONIC SPHERE A 3D simulation of the antenna arrays was attempted, but never completed due to complications that arose in the design. However, a plasmonic silver sphere in air was simulated and observed in 3D. In future work, this sphere can be developed into a 3D antenna on a silver surface and analyzed with periodic boundary conditions. The TE simulation of this sphere at 450 nm is shown in Figure 4 8. It displays a strong electric field on its surface and in the surrounding air. Figure A 1. Transverse electric (TE) simulation of the excited plasmonic silver sphere in air.

PAGE 101

101 REFERENCES [1 ] European Magnetic Resonance Forum/TRTF, http://www.magnetic resonance.org/ 2013, 2013 [2 ] Homeland Security News Wire, http://www.homelandsecuritynewswire.com/versatile terahertz technology could help aircraft land 2010, 2013 [3 ] S. Sy, S. Huang, Y. J. Wang, J. Yu, A. T. Ahuja, Y. Zhang and E. Pickwell MacPherson, Phys. Med. Biol., 2010, 55, 7587 7596. [4 ] M. Shur, Proc. ESSCIRC, 2005, 13. [5 ] D. Dragoman and M. Dragoman, Physica E, 2004, 24, 282 289. [6 ] P. H. Siegel, IEEE T Microw Theory 2002, 50, 910 928. [7 ] J. F. Federici, B. Schulkin, F. Huang, D. Gary, R. Barat, F. Oliveira and D. Zimdars, Semiconduct. Sci. Tec h. 2005, 20, S266 S280. [8 ] J. Grade, P. Haydon and D. van der Weide, Proc IEEE, 2007, 95, 1583 1591. [9 ] D. Cleary, Science, 2002, 297, 761. [10 ] F. Diep, Tech News Daily, http ://www.technewsdaily.com/16546 nypd terahertz scanner.html 2013, 2013 [1 ] H. Richter, A. D. Semenov, S. G. Pavlov, L. Mahler, A. Trediucci, H. E. Beere, D. A. Ritchie, M. Ortolani, U. Schade, K. S. I'Iin, M. Siegel, and H. W. Hubers, Proc. IEEE, 2007, 95, 1583. [12 ] M. Winnewisser, W. Gordy, K. V. L. N. Sastry and R. L. Cook, J. Chem. Phys., 1964, 41, 1687 &. [13 ] F. C. Delucia, E. Herbst, G. M. Plummer and G. A. Blake, J. Chem. Phys., 1983, 78, 2312 2316. [14 ] M. D. Marshall, A. Charo, H. O. Leung and W. Klemperer, J. Chem. Phys., 1985, 83, 4924 4933. [15 ] A. Charo and F. C. Delucia, J. Mol. Spectrosc., 1982, 94, 426 436. [16 ] B. Ferguson and X. C. Zhang, Nat. Mater. 2002, 1, 26 33. [17 ] C. Brenner, C. Friedrich and M. R. Hofmann, J. Infrared Millim. W. 2011, 32, 1253 1266. [18 ] R. A. Kaindl, D. Hagele, M. A. Carnahan, R. Lovenich and D. S. Chemla, Phys. Status Solidi B 2003, 238, 451 454.

PAGE 102

102 [19 ] R. A. Kaindl, M. A. Carnahan, D. Hagele, R. Lovenich and D. S. Chemla, Nature, 2003, 423, 734 738. [20 ] A. J. Fitzgerald, E. Berry, N. N. Zinovev, G. C. Walker, M. A. Smith and J. M. Chamberlain, Phys. Med. Biol., 2002, 47, R67 R84. [21 ] D. Mittleman ed., Sensing with Terahertz Radiation Springer Verlag, Germany, 2003. [22 ] R. M. Woodward, B. E. Cole, V. P. Wallace, R. J. Pye, D. D. Arnone, E. H. Linfield and M. Pepper, Phys. Med. Biol., 2002, 47, 3853 3863. [23 ] P. Knobloch, C. Schildknecht, T. Kleine Ostmann, M. Koch, S. Hoffmann, M. Hofmann, E. Rehberg, M. Sperling, K. D onhuijsen, G. Hein and K. Pierz, Phys. Med. Biol., 2002, 47, 3875 3884. [24 ] J. Q. Lu, M. S. Shur, J. L. Hesler, L. Q. Sun and R. Weikle, IEEE Electr. Device L ., 1998, 19, 373 375. [25 ] W. Knap, Y. Deng, S. Rumyantsev and M. S. Shur, Appl. Phys. Lett., 200 2, 81, 4637 4639. [26 ] H. Tanoto, J. H. Teng and S. A. Maier, Laser Focus World, 2012, 48, 57 61. [27 ] H. Pahlevaninezhad, B. Heshmat and T. E. Darcie, I EEE Photonic J. 2011, 3, 307 310. [28 ] A. Eshaghi, M. Shahabadi, L. Chrostowski and S. Kamal, J. Opt. Soc. Am. B 2012, 29, 3254 3258. [29 ] R. W. Boyd and G. L. Fischer eds., Nonlinear Optical Materials Elsevier, Oxford, 2001, p. 6237. [30 ] R. del Coso and J. Solis, J Opt Soc Am B, 2004, 21, 640 644. [31 ] P. E. Powers, Fundamental Nonlinear Optics CRC ; Taylor & Francis distributor, Boca Raton, Fla. 2011. [32 ] E. R. Brown, K. A. Mcintosh, K. B. Nichols and C. L. Dennis, Appl. Phys. Lett., 1995, 66, 285 287. [33 ] K. A. McIntosh, E. R. Brown, K. B. Nichols, O. B. McMahon, W. F. DiNatale and T. M. Lys zczarz, Appl. Phys. Lett., 1995, 67, 3844 3846. [34 ] M. Tani, P. Gu, M. Hyodo, K. Sakai and T. Hidaka, Opt. Quant. Electron., 2000, 32, 503 520. [35 ] F. Hindle, A. Cuisset, R. Bocquet and G. Mouret, Comptes Rendus Physique, 2008, 9, 262 275.

PAGE 103

103 [36 ] S. Pilla, Appl. Phys. Lett., 2007, 90, 161119. [37 ] K. A. Willets and R. P. Van Duyne, Annu. Rev. Phys. Chem., 2007, 58, 267 297. [38 ] H. Raether, Surface Plasmons Springer, Berlin, 1988. [39 ] S. A. Maier, Plasmonics : Fundamentals and Applications Springer, New York, 2007. [40 ] W. L. Barnes, A. Dereux and T. W. Ebbesen, Nature, 2003, 424, 824 830. [41 ] S. L. Brongersma and P. G. Kik eds., Surface Plasmon Nanophotonics Springer, The Netherlands, 2007. [42 ] E. C. Le Ru, E. Blackie, M. Meyer and P. G. Etchegoin, J Phys Chem C, 2007, 111, 13794 13803. [43 ] K. A. Willets and R. P. Van Duyne, Annu. Rev. Phys. Chem., 2007, 58, 267 297. [44 ] S. A. Maier, M. L. Brongersma, P. G. Kik, S. Meltzer, A. A. G. Requicha and H. A. Atwater, Adv Mater 2001, 13, 1501 +. [45 ] M. I. Stockman, Phys. Rev. Lett., 2004, 93, 137404. [46 ] S. I. Bozhevolnyi, V. S. Volkov, E. Devaux, J. Y. Laluet and T. W. Ebbesen, Nature, 2006, 440, 508 511. [47 ] E. Ozbay, Science, 2006, 311, 189 193. [48 ] G. Gallot, S. P. Jamison, R. W. McGowan and D. Grischkowsky, J. Opt. Soc. Am B, 2000, 17, 851 863. [49 ] K. L. Wang and D. M. Mittleman, Nature, 2004, 432, 376 379. [50 ] A. Ishikawa, S. Zhang, D. A. Genov, G. Bartal and X. Zhang, Phys. Rev. Lett., 2009, 102, 043904. [51 ] A. Y. Nikitin, F. Guinea, F. J. Garcia Vidal and L. Martin Moreno, Phys. Rev. B, 2011, 84, 161407. [52 ] Advanced Plasmonics, Inc., http://aplasmonics.com/ numerous patents and personal interactions, 2013. [53 ] T. Ryhanen, M. A. Uusitalo, O. Ikkala and A. Karkkainen eds., Nanotechnologies for Future Mobile Devices Cambridge University Press, United Kingdom, 2010. [54 ] H. Cao and A. Nahata, Opt. Express, 2004, 12, 1004 1010.

PAGE 104

104 [55 ] L. Solymar and D. and Walsh, Electrical Properties of Materials: Seventh Edition Oxford University Press, New York, 2004. [56 ] S. A. Maier and H. A. Atwater, J. Appl. Phys., 2005, 98, 011101. [57 ] J. A. Creighton and D. G. Eadon, J Che m. Soc. Faraday T. 1991, 87, 3881 3891. [58 ] A. Reeves ed., Introduction to Electrodynamics Prentice Hall, New Jersey, 1981. [59 ] M. Ohring, Engineering Materials Science, Volume 1 Academic Press, Inc., California, 1995. [60 ] R. A. Serway, Principles of Physics Saunders College Pub., Texas, 1998. [61 ] R. F. Egerton, Electron Energy Loss Spectroscopy in the Electron Microscope Springer, New York, 2011. [62 ] A. de Rooij, ESA J 1989, 13, 363. [63 ] W. M. Haynes ed., CRC Handbook of Chemistry and Physics CRC Press, Boca Raton, 2012. [64 ] D. Gr aham, K. Faulds and W. E. Smith, Chem. Commun. 2006, 4363 4371. [65 ] V. Brusic, M. Angelopoulos and T. Graham, 53rd Annu Tech Conf Soc Plast Eng 1995, 1397. [66] V. Brusic, G. S. Frankel, J. Roldan and R. Saraf, J. Electrochem. Soc., 1995, 142, 2591 2594. [67 ] V. Brusic, M. Angelopoulos and T. Graham, J. Electrochem. Soc., 1997, 144, 436 442. [68 ] D. E. Tallman, G. Spinks, A. Dominis and G. G. Wallace, J. Solid State Electr ., 2002, 6, 73 84. [69 ] L. Yang, D. H. Osborne, R. F. Haglund, R. H. Magru der, C. W. White, R. A. Zuhr and H. Hosono, Appl Phys. A Mater 1996, 62, 403 415. [70 ] R. Philip, G. R. Kumar, N. Sandhyarani and T. Pradeep, Phys. Rev. B, 2000, 62, 13160 13166. [71 ] S. Palomba and L. Novotny, Phys. Rev. Lett., 2008, 101, 056802. [72 ] P Genevet, J. Tetienne, E. Gatzogiannis, R. Blanchard, M. A. Kats, M. O. Scully and F. Capasso, Nano Lett. 2010, 10, 4880 4883. [73 ] M. Ren, E. Plum, J. Xu and N. I. Zheludev, Nature Commun. 2012, 3, 833.

PAGE 105

105 [74 ] G. Fan, S. Ren, S. Qu, Z. Guo, Q. Wang, Y. W ang and R. Gao, Opt. Commun., 2013, 295, 219 225. [75 ] COMSOL Mulitphysics, Plasmonic Wire Grating Tutorial, 2011, 2013 [76 ] H. Chung, N. M. Bradman, M. R. Davidson and P. H. Holloway, Opt. Engineering, 2008, 47, 118001. [77 ] J. C. Nabity, http://www.jcnabity.com/index.htm 2013, 2013 [78 ] N. M. Bradman, UF Dissertation, 2007 [79 ] B. Michel, A. Bernard, A. Bietsch, E. Delamarche, M. Geissler, D. Juncker, H. Kind, J. P. Renault, H. Rothuizen, H. Schmid, P. Schmidt Winkel, R. Stutz and H. Wolf, IBM J Res. Dev. 2001, 45, 697 719. [80 ] L. J. Guo, Adv Mater 2007, 19, 495 513. [81 ] A. Mckinley, R. H. Williams and A. W. Parke, J Phys. C Solid State 1979, 12, 2447 2463. [82 ] MicroChem Corp., http://microchem.com/pdf/PMMA_Data_Sheet.pdf 2001, 2013 [83 ] P. Rai Choudhury ed., Handbook of Microlithography, Micromahining and Microfabrication. Vol. 1, Microlithography SPIE Optical Engineering Press, Washington, 1997. [84 ] Kolmar Technologies, http://www.kolmartech.com/2color.htm 2005, 2013 [85 ] J. M. Blea, W. F. Parks, P. A. R. Ade and R. J. Bell, J. Opt. Soc. Am., 1970, 60, 603 &. [86 ] D. E. Mccarthy, J. Opt. Soc. Am., 1967, 57, 699 &. [87 ] ORIEL Instruments, http://www.newport.com/Merlin Digital Lock in Radiometry System/3 78848/1033/info.aspx 2013, 2013 [88 ] D. P. Fromm, A. Sundaramurthy, P. J. Schuck, G. Kino and W. E. Moerner, Nano Lett. 2004, 4, 957 961. [89 ] H. Fischer and O. J. F. Martin, Opt. Express, 2008, 16, 9144 9154. [90] A. Kinkhabwala, Z. Yu, S. Fan, Y. Avla sevich, K. Muellen and W. E. Moerner, Nat. Photonics, 2009, 3, 654 657. [91 ] J. Butet, K. Thyagarajan and O. J. F. Martin, Nano Lett. 2013, 13, 1787 1792.

PAGE 106

106 BIOGRAPHICAL SKETCH in Mons, Belgium. In 1989, she moved to the Philadelphia suburbs, where she spent the rest of her childhood. She graduated from Villa Maria Academy High School in 2005. During high school, she attended the Summer Engineering Experience at Drexel University ( SEED) and discovered and fell in love with the field of materials science and engineering after using an SEM to examine and then destroy a gold plated Cheerio She attended The Pennsylvania State University and graduated in 2009 with a Ba chelor of Science in materials s cience a nd engineering with a Minor in m athematics. Beginning her freshman year at Penn State, she worked for Dr. Darrell Schlom on novel com plex oxide thin films by pulsed laser deposition and molecular beam epitaxy for lead free piezoelect rics She also interned at The Aerospace Corporation in El Segundo, CA during the summers of 2008 and 2009, where she mainly worked with focused ion beam examination and preparation of various electronic samples. In 2009, Kathryn was admitted to Universit y of Florida on an Alumni Fellowship in the Department of Materials Science and Engineering. She began working for Dr. Mark Davidson and Dr. Paul Holloway in January 2010. While in Dr. Davidson and Dr. evices for THz emission and bio sensing She received her Master of Science in materials science and engineering in 2010. She earned her Ph.D. in materials s cience in e ngineering in August 2013.