AN ABSTRACT OF THE DISSERTATION OF
Zachary James Thompson for the degree of Doctor of Philosophy in Physics presented on August 19, 2015.
Title: Terahertz Imaging and Nonlinear Spectroscopy of Semiconductors using Plasmonic Devices
Abstract approved: Yun-Shik Lee
In this dissertation, a series of studies in the field of terahertz (THz) science are pre- sented, specifically using nonlinear THz spectroscopy. We exploit huge field enhancement and subwavelength confinement in plasmonic structures. There are three distinct projects which will be discussed: nonlinear THz spectroscopy using plasmonic induced transparency (PIT), THz-triggered insulator-metal transition (IMT) in nanoantenna patterned vanadium dioxide (VO2) films, and fabrication of sub-diffraction limit imaging bulls-eye structures. We used PIT structures to observe the high-field carrier dynamics in semiconductors, specifically in intrinsic, high resistivity silicon (high-ρ Si) and intrinsic gallium arsenide (GaAs). The PIT structures rely on the coupling of a ”bright mode” in a central half-wave dipole antenna to the ”dark mode” of the adjacent split-ring resonators. We employed these structures because of their sensitivity to carrier dynamics due to the sharp resonance of the ”dark mode.” We observed the response of the PIT oscillation to both low and high THz fields in the presence of an optical pump. Increasing the optical pump power, and therefore the number of carriers, resulted in the damping of the oscillation. With increasing THz field strength, we observed a field induced transparency from the intervalley scattering of the excited carriers and demonstrated THz control of the PIT oscillation. By changing the delay time between the THz and optical pulses, we demonstrated pulse shaping of the PIT waveforms. We demonstrated the THz-triggered insulator-metal transition (IMT) in nanoantenna patterned vanadium dioxide (VO2) films. Vanadium dioxide is a promising material for electronic and photonic applications due to its IMT transition lying near room temperature. We observed that the phase transition is activated on the sub-cycle time scale where strong THz fields drive the electron distribution far from equilibrium. We also observed a lowering the transition temperature of the IMT phase transition for both heating and cooling cycles in nanoslot antenna VO2 films with increasing THz fields and also a narrowing in the width of the observed hysteresis. Using the Fresnel thin-film coefficients, Drude model, and the resistivity in semiconductors we found the activation energy in the insulating phase and show that it can be lowered with THz fields. We employed THz time domain spectroscopy to extract the frequency dependence and to observe the transiently induced IMT from the strong THz fields. We attempted to fabricate sub-diffraction-limit imaging bulls-eye structures in the Oregon State University cleanroom. During the course of the project, recipes for two different types of photoresists, SU-8 2100 and SU-8 5, were developed. We observed lack of adhesion of the metal (Al) layer for the metal-dielectric interface. Lastly the removal of metal for the apertures posed additional problems. While this project did not ultimately succeed, we present an explanation of the issues associated with their fabrication and the steps necessary to complete fabrication. c Copyright by Zachary James Thompson August 19, 2015 All Rights Reserved Terahertz Imaging and Nonlinear Spectroscopy of Semiconductors using Plasmonic Devices
by
Zachary James Thompson
A DISSERTATION
submitted to
Oregon State University
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
Presented August 19, 2015 Commencement June 2016 Doctor of Philosophy dissertation of Zachary James Thompson presented on August 19, 2015.
APPROVED:
Major Professor, representing Physics
Chair of the Department of Physics
Dean of the Graduate School
I understand that my dissertation will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request.
Zachary James Thompson, Author ACKNOWLEDGEMENTS
I would like to begin by thanking my adviser, Dr. Yun-Shik Lee for his infinite patience and understanding during my time at Oregon State. I also would like to thank my group members Michael Paul, Andrew Stickel, Byounghwak Lee, and Ali Mousavian for their help inside and outside of the lab, I could not have done this without you guys. I especially thank Morgan Brown and Rick Presley for their invaluable guidance during the course of my fabrication projects and acting as sounding boards for any insane ideas I could conjure. I want to thank my parents, Rick and Sharon Thompson, for their endless words of encouragement and help during this process. I have the best parents in the world. Most importantly, I want to thank my wonderful wife Mikayla. She has kept me sane and dealt with any and all irrational reactions I’ve had to the smallest stimuli. Without her, I would be completely lost. For the rest of the people who have helped over the years (including my sister Sara AKA Fat Kid), I express my undying gratitude with the Del ’n Bones below. It is intended to represent the resilience and dedication required for the path I have chosen and to pay homage to those who are no longer with us. TABLE OF CONTENTS Page 1 Overview 1 1.1 History...... 1 1.2 Sources...... 2 1.3 Detection...... 3 1.4 Applications...... 4
2 Electromagnetic Waves in Nonlinear Media6 2.1 Linear Media and Wave Equation...... 6 2.2 Thin-Film Fresnel Formula...... 10 2.3 Harmonic Oscillator...... 14 2.4 Nonlinear Media...... 16
3 Terahertz Generation and Detection 20 3.1 Generation via Optical Rectification...... 20 3.1.1 Phase Matching...... 20 3.1.2 Zinc Telluride...... 22 3.1.3 Lithium Niobate...... 23 3.2 Terahertz Detection...... 26 3.2.1 Bolometer...... 27 3.2.2 Pyroeletric Detectors...... 28 3.2.3 Michelson Interferometry...... 28 3.2.4 Electro-Optic Sampling...... 29
4 Surface Plasmons and Surface Plasmon Polaritons 33 4.1 Theory of Surface Plasmons...... 33 4.2 Drude-Sommerfeld Model...... 33 4.3 Surface Plasmon Polaritons at Interfaces...... 34 4.3.1 Dispersion Relation...... 35 4.3.2 SPP Excitation...... 39 4.4 Excitation Methods...... 40 4.4.1 Otto Method...... 40 4.4.2 Kretschmann Method...... 41 4.4.3 Spatial Periodicity...... 41 4.5 Applications to Terahertz Science...... 42
5 Non-Linear Terahertz Spectroscopy using Plasmon Induced Transparency 43 5.1 Introduction...... 43 5.2 Plasmonic Induced Transparency...... 44 TABLE OF CONTENTS (Continued) Page 5.2.1 Coupling of Two Resonators...... 44 5.3 Coupling of Linear Antenna and Split-Ring Resonator...... 47 5.3.1 Fabrication of PIT Structure...... 52 5.4 Experimental Considerations...... 56 5.5 Initial Testing and PIT Observation...... 56 5.5.1 Resonant Cases in GaAs...... 57 5.5.2 Resonant Cases in Si...... 60 5.5.3 Off Resonant Cases in GaAs...... 64 5.5.4 Off Resonant Cases in Si...... 67 5.6 THz Field Effects of GaAs and Si PIT Structures...... 70 5.7 Optical Excitation in the Wake of a PIT Resonance...... 71 5.7.1 GaAs...... 72 5.7.2 Si...... 74 5.8 THz Control of PIT Resonance...... 75 5.9 Pulse Shaping of PIT Waveforms...... 77 5.10 THz pump-Optical Pump Experiments...... 78 5.11 Summary...... 80
6 Terahertz Field Induced Metal-Insulator Transition in Vanadium Dioxide 81 6.1 Introduction...... 81 6.2 Mott Insulators...... 81 6.3 Sample Structure...... 85 6.4 Experimental Considerations...... 86 6.5 Terahertz Field Induced Absorption...... 87 6.6 Hysteresis and Activation Energy...... 88 6.6.1 Resistivity Derivation...... 89 6.6.2 Resistivity and Activation Energy...... 92 6.6.3 Hysteresis Width...... 96 6.7 Transient Phase Transition...... 98 6.8 Summary...... 101
7 Sub-Diffraction Limit Nonlinear Imaging with Plasmonic Devices 102 7.1 Introduction...... 102 7.2 Background...... 102 7.3 Fabrication...... 104 7.4 Future Work...... 106 TABLE OF CONTENTS (Continued) Page 8 Conclusion 108
Bibliography 110
Appendix 124 A PIT Fabrication Recipe...... 125 B Bullseye Fabrication Recipe...... 127 LIST OF FIGURES Figure Page 1.1 The electromagnetic spectrum with the THz gap highlighted in blue.....1
2.1 Cartoon of the transmission...... 10 2.2 A cartoon of the harmonic oscillator model...... 14
2.3 Example of an harmonic and an anharmonic potential. The associated motion of a charge carrier is plotted on the left...... 16
3.1 Example of generation using optical rectification in a phase-matched medium. Where the light is propagating left to right, Eopt is the optical electric field, POR is the polarization of the material, and ET Hz is the THz electric field.. 21 3.2 A cartoon of the walk-off length. Figure 3.35 from Ref. [1]...... 22 3.3 Index of refraction plot. Note that the optical group index and THz phase index are matched at λopt ≈ 810 nm and νTHz ≈ 1.7 THz. Figure 3.37 from Ref. [1]...... 23
3.4 Cartoon of our tilted pulse front setup for LiNbO3. Note that the diffraction grating tilts the optical pulse front (red). The THz output pulse front is shown in orange...... 25 3.5 Diffraction angle plot...... 26 3.6 Schematic of a compound bolometer. Figure 4.25 from Ref. [1]...... 27
3.7 Schematic of a pyroelectric detector. Figure 4.28 from Ref. [1]...... 28 3.8 A cartoon of our Michleson interferometer using a bolometer...... 29 3.9 A cartoon of our EO sampling setup...... 31
4.1 Interface diagram...... 35
4.2 SPP dispersion curve...... 39 4.3 Otto configuration...... 40 4.4 Kretschmann configuration...... 41
4.5 Grating coupler, one example of a periodic array used to elicit a SPP resonance. 42
5.1 A) Parallel SRRs. B) Anti-parallel SRRs...... 44 LIST OF FIGURES (Continued) Figure Page 5.2 SRRs with 90o rotation. The figure illustrates the higher energy resonant case (A) where the magnetic dipoles are parallel and the lower energy resonant case where the magnetic dipoles are anti-parallel (B) and the associated line splitting (C)...... 45
5.3 The geometry for Az...... 49 5.4 The square loop cartoon...... 50
5.5 PIT resonance cartoon illustrating the coupling between the dipole antenna and adjacent SRRs...... 51 5.6 Here is an example unit cell of the PIT structure. Dimensions can be found in Table 5.3.1 (below)...... 53 5.7 Optical microscope pictures of the GaAs sample. The pictures on the left are the positive arrays. The pictures on the right are the negative arrays. The top row is the 0.6 THz arrays. The middle row is the 0.9 THz arrays. The bottom row is the 1.2 THz arrays...... 54 5.8 Optical microscope pictures of the Si sample. The pictures on the left are the positive arrays. The pictures on the right are the negative arrays. The top row is the 0.6 THz arrays. The middle row is the 0.9 THz arrays. The bottom row is the 1.2 THz arrays. The psychedelic blue color is due to the improper functioning of the white-balance on the microscope...... 55
5.9 The GaAs PIT 0.6 THz positive array when the THz polarization is parallel to the central antenna (0 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 57
5.10 The GaAs PIT 0.6 THz negative array when the THz polarization is perpen- dicular to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 57 5.11 The GaAs PIT 0.9 THz positive array when the THz polarization is parallel to the central antenna (0 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 58 5.12 The GaAs PIT 0.9 THz negative array when the THz polarization is perpen- dicular to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 58 5.13 The GaAs PIT 1.2 THz positive array when the THz polarization is parallel to the central antenna (0 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 59 LIST OF FIGURES (Continued) Figure Page 5.14 The GaAs PIT 1.2 THz negative array when the THz polarization is perpen- dicular to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 59 5.15 The Si PIT 0.6 THz positive array when the THz polarization is parallel to the central antenna (0 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 60 5.16 The Si PIT 0.6 THz negative array when the THz polarization is perpendicular to the central antenna (90 degree) with an incident THz field of 300 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 60 5.17 The Si PIT 0.9 THz positive array when the THz polarization is parallel to the central antenna (0 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 61 5.18 The Si PIT 0.9 THz negative array when the THz polarization is perpendicular to the central antenna (90 degree) with an incident THz field of 300 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 61 5.19 The Si PIT 1.2 THz positive array when the THz polarization is parallel to the central antenna (0 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 62 5.20 The Si PIT 1.2 THz negative array when the THz polarization is perpendicular to the central antenna (90 degree) with an incident THz field of 300 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 62 5.21 The GaAs PIT 0.6 THz positive array when the THz polarization is perpen- dicular to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 64
5.22 The GaAs PIT 0.6 THz negative array when the THz polarization is parallel to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 64
5.23 The GaAs PIT 0.9 THz positive array when the THz polarization is perpen- dicular to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 65
5.24 The GaAs PIT 0.9 THz negative array when the THz polarization is parallel to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 65 LIST OF FIGURES (Continued) Figure Page 5.25 The GaAs PIT 1.2 THz positive array when the THz polarization is perpen- dicular to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 66 5.26 The GaAs PIT 1.2 THz negative array when the THz polarization is parallel to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 66 5.27 The Si PIT 0.6 THz positive array when the THz polarization is perpendicular to the central antenna (90 degree) with an incident THz field of 300 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 67 5.28 The Si PIT 0.6 THz negative array when the THz polarization is parallel to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 67 5.29 The Si PIT 0.9 THz positive array when the THz polarization is perpendicular to the central antenna (90 degree) with an incident THz field of 300 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 68 5.30 The Si PIT 0.9 THz negative array when the THz polarization is parallel to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 68 5.31 The Si PIT 1.2 THz positive array when the THz polarization is perpendicular to the central antenna (90 degree) with an incident THz field of 300 kV/cm. The waveform is on the left and relative power spectrum is on the right.... 69 5.32 The Si PIT 1.2 THz negative array when the THz polarization is parallel to the central antenna (90 degree) with an incident THz field of 290 kV/cm. The waveform is on the left and relative power spectrum is on the right...... 69 5.33 The GaAs PIT 0.9 THz negative array at 90 degrees. The incident THz field strength was modulated...... 71 5.34 The Si PIT 0.9 THz negative array at 90 degrees. The incident THz field strength was modulated...... 71 5.35 Cartoon of the optical pump setup. A small hole is drilled through a focusing parabolic mirror through which the optical pump passes and overlaps the THz focus...... 72 5.36 The GaAs PIT 0.9 THz negative array at 90 degrees with an optical excitation 0.667 ps after the THz excitation. The legend shows the carrier concentration. 73 5.37 The GaAs PIT 0.9 THz negative array at 90 degrees with an optical excitation at 0.667 ps. The legend shows the carrier concentration...... 74 LIST OF FIGURES (Continued) Figure Page 5.38 The Si PIT 0.9 THz negative array at 90 degrees with an optical excitation at 0.667 ps. The legend shows the carrier concentration...... 74 5.39 The Si PIT 0.9 THz negative array at 90 degrees with an optical excitation at 0.667 ps. The legend shows the carrier concentration...... 75 5.40 The GaAs PIT 0.9 THz negative array at 90 degrees with an optical exci- tation at 0.667 ps. Increasing the incident THz field greatly modulates the transmission...... 76 5.41 The Si PIT 0.9 THz negative array at 90 degrees with an optical excitation at 0.667 ps. Increasing the incident THz field slightly modulates the transmission. 77
5.42 The GaAs PIT 0.9 THz negative array at 90 degrees with an optical excitation after the PIT excitation at the time listed in the legend. By delaying the optical pulse we were able to shape the transmitted THz waveform...... 78
5.43 The Si PIT 0.9 THz negative array at 90 degrees with an optical excitation after the PIT excitation at the time listed in the legend. By delaying the optical pulse we were able to shape the transmitted THz waveform...... 78
5.44 The GaAs PIT 0.9 THz negative array at 90 degrees with an optical excitation 0.667 ps after the PIT excitation. The high optical pump power damps the PIT oscillation...... 79
5.45 The Si PIT 0.9 THz negative array at 90 degrees with an optical excitation 0.667 ps after the PIT excitation. The high optical pump power damps the PIT oscillation...... 79
6.1 A. Orbital diagram for Vanadium when T is below Tc (unstrained) and above Tc (strained). B. Cartoon of the dimerization of valence electrons in neigh- boring Vanadium atoms...... 82 6.2 Cartoon of the nanoslot antennas...... 85
6.3 Plot of the THz field dependence for the nanoslot and bare VO2 (inset)... 87
6.4 Plot of the hysteresis curves for the nanoslot and bare VO2 (inset)...... 89 6.5 Plot of the normalized sheet conductivity comparing the bare and nanoslot VO2 data...... 93 6.6 Plot of the normalized sheet conductivity for the fit comparing the bare and nanoslot VO2 data...... 94 6.7 Plot of the resistivity as a function of temperature. The activation energy fits for temperatures between 35o and 55o C are over-plotted...... 95 LIST OF FIGURES (Continued) Figure Page 6.8 Plot of the hysteresis width as a function of incident electric field. The inset plot show the transition temperature for the nanoslot sample for increasing and decreasing temperature. The black lines are linear fits for the data.... 97 6.9 Waveforms for an incident THz field of 150 kV/cm at 45o C, 65o C, and 67o C. 98
6.10 Plot of the waveforms for incident THz fields of 150 kV/cm, 300 kV/cm, 630 kV/cm, and 850 kV/cm at 45o (a) and at 65o C (b). The power transmission spectra for each waveform is inset...... 99 6.11 Plot of the waveforms for 150 kV/cm at 67o C (temperature driven transition) and 850 kV/cm at 65o C (field driven transition) to help illustrate the lowering of the transition temperature. The yellow shaded area is the waveform for 150 kV/cm at 65o C...... 100
7.1 Example bullseye structure...... 103 7.2 Example of a 0.5 THz bullseye structure (left) and 1.0 THz bullseye structure (right). There are etching pits around the convex edges of the structures from the lack of S1818 adhesion and the resulting etching of the metal layer.... 106 Terahertz Imaging and Nonlinear Spectroscopy of Semiconductors using Plasmonic Devices
1 Overview
Terahertz radiation falls between the infrared (IR) and microwave regions of the elec- tromagnetic spectrum, colloquially know as the ”THz gap” (Fig 1.1). In nature, the THz generally corresponds to molecular rotational and vibrational modes. The photon energy at 1 THz is 4 meV which corresponds to thermal energy at a temperature of 48 K. The energy of this radiation is too low to excite atomic transitions which makes it an extremely useful tool to characterize materials since it acts as a nondestructive, non-contacting probe.
Figure 1.1: The electromagnetic spectrum with the THz gap highlighted in blue.
1.1 History
Terahertz science initially rose out of thermal detection. Room temperature corresponds to photon energies of 6 THz. Pioneering work was done by Rubens [2] in isolating this frequency. Planck recognized his work in 1922 by writing the following: ”Without the intervention of Rubens the formulation of the radiation law, and consequently the formulation of quantum theory, would have taken place in a totally different manner, and perhaps even not at all in Germany.” The next significant step in THz science was in 1965 when difference 2 frequency generation was used to create monochromatic 3 THz light. [3] Finally single-cycle THz sources were realized in 1990 [4] which lead to growth in the THz field. Now the field has become extremely accessible in recent years due to the advent of table-top sources and detectors.
1.2 Sources
We will briefly discuss several types of THz sources, including free-electron sources, THz lasers, photocurrent sources, and frequency conversion systems. We will begin with free-electron lasers (FEL). [5] These fall under free-electron sources. In this scheme, a population of electrons is excited using a short optical pulse. These excited electrons are then accelerated to relativistic speeds and passed through a magnetic array. This causes the electrons to oscillate in a sinusoidal pattern, thereby radiating narrowband THz radiation. Another example of a free electron source is a backward wave oscillator (BWO). [6] These work by projecting a beam of electrons into a counter propagating, slowly oscillating electromagnetic field. This causes a compression of the electron beam and the oscillation of the beam which emits and amplifies the THz radiation. A prime example of a THz laser is a quantum cascade lasers (QCL), where a material is engineered to have step potentials via periodic stacks of semiconducting materials. [7] An injected carrier will tunnel through the series of potential barriers, emitting a THz photon each time it tunnels. Photo-conductive (PC) antennas work in the following manner. [8] A semiconductor, generally gallium arsenide (GaAs), has two parallel strip-line antennas held at some potential difference. An optical pulse is used to excite carriers in the gap between the two antennas. The resulting motion of the electrons and holes gives rise to a time dependent current, emitting THz radiation. We will briefly discuss our method of generation, a frequency conversion system which 3 relies on optical rectification (OR). A short optical pulse of frequency ω is passed through a nonlinear crystal. The time-dependent polarization in the material is related to the intensity envelope of the optical pulse of frequency ω, which gives rise to a short, single-cycle THz pulse.
1.3 Detection
Detection of THz radiation falls into two categories: incoherent and coherent detection. In the former category, the detectors generally utilize thermal effects in some capacity. The latter uses methods very similar to the generation methods from the previous section and are used to acquire spectral information. Incoherent THz detectors are used for power measurements. Initially bolometers were used to detect thermal (THz) radiation by measuring the change in resistance across a small thermal mass when the mass is heated. [9] One of the issues of bolometers is they generally require liquid helium temperatures to detect THz radiation. Another method of thermal detection are golay cells. [10] Golay cells rely on the expansion of a small volume of gas to deform a flexible mirror and modulate a signal from a LED onto a photodetector. While golay cells are very sensitive detectors, they are generally large in size, reducing their utility. The rise and availability of micromachining has led to a reduction in their size. Lastly, we have pyroelectric detectors. [11] These rely on the heating of a crystal to change the instantaneous polarization. Pyroelectric detectors and bolometers will be discussed further in Sec. 3.2. Coherent THz detectors are used to extract frequency dependence from a transmitted signal. Photo-conductive switches work in much the same manner as PC antenna. First, an optical pulse is used to excite carriers between two strip-line antennas when there is no voltage bias between them. Then the THz pulse will hit the same spot as the optical did, inducing a current, which is measured. Electro-optic sampling uses a nonlinear optical crystal 4 in a similar fashion to optical rectification. The THz pulse is incident on the nonlinear optical crystal which induces a birefringence in the crystal. This physically means that the index of refraction in crystal changes depending on the incident polarization. This birefringence will rotate the polarization of the reference optical beam as it passes through the crystal. This rotation is proportional to the THz electric field. Electro-optic sampling is our primary method of extracting spectral data and will be discussed in-depth in Sec. 3.2.
1.4 Applications
The advent of table-top THz sources has led to an increase in THz research. Due to this increase in accessibility, there have been recent developments in THz science which have yielded promising applications. THz radiation is a useful tool for characterizing materials. High-speed wireless communication has been demonstrated and is being investigated by DARPA for secure communications. It also has great promise for security detection and imaging purposes. THz radiation can be used to characterize organic and semiconductor materials and devices. A prime example of research which has been conducted by our group for this purpose focused on graphene. Graphene has a huge response to THz radiation which allows for characterization of this novel, single layer material. THz radiation has been employed as a nondestructive, non-contacting probe. Due to graphene’s Drude-like response, the sheet conductivity can be extracted from the transmission. [12]. However it is noteworthy that when strong THz fields are applied there is an induced transparency. [13,14] The increase in demand for data transfer has driven wireless communication to THz frequencies. The lower bandwidths of the currently used GHz frequencies have created a push towards higher frequencies. [15] Recently transfer rates of up to 2.5 Gb/s have been observed at 0.625 THz, which illustrate the utility of this band. [16]Although the high transfer rates are attractive, they can only used for relatively short distance, on the order of meters, 5 due to power constraints. [17,18] Terahertz waves are an excellent tool for security detection and imaging. [19, 20] The THz regime is very attractive for security applications due to its non-ionizing nature from its low photon energy (4 meV). Material responses at THz frequencies correspond to molecular rotations. This leads to the ability to differentiate materials based on their spectral response. Owing to the general transparency of dielectrics in the THz regime, it allows for material detection inside packaging and differentiation of materials with similar optical properties due to their very different THz spectral responses. This allows for drug [21] and explosive [22] detection through non-destructive THz spectroscopy. Security imaging can utilize the THz response of materials. [23,24] Polar liquids (water), metals, plastics, and semiconductors all exhibit different responses to THz radiation. Polar liquids are highly absorptive which leads to the ability to differentiate between hydrated and dehydrated substances. Metals reflect nearly all incident THz radiation, leading to the easy detection of concealed weapons. Plastics have low absorption and low refractive index, leading to high transmission. Semiconductors have a high THz refractive index and low absorption. 6
2 Electromagnetic Waves in Nonlinear Media
2.1 Linear Media and Wave Equation
In order to understand the interaction of THz radiation and matter, our derivation must start at the very root of electricity and magnetism. This stems from the fact that the THz pulse used in our lab is generated via nonlinear optical processes (specifically optical recti- fication), which means we have to look at the polarization of materials. We will commence by writing Maxwell’s equations in matter. [25]
∂B ∇ × E = − (2.1) ∂t ∂D ∇ × H = J + (2.2) f ∂t
∇ · D = ρf (2.3)
∇ · B = 0 (2.4)
For linear media, we can write the displacement field (D-field) and auxiliary field (H- field) in the following manner:
D = 0E + P = E (2.5) 1 1 H = B − M = B (2.6) µ0 µ
Substituting these into Eq. 2.1 and Eq. 2.2, taking the curl, we find the generalized electromagnetic wave equations. 7
∂2E ∂ ∂P ∇ × ∇ × E + µ = −µ J + + ∇ × M (2.7) 0 0 ∂t2 0 ∂t f ∂t
∂2H ∂P ∂2M ∇ × ∇ × H + µ = ∇ × J + ∇ × − µ (2.8) 0 0 ∂t2 f ∂t 0 0 ∂t2
Using the fact that ∇ × ∇ × C = ∇ (∇ · C) − ∇2C and inserting the rest of Maxwell’s equations (Eq. 2.3 and Eq. 2.4), we arrive at the following:
∂2E 1 ∂ ∂P ∇2E − µ = ∇ρ + µ J + + ∇ × M (2.9) 0 0 ∂t2 f 0 ∂t f ∂t
∂2H ∂P ∂2M ∇2H − µ = −∇ × J − ∇ × + µ (2.10) 0 0 ∂t2 f ∂t 0 0 ∂t2
We can simplify these by using Ohm’s law, [26] assuming linear correlation between the
free volumetric current density (Jf ) and E via the electrical conductivity σ, and neglecting
charge fluctuations (∇ρf = 0). We will only proceed with the electric half of this deriva- tion since Maxwell’s equations are cyclic, therefore knowing the electric half determines the
magnetic half. Lastly we can assume the material is non-magnetically permeable (µ = µ0) because the electric response dominates the interaction. Repercussions of this are that M = 0.
∂2E ∂E ∂2P ∇2E = µ + µ σ + µ (2.11) 0 0 ∂t2 0 ∂t 0 ∂t2
(1) If we use P = 0χe E = ( − 0) E we arrive at Eq. 2.12. This specific form will be particularly useful when dealing with conductors.
∂2E ∂E ∇2E = µ + µ σ (2.12) 0 ∂t2 0 ∂t
However, a majority of materials being studies in this dissertation are generally dielectric 8
and insulating, we then ignore the conductivity term and rewrite the wave equation in the following form:
∂2E n2 ∂2E ∇2E = µ = (2.13) 0 ∂t2 c2 ∂t2
n2 1 q √ It should also be noted that µ0 = 2 = 2 , where n = = R is the index of c v 0 q refraction, c = 1 is the speed of light in vacuum, and v is the speed of the wave. The 0µ0 general solutions of this differential equation are linearly polarized, monochromatic plane- waves with wave vector k and angular frequency ω.
i(k·x−ωt) i(k·x−ωt) E (x, t) = E0e H (x, t) = H0e (2.14)
Using these solutions with Maxwell’s equations we can relate E and H using k and ω. Namely, the divergence yields that k · E = k · H = 0, meaning that the associated fields of the wave are perpendicular to the direction of propagation (transverse). Taking the curl
yields that k × E = ωµ0H. Lastly, we use Eq. 2.12 with Eq. 2.14 to obtain the dispersion relation.
2 2 k = µ0ω (2.15)
This formula relates the electric and magnetic properties at a given angular frequency to the propagation and dispersion of that wave in the medium. For our non-magnetic medium, we can relate the free-space wavelength (λ0) to the wave-vector in the following manner:
2πn ω k = = n (2.16) λ0 c
Now if the media of interest is a good conductor, the dispersion relation looks significantly different. The solutions to the wave equation in Eq. 2.14 can be inserted into Eq. 2.12, but 9
we must note that σ ω in good conductors, which allows us to ignore the first term of Eq. 2.12.
2 k ≈ iσµ0ω (2.17)
Another point of note is that k2 is purely imaginary, this means that each of the con- stituent components of k have the same magnitude.
rσµ ω Re |k| = Im |k| = 0 (2.18) 2
When using this in Eq. 2.14, it is easy to see that a wave propagating into a metal will exponentially decay from the imaginary portion of k. This yields an important result, the skin depth.
r 2 δs = (2.19) σµ0ω
The skin depth corresponds to the length at which the magnitude of the electric field decays to e−1. For comparison the skin depth for metals at THz frequencies is on the order of 0.1 µm, which is much smaller than the free space wavelength, 300 µm. In a lossy dielectric, we can equate the dispersion relations for a good conductor (Eq. 2.17) and a dielectric (Eq. 2.15) to find the refraction conductivity relation.
2 iσ R = n = (2.20) ω0
Lastly, we will use the time-averaged Poynting vector (Eq. 2.21) to find the radiation intensity. We do this because our bolometer measures the transmitted power of our THz beam (T , Sec. 2.2).
1 1 hSi = E × H∗ = v |E |2 kˆ (2.21) 2 2 0 10
The magnitude of the Eq. 2.21 yields the radiation intensity, typically measured in W/m2.
1 I = |hSi| = v |E |2 (2.22) 2 0
Lastly, we will relate the Poynting vector magnitude and the skin depth to find the penetration depth. For the purposes of this dissertation, we make use of the penetration depth, as it describes where the intensity decays to e−1. Therefore it relates the |E|2, the imaginary portion of the index of refraction κ, and the absorption coefficient α.
r δs c 1 1 δp = = = = (2.23) 2 2κω α 2σµ0ω
2.2 Thin-Film Fresnel Formula
Figure 2.1: Cartoon of the transmission.
Thin film samples can be said to exhibit Drude-like behavior if their spectral response is flat; meaning the spectral range of interest is far from resonance. This section specifically
pertains to our VO2 samples (specifically Sec. 6.6.2). Having a Drude-like response allows us to extract the sheet conductivity from the transmission data using the thin-film Fresnel 11
formula. In this derivation we assume that we are examining an isotropic and homogeneous thin-film at normal incidence. We further assume that there is no destructive interference
in the the thin film. This can be done since the VO2 film thickness is much less than the
λ wavelength (n2d 10 ). The sapphire substrate, on the other hand, is optically thick at 300 µm with n ≈ 3.1, which also allows us to assume there is no destructive interference in the substrate. We also ignore the absorption in the sapphire [27] and assume the pulses are well temporally separated.
ni − nj 2ni rij = tij = (2.24) ni + nj ni + nj
Now using the transmission and reflection coefficients from 2.24, we can start looking at the transmission through the thin film sample. We can view our transmission in the following way since we are dealing with a pulsed laser system: each transmitted pulse will contribute to the total transmission t, where t(n) corresponds to the nth transmitted pulse. During each trip through the material phase is acquired. The nth pulse exiting the material will have a
total phase of φn = (2n − 1) φ.
∞ X t = t(n) (2.25) n=1
iφd 3iφd 5iφd = t12t23e + t12r23r21t23e + t12r23r21r23r21t23e + .... (2.26) ∞ n iφd X 2iφd = t12t23e r23r21e (2.27) n=0
Since each element in the sum is less than one and monotonically decreasing, we can use the sum of a geometric series to calculate the quantity to which the series converges.
iφd t12t23e t = 2iφ (2.28) 1 + r23r12e d 12
Now we can simplify using Eq. 2.24 and applying the thin film conditions: only a small amount of phase is acquired when traversing the sample (φ 1 ,eiφ ≈ 1 + iφ) and the film thickness is much less than that of the wavelength (d λ).
4n1n2 t = −iφ iφ (2.29) (n1 + n2)(n2 + n3) e d + (n1 − n2)(n2 − n3) e d t13 (n1 + n3) = (2.30) n1n3 n1 + n3 − n2 1 + 2 iφd n2
n1n3 n1n3 Further, for a metallic thin film we can simplify using: 2 1 and (n3 − n1) n2 n2
n2. This is a direct result of the index of the thin film being much larger than either the index of air or the substrate. We must not forget to also utilize the refraction conductivity
d 2 relation (Eq. 2.20) and the simplification it allows us to make, in2φ2 = 2πi λ n2 = −Z0σd. 1 Where Z0 = = 376.7Ω, the vacuum impedance, and the conductivity, σ, can be re-written c0
in terms of the sheet conductivity, σs = σd. In the end we have the following result:
t (n + n ) t = 13 1 3 (2.31) n1 + n3 + Z0σs
However, we must also account for the reflections from inside the substrate. The solution for this is found by applying the sum of a geometric series as we did above in Eq. 2.28 and simplifying.
2iφd 2 4iφd r = r32 + t32r21t23e + t32r21r23t23e + ... (2.32) ∞ n 2iφd X 2iφd = r32 + t32r21t23e r21r23e (2.33) n=0 2iφd t32r21t23e = r32 + 2iφ (2.34) 1 + r21r32e d 2iφd r32 + r21e = 2iφ (2.35) 1 + r21r32e d 13
When inserting the constituent reflection and transmission coefficients and applying the thin-film approximation, this becomes slightly messier than the transmission.
(n − n )(n + n ) + (n + n )(n − n ) (1 + 2iφ ) r = 3 2 2 1 3 2 2 1 d (2.36) (n3 + n2)(n2 + n1) + (n3 − n2)(n2 − n1) (1 + 2iφd) n − n − Z σ = 3 1 0 s (2.37) n3 + n1 + Z0σs
Now that the most tedious part, in terms of derivations1, is behind us, we can press on to transmission through the substrate with and without the thin-film. We are applying this line of logic to bolometer (power/intensity) measurements, but the Fresnel equations are written in terms of the electric field, therefore we must take the norm-square of each term to find the power transmission.
2 2 twith = tt34 + tr34rt34 + tr34r t34 + ... (2.38)
2 2 twithout = t13t34 + t13r34r31t34 + t13r34r31t34 + ... (2.39)
2 2 2 2 2 2 2 4 4 2 Twith = t t34 + t r34r t34 + t r34r t34 + ... (2.40) 2 2 t t34 = 2 2 (2.41) 1 − r34r 2 2 2 2 2 2 2 2 4 2 Twithout = t13t34 + t13r34r31t34 + t13r34r31t34 + ... (2.42) 2 2 t13t34 = 2 2 (2.43) 1 − r34r31
Taking the ratio of Twith and Twithout allows us to determine the properties of the thin-film, where R is the relative power transmission.
2 2 2 Twith t (1 − r34r31) R = = 2 2 2 (2.44) Twithout t13 (1 − r34r ) 1Nothing is as tedious as TDS. 14
Lastly we can solve for the sheet conductivity of the thin-film by substituting in for the constituent parts of Eq. 2.44 and simplifying.
r 4 2 2 2 2 ! 1 2 2 Rn3 + 2n3n4 (2n1 + n4 (2 − R)) + n4 (4n1 + 4n1n4 + Rn4) σs = − n3 + 2n1n4 + n4 − 2n4Z0 R (2.45)
2.3 Harmonic Oscillator
Figure 2.2: A cartoon of the harmonic oscillator model.
The simplest way to describe nonlinear media is to start with the harmonic oscillator, specifically a damped-driven oscillator. We can view the ionic core as being immovable and tethered to the electron with via a spring. In this picture(Fig 2.2), an incident electric field is driving electrons, which oscillate in their respective potential wells. However, they will not oscillate indefinitely due scattering mechanisms which we condense into a damping term. For a Drude-Lorentz oscillator, we have the following equation of motion:
d2x dx q + Γ + ω2x = − E (t) (2.46) dt2 dt 0 m∗
where the charge carrier in question has effective mass m∗ and charge q. If we assume that the incident electric field is a monochromatic plane wave, we can easily find the solutions 15
using the following ansatz.
−iωt x = x0e xˆ (2.47)
Where the displacement from equilibrium x0 is given by:
q |E0| x0 = − ∗ 2 2 (2.48) m ω0 − ω − iΓω
Using this we can relate the displacement to the dipole moment (p = qx) and finally the macroscopic polarizability per unit volume of the material (P = Np, where N is the carrier density).
−iωt P = qNx = χe (ω) E0e xˆ (2.49)
We next need to use the displacement vector D in its two forms:
D = 0E + P = 0RE (2.50)
Now we can solve for the relative permittivity ( ). We can simplify by inserting the 0 2 Nq2 plasma frequency ω = ∗ and assume the medium is isotropic. This allows the relative p 0m permittivity to take the form of Eq. 2.51.
2 |P| ωp R (ω) = 1 + = 1 + 2 2 (2.51) 0 |E| ω0 − ω − iΓω
2 This is of particular usefulness since n = R, which allow us to rewrite the dispersion relation.
ω k (ω) = p (ω) (2.52) R c 16
2.4 Nonlinear Media
Figure 2.3: Example of an harmonic and an anharmonic potential. The associated motion of a charge carrier is plotted on the left.
The harmonic oscillator used in the previous section featured a perfectly harmonic potential, meaning that the potential energy of the charge carriers in the well is symmetric about its rest position. In order to see second order nonlinear effects, a noncentrosymmetric crystal must be used. This translates into the potential inside the crystal having some
1 2 2 anharmonicity, which in the simplest approximation takes the potential from 2 mω0x to 1 2 2 1 3 2 mω0x + 3 mαx . This non-linear term corresponds to multi-photon processes. We will incorporate this into the Drude-Lorentz model by adding the following term:
d2x dx q + Γ + ω2x + αx2 = E (t) (2.53) dt2 dt 0 m
We will now solve Eq. 2.53 by using a perturbative expansion and collecting terms, in this case, powers of λ. This is possible because the the series should converge since each successive term is much smaller than the previous, explicitly x(1) x(2) ... x(n). The λ term is present for book keeping purposes since this is a recursive solution and every term in this expansion depends on the preceding terms. It is worth solving this equation of motion for the general case since we will need terms for optical rectification and electro-optic sampling (Pockels Effect). 17
∞ X x (t) = λnx(n) (t) (2.54) n=1 The first order term, which does not include the nonlinear term, has the following equation of motion:
d2x(1) dx(1) q + Γ + ω2x(1) = E (t) (2.55) dt2 dt 0 m
We assume that there are two plane-waves in our derivation. We find the solution has the form of Eq. 2.57. This will allow us to explore all the solutions for two photon processes.
(1) (1) −iω1t (1) −iω2t x (t) = x (ω1) e + x (ω2) e + c.c. (2.56)
−iω1t −iω2t q E0e q E0e = ∗ 2 2 + ∗ 2 2 + c.c. (2.57) m ω0 − ω1 − iΓω1 m ω0 − ω2 − iΓω2
The second order term lacks the electric field term because the material is only being
driven at frequencies ω1 and ω2 and the resulting perturbation should only be due to the material’s response at those frequencies. This manifests itself in second order term by the the nonlinear term, x(1)2 is of order λ2, due to its dependence upon the first order terms.
2 (2) (2) d x dx 2 + Γ + ω2x(2) = −α x(1) (2.58) dt2 dt 0
Now we must be slightly more careful with the second order term in order to elicit the desired result. There will be several terms due to the interaction of the incident waves, specifically x(1)2 should have four distinct types of terms: second harmonic generation (SHG, Eq. 2.59), sum frequency generation (SFG, Eq. 2.60), difference frequency generation (DFG, Eq. 2.61), and optical rectification (OR, Eq. 2.62). 18
2 2 (1) −2iω1t (1) −2iω2t SHG = x (ω1) e + x (ω2) e + c.c. (2.59)
(1) (1) −i(ω1+ω2)t SFG = 2x (ω1) x (ω2) e + c.c. (2.60)
(1) ∗(1) −i(ω1−ω2)t DFG = 2x (ω1) x (ω2) e + c.c. (2.61)
(1) 2 (2) 2 OR = 2 x (ω1) + 2 x (ω2) (2.62)
Of these, OR is of particular interest because it governs how we generate our THz pulse. The fact that the OR term has zero frequency requires explanation of how this can generate THz. This explanation requires an alternative view of what is happening in the material, which complements the view of the harmonic oscillator. This view relies on the second order polarization of the material and we must recall the initial explanation of linear media and the relationships between D, E, and P. For nonlinear media, the D-field can be expressed as having both a linear and nonlinear component.
(1) (NL) (1) (NL) D = 0E + P + P = D + P (2.63)
Which leads to the macroscopic polarization of the material to be written perturbatively, in the same fashion as the solution to the harmonic oscillator.
P = P(1) + P(2) + P(3) + ... (2.64)